Conference Proceedings
22nd
International Conference
MEKON 2020 February 6th, 2020 Faculty of Economics, VSB — TU Ostrava
VSB – Technical University of Ostrava
Faculty of Economics
Proceedings of the 22nd International Conference
MEKON 2020
February 6th, 2020
Ostrava, Czech Republic
The conference is organised by:
VSB – Technical University of Ostrava,
Faculty of Economics
Proceedings of the 22nd International Conference MEKON 2020
Publisher: VSB – Technical University of Ostrava
Sokolská třída 33, 702 00 Ostrava 1, Czech Republic
Editors: Jiří Branžovský, Jakub Pavelek
Cover: Alžběta Gregorová
ISBN 978-80-248-4410-7
Copyright © 2020 by VSB – Technical University of Ostrava
Copyright © 2020 by authors of the papers
Publication has been supported by the Karel Englis Endowment Fund. Publication is not a subject
of language check. All papers passed a review process.
SCIENTIFIC COMMITTEE
doc. Ing. Vojtěch Spáčil, CSc.
Dean of the Faculty of Economics, VSB-TU Ostrava
doc. Ing. Lenka Kauerová, CSc.
Vice-dean for study affairs, Faculty of Economics, VSB-TU Ostrava
prof. Ing. Jana Hančlová, CSc.
Vice-dean for science, research and doctoral studies, Faculty of Economics, VSB-TU Ostrava
Ing. Karel Hlaváček, Ph.D.
Vice-dean for foreign affairs, Faculty of Economics, VSB-TU Ostrava
Ing. Aleš Lokaj, Ph.D.
Vice-dean for development, Faculty of Economics, VSB-TU Ostrava
doc. Ing. Lenka Fojtíková, Ph.D.
Department of European Integration, Faculty of Economics, VSB-TU Ostrava
doc. Ing. Petra Horváthová, Ph.D.
Department of Management, Faculty of Economics, VSB-TU Ostrava
doc. Ing. Igor Ivan, Ph.D.
Vice-rector for commercialization and cooperation with industry, VSB-TU Ostrava
Ing. Kateřina Kashi, Ph.D.
Chairman of Academic senate, Faculty of Economics, VSB-TU Ostrava
doc. Ing. Aleš Kresta, Ph.D.
Department of Finance, Faculty of Economics, VSB-TU Ostrava
prof. Ing. Martin Macháček, Ph.D. et Ph.D.
Department of Economics, Faculty of Economics, VSB-TU Ostrava
prof. Ing. Darja Noskievičová, CSc.
Department of Quality Management,
Faculty of Materials Science and Technology, VSB-TU Ostrava
prof. JUDr. Naděžda Rozehnalová, CSc.
Department of Law, Faculty of Economics, VSB-TU Ostrava
prof. Ing. Jan Sucháček, Ph.D.
Department of Department of Regional and Environmental Economics,
Faculty of Economics, VSB-TU Ostrava
prof. RNDr. Dana Šalounová, Ph.D.
Department of Mathematical Methods in Economics, Faculty of Economics, VSB-TU Ostrava
prof. Ing. Tomáš Tichý, Ph.D.
Department of Finance, Faculty of Economics, VSB-TU Ostrava
CONFERENCE GUARANTEE
prof. Ing. Jana Hančlová, CSc.
Vice-dean for sciene, research and doctoral studies, Faculty of Economics, VSB-TU Ostrava
CONFERENCE ORGANISING GUARANTEE
Ing. Jiří Branžovský
Department of Finance, Faculty of Economics, VSB-TU Ostrava
ORGANISING COMMITEE
Ing. Jiří Branžovský
Depatrment of Finance, Faculty of Economics, VSB-TU Ostrava
Ing. Jakub Pavelek
Department of Economics, Faculty of Economics, VSB-TU Ostrava
Suggested citation:
Author, A. 2020. Title of the paper. In Branžovský, J. and J. Pavelek (eds.). Proceedings of the 22nd
International Conference MEKON 2020. Ostrava: VSB – Technical University of Ostrava, pp. xxx-xxx.
ISBN 978-80-248-4410-7
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
Content
VENIAMIN BOLDYREV et al.
INFORMATION-COMPUTING SYSTEM FOR
DESIGNING AND CONSTRUCTION OF
INDUSTRIAL PAINTING LINES 1
JIŘÍ BRANŽOVSKÝ
THE STOCK MARKETS BEHAVIOR NEAR THE
OFFICIAL FEDERAL RESERVE SYSTEM'S
MEETINGS 10
VLADIMÍR BULKO
APPLICATION OF MEAN-REVERSION BINOMIAL
LATTICE APPROACH TO VALUATION OF
MORTGAGE IMPLICIT OPTION IN THE CZECH
MARKET
21
IVANA ČERMÁKOVÁ
USING GEOINFORMATION IN PUBLIC
ADMINISTRATION, CASE STUDY:
MORAVSKOSLEZKÝ REGION 32
ALINA CZAPLA INTER-ORGANIZATIONAL KNOWLEDGE
SHARING AND GAME THEORY 38
KATARZYNA CZERNÁ
COMPARISON OF EVALUATION OF INNOVATIVE
ACTIVITIES IN INNOVATIVE COMPANIES
WITHIN THE V4 COUNTRIES 47
PETRA DOLEŽELOVÁ
IMPACT OF UNILATERAL PREFERENTIAL
MEASURES OF THE EUROPEAN UNION, THE
UNITED STATES AND CHINA ON EXPORTS OF
THE LEAST DEVELOPED COUNTRIES
56
MERI DUDUCI
IMPLEMENTATION OF INDUSTRY 4.0: A
RESEARCH BASED ON THE EFFECTIVE
TRAINING OF HRM 66
IZABELA ERTINGEROVÁ
EVALUATION OF THE EFFICIENCY OF THE
SYSTEM OF SELECTED RESIDENTIAL SOCIAL
SERVICES FOR SENIORS IN THE CZECH
REPUBLIC
75
LUN GAO
ANALYSIS OF THE SPILLOVER EFFECT OF
STOCK MARKET RISK: BASED ON EVT-COPULA-
CVAR MODEL 85
DANIELA KHARROUBI
THE IDENTIFICATION OF FACTORS
INFLUENCING HUMAN RESOURCES
MANAGEMENT AND THE EVALUATION OF
THEIR INTENSITY: A CASE STUDY ON HUMAN
RESOURCES MANAGEMENT (HRM)
94
NATÁLIE KONEČNÁ
EVALUATION EFFICIENT PRICE OF
COMPENSATION OF SELECTED PUBLIC
TRANSPORT IN OLOMOUC REGION AND
MORAVIAN – SILESIAN REGION
105
FRANTIŠEK KONEČNÝ
EVALUATION OF CSR DISCLOSURE OF THE
BIGGEST COMPANIES IN CZECH REPUBLIC
WITH MCDM METHODS 116
FILIP LESSL MEASURING THE FINANCIAL PERFORMANCE OF
A COMPANY BASED ON SELECTED APPROACH 123
ONDŘEJ MIKULEC
IDENTIFYING FACTORS OF EMPLOYEE
TURNOVER WITH MULTIPLE
CORRESPONDENCE ANALYSIS 133
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
DAVID NEDĚLA DATA ANALYSIS AND TESTING WITH RESPECT
OF PORTFOLIO SELECTION PROBLEM 141
MICHAELA PETROVÁ
INSURABLE AND UNINSURABLE RISKS AND
THEIR CLASSIFICATION FROM THE
PERSPECTIVE OF A CZECH EXPORTER 154
LEE SABRINA
GOVERNANCE STRUCTURES OF MUNICIPAL
ENTERPRISES – EMPIRICAL STUDY OF
EFFICIENCY OF HOSPITALS 162
ADÉLA ŠPAČKOVÁ GENERALIZED LINEAR MODELS IN A MOTOR
HULL INSURANCE PORTFOLIO 175
ADRIÁN ŠPERKA et al. OPTIMALIZATION OF DIRECT COSTS OF THE
RAILWAYS OF THE SLOVAK REPUBLIC 184
TRAN VAN HAI TRIEU
DIGITAL TRANSFORMATION AND BUSINESS
PROCESS MANAGEMENT
IN CREATIVE INDUSTRIES: THE CASE OF FILM
PRODUCTION PROCESS
195
RIJAD TRUMIC AVOIDANCE OF COST INCREASES DURING
CHANGE MANAGEMENT 206
SUSANN WIECZOREK BUSINESS STUDIES IN TIMES OF CHANGE
(INDUSTRY 4.0) 215
XIAOJUAN WU
RESEARCH ON THE IMPACT OF
CHARACTERISTICS OF THE BOARD OF
DIRECTORS OF CHINESE APPLIANCE LISTED
COMPANIES ON CORPORATE SOCIAL
RESPONSIBILITY
223
MARTINA ŽWAKOVÁ
MULTI-CRITERIA DECISION MAKING USING THE
ENTROPHY METHOD APPLIED ON SELECTED
VARIABLES FROM THE AREA OF
DIGITALIZATION AND DEVELOPMENT IN THE
CENTRAL EUROPE TERRITORY
232
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
1
INFORMATION-COMPUTING SYSTEM FOR DESIGNING AND CONSTRUCTION OF
INDUSTRIAL PAINTING LINES
Boris Bogomolov1, Veniamin Boldyrev2, Valeria Elistratkina1, Vladimir Menshikov1, Yana
Seina2, Andrei Zubarev1,
1Department of innovative materials and corrosion protection, D. Mendeleev University of Chemical Technology
of Russia,
Miusskaya sqr. 9, Moscow 125047, Russian Federation
e-mail: [email protected]
2Department of Chemistry, Bauman Moscow State Technical University,
2nd Baumanskaya str. 5/1, Moscow 105005, Russian Federation
e-mail: [email protected]
Abstract
This article describes an example of creating an information and computing system for the design
and construction of industrial painting lines. The methodology of infological modeling of databases
based on queries is applied. The calculation of the main technological units of the system was made,
while it was possible to autonomously calculate any of the three subsystems. All source information
and calculation results are stored in a result file recorded on electronic media. An array of the initial
data of a specific drying chamber is also saved in the form of a file, which allows not to enter the initial
information when recalculating the chamber. Testing of the information-computing system for the
design and construction of industrial painting lines was carried out on several variants of painting lines.
Keywords
painting line, computer-aided design, infological modelling, information support, information-
computing system
JEL Classification
L86; L74; O32; O21
Information-computing systems application bases
The computer system for the design and construction of industrial paint lines includes an information
subsystem and three interconnected software blocks:
• expert system unit for designing a technological unit for surface preparation,
• block calculation chamber for applying powder paints,
• unit for calculating the chamber of radiation and convective drying of painted surfaces.
Information and computing system allow you to:
− to provide an integrated approach to the procedure for the automated design of paint lines;
− reduce design time by automating standard calculations and the rapid exchange of information
between program blocks;
− improve the quality of design by eliminating technical and design errors;
− to provide a search for a rational design solution, both due to the application of optimization
procedures, and due to the possibility of operational analysis of several alternative
technological solutions.
Figure 1 shows the functional information structure of an information-computing system.
The information subsystem is used to store information and exchange data necessary for solving design
and design problems, and includes: information about the processed product; a working database
containing all the necessary information for the operation of the software package; information on the
results of the calculation of individual technological units of industrial painting lines, transmitted to the
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
2
working database and output files of the system; reference and regulatory data necessary for design and
engineering calculations.
Figure 1. functionally information structure of an information and computing system for
the design and construction of industrial painting
Modern software applications in the chemical industries are complex information and computing
systems consisting of autonomous interconnected software blocks. In this case, special attention should
be paid to their information support, which includes three main groups of information arrays:
- source data files;
- files for storing and exchanging data within the software package;
- files of the results of work.
All these information arrays provide for the existence of procedures for their creation, filling,
organization and interconnection, requiring the development of special software blocks that are both
included in the software package and that work autonomously. The composition and structure of
information support is determined by the features of the applied problem and should be designed by
analogy with databases.
Expert system for painting lines
The technology of expert systems is one of the areas of a new field of research, which has received
the name of Artificial Intelligence - AI. Research in this area is focused on the development and
implementation of computer programs that can emulate (imitate, reproduce) those areas of human
activity that require thinking, a certain skill and accumulated experience. So, these include decision-
making tasks in the design and construction of industrial painting lines.
In the course of the evaluation of the designed paint line, the scope of available options is determined.
This allows you to get a reliable base for the preparation and implementation of long-term activities in
order to quickly return on investment.
Expert assessment in the design of industrial painting lines allows us to solve a number of urgent
problems and anticipate their appearance, for example:
• efficient use of energy and materials;
• compliance of the product with the specified standards and customer requirements;
• adaptation of production capacities;
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
3
• optimal compliance with legal requirements.
In addition to all of the above, the relevance of the problem of reducing costs is also the main goal of
the production management.
The expert system for designing the technological unit for surface preparation is formed on the basis
of the algorithmic and informational support of the design calculation of the technological unit for
surface preparation in paint and varnish production.
The expert system includes the following main steps:
• input of initial data using intelligent interface procedures;
• search for suitable technological schemes for surface preparation;
• the formation of the suspension from parts subjected to processing depending on the material and
overall dimensions;
• carrying out design calculations of processes included in the scheme;
• selection of rational options for the scheme and preparation of reporting documentation.
Programs are implemented on the principle basis of object-oriented programming and are
accompanied by a friendly intelligent interface, which ensures the versatility and effectiveness of
software. Consider some features of the implementation of its main stages.
The initial information of the expert system contains the characteristics of the processed parts,
including data on the material of the products, their dimensions and the nature of surface contamination,
and information on the available production facilities. For the expert system to work correctly, the
analyzed text information is entered using the menu constructed in accordance with the lists of typical
attribute values obtained by logical analysis of the domain information. All data is stored in a sequential
file of the project database with a unique name specified by the user for the designed technological unit.
When the project is called up again, all the information written to the file is read out and can be easily
adjusted.
At the next stage of the system’s work, the specified characteristics of the parts are compared with
the information in the database of typical technological schemes for surface preparation formed on the
basis of the standard. At the user's request, reference information on the composition and characteristics
of the technological stages is provided for each circuit. The user selects the schemes proposed by the
expert system or sets the number of any scheme recorded in the database.
For the selected technological scheme in the expert system, the following sequential operations are
performed: the choice of configuration and suspension configuration; calculation of the dimensions of
the sections of the preparation unit; technological calculation of sections. If the quality of the
technological decision being made is unsatisfactory at each stage of the algorithm, it is possible to return
to the previous stage of calculation, correct the initial information and repeat the design procedure.
Application information algorithm
For the design of information support for applied problems, the methodology of infological modeling
of databases based on queries was applied [1-3]. This technique was applied in the development of an
information and computer complex for the design and construction of industrial painting lines. The
following main stages of infological modeling of information support of the applied problem are
determined.
1. Determining the structure of a system-wide file, including an array of source data for solving the
problem. The file includes the technical task of the project, the characteristics of substances and
materials, the parameters of standard equipment, etc. The file structure is clearly defined, since the
information contained in it is necessary for all program blocks of the complex and is determined by the
format of the data read by the programs. The system-wide file is built on the basis of the “data storefront”
methodology [1-4], which is filled only with the information that is necessary to solve the applied
problem.
2. Development of a procedure for filling data marts and information sources. These sources include:
- databases of normative indicators, standards and reference data;
- files - the results of other software systems;
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
4
- data files filled in before the software starts using specialized dialog procedures (for example, entering
information about the technical task of the project). When filling out a data showcase, additional
operations are required for syntactic and semantic verification of data, analysis of dimension and
parameter values, and data verification. These operations are developed on the basis of an analysis of
the features of the applied problem, and within the domain can be transferred from one software package
to another.
3. Determining the composition of the source data files and the results of individual software modules
of the complex. The main goal of this data group is:
- preparation of information for repeated re-entry of parameters and their correction;
- preparation of these results for transmission to the document generation program;
- creating a file for the exchange of information between the programs of the complex. The peculiarity
of this block is that all information is created inside the corresponding program modules, but the data
generation procedure itself is determined by the features of the applied problem and is designed at the
stage of development of the infological model.
4. Formation of a library of standard documents, taking into account the metadata of the subject area, of
the results of the work of the program package.
The design of these blocks is based on the infological model for generating queries in the database and
the procedure for filling out standard documents (forms and database reports).
The use of infological modeling when creating information support for applied tasks allows you to
create file libraries, modules of typical information processing procedures, and document templates.
The first step in the design of a painting line is the formation of a surface preparation scheme for the
part before painting.
The choice of the circuit of the technological unit is from the list of circuits presented in the standard
in accordance with the characteristics of the machined parts [3-6]. At the user's request, reference
information on the composition and characteristics of the technological stages is provided for each
circuit. The user selects the schemes from those proposed by the expert system or sets the number of
any scheme recorded in the database.
For the selected technological scheme in the information-computer system, the following sequential
operations are performed:
- the choice of configuration and suspension;
- calculation of the dimensions of the sections of the preparation unit;
- technological calculation of sections.
If the quality of the technological decision being made is unsatisfactory at each stage of the algorithm,
it is possible to return to the previous stage of calculation, correct the initial information and repeat the
design procedure.
One of the most time-consuming and routine procedures for designing a surface preparation scheme
is to determine the suspension configuration, consisting of a set of products that are simultaneously
processed. Automatic determination of the limit dimensions of parts of one material by length, width
and height of the suspension is provided, which ensures the placement of any of the processed parts in
its volume.
The procedure for forming a set of suspensions is performed separately for each material of parts.
First of all, the total volume of each product is calculated, determined by the dimensions of the part and
their number. This is done in order to complete the suspension all the same parts were within the same
kit. If the total volume of the part is greater than the volume of the suspension, then the total number of
parts of the same type is divided into several suspensions of the same composition.
At the next stage, an approximate calculation of the dimensions of the circuit section occurs. The
same sizes are accepted for all sections, and the calculation of the width and height of the section is
performed on the basis of a typical configuration scheme [4-7]. The section length is calculated based
on the given conveyor speed and product processing time and taking into account the additional drain
zone, the length of which is 1.5-2 times longer than the length of the workpiece.
The technological calculation of the surface preparation unit is performed sequentially for all sections
included in the scheme. The main results of the calculation are the exact dimensions of the section, the
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
5
technological conditions for processing parts, and the characteristics of the pumping line of the flood
circuit. The computer system has several automatic sketching procedures that explain the design process
and display, as a rule, a structural solution. So, in fig. Figure 2 shows an automatically generated sketch
that allows the user to evaluate the obtained section configuration and, if necessary, change it by
correcting the data in the control form of the program block.
Figure 2. Sketch of the configuration of the section in accordance with the results of the
calculation of dimensions
The calculation results are presented in the form of a text file, which for a comprehensive description
of the circuit includes both the source information and the results of the work.
Further details are sent to the powder coating chamber.
The procedures of the program unit are designed to solve the following main design problems [6-10]:
• calculation of the aerodynamic operating conditions of the chamber, including the speed and flow
rate of air flows in the chamber, hydraulic resistance of all technological elements of the chamber,
with the given dimensions and design of the chamber;
• calculation of pressure drops in the ducts of the process unit, flow rates and flow rates, pipe fittings
resistances, with a known duct configuration and technologically specified air flow conditions;
• calculation of the characteristics of the cyclone and filter of the recovery unit, including the
determination of the speeds and hydraulic resistances of the devices;
• the calculation of their overall dimensions and characteristics of the equipment elements for a given
type of cyclone and filter;
• calculation of the operational characteristics of the fan and the selection of a suitable fan from
models manufactured by the industry;
• analysis of the mutual arrangement of the elements of the technological unit, performed according
to the sketch of the placement of circuit elements, in order to clarify the elevations of the installation
of filters and cyclones and assess the correctness of the given length of the duct sections.
The calculation unit of the technological unit of the camera for applying powder paints, like other
program modules of the system, is implemented using the principles of object-oriented programming.
Each element of the program complex is an independent program or information module, which allows
you to adapt the complex to solve various technological problems and expand the complex by
introducing new program objects. The software package is built in the form of software modules loaded
from the “calculation control unit”. Information exchange takes place: - using an array of global
variables; - a working database of a complex of programs; - using sequential source data files and
calculation results; - from databases of typical equipment (fans, cyclones, filters).
The developed software unit provides a fairly complete engineering calculation of the technological
unit of the camera for applying powder paints in an intelligent dialogue with the user. It is possible to
repeatedly correct the information and re-calculate the circuit.
The calculation results of the camera are recorded in the result files and partially sent to the working
database of the information-computing system for the design and construction of industrial painting
lines.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
6
The next step in the design of the painting line is the calculation of convective-radiation drying
processes, providing: - design calculation of convective-radiation drying; - search for a rational design
solution; - obtaining a design calculation protocol containing all the stages and results of solving the
design problem [7-9].
Figure 3. Sketch of the placement of the elements of the technological unit of the powder coating
chamber
In fig. Figure 4 shows a block diagram of an algorithm for calculating a radiation-convective drying
chamber, on the basis of which a complex of programs was developed.
At the first stage of the algorithm, input of initial information for calculating the camera takes place.
In this case, it is advisable to divide the information into groups: - camera characteristics; - details of the
part; - characteristics of emitters and air flow; - parameters of the drying process. This separation allows
you to vary the parameters in order to determine rational design decisions. So, for example, for the
existing process, you can select the number of emitters for the required temperature of the paint layer
and study the kinetics of heating the part.
To enter textual information used for routing calculations, a menu system is used that excludes
spelling and semantic errors and, as a result, errors in the calculation procedures of the program complex.
To enter regulatory information (information from directories) relational database tables are used.
Further, an additional calculation of the drying parameters in accordance with the initial data.
To control the correctness of the input source information, a camera sketch is used that is
automatically generated in accordance with the entered numerical values. In this case, you can
immediately see the possible mismatch of the specified dimensions of the camera and the dimensions
of the part, or a clear mismatch to the process of a given number of emitters. In this case, even before
the calculation, the necessary changes are made to the source data array. Then, the calculation of the
drying process is performed in order to determine the dynamics of changes in temperature of the air,
part and the paint layer, as well as the degree of curing of the paint or enamel layer. The algorithm
provides a constant sequence of steps: calculation of emitters (the stage is not automatically performed
when using only convective drying); calculation of the characteristics of the air flow of the chamber and
the main heat transfer coefficients. calculation of the dynamics of temperature changes in the chamber
for air, part and paint layer; calculation of the thermal balance of the camera, in accordance with which
the verification of the correctness of the calculations is performed.
After completing one stage of the program block, the next step of the algorithm becomes available.
As well as in other blocks of the information-computer complex, the design results are recorded in the
output files and in the working database, the information from which is necessary for repeated design
calculations and the design of the complex of technical documentation.
After the calculation is completed, the user is again in the input window. In this case, it is possible
either to terminate the program, or to correct the initial data with the subsequent repetition of the
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
7
calculation. The number of such repetitions is not limited. All initial information and calculation results
are stored in a result file recorded on electronic media. An array of the initial data of a specific drying
chamber is also saved in the form of a file, which allows not to enter the initial information when
recalculating the chamber.
Figure 4. Functional and informational structure of a program unit for designing a radiation-
convective drying chamber
Results
Testing of the information-computing system for the design and construction of industrial painting
lines was carried out on several variants of painting lines for managing a painting production projects at
NPO «Lakokoskrytie», Khot`kovo, Moscow region. The system is used by the scientific and design-
technological department, design bureau, engineering plant, commercial unit, which is part of the
organization. The system was applied to solve practical problems of developing the basic business
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
8
processes of the life cycle of a complex chemical process systems of dyeing industries (planning
production and choosing a rational technology; managing the design business process).
The developed information-computer system was used in practice when fulfilling the agreements
of NPO “Lakokosokrytie” on the development of compositions and technology for applying
nanomodified environmentally friendly hydrophobic paints and coatings, as well as the development of
compositions and technology for applying universal anti-corrosion paintwork to protect large-sized
metal structures and equipment.
When developing a universal automated complex for painting containers with radioactive waste
using an automatic remote control unit and elements of robotics. The specified unique complex was
installed and launched at the Voronezh NPP.
The system was used by NPO «Lakokoskoprytie» together with «Tagiltransmashproekt» and the
Czech company «GALATEK a.s.» for the development of a detailed design for a high-tech painting line
of freight cars of «Uralvagonzavod», which operates in Nizhny Tagil. A new painting line developed
using the system has made it possible to paint up to 16 thousand various modifications of freight cars
per year. In addition, in this project, an optimal ventilation system for the spray booths and a cleaning
unit for the air contaminated by the solvent vapor removed from the spray dryer were developed. The
designed new high-tech painting line is equipped with a gas purification unit operating on the principle
of reversible capture of organic substances in rotary adsorbers with subsequent desorption by hot air and
thermal afterburning. The developed installation provides a reduction in the concentration of pollutants
to the maximum permissible values. Apply the developed information-computer system for dyeing
industries to increase the effectiveness of the NPO «Lakokoskrytie», which in the period from 2010 to
2019 increased the company's income by 5 times.
References
[1] Bogomolov B.B., Bykov E.D., Men’shikov V.V., Zubarev A.M. (2017). Organizational and
technological modeling of chemical process systems. Theoretical Foundations of Chemical
Engineering, 51 (2), pp. 238–246.
[2] Bogomolov B.B., Boldyrev V.S., Zubarev A.M., Meshalkin V.P., Men’shikov V.V. (2019).
Intelligent logical information algorithm for choosing energy-and resource-efficient chemical
technologies. Theoretical Foundations of Chemical Engineering, 53 (5), pp. 709–718.
[3] Date C.J. (1999). An introduction to Database Systems. Reading, Mass.: Addison-Wesley.
[4] Bogomolov B.B., Men’chikov V.V., Bogoslovskii K.G., Bykov E.D., Shumova V.S. (2012).
Managing the design and operation of paint lines using business modelling. The Technology of
Pain and Varnish Coatings: A Collection of Scientific Works, Transaction of the Lakokraspokrytie
Research and Production Association, Moscow: Paint-Media, pp. 40.
[5] Bogomolov B.B., Men’chikov V.V., Bykov E.D., Bogoslovskii K.G. (2013). Modelling of
chemical process systems using the organizational and technological models of business processes.
The Technology of Pain and Varnish Coatings: A Collection of Scientific Works, Transaction of
the Lakokraspokrytie Research and Production Association, Moscow: Paint-Media, pp. 4.
[6] Averina Yu.M., Kalyakina G.E., Menshikov V.V., et al. (2019). Neutralisation process design for
electroplating industry wastewater containing chromium and cyanides. Herald of Bauman
Moscow State Technical University, Series Natural Sciences, 3, pp. 70-80.
[7] Omelchenko I.N., Lyakhovich D.G., Dobryakova K.V. (2019). The method of forming innovative
project portfolio in a project-oriented organization. Herald of the Bauman Moscow State Technical
University, Series Mechanical Engineering, 1, pp. 84-89.
[8] Omelchenko I.N., Lyakhovich D.G., Dobryakova K.V. (2019). Algorithm for innovative
development management of a project-oriented organization. Herald of the Bauman Moscow State
Technical University, Series Instrument Engineering, 1, pp. 129-134.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
9
[9] Korobets B.N. (2016). Models for technology programs within an intellectual property
management systems. Herald of the Bauman Moscow State Technical University, Series Natural
Sciences, 6, pp. 135-142.
[10] Bessarabov A.M., Kvasyuk A.V., Zaremba G.A., Kulov N.N. (2016). System studies of innovation
development in the business sector of chemical science. Theoretical Foundations of Chemical
Engineering, 50 (6), pp. 1001–1014.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
10
THE STOCK MARKETS BEHAVIOR NEAR THE OFFICIAL FEDERAL RESERVE
SYSTEM'S MEETINGS
Jiří Branžovský1
1Department of Finance, VŠB – Technical University of Ostrava,
Sokolská tř. 33, Ostrava 70200, Czech Republic
e-mail: [email protected]
Abstract
Objective of this paper is to analyse through an event-study the US stock markets around the Federal
Reserve System (Fed)’s official meetings of short-term interest rates changes, that are announced by
Federal Open Market Committee (FOMC). The main focus is on the performance of stock index around
the FOMC meetings, meaning the returns, volatility and trading volumes ahead of the monetary policy
meetings ex-ante in estimation window, the stock returns at the event window, and later ex-post in the
post-event window. The quantitative research is proceeded on the nearly 50-year period of the daily
data in order to find out the changes in behaviour over the time. Time series are observed on the
background of open market and open mouth operations during transparent and non-transparent period
of Fed, over different business cycles and particular Fed’ monetary goals periods. The hypothesis is to
find different results when the Fed lowered or increased its federal fund rates, or made no change. The
paper is about the rational expectations and informational efficiency of the US equity markets tracked
by S&P 500 price index and the US monetary policy, and how both affect each other.
Keywords
Federal Reserve System, interest rates, stock returns
JEL Classification
C01, C32, E44, G10
Introduction
The Federal Reserve System has been the monetary authority of the USA since its establishment in 1913
conducting its national monetary policy that would promote the maximum employment, stable prices
and moderate long-term interest rates.
The Fed was non-transparent until February 1994 when all the MP was running through open-market
operations while markets had not been aware of these transactions. Fed has started being transparent and
creditworthy since 1994 via the clear communication through “open-mouth operations” (e.g. official
public announcements, forward guidance, ...) that would help to support a decision-making process of
the consumers and corporations, reduce the economic uncertainty, while paralelly increasing the
effectiveness of the monetary policy itself. Official and public fed fund rate targeting has started since
August 1997.
It is believed that in long run financial markets are influenced by GDP and unemployment (as referred
by Taylor, 1995), but the short-term volatility is run by changes in interest rates, yield to maturities,
trade volumes or by market risk premium.
Literature Review
Central banks should be fully transparent. Based on several research papers the transparency lowers the
stock market and foreign exchange volatilities.
Fed is considered to be the global monetary policy authority, too big to fail influencing the whole globe.
Tessaromatis (1991) confirmed negative relationship between money supply shocks and stock returns
in the 1980s on the official announcement days as well as on the following day.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
11
Rudebusch (1995) studied non-transparent period of the Fed and identified one-to-two day response of
EFFR on the target fed fund rates.
Greenspan (1997) announced that the pain of over-inflated economy of the 1970s was relieved by
moving with fed fund rate more often, generally confirming since 1982.
Robertson & Thornton (1997) empirically evidenced that fed fund target rates are more difficult to
predict than effective inter-bank EFFR.
Bernanke & Mihov (1998) considered fed fund rates to be the most important monetary tool of Fed.
Kuttner (2001) analysed monetary shocks as inter-day changes in 3M EURIBOR futures one day ahead
and following FOMC date.
Selling (2001) considers financial markets efficient hence only monetary shocks can influence stock
returns.
Thornton (2004) was comparing fed fund target rates with market inter-bank FFR and identified their
closer relationship during transparent period of the Fed.
Bernanke & Kuttner (2005) found out that stock markets had been generally affected by unanticipated
MP (monetary shocks). They studied two regressors, fed fund rates as well as its 30-day futures. Second
one represented an alternative to the market participants´ expectations.
Ross (2012) observed that approximately 80 % of the monetary “shock” and stock return is generated
one day ahead of FOMC announcement days.
Kontonikas & MacDonald & Saggu (2013) identified stronger stock market reactions at MP during
crises periods and when economic conditions got worse. Moreover, there was a suprise positive
correlation between policy rates and stock returns during the initial phase of the Great Recession. Similar
results were obtained by Sirucek (2011) focusing on money supply only.
Unalmis (2015) evidenced a hike in stock market volatility on the FOMC dates specifically.
Haitsma & Unalmis & Haan (2016) in their event study found out that ultra-loosen expansionary MP
during the Great Recession led european and british stocks controversely to their declines.
Methodology and Data
This research topic analyses not only on how particular interest rate policies of the Fed have an impact
on the stock index returns, but additionally it distinguishes between specific periods, e.g. how
transparent Fed´ MP was in public, or what is the impact of its MP on stocks regarding to the economic
conditions via business cycles announced ex-post by National Bureau of Economic Research (further as
“NBER”).
Proposed transmission channel of interest rates affecting the stock prices and their returns is as follows:
lower rates and expansionary forward guidance result in greater corporate profits due to cheaper credit
links and loosen credit activities of the banks leading to greater investment capacity and households
credit consumption (interest rate & credit channels). Lower rates mean alternatively cheaper discount
rates increasing present values of future cash flows (asset prices channel).
Author has selected an empirical approach via event study (MacKinley, 1997) defining period before
FOMC as estimation window, official announcement day as an event window, and following period as
post-event window.
Author realises an importance of unanticipated monetary shocks that would influence the US economy,
hence refers the system to define the meaning of monetary tool whether it has expansive or restrictive
impulse.
Granger causality (1969) is the statistical predictive concept for stochastic linear measuring whether
lagged value of a stationary variable X does/doesn´t improve an explanation of another stationary
variable Y. It is not implication of that causality cause-result. It is in fact to see what proportion of Y is
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
12
explained by its time lags and then to see whether adding lagged values of X might improve their
relationship. If Y is Granger-caused by X, there are some information in variable X that help predict Y.
𝑥𝑡 = 𝑐1 +∑𝐴𝑖 . 𝑥𝑡−𝑖
𝑝
𝑖=1
+∑𝐵𝑖 . 𝑦𝑡−𝑖
𝑝
𝑖=1
+ 𝑒1𝑡 (1)
𝑦𝑡 = 𝑐2 +∑𝐴𝑖 . 𝑥𝑡−𝑖
𝑝
𝑖=1
+∑𝐵𝑖 . 𝑦𝑡−𝑖
𝑝
𝑖=1
+ 𝑒2𝑡
There have been selected two methods of monetary shocks decomposition, even though both take into
consideration changes in target federal fund rate (“FFR”) and changes in effective/market FFR
(“EFFR”).
MP shock1,t = (FFRt – FFRt-1) – (EFFRt – FFRt) (2)
MP shock2,t = (FFRt – FFRt-1) – (EFFRt – EFFRt-1) (3)
First monetary shock identification is based on the logic, what proportion of target FFR change was
predicted one day ahead (difference between target and effective FFR). Second shock is measured as
difference between day-over-day changes in FFR and changes in EFFR.
If MP shock negative, it is considered as expansionary as market participants significantly lowered their
interest rate expectations, more than the Fed really did, supporting their economic decisions to spend
and invest more. Oppositely once positive, shock is to have restrictive power slowing the economy
down.
Besides of traditional standard variance, semi-variance model was used too, as upside risks are not
always being considered to be true risks.
𝑠𝑡 = √1
p∑min (Ri − E(Ri); 0)2
𝑝
𝑖=1
(4)
Data
Observation is done on the long period of 12 203 working days from 4th January 1971 to 17th May 2019
of which 7 251 observations had been in the sample near FOMC official announcement dates. The data
were downloaded from the Fed´s official website, the FRED database and the Bloomberg financial
terminal.
All financial series X were recognized as trendy meaning those were non-stationary with a unit root,
hence first differentiation of naturally logarithms (DL_X) was used to obtain the stationarity in the first
order I(1). ADF and KPSS tests were performed for identifying this issue at 10% significance level.
The main and critical endogenous variable is world-widely known and observed US stock market S&P
500 price index returns (DL_SPX) of 500 largest and public-traded corporations in the USA.
Used regressors involved effective effective fed fund rate D_EFFR, ten-year US Treasury yields
D_TEN_YEAR_USTREASURIES, lower-bound fed fund target rate D_FFR_LOWER, narrow money supply
DL_M1, daily trading volumes of S&P 500 index DL_VOLUME, dummy variable by NBER setting up an
official periods of business booms and business crises ex-post DUMMY_NBER, and finally for extracting
a greater proportion of monetary residuum there was a necessity to set up some monetary shocks
impacting the market participants´ expectations of monetary policy.
Research was processed on the daily close stock prices of price market-capitalised and sectors-wide
S&P 500 index. Effective fed fund rate is a relevant indicator of market participants´ expectations of the
future MP that can be compared to fed fund target rate set up by FOMC ordinarily eight times a year
under Fed´ transparent period. Fed fund target rate directly influences effective fed fund rate which is
seasonally not-adjusted short-term nominal interest rate at which depository institutions trade federal
funds (balances held at Fed banks), with each other overnight (FRED, 2020), and indirectly influence
long-term rates.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
13
Chart 1: Raw time series (1971-2019)
Source: own calculations [Excel]
Chart 1 includes three time series since 4th January 1971 to 17th May 2019. Effective fed fund rate and
lower-bound fed fund target rate are both named in percentage (left axis), while stock price S&P 500
index is chartered on the right axis in points.
We went through all FOMC announcement days (T0, “10”), including also nine days ahead of them (T-
9 (“1”) to T-1 (“9”)) and five following days (T1 (“11”) to T5 (“15”)). When two FOMC meetings occurred
close to each other (and some days would be paralelly included into both time ranges, ex-post previous
FOMC and ex-ante the following FOMC), the meeting taking place later was preferred so that each
official meeting had at least five days ahead observed (unless that would mean zero ex-post days – in
that case one following date. The direction of fed fund rate and the stock index information have been
observed (returns, downside volatility, volume trades). Unofficial changes of fed fund rates were not
those planned ahead, there was no study of days around these dates of rate changes. One of the
observations is that market rate represented by effective FFR was in some particular times even lower
than target FFR set up by Fed. Fed fund target rate has started been modified in multiples of 25 bp since
1989.
Chart 2: Observations around FOMC announcement days (mentioned as number 10)
Source: statistical software eViews
In the Chart 2 there are number of observations of the dates around the FOMC official announcement
dates (“10”), approximately two weeks ahead and one week afterwards. There were 530 observed
FOMC dates, and their total number is decreasing at both sides as the chance of another (next or
0
500
1000
1500
2000
2500
3000
3500
0
5
10
15
20
25
04.0
1.19
71
04.0
1.19
73
04.0
1.19
75
04.0
1.19
77
04.0
1.19
79
04.0
1.19
81
04.0
1.19
83
04.0
1.19
85
04.0
1.19
87
04.0
1.19
89
04.0
1.19
91
04.0
1.19
93
04.0
1.19
95
04.0
1.19
97
04.0
1.19
99
04.0
1.20
01
04.0
1.20
03
04.0
1.20
05
04.0
1.20
07
04.0
1.20
09
04.0
1.20
11
04.0
1.20
13
04.0
1.20
15
04.0
1.20
17
04.0
1.20
19
EFFR Low SPX
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
14
previous) FOMC meeting being in these scales was increasing. There had been in total 55 out-of-sample
changes in FFR, all of them in non-transparent period. NBER evaluated ex-post an observed period of
12 202 days as follows: 10 695 boomy days (86.4 % of time) and 1 507 “rainy” days (13.6 %).
From boxplots (Chart 3) is apparent that the largest volatility of returns is observed around expansionary
FOMC meetings, the lowest controversely with restrictionary meetings, but skewed more upside than
downside.
Chart 3: Boxplot of S&P 500 daily returns under different MP, excluded further observations
Notes: Each column represent average mean returns of stock price index over observed time period, based on
change in target FFR. Inner period includes days around FOMC official meetings, while outer period changes
are those outside.
Source: own calculations [Excel]
Results
ADF and KPSS tests were proceeded in order to address issue of unit roots at 10% significance level1. Stationary time series were generally not identified to be cross-correlated except of MP shock II with
effective FFR differences (91 %).
The Granger causality table reveals that monetary variables do Granger cause stock returns, but not
inversely. Obviously, different interest rates influence each other on the money markets, as well as fixed-
income markets.
Table 1: Granger causality among monetary time series
1 This paper works with * at 10%, ** at 5% and *** at 1% statistical significance level.
MP_SHOCK_I does not Granger Cause DL_SPX 12200 1.51226 0.2205
DL_SPX does not Granger Cause MP_SHOCK_I 1.42612 0.2403 MP_SHOCK_II does not Granger Cause DL_SPX 12200 7.13903 0.0008***
DL_SPX does not Granger Cause MP_SHOCK_II 2.02327 0.1323
DUMMY_NBER does not Granger Cause DL_SPX 12200 3.37517 0.0342**
DL_SPX does not Granger Cause DUMMY_NBER 0.25811 0.7725
D_EFFR does not Granger Cause DL_SPX 12200 11.3472 1.E-05***
DL_SPX does not Granger Cause D_EFFR 1.55798 0.2106 D_TEN_YEAR_USTREASURIES does not Granger Cause DL_SPX 12200 16.6479 6.E-08***
DL_SPX does not Granger Cause D_TEN_YEAR_USTREASURIES 1.27210 0.2803 D_FFR_LOWER does not Granger Cause DL_SPX 12196 0.98226 0.3745
DL_SPX does not Granger Cause D_FFR_LOWER 0.09294 0.9112
DL_M1 does not Granger Cause DL_SPX 11189 3.00225 0.0497**
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
15
Sources: statistical software eViews
In total 426 cases, the Fed had made no changes in its policy rate, in 48 cases it decreased FFR while in
56 cases increased (once even during the economic crisis on 24 Sept 1982). Surprisingly, window study
over 530 official FOMC meetings revealed shocking information about the economic theory that
expansive MP actually caused the tumble in stock prices and restrictive MP led to increase in stock
returns – this concept was valid only during economic downturns, while during economic booms it was
vice versa.
Table 2: Change in MP based on business cycle and different MP
Source: own calculations [Excel]
In terms of the stock mean returns, if Fed had taken a restrictive tool, the stock markets had been
producing positive returns already two days ahead and even on the Day, while during three days
afterwards handing over previous profits. This could imply the theory that market participants have no
negative anticipation ex-ante, but restrictive surprise then support investors in their selling out the stocks
right afterwards.
Table 3: S&P 500 in-sample daily mean returns near FOMC dates under different MP
Source: own calculations [Excel]
Curiously, mean stock returns, if rates went up, have generally exceeded those returns if rates fell down,
three days ahead up to the official announcement dates, but this relation has inversed ex-post the FOMC
announcement when truely lowering the rate leads to higher returns ex-post and oppositely.
Inter-FOMC meetings average daily return during expansionary policy and outside of our observed
samples T-9 to T+5 is 0.342 %, significantly higher than anytime else, compared to population
expansionary average return 0.030 %, or to sample expansionary return 0.018 %. That identifies the fact
that unanticipated MP during unofficial meetings had significant power to influence the financial
markets.
Business cycles
Boom Crisis Celkem Number of FOMC mtg Celkem Average SPX return
Change in FFR Number of FOMC mtg Average SPX return Number of FOMC mtg Average SPX return
Down 33 -0,202% 15 0,104% 48 -0,106%
Unchanged_inner 367 0,174% 59 -0,060% 426 0,142%
Up 55 0,202% 1 -0,397% 56 0,191%
Total 455 0,150% 75 -0,032% 530 0,124%
DL_SPX does not Granger Cause DL_M1 2.54529 0.0785 DL_VOLUME does not Granger Cause DL_SPX 12196 0.62615 0.5347
DL_SPX does not Granger Cause DL_VOLUME 6.74925 0.0012***
D_TEN_YEAR_USTREASURIES does not Granger Cause D_EFFR 12200 3.52096 0.0296**
D_EFFR does not Granger Cause D_TEN_YEAR_USTREASURIES 32.1849 1.E-14***
D_FFR_LOWER does not Granger Cause D_EFFR 12196 12.6457 3.E-06***
D_EFFR does not Granger Cause D_FFR_LOWER 4.69516 0.0092***
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
16
Chart 4: S&P 500 daily mean returns around FOMC dates under different MP
Source: own calculations [Excel]
Stock markets´ daily volatility of the returns, measured by standard deviation (Table 4), spiked
especially on the official FOMC announcement days, regardless of monetary policy rate change. A bit
surprisingly, variance during expansionary periods was rather significantly higher than during no change
or even contractionary periods. Particularly, if Fed lowered the rates, the stocks volatility jumped
enormously one day ahead and on the official date.
Table 4: S&P 500 daily volatility around FOMC dates
Source: own calculations [Excel]
Opposite to results of the Table 2, Chart 5 shows that in the sample during the days around FOMC
meetings any change in target FFR led to cumulative decrease of stock returns during economic booms,
and inversely both changes produced higher cumulative returns during crises.
Chart 5: S&P 500 in-sample daily mean cumulative returns under different business cycles and
MP
Source: own calculations [Excel]
In-sample stock volatility was obviously higher during economic crises, especially periods when Fed
made no updates in target FFR.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
17
Chart 6: S&P 500 in-sample daily returns volatility in different business cycles and MP
Source: own calculations [Excel]
Chart 7: S&P 500 downside volatility of daily returns under different MP
Source: own calculations [Excel]
For volatility measurement, a semi-variance model was selected too. The shape of semi-variance curves
copies similarly the widely used variance model, results show however smaller risks. Findings reveal
that semi-variance pops on the Day if Fed lowered the rates. Semi-volatility of returns is higher for
expansionary MP of the Fed.
Daily trading volumes are abnormally high on the official Days, even more if FOMC decided to
implement more loosen monetary tools. From the charting is nearly apparent some Week effect when
Wednesdays have more closed trades than other weekdays.
Chart 8: S&P 500 daily mean trading volumes around FOMC dates under different MP
Source: own calculations [Excel]
There are both monetary shocks present (left axis) on the histogram below over the observed period of
time, with daily stock returns too (right axis) with further observations excluded. The magnitude of the
shocks is related to the environment of fed fund rates then, as well as to the less transparent period when
the Fed was not focusing that much on forward guidance and qualitative speeches of governors and
others.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
18
Chart 9: MP shocks and stock returns
Sources: statistical software eViews
Conclusions
Author in his previous research papers has been studying and analysing factors influencing stock
markets via regressions, VAR and VEC models. From those the main conclusion was that it is mostly
real macroeconomic variables that play statistically significant role while monetary policy´ tools not that
much, even though Granger causality table in this paper shows statistically significant one-way causal
relations from monetary variables to the stock returns. However, in this paper author focused on the
relationship between particularly interest rates set up by Federal Open Market Committee and stock
returns that seem to vary over different periods of time, which is the key research of this paper.
Moreover, stock returns are significantly higher during inter-FOMC meetings than during the observed
periods of FOMC with nine days ahead and five days afterwards.
Findings here focused mostly on the event study of the Fed windows under different MP and business
cycle. There are many more alternatives to track, e.g. periods of transparency of the Fed, or structural
changes of primary MP targets that were not feasible to mention here in this research paper, but another
future ones. Approximately three fifths of the time had been considered as in-the-sample close FOMC
official meetings. The week effect was recognized on the daily trading volumes when in-sample –
Wednesdays are the top daily weekdays for their trading activity.
The largest volatility of returns is observed around expansionary FOMC meetings, the lowest
controversely with restrictionary meetings, but skewed more upside than downside. Findings reveal that
semi-variance of stock returns pops on the Day if Fed lowered the rates. Surprisingly, window study
over 530 official FOMC meetings revealed shocking information about the economic theory that
expansive MP actually caused the tumble in stock prices and restrictive MP led to increase in stock
returns – this concept was valid only during economic downturns, while during economic booms it was
vice versa. Additionally, in the sample around FOMC meetings any change in target FFR led to
cumulative decrease of stock returns during economic booms, and inversely both changes produced
higher cumulative returns during crises.
If Fed lowered the rates, the stocks volatility jumped enormously one day ahead and on the official date.
Acknowledgement
This paper has been elaborated in the framework of the grant programme „Support for Science and
Research in the Moravia-Silesia Region 2018" (RRC/10/2018), financed from the budget of the
Moravian-Silesian Region.
The author also gratefully acknowledges financial support from the VSB – Technical University of
Ostrava SGS grant project no. SP2020/116 (Economic policy challenges in developed countries). This
article has been supported by SGS grant from VSB – Technical University of Ostrava no. SP2020/116
(project ‘Economic policy challenges in developed countries’).
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
19
References
[1] Bernanke, B. and K. N. Kuttner, 2005. What Explains the Stock Market's Reaction to Federal
Reserve Policy? The Journal of Finance [online]. 60(3), 1221-1257 [cit. 2020-01-02]. DOI:
10.1111/j.1540-6261.2005.00760.x. ISSN 00221082. Available from:
http://doi.wiley.com/10.1111/j.1540-6261.2005.00760.x
[2] Bernanke, B. and I. Mihov, 1998. Measuring Monetary Policy. The Quarterly Journal of
Economics [online]. 113(3), 869-902 [cit. 2020-01-22]. DOI: 10.1162/003355398555775. ISSN
0033-5533. Available from: https://academic.oup.com/qje/article-
lookup/doi/10.1162/003355398555775
[3] Bernanke, B. and K. N. Kuttner, 2005. What Explains the Stock Market's Reaction to Federal
Reserve Policy? The Journal of Finance [online]. 60(3), 1221-1257 [cit. 2020-01-02]. DOI:
10.1111/j.1540-6261.2005.00760.x. ISSN 00221082. Available from:
http://doi.wiley.com/10.1111/j.1540-6261.2005.00760.
[4] Board of Governors of the Federal Reserve System: About the Fed [online], 2019. Washington,
D.C., USA [cit. 2019-11-29]. Available from: https://www.federalreserve.gov/aboutthefed.htm
[5] FRED. Federal reserve economic data – FRED – St. Louis Fed. [online]. Federal reserve bank of
St. Louis, 2019. [cit. 2019-06-20]. Available from www:
<https://fred.stlouisfed.org/series/fedfunds>.
[6] Granger, C. W. J., 1969. Investigating Causal Relations by Econometric Models and Cross-
spectral Methods. Econometrica [online]. 37(3) [cit. 2020-01-22]. DOI: 10.2307/1912791. ISSN
00129682. Available from: https://www.jstor.org/stable/1912791?origin=crossref
[7] Greenspan, A., 1997. Remarks by Chairman Greenspan at the 15th Anniversary Conference of the
Center for Economic Policy Research at Stanford University, September 5.
[8] Haitsma, R., D. Unalmis and J. de Haan, 2016. The impact of the ECB's conventional and
unconventional monetary policies on stock markets. Journal of Macroeconomics [online]. 48, 101-
116 [cit. 2020-01-02]. DOI: 10.1016/j.jmacro.2016.02.004. ISSN 01640704. Available from:
https://linkinghub.elsevier.com/retrieve/pii/S0164070416000276
[9] Kuttner, K. N, 2001. Monetary policy surprises and interest rates: Evidence from the Fed funds
futures market. Journal of Monetary Economics [online]. 47(3), 523-544 [cit. 2020-01-02]. DOI:
10.1016/S0304-3932(01)00055-1. ISSN 03043932. Available from:
https://linkinghub.elsevier.com/retrieve/pii/S0304393201000551
[10] Kontonikas, A., R. MacDonald and A. Saggu, 2013. Stock market reaction to fed funds rate
surprises: State dependence and the financial crisis. Journal of Banking & Finance [online]. 37(11),
4025-4037 [cit. 2020-01-02]. DOI: 10.1016/j.jbankfin.2013.06.010. ISSN 03784266. Available
from: https://linkinghub.elsevier.com/retrieve/pii/S0378426613002987
[11] Robertson, J. and D. Thornton, 1997. Using federal funds futures rates to predict Federal Reserve actions,
Review from Federal Reserve Bank of St. Louis, Nov. 45-53
[12] Rudebusch, G. D., 1995. Federal Reserve interest rate targeting, rational expectations, and the
term structure. Journal of Monetary Economics [online]. 35(2), 245-274 [cit. 2020-01-02]. DOI:
10.1016/0304-3932(95)01190-Y. ISSN 03043932. Available from:
https://linkinghub.elsevier.com/retrieve/pii/030439329501190Y
[13] Sellin, P, 2002. Monetary Policy and the Stock Market: Theory and Empirical Evidence. Journal
of Economic Surveys [online]. 15(4), 491-541 [cit. 2020-01-02]. DOI: 10.1111/1467-6419.00147.
ISSN 0950-0804. Available from: https://onlinelibrary.wiley.com/doi/abs/10.1111/1467-
6419.00147
[14] Sirucek, M, 2011. Impact of monetary policy on US stock market. Trends economics and
management, Vol. V, No. 09 (September 02, 2011): 53-60.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
20
[15] Tessaromatis, N. P., 1991. Money supply announcements and stock prices: the UK evidence. ISSN:
1105-8919, Vol. 41, Edition: 4, Page: 408-419
[16] Thornton, D. L., 2014. Monetary policy: Why money matters (and interest rates don’t). Journal of
Macroeconomics [online]. 40, 202-213 [cit. 2020-01-02]. DOI: 10.1016/j.jmacro.2013.12.005.
ISSN 01640704. Available from:
https://linkinghub.elsevier.com/retrieve/pii/S0164070414000044
[17] Unalmis, D. a I. Unalmis, 2015. The Effects of Conventional and Unconventional Monetary Policy
Surprises on Asset Markets in the United States [online]. Munich [cit. 2020-01-04]. Available
from: https://mpra.ub.uni-muenchen.de/62585/
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
21
APPLICATION OF MEAN-REVERSION BINOMIAL LATTICE APPROACH TO
VALUATION OF MORTGAGE IMPLICIT OPTION IN THE CZECH MARKET
Vladimír Bulko1
1Department of Finance, VŠB – Technical University of Ostrava,
Sokolská tř. 33, Ostrava 70200, Czech Republic
e-mail: [email protected]
Abstract
In the moment of a mortgage approval a Czech bank virtually writes an implicit option to a client who
either exercises it by signing the Mortgage contract or the option expires worthless. Described implicit
option is valued by the means of binomial lattice framework with utilization of the mean-reversion
approach to valuing of interest-rate backed derivatives. The method of valuation is applied to one of the
Czech banks mortgage data. The paper introduces the methodology and model framework with the main
focus on the application on the Czech specific data and analysis of the results. Goal of the paper is to
apply mean-reversion binomial lattice approach to valuation of interest-rate backed options to a
mortgage origination implicit option in the Czech market. The paper is concluded by a recommendation
to the Czech banks whether the used methodology yields significant results for this particular implicit
option.
Keywords
Mortgages, Options, Binomial lattice, Mean reversion
JEL Classification
C58, C63, D81, G12, G21
1 Introduction
The paper aims to apply standard financial methods to value an implicit option written to clients when
their mortgage is originated. The Czech mortgage data are used as an empirical background. The
profitability of mortgages is a complex problem to solve and this paper focuses on only a short part of
the mortgage’s life – the origination. Origination of the mortgage is defined here as a time between
approval of the mortgage with fixed interest rate (IR) and signing of the mortgage contract. The time
gap creates a risk for the bank. When the contract is finally signed the market rates might be very far
from the interest rate signed in the contract. The risk can be propagated into the mortgage’s profitability
via two channels: cost of funding (increased market rate increases cost of funding the mortgage and so
decreases profit margin) and opportunity cost (bank could have allocated its capital to current mortgages
with higher rates rather than yesterday’s, so increasing implied cost of capital and again decreasing profit
margin). To include this risk to the profitability it must be properly valued.
Relevant literature on mortgage-backed options valuation is scarce. The authors focus either on
valuation of a whole mortgage and its derived securities represented by Kau et al. (1987) and Calvo-
Garido and Vázquez (2017) or on more technical aspects of valuing the mortgage prepayment and
default options such as Hürliman (2011).
As literature directly related to the valuation of the risk described above was not found, the paper aims
to fill the gap by describing a possible approach and outcomes. The risk is valued by a construction of
the specific implicit option and valuing it by means of mean-reversion modeling and binomial tree
approach.
The first part of the paper presents the investigated implicit option, mean-reversion framework to model
the underlying asset’s price, binomial lattice approach to valuation of the option and at the end a brief
description of the data used. Second part focuses on estimation of necessary parameters and valuation
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
22
of the implicit option. Concluding paragraph focuses on discussion of the results and recommendation
for further investigation.
2 Methodology and data
2.1 Mortgage origination implicit option
The investigated implicit option is written by the bank to a client in the moment of mortgage approval.
The Implicit option gives a right to a client to buy one unit of mortgage contract, hence it is a CALL
option, expires in 30 days and is of the American type because a client can exercise the option by signing
the mortgage contract any time since writing (mortgage approval). Strike price (IR to be signed) for the
implicit option is determined before a mortgage approval2. A rational client considers exercising the
option only unless he/she is offered a better rate from a competing bank. There exist also non-interest-
based costs (such as opening fees, discounts for current account usage, long-term loyalty benefits, etc.)
attributed to signing a mortgage contract in a specific bank which a rational client has to consider,
however from a mortgage life-time point of view these costs are negligible and therefore this paper
omits those in the decision rule. To decide whether the option is in-the-money, the data on all offers for
all clients at all times would have to be available however we have to rely on average mortgage market
interest rates (AMMIR).
2.2 Mean-reversion approach to model AMMIR
The decision rule is directly based on the dynamics of the AMMIR in time and therefore before
embarking on valuing the implicit option, AMMIR must be modelled at the first place. AMMIR as a
price is an IR which is experienced to behave differently to for example stock prices in a way that its
trajectory tends to reverse back to some theoretical mean. Intuition behind is that if the IR could rise
infinitely (as stock prices), this would have stopped most of the current economic activity because it
would have made today’s consumption very expensive relative to tomorrow’s consumption and all
surpluses would have been invested. In other words, IR is a price which intermediates inter-temporal
relationship between consumption today and tomorrow. This intuition coupled with hard data leads
economists to believe that interest rates in the long-run tend to hover around some theoretical steady
state (mean) which ensures inter-temporal general equilibrium.
Family of so-called mean-reversion models can be expressed by a differential equation
𝜹𝒓 = 𝜿(𝝁 − 𝒓)𝜹𝒕 + 𝝈𝒓𝜸 𝜹𝒛 (1)
where r stands for modelled IR (AMMIR in our case), 𝜅 is a rate of mean-reversion, µ is a theoretical
mean to which IR tends to return, σ controls a magnitude of randomness entering the system, γ express
an elasticity between IR change and the level of IR and δz is a standard Brownian motion. The equation
(1) can be reformulated to a form better suited for estimation
𝒓𝒕+𝟏 − 𝒓𝒕 = 𝜶 + 𝜷𝒓𝒕 + 𝜺𝒓,𝒕+𝟏 (𝟐)
𝑬𝒕[𝜺𝒓,𝒕+𝟏𝟐 ] = 𝝈𝟐𝒓𝒕
𝟐𝜸 (𝟑)
where 𝛼 = 𝜅 ∙ µ ∙ δ𝑡 and 𝛽 = −𝜅 ∙ δ𝑡, hence µ = −𝛼
𝛽 . Based on restrictions placed on different
parameters of the equation (1) we can distinguish number of well-known mean-reversion models. In this
2 Strike price of the implicit option is determined in the moment of mortgage IR offer by a banker to a client. The
offer tends to happen 2 to 4 weeks before the mortgage is approved, so there is already some time for the underlying
asset’s price to deviate from the offered fixed IR in upward direction. Result is that in the moment of writing, the
implicit option can be already deeply in the money (a negative open position for a bank). It is important to mention
that the offered fixed IR is not an object of approval and a bank have absolutely no means of adjusting it in upward
direction after the offer is given.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
23
paper however, only Vasicek (1977) model is investigated due to its frequent usage and parameter
restriction 𝛾 = 0 which removes a variance endogeneity from the model and hence yielding σ equal to
standard deviation of i.i.d. process residuals of the econometric interpretation (2).
2.3 Binomial model framework
As the implicit option is the American type, we cannot use the canonical Black-Scholes-Merton (BSM)
differential equation for valuation and the more robust Binomial model framework is utilized. This paper
uses description and notation of the binomial method from (Higham, 2002), however it is ultimately
based on original work from (Cox et al., 1979).
Let S be the price of the underlying asset in time t = 0 when an option is written. Holder may exercise
the option by buying the asset any time until expiration at time T. Between writing of the option and its
expiration there is discrete time assumed with a step δt. Any point in time t can be described as 𝑡 =(𝑖 − 1)𝛿𝑡; 𝑖 ∈ [1,𝑀 + 1] and therefore 𝑇 = 𝑀𝛿𝑡 . The crucial block of the binomial method is an
assumption that between successive points in time the asset’s price can move only in two directions –
up or down. Probability of an upward movement is p and of the downward movement is 1-p. Let u
denote the magnitude of the upward movement with condition 𝑢 > 1 and d the magnitude of the
downward movement with condition 𝑑 ∈ [0,1), we can derive price in the time 𝑡 = 𝛿𝑡 as either uS or
dS. Similarly, the asset’s price in time 𝑡 = 2𝛿𝑡 is derived to get three possible prices u2S, udS or d2S.
Generally, for any time 𝑡 = (𝑖 − 1)𝛿𝑡 there is 𝑖 possible asset prices which are denoted as
𝑆𝑛𝑖 = 𝑢𝑖−𝑛𝑑𝑛−1𝑆; 𝑛 ∈ [1, 𝑖]; 𝑖 ∈ [1,𝑀 + 1] (4)
and form the recombining binomial tree.
Let E be the strike price of the option which is a price the option holder can buy the underlying asset for
at any time until expiration at time T. Expiration time 𝑡 = 𝑡𝑀+1 = 𝑇 is special in a sense that at this
moment the option must be either exercised with profit 𝑆𝑛𝑀+1 − 𝐸 or expires worthless. Hence the value
of the option at expiration can be derived as
𝑉𝑛𝑀+1 = 𝑚𝑎𝑥(𝑆𝑛
𝑀+1 − 𝐸, 0) (5) Goal of the binomial method is to find the value of the option at t = 0 and this is achieved by recurring
weighting of the option values by probability and time. We know the possible option values in expiration
𝑉𝑛𝑀+1 from (5) and financial modelling theory states that we can find any 𝑉𝑛
𝑖 with recurrence equation
𝑉𝑛𝑖 = e−ρδt(𝑝𝑉𝑛+1
𝑖+1 + (1 − 𝑝)𝑉𝑛𝑖+1) ; 𝑛 ∈ [1, 𝑖]; 𝑖 ∈ [1,𝑀] (6)
where 𝜌 denotes a risk-free interest rate representing an opportunity cost of holding the option. Since
the option we price here is American type, it is possible to exercise the option in any point in time,
therefore equation (6) is not enough. Rational holder of an American type option must consider both the
expected price of the underlying asset in expiration and importantly also its current price. Such holder
considers in each point in time either making an instant profit if 𝐸 > 𝑆𝑛𝑖 or waiting another fraction of
time in expectation of making a profit then. This intuition leads us to the recurrence equation
representing our desired decision rule
𝑉𝑛𝑖 = 𝑚𝑎𝑥(𝑚𝑎𝑥(𝑆𝑛
𝑖 − 𝐸, 0), e−ρδt(𝑝𝑉𝑛+1𝑖+1 + (1 − 𝑝)𝑉𝑛
𝑖+1)); 𝑛 ∈ [1, 𝑖]; 𝑖 ∈ [1,𝑀] (7)
2.4 Binomial model parameters
To utilize the Binomial method for the problem stated in this paper we need to explicitly determine the
parameters p, u, d, 𝜌, M, E and S.
Price of the underlying asset at writing of the option S is defined in this paper as the AMMIR in the
month of the implicit option writing (month of mortgage approval) and will be modeled using mean-
reversion approach described above. Strike price E is obtained from the data on real mortgage offers
described in the next subchapter.
Technical parameter M must be chosen high enough for the binomial method to converge fast enough
to BSM results, however low enough for the method to be numerically stable and computationally
reasonable. The convergence is tested by changing the nature of the option to European type and
comparing results of the Binomial method with the results of BSM model for this testing option. As the
implicit option has expiration of 30 days, the multiples of 30 are tested as proposed values for M.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
24
Opportunity cost of holding the mortgage origination implicit option for a client is negligible because
except for spending time fetching the offer a client does not directly allocate any capital into the
transaction yet. This is underlined by a common practice of clients to collect number of mortgage rate
offers from different banks to compare and chose the one with the lowest offered rate – such practice
literally cost them nothing in neither direct cost nor invested capital. On the other hand, the writer of the
implicit option (a bank) invests capital to the transaction from the very beginning, for instance by
allocating the time of the banker and the back-office specialists to prepare the mortgage or maybe even
more importantly by taking the mere risk of the rates moving in the undesired direction. A bank could
have put its capital to the account at Czech national bank (CNB) and earn risk-free interest from just
keeping the capital there. Essentially a bank is deciding between the writing of the implicit option and
depositing its capital to CNB and consequently the CNB risk-free rate should be priced into the value
of the written option. Therefore, risk-free interest-rate 𝜌 is for the purpose of this paper chosen to be
average 2W REPO rate announced by the CNB in the month of the implicit option writing.
Mean-reversion Binomial model differs to standard Binomial model in the sense that in each node there
must be a consideration on the probability and/or magnitude of the upward and downward movements
to capture the mean-reversion setting of the underlying asset compared to standard parameters fixed
throughout all nodes. This paper determines parameters u and d implicitly based on work from Bastian-
Pinto (2015), such that the price of underlying asset in each node is formulated as
𝑆𝑛𝑖 = 𝜇 + (𝑆1
1 − 𝜇)𝑒−𝜅(𝑖−1)𝛿𝑡 + (𝑖 − 2𝑛 + 1)𝜎√𝛿𝑡 (8) where the first term of the sum 𝜇 is the steady-state level of the underlying asset, the second term
represents the mean-reversion setting and the third term brings time determined volatility into the system
with (𝑖 − 2𝑛 + 1) representing difference between number of upward and downward movements.
Probability of the upward movement p follows standardly described behavior
𝑝 =𝑒𝜅𝛿𝑡 − 𝑑
𝑢 − 𝑑; 𝑝 ∈ [0,1] (9)
2.5 Data description
Data are obtained either from the public sources (such as 2W REPO, PRIBOR) or are the courtesy of
one of the Czech banks (such as AMMIR, IRS or data for mortgage contracts).
AMMIR (known as “FINCENTRUM HYPOINDEX”) is constructed by NEWVALUES (2020) as a
weighted average interest rate pooling average rates of different mortgage fixes and loan-to-value
measures and weighting them by originated volumes. Throughout the whole Czech banking market this
AMMIR is considered as the best proxy for the average mortgage market rate available. The CNB also
publishes its own average mortgage market rate however this rate does not coincide with the banking
industry definition in a way that CNB rate on one hand omits unpurposed mortgages (and considers
them as consumer loan) and on the other hand includes all building savings loans which has considerably
higher rates than standard bank mortgages. This creates an upward bias of the CNB metric and that is
why we decided to use rather the widely acknowledged but paid metric before the freely published
however biased one.
Source of publicly published rates 2W REPO and PRIBOR is CNB (2020). 2W REPO monthly
representation is created by taking last day of month value to keep the sharp transitions in between
months when CNB monetary policy change occurs. PRIBOR is used as a daily average of the month
provided by CNB.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
25
Figure 1. Czech mortgage market important interest rates
(source: Author)
Data used to determine the strike prices E and are anonymized real data on 12941 mortgage contracts
which were approved two consecutive years. These data are the courtesy of one of the bigger mortgage
providers at the Czech market. Interesting fact from the data is that on average 8% of mortgage
origination implicit options does not result in a signed contract and median time to sign the contract
(exercise) is 20 days. Furthermore if we compare approved IR (strike E) with the AMMIR of the same
month (underlying asset’s price at t=0) we can distinguish 3 standard types of implicit options written
together with a mortgage contract: in-the-money (ITM), at-the-money (ATM) and out-of-the-money
(OTM). Figure (2) shows the share of ITM in all implicit options by month (few ATM options are
included into OTM in the figure (2)). Throughout the data period, there were 36% of ITM options
written.
0,00%
1,00%
2,00%
3,00%
4,00%
5,00%
6,00%2
00
9/0
1
200
9/0
5
200
9/0
9
201
0/0
1
201
0/0
5
201
0/0
9
201
1/0
1
201
1/0
5
201
1/0
9
201
2/0
1
201
2/0
5
201
2/0
9
201
3/0
1
201
3/0
5
201
3/0
9
201
4/0
1
201
4/0
5
201
4/0
9
201
5/0
1
201
5/0
5
201
5/0
9
201
6/0
1
201
6/0
5
201
6/0
9
201
7/0
1
201
7/0
5
201
7/0
9
201
8/0
1
201
8/0
5
201
8/0
9
201
9/0
1
201
9/0
5
201
9/0
9
CNB 2W REPO AMMIR PRIBOR 1M IRS 5Y
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
26
Figure 2. Share of options by value at writing
(Source: Author)
3 Vasicek model parameters estimation
Due to linearity of the econometric interpretation of Vasicek model (with 𝛾 = 0) in equations (2) and
(3), the standard ordinary least squares (OLS) estimator is utilized. Data have monthly frequency and
model is estimated on periods 2009/01:2019/11 and 2014/01:2019/11. The latter period is expected to
result in more reasonable parameters for the implicit option valuation due to trajectory of the AMMIR
in the Figure (1).
3.1 Estimation results
F-statistic (p-value 0.028**) rejects the hypothesis of the full period model’s insignificance, however
the insignificance of the restricted period model is not rejected (p-value 0.14). Both White and Breusch-
Pagan tests did not reject hypothesis of heteroscedasticity for standard OLS estimator, therefore
heteroscedasticity-consistent standard errors are used. Furthermore, based on the Breusch-Godfrey test,
residuals with lag (-2) are identified as significantly correlated with dependent variable pointing to an
autocorrelation problem. Autocorrelation problem can be solved with better specification of the model
via adding dependent variable lags of the higher order to the model, although this would considerably
alter the Vasicek model (1977) and is therefore only noted in this paper. Lastly the hypothesis of
normality of residuals is also rejected.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
ITM OTM
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
27
Table 1. Vasicek model estimated parameters
estimation data
period metric
calculated parameters estimated parameters
𝜅 µ α β σ
2009/01:2019/11
value 0.212936 0.017365 0.000308138 -0.0177447 0.000666
lower 0.402497 -0.005687 -0.000190743 -0.0335414 not relevant
upper 0.023376 0.414289 0.000807019 -0.00194796 not relevant
p-value not relevant not relevant 0.2239 0.028** not relevant
2014/01:2019/11
value 0.448261 0.020726 0.000774215 -0.0373551 0.00056774
lower 1.046719 -0.003726 -0.00032504 -0.0872266 not relevant
upper -0.150196 -0.149682 0.00187347 0.0125163 not relevant
p-value not relevant not relevant 0.1645 0.1397 not relevant
(source: Author)
Lower and upper terms in Table (1) refer to values of the lower and upper bounds of the 95% confidence
interval of the parameters’ estimation. Number of * with the p-value represents statistical significance
of a parameter based on standard t-test or F-test (* >90%, ** >95%, *** >99%).
3.2 Estimated parameters discussion
Presented results show that the Vasicek model can only very weakly explain the behavior of the AMMIR
in the Czech market. Despite estimated parameters α and β yields reasonable parameters 𝜅 and µ, when
the 95% bands are taken into consideration the parameters can be anywhere including the negative
values.
Parameter 𝜅 estimated value from the full period model is more reasonable as it can hover between 2%
and 40% rate of return to mean in a one month. On the other hand, the restricted period model yields
parameter 𝜅 which can become even negative 15% and this would violate the mean-reversion settings
and would lead to exponential behavior of the AMMIR.
Results for long-term mean µ follow suite in having reasonable estimated value with however
unreasonable upper and lower bounds. The full period model estimates this parameter to be at 1.74%
and the restricted period model at 2.07%. This is reasonable because AMMIR hovered close to 2% in
the past 5 years and trended towards the 2% level in the 5 years before that (Figure 1). Looking at the µ
upper and lower bounds for both models however this shows the very weak ability of the Vasicek model
to interpret Czech AMMIR. For the full period model long-term mean AMMIR could be between -
0.57% and +41.43% and from the restricted period model the estimated value is not even included inside
the bounds.
Taking into consideration the results and their significance, the paper uses estimated parameters from
the full period model despite the expectation expressed at the beginning of this chapter.
4 The implicit option valuation
The results of valuation of the mortgage origination implicit options are the weighted average of the
values for each implicit option from the data. The contracted volume of the mortgage is used as a weight,
making value of option for bigger volume mortgage more relevant in the results. Two metrics represent
the results:
• relative price for a client - option value in the form of per annum (p.a.) IR - it is a natural result
of the calculations because the underlying asset is an IR,
• absolute price for a client - the value of the option in the Czech currency unit (CZK) calculated
as a one-month (30 days maturity option) interest income that would be paid to a bank for buying
the option for a concrete mortgage, in other words mortgage volume multiplied by relative price
for a client per month
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
28
Due to the fact that the share of ITM options at origination is high compared to OTM, the results are
mostly comprised of the intrinsic value of the options. Therefore, the author decided for the sake of
presenting the ability of the Binomial method to yield reasonable results to show also results of the
computation as if all the options where originated as ATM.
For computational efficiency purposes the parameter M was chosen to be 60 as it yields results close to
the high multiples of 30, such as M=1200.
4.1 Results with Vasicek model estimated parameters
With Vasicek full period model estimated parameters (𝜅=21.3%; µ=1.74%; σ=0.0666%) the average
value of the implicit options written during the investigated period was 0.0592% (in basis points
5.92 bps) and the absolute price which a client should pay to buy the option is CZK 258. Figure (3)
shows the distribution of the options values, where the majority of the option values is below 4bps.
With assumption of all options being ATM at writing the average weighted value of the option is
0.95 bps and the average price for the client is CZK 136. The most part of the difference is the intrinsic
value of the options originated as ITM. On the other hand, also some options which were deep OTM
became ATM and may have got positive value at origination bringing more mortgages into the
weighting average and that is why the difference between absolute prices and relative prices does not
correspond.
Figure 3. Options value distribution under Vasicek full period model parameters
(source: Author)
4.2 Parameter sensitivity analysis
Due to the fact that Vasicek model produced statistically insignificant results, it is necessary to
investigate the sensitivity of results on different settings of parameters. For this analysis the author has
chosen parameters 𝜅, µ and E or more precisely parameter 𝜅 and differences between the price of the
underlying asset at origination S of the option and µ and E. The rest of parameters is fixed as follows: S = 2.5%; 𝜌 = 2%; σ = 0.0666%.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
29
Table 2. Value of implicit options under different parameter assumptions
𝜅 =10% S - µ
𝜅 =50% S - µ
1% 0,50% 0% -0,50% 1% 0,50% 0% -0,50%
S - E
0,50% 50,33 bps 50,33 bps 50,33 bps 50,33 bps
S - E
0,50% 50,33 bps 50,33 bps 50,33 bps 50,33 bps
0,20% 20,38 bps 20,38 bps 20,38 bps 20,38 bps 0,20% 20,38 bps 20,38 bps 20,38 bps 20,38 bps
0% 0,99 bps 0,99 bps 0,99 bps 0,99 bps 0% 0,96 bps 0,98 bps 0,99 bps 0,99 bps
-0,20% 0 bps 0 bps 0 bps 0 bps -0,20% 0 bps 0 bps 0 bps 0 bps
𝜅 =100% S - µ
𝜅 =500% S - µ
1% 0,50% 0% -0,50% 1% 0,50% 0% -0,50%
S - E
0,50% 50,33 bps 50,33 bps 50,33 bps 50,33 bps
S - E
0,50% 50 bps 50 bps 50,33 bps 52,27 bps
0,20% 20,38 bps 20,38 bps 20,38 bps 20,38 bps 0,20% 20 bps 20 bps 20,38 bps 22,32 bps
0% 0,87 bps 0,96 bps 0,99 bps 0,97 bps 0% 0 bps 0 bps 0,99 bps 2,35 bps
-0,20% 0 bps 0 bps 0 bps 0 bps -0,20% 0 bps 0 bps 0 bps 0 bps
(Source: Author)
Regarding rate of mean-reversion 𝜅 parameter it is obvious from the table (2) that its impact on the
implicit option value is negligible unless it becomes high or the investigated option was ATM at its
writing. Despite having small impact on the value of the option, the expected trajectory is strongly
impacted by the magnitude of 𝜅 (Figure 4). The same story can be told about sensitivity to S - µ. These
two parameters create mean-reversion setting for the AMMIR which is very weak in the model setup-
up presented in this paper.
On the other hand, value of the option is mostly sensitive to S – E as this relationship shows whether
the option is ITM, ATM or OTM in the moment of origination, hence determining its intrinsic value.
Extrinsic value for the ITM option comprises only a small part of the option’s value, although is
pronounced for ATM options.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
30
Figure 4. Binomial trees under different sets of parameters
(Source: Author)
5 Conclusions
In this paper we applied a mean-reversion binomial approach for valuation of the implicit option written
when a mortgage is approved to the client on the Czech market data. To capture mean-reversion setting
of the underlying asset the renown Vasicek model (1977) was used, its parameters were estimated on
11 years of monthly data utilizing standard econometric techniques. The estimates were however
statistically insignificant and we have to conclude that the Vasicek model with original set-up is
unsuitable for application on the Czech market. Despite statistical insignificance of the parameters, their
values were reasonable and therefore were used to value the implicit option via binomial model.
Application of the binomial approach was successful in yielding a reasonable result. Finally, a sensitivity
of option values on changes of parameters was investigated and we came to conclusion that the binomial
model is overall robust except for the impact of the option’s strike versus underlying assets price setting
at its writing. In conclusion the binomial approach is a strong tool to value the implicit options however
other mean-reversion models has to be investigated to improve the ability of the overall model to capture
the mean-reversion setting of the IR’s in the Czech market.
References [1] AHMADI, Z. et al. (2020). A lattice-based approach to option and bond valuation under mean-
reverting regime-switching diffusion processes. Journal of Computational and Applied
Mathematics, 363, pp. 156-170.
[2] BASTIAN-PINTO, Carlos et al. (2010). A Non-Censored Binomial Model for Mean Reverting
Stochastic Processes. Proceedings 14. Annual international conference on real options.
[3] BASTIAN-PINTO, Carlos de Lamare. (2015). Modeling Generic Mean Reversion Processes with
a Symmetrical Binomial Lattice - Applications to Real Options. Procedia Computer Science, 55,
pp. 764-773.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
31
[4] CALVO-GARRIDO, Maria and Carlos VÁZQUEZ. (2018). Mathematical analysis of obstacle
problems for pricing fixed-rate mortgages with prepayment and default options. Nonlinear
Analysis: Real World Applications, 39, pp. 157-165.
[5] CNB. (2020). PRIBOR rates - monthly and yearly averages. [online database]. Czech Republic:
Czech national Bank. Available at: <https://www.cnb.cz/en/financial-markets/money-
market/pribor/fixing-of-interest-rates-on-interbank-deposits-pribor/averages_form.html>.
[6] COX, J. C., S. A. Ross and M. Rubinstein. (1979). Option pricing: A simplified approach. Journal
of Financial Economics, 7, pp. 229-263.
[7] HAHN, Warren J. and James S. DYER. (2008). Discrete time modeling of mean-reverting
stochastic processes for real option valuation. European Journal of Operational Research, 184(2),
pp. 534-548.
[8] HIGHAM, Desmond J. (2002). Nine Ways to Implement the Binomial Method for Option
Valuation in MATLAB. SIAM Review, 44(4), pp. 661–677.
[9] HULL, John. (2018). Options, futures, and other derivatives. 10th ed. Upper Saddle River:
Pearson Prentice Hall. Prentice Hall series in finance. ISBN 978-9-35-286659-5.
[10] HÜRLIMANN, Werner. (2012). Valuation of fixed and variable rate mortgages: binomial tree
versus analytical approximations. Decisions in Economics and Finance, 35(2), pp. 171-202.
[11] KAU, James B. et al. (1987). The valuation and securitization of commercial and multifamily
mortgages. Journal of Banking and Finance, 11, pp. 525-546.
[12] KHRAMOV, Vadim. (2013). Estimating Parameters of Short-Term Real Interest Rate Models.
IMF Working Papers. WP/13/212.
[13] MUNNIK, Jeroen F. J. de. (1996). The valuation of interest rate derivative securities. New York:
Routledge. ISBN 0-415-13727-6.
[14] NEWVALUES. (2020). STATISTIKY HYPOEXPERT. [online database]. Czech Republic:
NEWVALUES s.r.o.. Available at: <https://new-values.com/hypoexpert/>.
[15] VAŠÍČEK, Oldřich. (1977). An equilibrium characterization of the term structure. Journal of
Financial Economics, 5, pp. 177-188.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
32
USING GEOINFORMATION IN PUBLIC ADMINISTRATION, CASE STUDY:
MORAVSKOSLEZKÝ REGION
Ivana Čermáková1
1Department of Applied Informatics, VŠB – Technical University of Ostrava,
Sokolská tř. 33, Ostrava 70200, Czech Republic
e-mail: [email protected]
Abstract
Provides information to citizens through web services is increasing trend in the public administration.
One of the key areas of the sharing information are spatial data. These data can be used for urban
planning decision making, passports (study regarding vegetation), traffic monitoring and so on. In these
days data can be provided immediately. Because of the UAV or satellites. Availability of the spatial
datasets accelerates the possibilities of public administration dealing. The article is proposed like a case
study. Area of interest is Moravskoslezský Region in the Czech Republic. Using the geoinformation in
the public administration, problems with implementation and possibilities of the next development are
contained also.
Keywords
Geoinformation, Moravskoslezský Region, Spatial Data, Public Administration
JEL Classification
H70, H83, O18, O38
Introduction
Information are the key part of the modern society. In these days, the information has an immeasurable
value. With technological development it´s much easier observe the geoinformation - information
regarding landscape and their changes. It means, that the society wants to be informed about a landscape,
their changes and the possibilities of the changes. Using the Geographic Information Systems (GIS) and
geoinformation saving time and money. These are part of the reasons why the public administration
started using the geoinformation. The global data providing is most using way in public administration.
But the accessary and the focus of the data wasn´t enough for the needs of the regions. So, most of the
regions or big cities created their own source of spatial data. The using of these information’s are
different and very extensive. So, Moravskoslezský region is chosen like area of the interest for the thesis.
The thesis is focused on source and using of spatial data on the regional level, namely Moravskoslezký
region in the Czech Republic. Problematic of geoinformation and their using in regional level is included
also. Possibilities of using geoinformation in future are included also.
Literature Review
The term geoinformation can be conceptualize in a number of ways. information used for geographic
services, geoinformatics, spatial position and spatial monitoring is one of the most used definition
explain geoinformation by Shaytura (2018). Geoinformation can be explain like information that support
the discipline of photogrammetry and remote sensing also by Lazaridou and Patmios (2012). So,
geoinformation can be explain like information which have some spatial context and can be used for
information systems work with the spatial data.
Geoinformation at regional level
The area of interest is the Moravskolezský region in the Czech Republic. So, the regional providers of
the data are contained first. Sometimes it means that the providers are capital cities of the regions,
sometimes are providers whole regions. Prague, the capital city of the Czech Republic, provides
geoinformation via different webpages. The first is geoportal. The geoportal has a concept of most using
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
33
services on the first page, e.g. download the documents or maps, import geodetic documentation,
provide geodata, e-import, e-export and open data. The maps on the first page are available only in Java.
Then, the geoportal is divided to seven parts, but for the geoinformation vie are important only these
parts: maps, data a service. The section maps contain base online map, archive map, traffic map
topographic map and maps focused on important areas. Section data contains data, metadata and
searching in open data. Section service provide browsing, quering and searching services. The quering
and serching services aren´t available now by Prague Institute of Planning and Development (2013).
Map of passports (green vegetation) is another Prague´s web page providing geoinformation. The web
page allows see the passports in the area and users can chose another spatial information visible in map,
e.g. bicycle kickstand, areas where people can buy bioproducts, no-smoking areas or baby-friendly
areas. 1 792 places in Prague are listed now by Automat (2010). Prague has geoportal focused on crisis
management also. The geoinformation are distributed by portals of each city district, e.g. District no. 8
provide the floods map and integrated saving system map by MČ Praha 8 (2012).
Hradec Králové, the capital city of Královehradecký region, provide the passport geoportal also. The
map allows see the base map or orthophoto map through 2011 and 2019. The users can create their own
notes to the map, distance measuring or export the map. The map is very detailed and for example each
tree has identification number and basic notes by T-Mapy (2017). Hradec Králové has portal of crisis
management also by T-Mapy (2017). The map includes information about position of sirens, offices of
integrated saving system and other necessary known object in the case of dangers. Hradec Králové
provide ternary map, urban planning map, public administration forests map, environmental map, social
and business map, barrier free map also. The Hradec Králové city try to use the whole potential of the
geoinformation in public administration. So, this is the reason why the area of interest of this geoportal
is so wide and supply more topics and possibilities than the capital city of the Czech Republic.
Středočeský region, region close to Prague city, has crisis portal which inform about actual situation and
potentional hazards through the map by Středočeský kraj (2015). Středočeský district provide traffic
map, environmental map, distortion of various substances map, urban planning map, sports map, gap
donation map, investment map and library map. Maps inform about the topics only. Create some analysis
or find out some connection isn´t obtained in the geoportal.
Jihomoravský region has concept of brownfield like map of brownfields, where citizens can download
the documentation about each project by RRAJM (2019). The Brno city, capital city of the Jihomoravský
region, the problematic of brownfields presents through portal contains base map. The base map can be
change to the aerial map. The information from cadastre and other information are visible after click on
the brownfield by Statutární město Brno (2019). The Statutární město Brno provide the historical
orthophoto map, barrier free map, map of closures, catchment of nursery schools map, environmental
map, cemetery maps, projects map, companies map and map of ashbin for recycling on the geoportal
also. Some maps allow measuring and creating of analysis. The geoportal is often used for need of public
administration, e.g. urban planning or monitoring of development of industry zones by Statutární mesto
Brno (2019).
Plzeňský region has geoportal which provide orthophoto map, cadastral map, digital technical map,
urban planning map and base information about the maps by Geoportál Plzeňského kraje (2014). Java
is need for displaying each map. Find the address or cadastre information, draw the easy path or
displaying different layers is available. But the possibilities of the geoportal aren´t wide and it´s visible
that the geoportal isn´t often used for needs of the public administration. Because of data from 2014 and
providing only static information.
Karlovarský region provide geoinformation through geoportal also by Geoportal (2014). But in these
day isn´t available. Liberecký region isn´t available from October 2019 available also by Geoportál
Libereckého kraje (2014).
Ústecký region provide geoinformation through digital maps portal by Geoportál Ústeckého kraje
(2014). The concept is created similar like in the Plzeňský region. Both are created by the same
company. While static information are used in Plzeňský region, actual information are used in Ústecký
region, e.g. map of winter maintenance plan for season 2019-2020 is available. The classic topics: urban
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
34
planning map, environmental map, traffic map and so on are contained also. Watchtowers map, culture
map, family silver map, beer and wine map and shipment map are important for the visitor.
Pardubický region provides geoinformation through public administration webpage of the region by
Pardubický kraj – Mapy (2018). Java is need for the displaying the maps. The webpage provides online
maps (available on their servers) and Web Map Services (WMS). WMS are available datasets which
can be used in GIS for base data or analysis through connection with the provider´s server. Connection
with the Internet through whole data processing is necessary. Without it, data can´t be visible. Base map,
administrative map, urban planning map, environmental map, traffic map, culture map and public
administration objects map is available via Pardubický region WMS. Online maps on the server should
displaying base geographic information in the region.
Jihočeský region provide geoinformation through digital maps portal by Geoportál Jihočeského kraje
(2014). The concept is created similar lik Plzeňský and Ústecký region by T-Mapy company. Cadastral
map, urban planning map, education and social map, traffic map, reference map, environmental map
and other maps are contained. The portal is used for need public administration and the contained
information are actual (last actualization was provided 2 January 2020). Immediately shared data
through maps (e.g. places of available shared bicycles or free parking) aren´t used.
Vysočina region provide geoinformation through geoportal by Geoportal (2002). The geoportal contains
maps, application, map services, data supply, metadata and datasets. Most important are map services
and applications. Map services contains basic maps e.g. cadastral. Applications are maps contained
immediate data. Some maps are created by Vysočina region, some are connected with specialized
servers (e.g. traffic map regarding actual traffic situation in the region is connected with
dopravniinfo.cz). Application contains traffic map, culture map, floods map, development plan map and
tourism map. The tourism map can be important for visitors because of the supply of potential sight-
seeing and informing about actual actions in the region.
Zlínský region provides map services portal by DMVS-ZK (2014). The portal is divided to next sections:
maps, documents and background information and metadata. Most important are maps. The section
maps are divided to: brownfields, floods, anti-floods restrictions, water piping development plan,
canalization development plan, orthophoto, manufacture areas, land-use by INSPIRE and administrative
agency. All maps are available in special webpages or can be distributed by WMS. All the data are
actual, but immediate data aren´t provided.
Olomoucký region provides geoinformation through maps in the public administration webpages by
GIS mapy (2011). The maps are visible like PDF documents. The focus of the maps is next: education,
ethnic groups, dependency on various things and substances and social excluded areas.
Geoinformation in Moravskoslezský region
Public administration of Moravskoslezký region provides geoinformation through various webpages. In
these days, one of the important topics is using of brownfields. Moravskoslezký region has geoportal
where citizens can see the map of brownfields. The regenerated objects are visible also. The citizens can
join to effort reconstruct the brownfields or make a proposal how the objects can be reconstructed and
which function the objects should have after Moravskoslezký kraj (2019).
The main of geoinformation is distributed via Moravskoslezký region webpages: section Maps. The
section Maps is divided to six parts: Actual Information, Basic Maps, Urban Planning, Investment and
Property, Environment and Tourism by Mapy (2019). The section Actual information contains new map
application regarding urban planning, data sources (WMS, datasets available for downloading and so
on), links for other public administration map servers and GIS (there are only base information about
the GIS and which product use the regional public administration). The section Basic Maps contains
zonation dividing of area, aerial photos, historical maps and war tombs. The section Urban Planning is
dividing to territory municipalities plan, urban development policy, analytics background, analytics
background of municipalities, territory administration of construction administration, development of
settled areas, cemetery and counting of traffic intensity. The section Environment contains case study
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
35
of ecological stability, small water management actions, development plan of water pipes and
canalization, floods areas, prevention of breakdown and geology. The section Investment and Property
is divided to real estate and areas of building. The section Tourism contains open churches, skiing areas,
areas suitable for swimming and other water bodies, watchtowers, auto camp, culture and brewery.
The Ostrava city, the capital city of the Moravskoslezský region, provide their own geoportal by Mapový
portál města Ostravy (1999). The geoportal is divided to three sections Most used maps, Urban Planning
and Notices and Ordinance. The section Most used maps contain cadastral map, open data, election
places, cycling traces, historical maps, education map, free barrier map, WMS and environmental map.
The section Urban Planning contains urban plan, utility report, analytics background and building of the
year. The section Notices and Ordinance contains price map, nonsustaining communications, parking
zones, isolated trash, punishment of alcoholic drinks, concours using fee, areas available for free running
dogs and punishment of advertisement sharing. Immediate information like available park places in
chosen location or available closest share bicycle aren´t contained in the geoportal. But companies
provided these services have this information on their webpages.
Comparison of using geoinformation on regional level
The thesis is focused on using geoinformation in public administration in Moravskoslezský region. So
commonly, this information should be compared with other regions in the same state. This is the reason
why in previous chapter are listed all regions of the Czech Republic.
The using geoinformation can be divided to two parts: Static Information and Actual Information. The
section Static Information means the static information are distributed via webpages to citizens and
visitors. Like static information are explain information about features which are constant. Typical
representant of this group can be administrative dividing of municipalities in the region. Meanwhile, the
second section, section Actual Information, contains actual information. The actual information doesn’t
mean immediate information in this case. It means that the people are inform about actual topics and
have actual data. In some case, the public administration only creates the geoportal, because it must had
been created and now it´s not used, the data aren´t actual and so on. The typical representative of this
group can be canalization development plan or development of industry zones.
The Moravskoslezký region provide both of the listed parts. When compare the Moravskoslezský region
with other regions it´s clear the region is in the better half. Because of some regions don´t provide any
data (or provide the data only when the portal or webpage was established) in these days. The
Moravskoslezský region provide basic and actual data what many regions don´t provide. But immediate
data don´t use. So, it´s clear, that the region is in the better half, but because of different focus of maps
and topics it can´t be comparable concrete.
Discussion
Increasing interest of spatial information needs the public administration provide the data also.
Meanwhile some regions and municipalities behave to the new trend when use the geoinformation not
only for distributing the static data but for planning and immediate sharing the data, some regions totally
ignore the trend. The only provide the base data because the must in the time when the European Union
(EU) project INSPIRE was running. And then just ignore it. It´s questing if it´s because they don´t have
the specialist on GIS, haven´t enough money for sustainable working system and innovation or they just
don´t want it. In case of small municipalities, it´s clear mainly the financial problem because of money
budget. But what is the problem for the big municipalities? It´s find out that using geoinformation save
time and money. And the geoinformation can be used for decision making e.g. urban planning. It´s sure
that the trend of geoinformation sharing will be continued and main aim will be sharing immediate
information.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
36
Conclusion
The geoinformation and their using is increasing in these days. First time the private sector uses it, but
the public administration uses it in these days also. Because the citizens want the spatial information
and public administration can use this information not only for sharing information about some topic.
But they can use it for needs of public administration which save time and money. By geoinformation
can be find out the information which will other undetectable. Typical representative of this is where to
build a watchtower if we want to have the best view.
The thesis is focused on regional level of using geoinformation in public administration. The case study
is focused on Moravskoslezký region in the Czech Republic. All regions are listed for next comparing.
In the thesis the focus is on which geoinformation are distributed to the citizens and visitors of the region
and how. The two big groups of distributed geoinformation are find out: Static Information and Actual
Information. The group of Static Information represents portal or webpages where only static
information are represented, e.g. administrative dividing of the municipalities in the regions. It´s
information which aren´t often changed. The group of Actual Information can´t be compared with the
term immediate information. In this case it means that the citizens and visitors have information about
actual topics and the statement of the topics. The typical representative can be canalization development
plan. It was found out that the Moravskoslezký region provide both of the types of the geoinformation.
Moravskoslezský region is listed in the better half of the regions because not many regions have both of
geoinformation. Moravskoslezký region distributed the geoinformation through three webpages. The
first is webpage regarding brownfields. The second, the most important for citizens, is Moravskoslezký
region webpage in the section Maps. The section contains next areas of interest: Actual Information,
Basic Maps, Urban Planning, Investment and Property, Environment and Tourism. The last webpage is
geoportal of Ostrava city, the capital city of Moravskoslezský region. There is the important actual
information for citizens of the city. The immediate geoinformation are only one which aren´t provided
by public administration in the Moravskoslezký region. But the geoinformation are there, e.g. shared
bicycles, but the private companies provided them in these days. But this information are the most
important which citizens want to know. So, there is the highest potential for development of using
geoinformation in Moravskoslezký region.
Acknowledgement
This research was financially supported by the VSB – Technical University of Ostrava.
References
[1] Automat. (2010). Zelená mapa Prahy. [Online]. Available at: <http://zelenamapa.cz/>.
[2] DMVS-ZK. (2014). Portál mapových služeb. [Online]. Available at: <https://gis.kr-zlinsky.cz>.
[3] Geoportal. (2002). Geoportál DMVS Kraje Vysočina. [Online]. Available at: <http://geoportal.kr-
vysocina.cz/web/>.
[4] Geoportal. (2014). Geoportál DMV Karlovarského kraje. [Online]. Available at:
<https://geoportal.kr-karlovarsky.cz/web/>.
[5] Geoportál Jihočeského kraje. (2014). Portál digitální mapy veřejné správy Jihočeského kraje.
[Online]. Available at: <https://geoportal.kraj-jihocesky.gov.cz/>.
[6] Geoportál Libereckého kraje. (2014). Geoportal. [Online]. Available at: <https://geoportal.kraj-
lbc.cz/>.
[7] Geoportál Plzeňského kraje. (2014). Portál digitální mapy veřejné správy Plzeňského kraje.
[Online]. Available at: <https://geoportal.plzensky-kraj.cz/gs/>.
[8] Geoportál Ústeckého kraje. (2014). Portál digitální mapy veřejné správy Ústeckého kraje. [Online].
Available at: <https://geoportal.ustecky-kraj.cz/gs/>.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
37
[9] GIS Mapy. (2011). Olomoucký kraj. [Online]. Available at: <https://www.olkraj.cz/gis-mapy-
aktuality-211.html>.
[10] Lazaridou, M. A. and E. N. Patmios. (2012). Photogrammetry-remote sensing and geoinformation.
22nd Congress of the International Society for Photogrammetry and Remote Sensing, ISPRS 2012,
Australia, pp. 69-71.
[11] Mapový portál města Ostravy. (1999). Mapový portál. [Online]. Available at:
<https://mapy.ostrava.cz/>.
[12] Mapy. (2019). Moravskoslezský kraj. [Online]. Available at: <https://www.msk.cz/mapy/>.
[13] MČ Praha 8. (2012). Městská část Praha 8. [Online]. Available at: <https://m.praha8.cz/Mapy-
krizoveho-rizeni.html>.
[14] Moravskoslezský kraj. (2019).
[15] Pardubický kraj – Mapy (2018). Pardubický kraj. [Online]. Available at:
<https://www.pardubickykraj.cz/gis/>.
[16] Prague Institute of Planning and Development. (2013). Geoportal Praha. [Online]. Available at:
<http://www.geoportalpraha.cz/>.
[17] RRAJM. (2019). Brownfieldy. [Online]. Available at: <https://www.brownfieldy-jmk.cz/>.
[18] Shaytura, S. V. et al. (2018). Geoinformation services in spatial economy. International Journal
of Civil Engineering and Technology. 9(2), pp. 829-841.
[19] Statutární město Brno. (2019). Mapový portál Brno. [Online]. Available at:
<http://gis.brno.cz/portal/>.
[20] Středočeský kraj. (2015). Středočeský kraj: Mapové aplikace. [Online]. Available at:
<https://gis.kr-stredocesky.cz/JS/MAPY/>.
[21] T-Mapy. (2017). Mapové aplikace. [Online]. Available at: <http://geoportal.mmhk.cz/portal/>.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
38
INTER-ORGANIZATIONAL KNOWLEDGE SHARING
AND GAME THEORY
Alina Czapla1
1Department of Organizational Relationship Management, University of Economics in Katowice,
1 Maja 50, 40-287 Katowice, Poland ,
e-mail: [email protected]
Abstract
Knowledge management is in the center of attention of many researches and knowledge sharing (KS)
is currently the subject of many scientific studies. However, the vast majority of them refer to sharing
knowledge within the organization. This study focuses on inter-organizational knowledge sharing.
There are many similarities between inter-organizational knowledge sharing (IKS) and the strategic
game, so exchange of knowledge between organizations has been analysed in the framework of game
theory (GT). The analysis showed, that game theory is a useful tool for describing KS and can help
managers in making decisions. However, treating IKS as a strategic game is not always beneficial,
sometimes it is better to establish mutual cooperation.
Keywords
Knowledge sharing, Inter-organizational knowledge sharing, Game theory.
JEL Classification
D80, C570.
Introduction
The importance of knowledge and knowledge management has been constantly emphasized in recent
decades. The benefits of sharing knowledge have been highlighted. Increased competitiveness and
innovativeness have been considered to be particularly important. Although KS was mainly analysed as
an activity within the organization, inter-organizational knowledge sharing is currently gaining
importance. The possibility of obtaining knowledge from external sources can bring many benefits to
an organization. KS also carries risks, so the decision “share knowledge” or “not share knowledge” with
another organization should be made carefully.
The decision support tool was recently created by mathematicians. Game theory not only helps to model
real situations, but also is useful to find the right strategy. This theory has already found applications in
many areas, but its usefulness in economics and management seems to be particularly important. That
is why IKS is modelled in this study using game theory.
Methodology of using non zero-sum games to improve decision making in choosing courses of action
was proposed to make strategic decisions. Decision to "share knowledge" or "not share knowledge" with
another organization was treated as a solution of strategic game. Game theory rules and the payoff matrix
were used to solve this game. The solutions were examined and discussed.
From the considerations it follows that the possibility of cooperating in the field of inter-organizational
knowledge sharing should be considered before applying the game theory approach. In some situations
cooperation instead of competition will allow organizations to achieve optimal benefits. If it is not
possible, then treating IKS as a strategy game can be helpful for managers in making decision. By
applying mathematical rules we can find the game solution - the right strategic decision. However, the
concept of using game theory to make strategic decisions regarding IKS has also some disadvantages.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
39
Framework
Organizational Knowledge
Literature reveals many different definitions and perspectives on knowledge (Small and Sage,
2005/2006, p. 153). What is knowledge? There were many answers and many arguments used in
supporting them, but none of those theories has been accepted so far as being fully satisfactory (Bolisani
and Bratianu, 2018, p. 2). Researchers and practitioners have failed to agree on a definition of what
constitutes knowledge (Biggam, 2001, p. 6). Although scientists have not managed to develop a clear
definition of knowledge to this day, they agree that this is more than just data or information.
Also organizational knowledge is much talked about but little understood (Tsoukas and Vladimirou,
2001, p. 973). It is defined in various ways, for example: organizational knowledge is the collection of
knowledge, which exists in the organization that has been derived from current and past employees
(Jones and Leonard, 2009, p. 29); organizational knowledge is the set of collective understandings
embedded in a firm, which enable it to put its resources to particular uses (Tsoukas and Vladimirou,
2001, p. 981; Penrose, 1959); organizational knowledge is a dynamic process, of an essentially and
inherently social and interactive nature, which demands active and committed participation and
involvement by people (Cardoso et al., 2012, p. 267).
Organizational knowledge is much more than the sum of knowledge of its individual members.
Knowledge is organizational simply by its being generated, developed and transmitted by individuals
within organizations. Furthermore, knowledge becomes organizational when individuals operate
according to the general rules developed by the organization (Tsoukas and Vladimirou, 2001, p. 979).
Organizations are the sites of cultural knowledge, that provide an organized system with a distinct
identity and enable its members to act in coordinated ways (Tsoukas, 2011, p. 13).
Both scientists and practitioners agree that organizational knowledge is a valuable resource. Knowledge
represents the most important resource in creating the competitive advantage (Bratianu, 2015, p. 131).
Organizational knowledge is identified as one of the contributing factors to organizational
competitiveness (Pangil and Nasurddin, 2013, p. 349). It is perceived as the primary source of the
creation of value (Shaheen, 2017, p. 24). If the know-that or know-what knowledge is visible and can be
easily imitated by the other competitors, the know-how knowledge is invisible and can be considered
the backbone of the organizational knowledge (Bratianu, 2015, p. 131).
Knowledge Sharing
Knowledge management and especially knowledge sharing are currently the subject of many scientific
studies. The vast majority of them focus on sharing knowledge within the organization. Knowledge
sharing is basically the act of making knowledge available to others (Ipe, 2003, p. 341). If we consider
the internal exchange of knowledge, it concerns the exchange of information and know-how between
employees, teams or departments. Such exchange is considered to be very beneficial for an organization,
although individuals may sometimes suffer losses. Nevertheless, the quality of knowledge sharing is the
major factor that facilitated individual creativity (Lee, 2018, p. 10).
Knowledge sharing is affected by multi-level factors: organizational level, team level and individual
level factors; some will promote knowledge sharing, and some will have a negative impact (Zheng,
2017, p. 51). There are four enablers for KS: technology that supports KS, culture that influences the
attitude towards KS, organizational structure that affects KS style and motivation that determines KS
strategy (Xu et al., 2014, p. 14). KS depends on the nature of knowledge, the motivation to share, the
opportunities to share, and the culture of the work environment (Ipe, 2003, p. 351). The significant
drivers of KS are: enjoy helping others, monetary rewards, management support, change of knowledge
sharing behavior and recognition. The significant identified barriers to knowledge sharing are: change
of behavior, lack of trust and lack of time (Razmerita et al., 2016, p. 1).
KS does not only mean reorganization and effective transfer of knowledge, skills, and information, but
it also indicates the creation of new knowledge and innovative ideas (Lee, 2018, p. 3). Knowledge
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
40
sharing has various positive effects on organizations. In a knowledge economy, effective sharing of
knowledge makes businesses function more effectively (Safari and Soufi, 2014, p. 13). KS increases the
effectiveness and quality of work to improve performance for the benefit of the organizations (Mohajan,
2019, p. 57). It influences the creativity of the team (Men et al., 2017, p. 1). Knowledge sharing
orientation significantly and positively impacts the business performance (Vij and Farooq, 2014, p. 17).
Inter-organizational Knowledge Sharing
Inter-organizational knowledge sharing confronts firms with a paradox of dealing with contradictory
requirements. On the one hand, KS can give firms new business opportunities, on the other, partners can
lose the uniqueness of their companies' knowledge. We can observe the competitive paradox of inter-
organizational knowledge sharing: how to reap the benefits of cooperating without losing one’s own
advantage ( Loebbecke et al., 2016, p. 5).
IKS and achieving innovation through IKS are critical for organizational survival (Tsai, 2016, p. 1402).
IKS can bring many benefits to the organization. Enhancement of effectiveness and efficiency by
spreading good ideas and practices are main advantages of knowledge sharing between
companies (Safari and Soufi, 2014, p. 21). Other benefits of inter-organizational knowledge
sharing are access to competitive knowledge, increased company’s competitive advantage,
synergy effects in the creation of know-how. Joint knowledge resources foster innovation, learning,
and knowledge creation (Ilvonen and Vuori, 2013).
IKS can prove to be worthwhile only when it is a joint activity between partners in which every party
attempts to create more value together than what they would be able to create individually. However,
inter-organizational knowledge sharing can be not only fruitful but also threatening (Safari and Soufi,
2014, pp. 13-14). IKS carries risks for example knowledge spill-over, opportunistic behavior, conflicts
with partners, risk of lack of balance between competition and co-operation (Ilvonen and Vuori, 2013).
Game Theory
Game theory is a young branch of mathematics that often supports decision making. It is often used in
optimization techniques. Games that are the focus of this theory illustrate various real decision-making
situations. In a simplified mathematical model, with the help of logical reasoning and mathematical
rules, GT allows to find a game solution. That is why game theory is helping to solve strategic problems.
Game theory is used by practitioners from various fields, including biology, psychology, international
relations and philosophy (Watson, 2002, p. 2). It is helpful in the development of computer science
(Halpern, 2007), cybernetics (Kazimierczak, 1973) or artificial intelligence (Tennenholz, 2002). It is
used in geology (Krzak, 2013), politics (McCarty and Meirowitz, 2007), jurisprudence (Załuski, 2013),
sociology (Burns et al., 2017) and military sciences (Fox, 2016). These are just examples of areas where
GT supports practitioners or forms the basis of some research.
However, the use of game theory in economics and management deserves special attention. Game theory
models are used in finance (Allen and Morris, 2014), accounting (Kanodia, 2014), marketing (Moorthy,
2014), management (Li and Whang, 2014). Watson (2002) describes the application of game theory in
the organization of markets, trade and negotiations. Drabik (2009, p. 28) underlines that GT models
many economic processes such as production, transport, distribution of goods, economic growth as well
as competition and cooperation.
Methodology
Methodology of using non zero-sum games to improve decision making in choosing courses of action
was proposed to make strategic decisions. Decision to "share knowledge" or "not share knowledge" with
another organization was treated as a solution of strategic game. Game theory rules and the payoff matrix
were used to solve this game. The solutions were examined and discussed.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
41
The concept of using game theory to make strategic decisions regarding IKS has undergone critical
analysis. Weaknesses of this approach have been identified. The advantages of this approach have also
been pointed out.
Results
Inter-Organizational Knowledge Sharing as a Strategic Game
For organization the decision to "share knowledge" or "not share knowledge" with another one is a part
of strategy. Inter-organizational knowledge sharing can be very beneficial, but it's also risky. There are
many similarities between knowledge sharing and strategy games. They involves two or more persons,
organizations or players; the player or organization chooses one of the possible strategies; the strategy
leading to the highest payoff is chosen (Chua, 2003, p. 120). In this paper KS between two organizations
is considered. This economic situation is presented as a a two-player non-zero sum game. Each
organization (player) chooses one of two strategies. The payoff matrix is presented in table 1.
Table 1. Payoff matrix of knowledge sharing between two organizations
Player/organization 2
Share knowledge Not share knowledge
Player/organization 1 Share knowledge (a1,a2) (b1,c2) Not share knowledge (c1,b2) (0,0)
Source: Own elaboration
It was assumed that if both organizations decide to choose the "not share knowledge" (NSK) strategy,
their payoffs will be 0. Due to the fact that knowledge is a valuable resource, it was also assumed that if only one organization decides to “share knowledge” (SK), then the payoff of the second one will be
positive: ci>0, where iє{1,2}. (1)
For the same reason:
ai>bi, where iє{1,2}. (2) It means that the payoff of an organization that shares knowledge is larger when the second player also
shares knowledge. Generally speaking, we assume that acquiring knowledge from outside is always
beneficial for the organization. Importantly, payoffs are not only financial profits but also all other
benefits or losses. They determine the utility of each strategy for a given player.
Game Solution
To find a solution to the game, the concept of dominant strategy is helpful. Strategy is called dominant,
when always produces a higher payoff, regardless of what strategies are used by the other players
(Harrington, 2009, p. 56). Weakly dominant strategy is defined analogically, but only one payoff has to
be higher, others can be greater than or equal. Game theory assumes the rationality of all players, which
in practice means that if a given player has a dominant or weakly dominant strategy, then he will apply
it. Additionally, common knowledge about the rationality of all players is assumed: all the players are
rational; all the players believe that all the players are rational; all the players believe that all the players
believe that all the players are rational; and so on; and so on (Heifetz, 2012, p. 49). If none of the players has a dominant or weakly dominant strategy, then the game's solution is the most
favorable Nash equilibrium (named after John Nash, who first described it). Such equilibrium is a set of
strategies, in which each player's strategy maximizes his payoff, given the strategies used by the other
players (Harrington, 2009, p. 90). In other words, no player has anything to gain by changing only his
own strategy. The rational player chooses the strategy for which his payoff is the largest. The solution of the game
will therefore depend on the relationship between the payoffs of i-th player. To find a solution to the
game, all possible cases should be considered. the solution to the game titled “Inter-organizational
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
42
knowledge sharing”, obtained by searching for dominant strategies and Nash equilibria, is presented in
the table 2.
Table 2. Game solution
Payoff values Game solution
bi≥0, iє{1,2} (SK, SK)
{bi < 0ai > ci
, iє{1,2} (SK, SK)
{bi < 0, 𝑖є{1,2} a1 ≤ c1or a2 ≤ c2
(NSK, NSK)
{b1 ≥ 0 and b2 < 0
a2 > c2
(SK, SK)
{b1 ≥ 0 and b2 < 0
a2 < c2
(SK, NSK)
{b1 ≥ 0 and b2 < 0
a2 = c2
(SK, SK or NSK)
{b2 ≥ 0 and b1 < 0
a1 > c1
(SK, SK)
{b2 ≥ 0 and b1 < 0
a1 < c1
(NSK, SK)
{b2 ≥ 0 and b1 < 0
a1 = c1
(SK or NSK, SK)
Source: Own elaboration
To show an example of used reasoning, let's consider the following situation:
{bi < 0ai ≤ ci
, where iє{1,2}. (3)
Then in the payoff matrixes of both players we can find the dominant strategy or weakly dominant
strategy:
[a1 b1𝐜1 0
] and [a2 𝐜2b2 0 ]. (4)
So the solution of the game is (NSK,NSK). This does not mean, however, that this solution is best for
players. If:
ai>0, where iє{1,2} (5) then strategy (SK,SK) would be better for both organizations. Although the reasoning is correct, the
result is not always optimal. The reason is that there is a game going on between players. We have not
considered in the model that cooperation can be more beneficial for organizations.
Prisoner’s Dilemma
Game theory explains this specific strategic situation through simply story called the prisoner's dilemma,
which is often described in the literature (for instance Heifetz, 2012; Geckil and Anderson, 2010;
Harrington, 2009). Two criminals are suspected of having committed a crime together. The police
caught them, but evidence is lacking. Each suspect is promised that if he confesses he will go free, while
his colleague will receive a five-year prison sentence. If both confess, each will spend three years in
prison. If none confess, they will spend one year in prison. Table 3 shows the payoff matrix.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
43
Table 3. Payoff matrix in Prisoner’s Dilemma
Prisoner/player 2
Not confess Confess
Prisoner/player 1 Not confess (-1,-1) (-5,0)
Confess (0,-5) (-3,-3)
Source: Own elaboration (based on Heifetz, 2012, p. 28).
In the payoff matrixes of the players we can find a dominant strategies
[−1 −50 −3
] and [−1 0−5 −3
]. (6)
Because:
{0 > −1,−3 > −5
(7)
so strategy “confess” dominates strategy “not confess”. If the suspect confesses and his colleague does
not, he will be free (it is better than not to confess and spend a year in prison). If the suspect confesses
and his colleague also, he will spend three years in prison instead of five. So the solution to the game is
therefore the decision "confess" by each prisoner.
Both prisoners are aware that they’d better not say anything, but they cannot communicate and therefore
cannot rely on agreements between them. It is a situation in which individuals receive less than in the
case of cooperation.
Discussion
The above example shows that using game theory to make strategic decisions about inter-organizational
knowledge sharing is not always beneficial for the organizations. As with the prisoner's dilemma, it is
recommended to use game theory to make decisions about IKS when organizations cannot communicate
with each other. Otherwise, it is always best to consider cooperation first. Modified approach to IKS is
shown in Figure 1.
Figure 1. Decision path for IKS
Source: Own elaboration
The concept of using game theory to make strategic decisions regarding IKS has also another
disadvantages. The first problem is the assumption that we know players' payoffs. In reality such data
is not known or is only known approximately. The second assumption - that players are rational also
often is not fulfilled in reality. Managers often do not know the basics of game theory and act according
to their knowledge, but not always rationally. Similarly, in the case of common knowledge about
rationality. From people unfamiliar with the basics of game theory, one cannot expect common
knowledge of rationality. Therefore, IKS modeling through a strategic game should be used with
caution. Additionally, model is always a simplification, it does not fully describe reality.
To find the best strategy for
knowledge sharing
If it is possible to set the terms of
cooperation
Set mutually beneficial terms of
IKS
If cooperation is not possible, use game theory and
Decide to share knowledge, when
it is profitable
Otherwise, decide not share
knowledge
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
44
On the other hand, game theory allows to assess the situation rationally, analyse possible payoffs and
choose the right strategy. It helps to understand decision making mechanisms. Knowledge of the basics
of this theory simplifies management and leads to better results.
The solution of “inter-organizational knowledge sharing” game is clearly indicated in this study. This
can be helpful for practitioners. Knowledge as a valuable resource should be more often obtained from
external sources.
Conclusions
The decision whether to share knowledge with other organizations or not is a typical strategic situation
that can be described using game theory. Game theory can be helpful in analyzing the possible benefits
and losses resulting from knowledge sharing. Nevertheless, one should be aware that game theory
illustrates a simplified model of reality, so it can only help managers make decisions. Nevertheless, the
possibility of cooperating in the field of inter-organizational knowledge sharing should be considered
before applying the game theory approach. In some situations cooperation instead of competition will
allow organizations to achieve optimal benefits.
This study has limitations. Mainly, IKS between two organizations was considered. The analysis of
knowledge sharing by more organizations should be the subject of future research. Additionally, the
theoretical approach was proposed in this paper. Case studies of real situations could verify the
correctness of the conducted considerations.
Despite the fact that the results can’t be uncritically applied in practice, knowledge of the basics of game
theory can help managers to find rational IKS strategy.
References
[1] Allen F. and Morris S. (2014). Game Theory Models in Finance, [in:] Chatterjee K. and Samuelson
W. (Eds.), Game Theory and Business Applications. Springer, pp. 17-42.
[2] Biggam J. (2001). Defining Knowledge: an Epistemological Foundation for Knowledge
Management. Proceedings of the 34th Hawaii International Conference on System Sciences, pp.
1-7.
[3] Bolisani E. and Bratianu C. (2018). The Elusive Definition of Knowledge, [in:] Bolisani E. and
Bratianu C. (Eds.), Emergent knowledge strategies: Strategic thinking in knowledge management.
Cham: Springer International Publishing, pp. 1-22.
[4] Bratianu C. (2015). Organizational Knowledge Dynamics: Managing Knowledge Creation,
Acquisition, Sharing, and Transformation. Hershey: IGI Global.
[5] Burns T. R., Roszkowska E., Corte U. and Machado N. (2017). Sociological Game Theory:
Agency, Social Structures and Interaction Processes. Optimum. Studia Ekonomiczne, 5(89), pp.
187-199.
[6] Cardoso L., Meireles A. and Peralta C. F. (2012). Knowledge management and its critical factors
in social economy organizations. Journal of Knowledge Management, 16(2), pp. 267-284.
[7] Chua A. (2003). Knowledge sharing: A game people play. Aslib Proceedings, 55(3), pp. 117-129.
[8] Drabik E. (2009). Kilka uwag o formalnych zasadach matematycznego modelowania zjawisk
ekonomicznych i interakcji społecznych. Ekonomika i Organizacja Gospodarki Żywnościowej,
79, pp. 23-37.
[9] Fox W. P. (2016). Applied Game Theory to Improve Strategic and Tactical Military Decisions.
Journal of Defense Management, 6(2), pp. 1-7.
[10] Geckil I. K. and Anderson P. L. (2010). Applied Game Theory and Strategic Behavior. USA: CRC
Press.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
45
[11] Halpern J. Y. (2007). Computer Science and Game Theory: A Brief Survey. Available at:
<https://arxiv.org/pdf/cs/0703148.pdf>.
[12] Harrington J. E. (2009). Games, strategies, and decision making. New York: Worth Publishers.
[13] Heifetz A. (2012). Game Theory. Interactive Strategies in Economics and Management. New
York: Cambridge University Press.
[14] Ilvonen I. and Vuori V. (2013). Risks and benefits of knowledge sharing in co-opetitive knowledge
networks. International Journal of Networking and Virtual Organisations, 13(3), pp. 209 – 223.
[15] Ipe M. (2003). Knowledge Sharing in Organizations: a Conceptual Framework. Human Resource
Development Review, 2(4), pp. 337–359.
[16] Jones K. and Leonard L. N. K. (2009). From Tacit Knowledge to Organizational Knowledge for
Successful KM, [in:] King W. R. (Ed.), Knowledge Management and Organizational Learning.
Springer Science+Business Media, LLC, pp. 27-39.
[17] Kanodia C. (2014). Game Theory Models in Accounting, [in:] Chatterjee K. and Samuelson W.
(Eds.), Game Theory and Business Applications. Springer, pp. 43-80.
[18] Kazimierczak J. (1973). Teoria gier w cybernetyce. Warszawa: Wiedza Powszechna.
[19] Krzak M. (2013). Teoria gier w geologii gospodarczej. Kraków: Wydawnictwa AGH.
[20] Lee J. (2018). The Effects of Knowledge Sharing on Individual Creativity in Higher Education
Institutions: Socio-Technical View. Administrative Sciences, 8(21), pp. 1-16.
[21] Li L. and Whang S. (2014). Applications of Game Theory in Operation Management and
Information Systems, [in:] Chatterjee K. and Samuelson W. (Eds.), Game Theory and Business
Applications. Springer, pp. 103-136.
[22] Loebbecke C., van Fenema P. C. and Powell P. (2016), Managing Inter-Organizational Knowledge
Sharing. The Journal of Strategic Information Systems, 25 (1), pp. 4-14.
[23] McCarty N. and Meirowitz A. (2007). Political Game Theory: An Introduction. Cambridge
University Press.
[24] Men C., Fong P. S. W., Luo J., Zhong J. and Huo W. (2017). When and how knowledge sharing
benefits team creativity: The importance of cognitive team diversity. Journal of Management &
Organization, pp. 1-18.
[25] Mohajan H. K. (2019). Knowledge Sharing among Employees in Organizations. Journal of
Economic Development, Environment and People, 8(1), pp. 52-61.
[26] Moorthy S. (2014). Marketing Applications of Game Theory, [in:] Chatterjee K. and Samuelson
W. (Eds.), Game Theory and Business Applications. Springer, pp. 81-102.
[27] Pangil F. and Nasurddin A. M. (2013). Knowledge and the Importance of Knowledge Sharing in
Organizations. Conference on Business Management Research Universiti Utara Malaysia, pp.
349-361.
[28] Penrose E. (1959). The Theory of the Growth of the Firm. New York: Wiley.
[29] Razmerita L., Kirchner K. and Nielsen P. (2016). What Factors Influence Knowledge Sharing in
Organizations? : A Social Dilemma Perspective of Social Media Communication. Journal of
Knowledge Management, 20(6), pp. 1-31.
[30] Safari H. and Soufi M. (2014). A Game Theory Approach for Solving the Knowledge Sharing
Problem in Supply Chain. International Journal of Applied Operational Research, 4(3), pp. 13-
24.
[31] Shaheen O. (2017). Knowledge and knowledge management in organization: identifying the
critical role of IT in the knowledge management process. World Scientific News, 87, pp. 24-48.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
46
[32] Small C. T. and Sage A. P. (2005/2006). Knowledge management and knowledge sharing: A
review. Information Knowledge Systems Management, 5, pp. 153-169.
[33] Tennenholz M. (2002). Game theory and artificial intelligence, [in:] d’Inverno M., Luck M., Fisher
M. and Preist C. (Eds.), Foundations and applications of multi-agent systems. Springer, pp. 49-
58.
[34] Tsai A. (2016). The effects of innovation by inter-organizational knowledge management.
Information Development, 32(5), pp. 1402-1416.
[35] Tsoukas H. and Vladimirou E. (2001). What is organizational Knowledge. Journal of Management
Studies, 38(7), pp. 973-993.
[36] Tsoukas H. (2011). Representation, Signification, Improvisation – A Three-Dimensional View of
Organizational Knowledge, [in:] Canary H. E. and McPhee R. D. (Eds.), Communication and
Organizational Knowledge. New York and London: Routledge, pp. 10-19.
[37] Vij S. and Farooq R. (2014). Knowledge Sharing Orientation and Its Relationship with Business
Performance: A Structural Equation Modelling Approach, The IUP Journal of Knowledge
Management, 12(3), pp.17-41.
[38] Watson J. (2002). Strategia. Wprowadzenie do teorii gier. Warszawa: Wydawnictwo Naukowo-
Techniczne.
[39] Xu J., Quaddus M. and Gao X. (2014). Towards a Knowledge Sharing Model for Small
Businesses. The International Technology Management Review, 4(1), pp. 12-26.
[40] Załuski W. (2013). Game theory in jurisprudence. Kraków: Copernicus Center Press.
[41] Zheng T. (2017). A Literature Review on Knowledge Sharing. Scientific Research Publishing, 5,
pp. 51-58.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
47
COMPARISON OF EVALUATION OF INNOVATIVE ACTIVITIES IN INNOVATIVE
COMPANIES WITHIN THE V4 COUNTRIES
Katarzyna Czerná1
1Department of Business Administration, VŠB – Technical University of Ostrava,
Sokolská tř. 33, Ostrava 70200, Czech Republic
e-mail: [email protected]
Abstract
Conditions of a current competitive environment urge management to innovation strategy, a key for
obtaining a significant market position. For efficient implementation of innovations, evaluation of the
best practice and its inclusion in entrepreneurship are crucial. The selection of countries for innovation
activities evaluation is based on the assumption of similar economic, cultural developments and mutual
connection based on international cooperation within V4. Even the innovation performance of the
selected countries is evaluated within EU28 countries according to Eurostat data and the standardized
CIS 2016 Innovation Questionnaire are used for collecting data on innovation activities of enterprises,
there are quite significant differences caused primarily by mandatory and optional questions. The paper
aims to compare the way of evaluation of innovative activities in innovative companies within the V4
countries. At the same time, the evaluation of innovative activities through data of innovative enterprises
available from statistical offices will be carried out. Evaluation of innovation activities will primarily
focus on their subject, the costs of the implementation and the obstacles to the implementation of
innovation activities. A hierarchical clustering analysis, Ward´s approach with Euclidean distance, and
regression analysis will be used to evaluate and compare the results for innovative enterprises in the
selected countries.
Keywords
innovation activities, innovative companies, manufacturing industry, regression analysis, V4.
JEL Classification
M10, O31, O32
Introduction
In today's highly competitive business environment, innovation activities are essential for the extension
and subsequent business development. Their correct implementation can secure the market position,
economic growth and long-term competitive advantage, which many managers are already trying to
achieve as an innovator on the market (Vance, 2015).
The key question is how can it be done? One way is to follow leaders in the industry, analyse their
working practices, and then apply best practice examples in their own environment with necessity in
finding their way of transferring the innovative idea into practice, consequently avoiding the copying of
others (Peterková, Ludvík, 2015). Repeating for Peterková and Wozniaková (2015), innovation should
not terminate in itself but should lead to more high-level performance, better makings, and greater
capability. Tidd and Bessant (2016) add the wellspring for national industrial growth.
At the same time, however, businesses have to face the progress of globalization, the increasing
frequency of change, and its distracting nature. These elements also bring advantages, which include, in
particular, the possibility of comparing their own innovative activities with foreign principles of
operation, of course, in the case of countries with similar economic and social development. The
Community Innovation Survey (CIS, 2016) is used within the EU to facilitate comparison, carried out
with two years' frequency by EU member states and a number of ESS member countries. The CIS stands
for a harmonized survey of innovation activity in enterprises, created to provide information on the
innovativeness of sectors by type of enterprises, on different types of innovation and on various aspects
of the development of innovation, such as objectives, the sources of information, public funding, the
innovation expenditures, etc. The CIS provides statistics broken down by countries, types of innovators,
economic activities and size classes.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
48
Even the innovation performance is evaluated according to standardized CIS, there are quite significant
differences caused primarily by mandatory and optional questions. The paper aims to compare the way
of evaluation of innovative activities in innovative companies within the V4 countries. These countries
were chosen not only for similar economic and social developments but also for their strong
interconnection and possible cooperation.
The Visegrád Group, Visegrád Four, or V4, means a cultural and political alliance of four states from
Central Europe, naimly the Czech Republic, Hungary, Poland, and Slovakia, that are members of the
European Union (EU) and NATO . Its purpose is the advancing army, cultural, economic and energy
cooperation with one another simultaneously with furthering their integration in the EU (Visegrad
Group, 2019).
At the same time, the evaluation of innovative activities through innovative enterprises available from
statistical offices will be carried out. Evaluation of innovation activities will primarily focus on their
subject, the costs of implementation and obstacles to the implementation of innovation activities. A
hierarchical clustering analysis, Ward's approach with Euclidean distance, and regression analysis will
be used to evaluate and compare results for innovative enterprises in selected countries. The research
results are analysed using the IBM SPSS Statistics 21 statistical program.
Literature Review
In internal and foreign literature exist distinct ways of defining the key concept of innovation. Although,
authors usually repeat the changes towards something new. Firstly, Schumpeter (1987) at his time
recognized only absolute innovation in which he meant the launching of a new product or existing
product with new features, the introduction of a new production process, the opening of a new market,
the use of new raw materials or the creation of a new production organization. For functional reasons,
however, the understanding of innovation has to be increased to changes of all sorts, which are new
internally, for a singular business. This concept is visualized in relative innovations and the distance of
a new product or other new factors from the original position before innovation. According to Valenta
(2001), there are nine innovation codes that are divided into three groups: rationalization, qualitative
innovation, and a technological breakthrough.
The OECD definition (2018) declaims the introduction of a new or significantly improved product and
the use of a new or significantly improved process in the inner environment of the production company
which is completed by Tidd and Bessant (2016) with need develop them into practical use. Concerning
the authenticity of the implemented changes, Kuratko (2009) characterizes four types of innovation:
invention, extension, duplication, and synthesis. Innovations are varying from brand new products,
services or processes to the sequence of existing concepts into new forms of use.
Bessant and Tidd (2014); Green (2005); Ireland et al. (2011) see the innovation spectrum from minor
incremental improvements (incremental innovations) to radical changes (radical innovations) that
change the way we think and use.
For benchmarking within EU countries, innovation activities are monitored through a statistical
sampling survey, which is governed by the international Oslo Manual (OECD, 2018). Here can be found
a basic categorization to technical and non-technical innovations. The first group includes the product
(the introduction of new or significantly improved products or services) and process innovation (the
introduction of new or significantly improved production or even delivery methods). Non-technical
innovations include marketing (introducing a new marketing method that includes significant changes
in product design or packaging, product placement, product support, or valuation) and organizational
innovations (introducing a new organizational method incorporate business practices, job organization,
or on external relationships). In the Eurostat conception (2014), an enterprise is estimated to be an
innovative company when it introduced one of the foregoing innovations, i.e., product, process,
marketing and organizational, during the period. The evaluation of innovation activities of business units
in EU countries was carried out through a statistical sample survey based on the International Oslo
Manual 2005 prepared on the initiative of the OECD and in line with Commission Implementing
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
49
Regulation (EU) No 995/2012 of 26 October 2012. The goal of the statistical survey is to present
globally comparable data on innovation environments and innovative business activities. The
harmonized model questionnaire of Eurostat for the EU Innovation Survey CIS 2016 (Community
Innovation Survey 2016) for the 2014-2016 reference period is used to collect innovation data on
enterprises.
The structure of the CIS 2016 questionnaires is consistently built and contains 15 research areas, with
questions that are mandatory or voluntary in each area. These are 1. General information about the
enterprise (4 questions); 2. Product innovation / good or service / (4 questions); 3. Process innovation
(2 questions); 4. Ongoing or abandoned innovation activities for product or process innovations (1
question); 5. Innovation activities and expenditures for product and process innovations (2 questions);
6. Public financial support for product and process innovation activities (1 question); 7. Sources of
information and co-operation for product and process innovations (3 questions); 8. Organizational
innovation (1 question); 9. Marketing innovation (1 question); 10. Factors hampering innovation
activities (1 question); 11. Effect of legislation and regulations on innovation activities (2 issues); 12.
Non-innovators (3 questions); 13. Intellectual property rights (1 question); 14. Innovations in logistics
(5 questions); 15. Basic economic information on business (4 questions).
Based on the results of the questionnaire, first innovation indicators and their components can be
recognized. These are major economic indicators, the position of the enterprise in international markets,
the number of innovating enterprises according to the type of innovation, the costs of technical
innovation activities according to the type of costs, revenues for innovative products and services
according to the degree of innovation, cooperation on innovation activities, technical innovations, results
of innovative activities, obstacles to the implementation of innovation activities. The innovation
indicators are monitored according to the ownership of the company (domestic and foreign control), size
of the enterprise (small, medium, large), prevailing economic activity (CZ-NACE classification) and
regional classification (CZ-NUTS 2 classification).
For measuring innovation performance at the international level are used composite and straightforward
indicators. One of the single indicators is a knowledge intensity indicator that serves to determine
innovation performance. It is calculated as the ratio of total R&D expenditure (GERD) and gross
domestic product (GDP). The composite indicators include the Summary Innovation Index (SII), Global
Innovation Index (GII) and Innovation Output Indicator (IOI). Summary Innovation Index allows a
comparison of EU member states' innovation and selected third countries. It consists of four indicator
areas - Framework Conditions, Investments, Innovation Activities, Impacts. These areas contain ten
sub-innovation groups and consist of 27 different weight indicators. According to the achieved value of
SII, the rated countries are divided into four groups - Innovation Leaders (score more than 20% above
EU average), Strong Innovators (score between 90-120% of EU average), Moderate Innovators (score
between 50-90% of EU average), Modest Innovators (50% of EU average).
In 2003, Chesbrough, supported by Herzog (2011), Pitra (2006), noticed that companies from different
high-tech sectors have shifted how innovation is realized. These companies have moved their efforts
from a closed innovation model based on their corporate research to an open innovation model of using
their own but also external ideas and technologies. Bessant et al (2012) further highlight the value of
building an innovation network for cooperation which can provide a way of getting access to different
resources through a shared exchange process. Also, collective learning offers exchange practices,
challenging models. Partners encourage each other with new perspicacity and ideas that lead to routine
experimentation. Collective risk-taking play also a key role in innovation networking. Collaboration can
work with other businesses within a group, with suppliers, customers, but also with competitors from
the same industry. As Birskinshaw (2007) highlights, as partners can stand consultants in the field of
research and development, universities, public research institutions or private research organizations.
It follows from the above, one and generally valid concept of the innovation can´t be defined. Innovation
is broadly understood as a change, in any area of social life. When it comes to innovation in business
practice, it is advisable to prefer a narrower definition of the concept, with innovation seen not just a
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
50
change in products and services but also a change in the circumstances and ways in which they get and
keep on the market.
Methodology and Data
The necessary data for evaluating the innovation activity of enterprises in the V4 countries were obtained
from the National Statistical Offices of each country, that means Czech Statistical Office, Polish
Statistical Office (GUS), Statistical Office SR and Hungarian Central Statistical Office, where, at two-
year intervals, a statistical survey on innovation activities of enterprises respecting the OECD
methodological principles in the Oslo Manual is conveyed. The evaluation of innovation activities is
based on the differences and similarities in CIS in the area of the enterprise size data.
At the same time, this research aimed to estimate the realization of innovative activities in the
manufacturing industry in small and medium-sized and large enterprises based in the selected countries.
Based on the literature, the aim of the paper and previous author´s research the following hypotheses
were defined:
• Hypothesis 1: There is a dependency between the size of the company and the type of realized
innovation.
• Hypothesis 2: Increasing market share as a result of the introduction of innovation is related to the
relative novelty of the product (new for the enterprise only).
Defined hypotheses were judged in IBM SPSS Statistics Software by using regression analysis. The
results were surveyed for small and medium-sized enterprises with up to 249 employees and large
enterprises with 250 and more employees. The fundamental tool was a questionnaire with the principal
innovative signs: data on the innovations carried out according to innovative types, the novelty of the
product innovations, cooperation on innovation activities.
The hierarchical cluster analysis, Ward technique with Euclidean distance, is used for basic comparison
of the results for innovating enterprises in the selected countries. Cluster analysis commits to a set of
methods intended to analyse multidimensional data to classify a plurality of objects into several
relatively homogeneous subsets, known as clusters. Objects within clusters are as close as possible, and
objects belonging to different clusters are as different as possible. Conventional clustering methods
include the Ward method - at each step for each pair of deviations, the increment of the sum of the
squares of the deviations resulting from their merger is computed, and then the clusters corresponding
to the minimum value of this increment are combined. A binary tree, a dendrogram can represent
clustering with this method. Ward's method is suitable for working with objects that have the same
dimension of variables. Ward's minimum variance method is the most commonly used in management
(Charry, Coussement, Demoulin, Heuvinck, 2016).
Empirical Results
Based on the European Innovation Scoreboard 2017, the achieved values of SII in the selected countries
are visible in Figure 1.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
51
Figure 2. Summary Innovation Index and components in V4 countries
Source: own research based on European Commission (2018)
Slovakia is perceived as a moderate innovator with SII 69,1. Sales impacts and Employment impacts
are the strongest innovation dimensions and Slovakia scores particularly well on Sales of new-to-market
and new-to-firm product innovations, Employment fast-growing enterprises of innovative sectors, and
Medium and high-tech product exports. Finance and support, Intellectual assets and Attractive research
systems are the weakest innovation dimensions. Overall, Slovakia’s lowest indicator scores comprise
Venture capital expenditures, PCT patent applications, and Lifelong learning.
Also, Hungary belongs to the group of moderate innovators (SII 69) with employment impacts and the
innovation-friendly environment as the strongest dimensions. The highest performance is visible in
section employment fast-growing enterprises of innovative sectors, medium and high-tech product
exports, and broadband penetration. On the other hand, innovators, intellectual assets and finance and
support are the weakest innovation dimensions. Hungary’s lowest indicator scores remain in design
applications, SMEs innovating in-house, and SMEs with marketing or organizational innovations.
The Czech Republic (SII 89,4) and Poland (61,1) are ranked also among the most numerous groups of
countries. The strength of the innovation system in Poland is based on the sales impact, employment
impacts, and an innovation-friendly environment. Shortcomings are perceived in finance and support,
innovators, intellectual assets and linkages. The strengths of the innovation system in the Czech
Republic endure in corporate investment, employment impacts and sales. Weaknesses are in intellectual
property, ties, and innovators. It is clear that the Czech Republic has the highest SII of the monitored
countries, although it lags behind the EU average by 19.4 points. At the same time, all four countries
were found to be lagging behind innovation leaders, including Sweden (SII 148), Finland (SII 146),
Denmark (SII 141) or the Netherlands (SII 135).
In the research, the author of the paper further focused on comparing data from 2016, as it is the latest
available data in all countries for the same period (in Poland since last year they publish data in an annual
period, ie two-year results cannot be obtained, in Hungary the last available data are from 2016). Based
on a comparison of the structure of the CIS questionnaire in individual V4 countries, it was found that
the available information has a more detailed character in the Czech Republic, Poland, and Slovakia.
Within the Hungarian data in the English-language version, tables are available in the Internet interface
regarding the share of innovative enterprises by staff categories and by NACE, enterprises with
technological innovation by developer and type of innovation by staff categories, enterprises by product
0,0
50,0
100,0
150,0
200,0
250,0
EU Czechia Hungary Poland Slovakia
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
52
and / or process innovation by type of innovative organization, enterprises with technological innovation
by developer and type of innovation by NACE, distribution of turnover of product innovator enterprises,
enterprises by type of product and / or process innovation, type of product and / or process innovation
by staff categories, share of innovative enterprises by cooperative partners and staff categories and
NACE. A separate category in Hungarian statistics is data related to the use of patents in innovative
business or financing of innovation by the state and the European Union.
On the other hand, in other V4 countries, it is possible to find information concerning the costs of
technical innovation including their structure and sales of innovative products, at the same time divided
by individual NACE branches but also by individual territorial division. It is also possible to find
detailed data concerning only technical (product and process) or non-technical innovations, again with
respect to individual fields of NACE. Attention is paid to technical innovations, their development or
results. To a certain extent, Poland, where there is no statistical data on how to develop innovative
products placed on the market, is outside the group of these three countries. Also, the Polish Statistical
Office does not disclose how enterprises in the manufacturing industry cooperate with other innovation
entities according to the country of the cooperating partner.
Another difference is the very nature of the data. In the Czech Republic and Slovakia, in addition to the
percentage calculations of the individual metrics, absolute figures, ie the number of enterprises in each
category, are reported. In Poland and Hungary, only the relative figures, ie the percentages of the
individual categories, can be found on the official website of the statistical offices.
Slovak and Hungarian institutions work mainly with dissemination databases and summary tables, also
known as STADat. The data are presented in the form of reports in annual time series in territorial
structures for the Slovak Republic, regions, regions and districts. Report output can be exported to data
formats: PDF, XML, XLSX, XLS, CSV. STATdat. is a public data database based on IBM Cognos BI
technology. In the Czech Republic and Poland, Excel files are published directly on the statistical offices'
websites, which must be downloaded and edited according to their wishes.
Verification of hypothesis 1
During the analysis of relationships in the environment of innovating companies in the V4 countries the
question about the relation between the size of the company and the type of realized innovation arose.
Because of that, hypothesis 1 assumes that this dependence is visible across countries. To determine the
dependence of these variables, statistical testing was performed using hierarchical regression analysis.
As a dependent variable, the type of implemented innovation was chosen, the independent variable is
the size of the company. Multicollinearity in variables was investigated using the Pearson correlation
coefficient, the tolerance values and mean values of VIF. Multicollinearity was not confirmed. Based
on the Durbin-Watson test values (2,974), the assumption of error independence was confirmed.
Dependence was confirmed by β = 1.218; p (0.048) <0.05. Large and medium-sized companies favour
the implementation of technical innovations, while small companies prefer non-technical innovations
due to limited resources.
Verification of hypothesis 2
Moreover, deliberations directed to the question of the absolute novelty of the product, a product that is
new for the whole market, not just for the organization and its relationship with the method in which
product innovation evolves especially self-managed type. Hypothesis 2 implies that this dependency
exists. Repeatedly, hierarchical regression analysis was used to define the dependence of these two
variables when the dependent variable was the absolute novelty of the product. Based on the result, we
can state that multicollinearity through VIF values was not detected. Based on the Durbin-Watson test
values (2,974), the assumption of error independence was confirmed. By performing statistical testing
by regression analysis, it was found that this dependence exists, p (0.044) <0.05.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
53
Cluster analysis
Four variables have been involved in the cluster analysis (the type of innovation, product novelty, type
of collaboration and developer of innovation).
The Euclidean metric calculated the proximity between profiles. The object combination was compared
with the Ward method. The Ward's minimum variance method creates clusters that minimize variance
within each cluster. For each cluster, the mean is calculated. In each cluster, the observations are
compared to the mean for each variable. Observations or/and clusters are combined so that the variance
within the final cluster solution is minimized. Ward's minimum variance method is the most commonly
used in management. Individual business entities are identified by numbers, namely 1 - small Czech
enterprises, 2 - small Polish enterprises, 3 - small Slovak enterprises, 4 - small Hungarian enterprises, 5
- medium Czech enterprises, 6 - medium Polish enterprises, 7 - medium Slovak enterprises, 8 - medium
Hungarian enterprises, 9 - large Czech enterprises, 10 - large Polish enterprises, 11 - large Slovak
enterprises, 12 - large Hungarian enterprise. The entire clutter process displays the dendrogram in Figure
2.
Figure 2. Dendrogram – cluster analysis
Source: own research based on IBM SPSS Software
It can be said that using the Ward method three clusters of enterprises were created. The first cluster
consists only of enterprises based in the Polish territory, and all size classes of enterprises are described.
All these businesses, regardless of their size, share the fact that the most significant issue is the lack of
decision-making about the development of new products, as well as the greatest cooperation with
suppliers, i.e., an external source.
The second cluster consists only of enterprises with headquarters in the Czech Republic and Slovak, and
all size-types of enterprises represent it. All of these businesses, regardless of size, have a common
launch of new products for the enterprise only, and mainly realize product innovations. Also the first
cluster lays within the second cluster, when the similarity of large enterprises implementation the most
innovative products and services both new on the market and new ones only for businesses appeared.
The last cluster is represented only by Hungarian companies based on co-operation arrangements on
innovation activities at product or process innovative enterprises.
Conclusion
The first part of the paper was based on a comparison of individual V4 countries based on the
organization of the European CIS. This is a harmonized questionnaire aimed at collecting and analysing
data within the European Union on innovative business activities. Although this questionnaire is
harmonized, not all questions are exhaustively given, there is some freedom for each country. This step
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
54
aimed to allow each country in the questionnaire to cover the specific features of its economy.
Unfortunately, in this way, a thorough cross-country analysis within the EU is impossible to some extent.
In the Czech Republic and Slovakia, the structure of data is almost identical; Data on obstacles or
funding are available, both in relative and absolute terms. The absolute expression is missing in the case
of Polish and Hungarian statistics. At the same time, it is not possible to trace detailed data in Poland in
terms of how technical innovation or cooperation is developed. While Hungary provides interactive
tables on its official statistical website on the internet interface, their number in English is limited to
basic metrics related to innovative activities.
Another part of the paper focused on finding the truth of the tentative hypotheses, which were defined
not only based on literary research but also based on its previous research as follows: increasing market
share as a result of the introduction of innovation is related to the relative novelty of the product (new
for the enterprise only). Both hypotheses using regression analysis have been confirmed, implying first
of all that are companies favor the implementation of technical innovation, while small companies prefer
non-technical innovation due to limited resources, independently of the analysed state. At the same time,
it has been proven that in all V4 countries increasing market shares is the result of implementing a
relative product change. This solution is safer and, in the case of radical novelty, companies are afraid
of initial distrust of customers and hence a decline in market share. Although this share can be increased
after the first phase has been overcome, companies prefer constant growth over a steep rise in market
shares.
In the last part of the paper, the comparison of all V4 countries was performed by performing a cluster
analysis with four variables, namely the type of innovation, product novelty, type of collaboration and
developer of innovation. These variables were selected because they are available in all countries under
review. Based on the analysis, it was found that 3 clusters had been created with respect to the states:
Poland (1); Slovakia, Czech Republic (2); Hungary (3), it is clear that individual economies retain
certain characteristics, and the connection between Slovakia and the Czech Republic is logical to
determine the long-term interconnection in all respects. solutions according to the size of companies, as
one might expect.
Research has been limited by the nature of available data, which is very sensitive. The result of the paper
also implies the need for greater unification of the CIS structure, the creation of an exhaustive corpus,
which would include both absolute and relative values, which would allow a more in-depth examination
of innovative business activities not only in academia and science but especially in business. In this way,
one could provide some insight into the best practices that work abroad and that would be applicable to
the domestic economy.
References
[1] Bessant, J., A. Alexander, et al. (2012). Developing innovation capability through learning
networks. Journal of Economic Geography 12, pp. 1087-1112.
[2] Birkinshaw, J., J. Bessant, and R. Delbridge. (2007. Finding, forming, and performing: Creating
networks for discountinouous innovation. California Management Review, 49(3), pp. 67-83.
[3] Charry, K., Coussement, K., Demoulin, N., Heuvinck, N. (2016). Marketing Research with IBM
SPSS Statistics. New York: Routledge.
[4] ČSÚ. (2018). Inovační aktivity podniků 2014-2016. [online database]. Praha: Český statistický
úřad, Available at: <https://www.czso.cz/csu/czso/inovacni-aktivity-podniku-2014-2016>.
[5] Drucker, P. (1985). Innovation and Entrepreneurship. New York: Harmer & Row.
[6] European Commission. (2018). European Innovation Scoreboard. [online database]. Available at:
https://interactivetool.eu/EIS/EIS_2.html#a.
[7] Eurostat. (2014). Innovation statistics. [online database]. Luxembourg: European Commision,
Available at:https://ec.europa.eu/eurostat/statistics-
explained/index.php?title=Innovation_statistics#Further_Eurostat_information.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
55
[8] GUS (2018). Działalność innowacyjna przedsiębiorstw 2014-2016. [online database]. Warszawa:
Główny Urząd Statystyzny. Available at: https://stat.gov.pl/obszary-tematyczne/nauka-i-
technika-spoleczenstwo-informacyjne/nauka-i-technika/dzialalnosc-innowacyjna-
przedsiebiorstw-w-polsce-w-latach-2014-2016,14,6.html.
[9] Herzog, P. (2011). Open and Closed Innovation. Wiesbaden: Gabler Verlag.
[10] HUNGARIAN CENTRAL STATISTICAL OFFICE (2019). Science and technology. [online].
Budapest: HCSO. Available at: http://statinfo.ksh.hu/Statinfo/index.jsp.
[11] Kuratko, D. F. (2009). Entrepreneurship: Theory, process, practice. Mason: South-Western
Cengage Learning.
[12] Oslo Manual. OECD, 4th Edition, 2018. Available at: <https://read.oecd-ilibrary.org/science-and-
technology/oslo-manual-2018_9789264304604-en#page1>
[13] Peterková, J. and L. Ludvík. (2015). Řízení inovací v průmyslovém podniku. SAEI, vol. 42.
Ostrava: VŠB-TU Ostrava.
[14] Peterková, J. and Z. Wozniaková. (2015). The Czech innovative enterprise. Journal of Applied
Economic, 10/(2), 243-252.
[15] Pitra, Z. (2006). Management inovačních aktivit. Praha: Professional Publishing.
[16] Schumpeter, J. A. (1987). Theory of Economic Development. The New Palgrave, 1987.
[17] Štatistický úrad Slovenskej republiky (2019). Databázy. [online]. Bratislava: ŠÚS. Available at:
https://slovak.statistics.sk/wps/portal/ext/Databases.
[18] Tidd, J. and J. Bessant (2016). Managing Innovation: Integrating Technological, Market and
Organizational Change. Chichester: Wiley.
[19] Valenta, F. (2001). Inovace v manažerské praxi. Praha: Velryba.
[20] Vance, A. (2015). Elon Musk: How the Billionaire CEO of SpaceX and Tesla is Shaping our
Future. New York: HarperCollins Publishers.
[21] Veber, J. et al. (2016). Management inovací. Managing Innovation. Praha: Management Press.
[22] Visegrad Group (2019). About the Visegrad Group. [online]. Available at:
http://www.visegradgroup.eu/about.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
56
IMPACT OF UNILATERAL PREFERENTIAL MEASURES OF THE EUROPEAN UNION,
THE UNITED STATES AND CHINA ON EXPORTS OF THE LEAST DEVELOPED
COUNTRIES
Petra Doleželová1
1Department of European Integration, VŠB – Technical University of Ostrava,
Sokolská tř. 33, Ostrava 70200, Czech Republic
e-mail: [email protected]
Abstract
The main aim of the paper is to find out how the exports of the least developed countries (LDCs) have
evolved in terms of the commodity and geographical structure since the introduction of the main
preferential schemes for LDCs - Everything but Arms (EBA) of the European Union (EU) and African
Growth and Opportunity Act (AGOA) of the United States (US) and later Chinese duty-free, quota-free
program. To identify the changes, we carried out two different cluster analyses based on the data from
2000 and 2018 in which the LDCs were sorted into groups based on the similarity of their exports. The
changes in the position of individual LDCs within these groups indicate the changes in their export
structure. The results do not suggest that preferential schemes have contributed to a greater
diversification of LDCs exports or an increase in the proportion of processing-intensive products in
them. However, there have been significant changes in the geographical focus of LDCs exports within
the European Union, the United States and China.
Keywords
China, European Union, Least developed countries, Nonreciprocal trade preference, United States.
JEL Classification
F10, F15, F40, O10,
Introduction
Poverty is one of the most complex and widespread global problems humanity has ever faced. During
years the trade has proven to be one of the most effective instruments for eradicating poverty. Therefore,
rich countries help developing countries to market their products on the world market and benefit from
engaging in international trade. For this purpose, developed and some developing countries grant
unilateral trade preferences to the poor countries. Although there are currently around fifty countries
providing nonreciprocal trade preferences, this paper focuses on the preferential schemes of the three
main global actors, i.e. the European Union, the United States and China. The first part of this paper
provides a brief overview of preference schemes for LDCs provided by these economies. Although these
preferences have been granted for a long time, to this point, it is not clear whether they achieve their
purpose of promoting exports of beneficiary countries. Selected studies dealing with the impacts of
preferential schemes are closer discussed in the literature review.
Trade preferences are usually provided with the two main goals: to increase export volumes for
developing countries and thereby boost their export earnings, and to facilitate export diversification
(Persson, 2013). However, the majority of authors, including those mentioned in the literature review,
who sought to measure the impact of these preferences, focused only on the impacts on the trade volume
of the beneficiary countries. Compared to these studies, the main goal of this paper is not to assess the
impact on the volume of exports of the LDCs, but to map the changes in the structure of these exports
that have occurred since the introduction of the preferential schemes.
The main aim of the paper is to find out how the exports of the least developed countries have evolved
in terms of the commodity and geographical structure since the introduction of the main preferential
schemes for LDCs - Everything but Arms of the European Union and African Growth and Opportunity
Act of the United States and later Chinese duty-free, quota-free program. We also try to find out whether
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
57
the geographical focus of exports of the least developed countries within the three selected economies,
i.e. the European Union, the United States, and China, has changed over the years especially after the
introduction of the preferential system of China in 2010.
The paper will, therefore, provide answers to the two research questions. The first research question is
whether the beneficiary LDCs' exports have evolved over the time towards exports less concentrated
and exports of the higher value-added products.
The second research question is whether the introduction of a program for the LDCs by China in 2010
caused some LDCs to shift part of their exports from the European Union and the United States to China.
In order to determine how the structure of the LDCs´ exports have changed following the introduction
of the preferential schemes in question, several groups were created within the LDCs based on the export
data from 2000 and 2018. Groups of the LDCs were created based on the similarity of their exports. In
each group, all the countries included share the same main characteristics and therefore are similar in
terms of the export structure and its geographical focus. Conversely, countries belonging to different
groups are very different in their export structure and geographic focus.
Background
Empirical evidence supports the idea that the expansion of trade is one of the most proven means to
boost the growth and development of developing countries (Grossman, Helpman, 2015). Therefore, all
countries even the least developed ones should have a chance to engage in international trade and benefit
from it. But being successful in competing with others in the world market can be difficult for some
countries, especially the least developed ones. The idea that developing countries should receive “special
and differential treatment” in the trade area originated from the General Agreement on Tariffs and Trade
(GATT) in the early 1970s. This special treatment can take several different forms, although its most
well-known form is the Generalized System of Preferences (GSP). Under this scheme, developed
countries apply concessional measures towards developing countries in the form of unilateral trade
preferences (Pareja et al., 2016). As the word unilateral implies, these preferences are provided by a
preference-granting country to a developing country without any reciprocal preferences for the donor’s
exports. The expected result of these measures is an increase in exports of beneficiary countries towards
the preference-giving country. These preferences may take the form of duty-free access to the donor’s
market or substantially lower than the normal Most-favoured-nation tariffs. The list of affected products
varies from several dozens to thousands of items for different preferential schemes.
Unilateral preferences have been applied since the early 1970s and are currently part of the trade policies
of all developed countries. Most of these countries have also introduced more privileged preference
programs that can be targeted either at developing countries located in a particular region or countries
with a high degree of underdevelopment.
One of the longest applied and most comprehensive preference schemes is the Generalized System of
Preferences of the European Union. The first GSP scheme of the European Community was applied in
an initial phase between 1971–1981 and has been subsequently renewed several times. At each renewal,
the GSP was also revised in terms of the range of products covered, quotas and ceilings as well as the
lists of beneficiaries and conditions for export of agricultural products (Aiello, 2010). The General
System of Preferences of the European Union is one of the most studied preferential schemes, especially
its Everything but Arms initiative. However, as we mention in more detail in the literature review,
prevalent is the empirical literature claiming that the European Union´s GSP fails to achieve its
objectives in terms of enhancing the trade flows of beneficiaries towards EU markets (Cipollina and
Salvatici, 2007).
Everything but Arms initiative became part of the European Union´s preferential scheme on March the
5th, 2001. The Everything but Arms is specifically targeted on the least developed countries and
compared to other preferences under the GSP, the Everything but Arms has an unlimited period of its
implementation. Under the Everything but Arms all products from the least developed countries except
for arms and munitions, have duty-free access and access without any quantitative restrictions to the
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
58
market of the European Union. Expectations from the EBA were from the beginning high even though,
as Brenton (2003) proved, the vast majority of imports from the least developed countries have already
entered the EU without duty and quotas before the EBA´s implementation. Currently is 7,200 products
covering also agricultural products including sensitive ones eligible for the EBA initiative.
The second most frequently studied preferential scheme is the African Growth and Opportunity Act that
came into force in 2001 intending to enforce trade and investment of sub-Saharan African countries in
the United States. This cooperation is supposed to stimulate economic growth and help the countries of
sub-Saharan Africa to integrate into the world economy (AGOA, 2018). However, likewise the EBA,
the African Growth and Opportunity Act is not entirely flawless and is criticized for several reasons.
For example, Fayissa and Tadesse (2008) point to the fact that exports from African countries are mainly
dominated by petroleum products with relatively low value-added.
The developed countries are no longer the only ones that provide unilateral preferences. Recently,
several developing countries have also introduced their preferential schemes. One such a country is
China, that started in 2001 to grant duty-free treatment to developing countries that have good diplomatic
relations with China. Since then, China has been gradually working towards an increase in the product
coverage of its LDC scheme. The Chinese duty-free, quota-free market access program for LDCs
entered into force in 2010 covering 95 percent of China’s total tariff lines.
Literature Review
Although unilateral preferences have been applied by developed countries for a long time, the evidence
of their effectiveness is inconsistent. We can divide studies on the impact of trade preferences into two
main groups: studies confirming the effect of trade preferences and studies denying that these
preferences somehow affect the trade of developing countries.
Aiello (2010) finds positive effects of preferences on LDC´s export to OECD countries on three different
levels of data aggregation: total exports, total agricultural exports, and export flow for ten groups of
agricultural products at 2-digit level. In line with Aiello´s findings, Thelle (2015) finds that GSP
preferences have contributed to an export increase of covered products by up to 5%, compared to the
pre-preference export level. Thelle also points out that preferences under the Everything but Arms
scheme have generated higher export responses than preferences under the GSP General Arrangement
or GSP+ scheme.
Ornelas (2018) acknowledges the positive effect of preferences on trade but with some limitations. He
claims that nonreciprocal preferences boost the exports of the least developed countries, but only if these
countries are members of the World Trade Organization (WTO). However, non-reciprocal preferences
help non-LDCs promote foreign sales only if they are not members of the WTO.
As mentioned above, the EU preferential system is often the subject of studies assessing the
effectiveness of preferences. Cernat (2004) in his study focuses solely on the impacts of the Everything
but Arms on third developing countries and the LDCs. The study shows moderate trade gains from the
EBA initiative with the largest gains being recorded for sub-Saharan Africa. Only minor impact of EU´s
GSP on the trade of beneficiary countries is found also by Cipollina et al. (2013) with preferences having
a significant impact only in some sectors such as ceramics and glassware, textiles and footwear and for
specific exporters. So far, studies showing the significant impact of EU preferences on LDCs exports
are very scarce.
The effectiveness of the Generalized System of Preferences is questioned also by Herz and Wagner
(2011) who in their study draw attention to the short duration of effects. They state that the GSP tends
to foster developing countries' exports in the short-run but hampers them in the long-run. They also point
to the fact that the GSP granting countries are initially able to promote their exports, since the GSP
recipients import inputs mainly from the GSP granting country.
Gradeva and Martínez-Zarzoso (2010) on the example of the African, Caribbean and Pacific LDCs show
that eligibility for the EU´s Everything but Arms scheme alone does not contribute to the increase of the
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
59
exports of these countries because there cannot be found any substantial improvements in their export
performance. They also address the issue of replacing development assistance with non-reciprocal
preferences, which they consider to be highly questionable.
Following the introduction of the AGOA by the United States in 2001, the attention of experts shifted
also in this direction. Unfortunately, in this case, too, there is prevalent empirical evidence suggesting
little or no significant impact. But of course, studies in favour of the AGOA can also be found. The
findings in the study of Kassa (2019) show that most of the eligible countries registered gains in exports
due to the African Growth and Opportunity Act. However, the gains were relatively unevenly
distributed, with exports of oil and other minerals making up the largest part of the growth in exports.
The study by Wamisho (2015) indicates that the AGOA trade preferences do not have a statistically
significant impact on sub-Saharan Africa´s agricultural exports. Fernandes (2018) finds the positive
impact of the AGOA on the export of the least developed countries in Africa. He shows on a sample of
African countries exports to the US at HS 6-digit level in 26 years that the biggest boost from the AGOA
to African countries’ exports was for apparel products.
Since China's preferential system has been provided for the shortest time of the three preferential
systems in question, there are very few studies examining it. Here we can mention the study of Minson
(2007) who examines its potential and weaknesses.
There are studies in which the preferential systems are not only evaluated but also compared to each
other. Coulibaly (2017) examines the impacts of the AGOA and the EBA on the LDCs located in Africa
over the period 2001-2015. Although he finds positive impacts of these preferential schemes, he also
states that not all African countries have benefited from them, such as some West African countries.
Klasen´s (2016) study assesses the impact of specific preference regimes of different economies on the
exports of LDCs. Out of the nine different preferential systems examined, a positive and significant
impact on exports has been proven only in the case of GSP granted by Canada, Australia, and the
European Union.
Methodology and Data
The final groups of the LDCs were formed based on the results of two different cluster analyzes. Cluster
analysis is a multivariate method which purpose, as explained by Bijnen (1973), is „to group and
distinguish comparable units, and separate them from differing units.“ Cluster analysis aims to classify
objects based on given variables into several groups or as Sinharay (2010) put it: to group similar
observations into a number of clusters based on the observed values of several variables for each
individual.
The resulting clusters are defined through an analysis of the given data, where the similarity of the cases
within cluster and dissimilarity between groups is maximized.
The methods of cluster analysis can be divided into two main groups: hierarchical methods and non-
hierarchical methods. The algorithms in the hierarchical clustering are based on gathering the most
similar two objects in a cluster. Such a process is very extensive because all objects must be compared
before every clustering step. In contrast to non-hierarchical methods, the hierarchical clustering creates
a hierarchy of clusters and does not require specifying the number of clusters before carrying out the
analysis. In our case, we had not determined the exact number of clusters we required, therefore we
could work with hierarchical methods. Furthermore, the results of hierarchical clustering can be easily
visualized by a two-dimensional graph called a dendrogram.
Hierarchical methods can be further classified as agglomerative or divisive methods. In this paper, we
use agglomerative hierarchical clustering. Although we can find a number of different agglomerative
hierarchical clustering techniques, they are all based on one single approach. Agglomerative
Hierarchical Clustering is an iterative classification multi-step method. At the beginning of each
agglomeration hierarchical analysis, all objects in the analysis begin as separate clusters. In the first step,
the dissimilarity between the N objects is calculated. Based on the rule of minimization of agglomeration
criterion first two objects are clustered together thus creating a class comprising these two objects. Then
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
60
again using agglomeration criterion the dissimilarity between this cluster and other, now N - 2, objects
are calculated. Two objects or classes of objects for which when clustering the agglomeration criterion
is minimal are then clustered together. This process is then repeated, reducing the number of clusters in
every iteration. At the end of the process, we have only one cluster left in which all objects are included.
As mentioned above, the graphical output of hierarchical clustering is a dendrogram. A dendrogram is
a tree-shaped diagram displaying the clusters formed at each step of the algorithm together with their
similarity levels. With the help of a dendrogram, the optimal number of clusters is selected from all
possible cluster solutions. Dendrogram allows determining the level at which to cut the tree diagram to
generate a suitable number of groups.
To identify the changes in LDCs exports to three preference-granting economies in a given period 2000
- 2018, it was necessary to generate groups of countries in two different years, at the beginning and the
end of the period. The objects researched in our analysis are 41 least developed countries, the rest of the
LDCs were excluded from the analysis due to lack of actual data.
The variables based on which the countries were divided into individual groups is the volume of exports
of LDCs to the EU, the United States, and China within individual product categories. The different
number of groups, different characteristics and mainly the change in the position of individual LDCs
within these groups allowed us to identify how the patterns of LDCs´ exports have changed since 2000.
These product categories are based on the standard international trade classification on a one-digit level.
These categories are:
• food, drinks and tobacco,
• raw materials,
• energy products,
• chemicals,
• manufactured goods classified chiefly by material,
• machinery and transport equipment,
• other manufactured goods,
Category of energy products was from analysis excluded due to the lack of actual data, therefore we
worked with six product categories. Three main export flows of LDCs were used in the analysis: exports
to the European Union, the United States of America and China. Each export flow was divided into six
product categories. This means that we had 18 variables based on which individual clusters of countries
were created. The first six variables are the volume of exports in each product category to the EU, the
next six variables are the volume of exports in each category to the USA and the last six variables are
the volume in each category exported to China. All export-related data were taken from UNCTADstat
– the statistical database of the United Nations Conference on Trade and Development.
As a linkage method for evaluation of similarity between clusters, we used Ward´s method since this
method is most appropriate for quantitative variables. Ward´s method seeks to join the two clusters
whose merger leads to the smallest within-cluster sum of squares (Moral, 1980). Field (2000) describes
Ward´s method as follows: “The difference between each case within a cluster and that average
similarity is calculated and squared. The sum of squared deviations is used as a measure of error within
a cluster. A case is selected to enter the cluster if it is the case whose inclusion in the cluster produces
the least increase in the error.” Ward method is calculated as
∆(𝐴, 𝐵) = ∑ ‖𝑥𝑖⃗⃗ ⃗ − 𝑚𝐴⋃𝐵⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ ⃗⃗ ⃗‖𝑖∊𝐴⋃𝐵2− ∑ ‖𝑥𝑖⃗⃗ ⃗ − 𝑚𝐴⃗⃗ ⃗⃗ ⃗‖𝑖∊𝐴
2− ∑ ‖𝑥𝑖⃗⃗ ⃗ − 𝑚𝐵⃗⃗⃗⃗⃗⃗ ‖𝑖∊𝐵
2 (1)
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
61
= 𝑛𝐴𝑛𝐵
𝑛𝐴+𝑛𝐵‖𝑚𝐴⃗⃗ ⃗⃗ ⃗ − 𝑚𝐵⃗⃗⃗⃗⃗⃗ ‖
2 (2)
where 𝑚𝑗⃗⃗ ⃗⃗ presents the centre of cluster j, and nj is the number of points in it. ∆ is the merging cost of
combining clusters A and B.
As a distance measure we used Square Euclidian Distance, which is measure proposed for the Ward´s
method and also the most common measure used in cluster analysis when working with interval data.
According to Sakthivel (2015) squared Euclidean distance is the sum of the squared differences between
scores for two cases on all variables calculated as
𝑑 (𝑖, 𝑗) = ∑ (𝑋𝑖𝑘 − 𝑋𝑗𝑘)2𝑛
𝑘=1 (3)
where i = Xin and j = Xjn are two n dimensional data objects.
Empirical Results
In the first cluster analysis based on the export data of the least developed countries in 2000, the six
groups of countries were generated. Then according to the results of the second cluster analysis based
on data from 2018, the countries were again divided into groups. This time, however, nine groups were
identified as the optimal number.
As can be seen in the Table 1, the five groups in 2018 had the same characteristics as those of the year
2000. The individual pairs of these groups shared the same product category that contributed most to
their exports and the same destination of these exports.
We can also see that in 2000 the European union was the most important export market for all the LDCs
excluding 6 countries whose exports were mainly focused on other manufactured goods going to USA.
The exports of the LDCs in 1st groups were strongly concentrated in both years. More than 50% of the
total EU-US-China exports of these countries were made of the raw materials exported to the European
Union. Moreover, when taking in account all product categories, almost 90% of the total exports of these
LDCs went to the European Union.
Both 2nd groups in 2000 and 2018 comprise of countries whose EU-US-China exports were dominated
mainly by machinery and transport equipment and miscellaneous manufactured articles directed to the
European Union. Therefore, we can say that these countries concentrated mainly on exports of high
value-added products.
Countries in the 3rd groups in 2000 and 2018 also focused on exports of products with higher added
value. The vast majority of their exports were made up of manufactured goods classified chiefly by
material and were directed to the European Union.
Both 4th groups included countries which exported predominantly food, drinks and tobacco to the
European union.
Countries in the 5th groups were also strongly oriented towards export to the European Union. These
countries exported mainly product from two product categories, i.e. food, drinks and tobacco and raw
materials most of which were exported to the European Union.
As we can see the structure of these groups has changed significantly and only eight countries remained
in the same group in both years, i.e. Burkina Faso, Chad, Botswana, Burundi, Malawi, Senegal, Uganda,
and Yemen. Which means that their export structure in 2000 and 2018 was similar.
The 6th, the last group in 2000 consists of countries whose EU-USA-China exports were more than fifty
percent composed of miscellaneous manufactured articles destined to the USA. Based on the data from
2018 there was not generated any group similar to this one. This means that in 2018 the largest share of
these countries' exports was made up of other products than miscellaneous manufactured articles or the
largest share of exports went to the EU or China instead of the United States. For example, in
Bangladeshi and Cambodia, the largest share of exports in 2018 was again made up of miscellaneous
manufactured articles but exported to the European Union instead of the US. Therefore, we can say that
these two countries have shifted part of their exports to the European Union from the United States.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
62
The other groups created in 2018 have no equivalent among the groups in 2000 and have very different
main features. This means that some countries have changed focus of their exports and thus disconnected
from their original groups from 2000 and created completely new ones in 2018.
These were mainly groups of countries whose largest share of exports went to China in 2018. This was
particularly the case for the eighth group in 2018, that included seven countries for which raw materials
to China accounted for the largest share of exports. The same goes for the sixth group, which included
countries whose EU-US-China export was largely made up of manufactured goods from the sixth
product category heading to China. It can be seen in the table 1 that the 6th group actually originated
from countries that shifted a significant part of their exports of manufactured goods from the European
Union to China. This means that in 2018 these countries concentrated their exports more to China instead
of the European Union.
The 7th group contained countries whose EU-US-China exports in 2018 consisted of more than 60% of
raw materials going to China. This group was of medium size and contained five countries. Although
this group of countries had the same product and geographic focus of the largest share of exports as the
8th group, these two groups are, in fact, different. When considering all products categories, the 8th group
exported in total most to the European Union but the 7th group exported most to China.
The last 9th group created in 2018 included countries that in 2000 originally exported the largest part of
their exports consisting mainly of food, beverages, tobacco and raw materials to the European Union.
In 2018, however, the largest share of these countries' exports was directed to the United States.
Table 1: Groups of LDCs created based on data from 2000 and 2018
Source: own creation
The Table 1 heading shows the numbers indicating the product group and destination of the largest share
of LDCs exports. For example, the EU 2 + 4 column lists LDCs whose largest share of exports were
raw materials destined for the European Union
Conclusion
The paper aimed to identify how the introduction of preferential systems for the least developed
countries by the European Union, the United States and later China influenced the structure and
geographical focus of LDCs exports to these economies.
Afghanistan 1 Bhutan 2 Botswana 3 Burundi 4 Eritrea 5 Bangladesh6
Benin 1 Djibouti 2 Central African Republic3 Ethiopia 4 Kiribati 5 Cambodia 6
Burkina Faso 1 Lao People's Dem. Rep.2 Dem. Rep. of the Congo3 Malawi 4 Samoa 5 Lesotho 6
Chad 1 Madagascar2 Gambia 3 Mozambique4 Solomon Islands5 Maldives 6
Guinea 1 Sierra Leone2 Zambia 3 Rwanda 4 Somalia 5 Myanmar 6
Mali 1 Senegal 4 Togo 5 Nepal 6
Mauritania 1 Timor-Leste 4 Yemen 5
Niger 1 Uganda 4
Vanuatu 1 United Republic of Tanzania4
Burkina Faso 1 Bangladesh 2 Bhutan 3 Burundi 4 Afghanistan 5 Dem. Rep. of the Congo6 Eritrea 7 Benin 8 Kiribati 9
Chad 1 Cambodia 2 Botswana 3 Djibouti 4 Ethiopia 5 Zambia 6 Gambia 7 Central African Republic8 Samoa 9
Somalia 1 Mozambique3 Malawi 4 Lesotho 5 Guinea 7 Mali 8 Timor-Leste 9
Maldives 4 Madagascar 5 Lao People's Dem. Rep.7 Mauritania 8 Vanuatu 9
Senegal 4 Myanmar 5 Solomon Islands7 Niger 8
Uganda 4 Nepal 5 Sierra Leone 8
Rwanda 5 Togo 8
United Republic of Tanzania5
Yemen 5
2018
2000
USA 0+1China 2 + 4China 2+4China 6USA 8EU 0+1, 2+4EU 0+1EU 6EU 8EU 2+4
Total: China Total:EU
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
63
Countries have not seen any shift towards increasing the share of products with higher added value in
their exports. On the contrary, the number of countries whose largest part of exports consisted of these
products decreased compared to 2000.
The preferential schemes should help LDCs to better assert themselves on the world market and
gradually expand their product portfolio so that they are not dependent on exports of primary
commodities. Exporting high value-added products brings much more to LDC economies than primary
commodity exports. Nevertheless, the export of LDCs has not evolved within the product structure, and
still mostly raw materials, agricultural products, food, and beverages are exported to the preference-
granting economies. Therefore, we can say that the results do not suggest that preferential schemes for
LDCs of the European Union, the United States, and China have contributed to a greater diversification
of LDCs exports or an increase in the proportion of processing-intensive products in them.
However, there have been significant changes in the geographical focus of LDCs exports within the
European Union, the United States and China. In 2000, of the three preferential economies, the European
Union was clearly the largest export market for LDCs. More precisely, the European Union took the
largest share in exports of 34 LDCs. In 2018, however, the EU occupied the largest share of exports in
only 24 least developed countries. This finding is consistent with the fact that the European Union's
share in world trade is gradually decreasing. The share of the United States in LDCs exports has also
decreased since 2000, but not as significant as that of the European Union. While in 2000 China was
not the largest export market for any LDC within EU-US-China exports, in 2018 the largest share of
export of twelve countries was exported to China. Therefore, we can say that China's share in LDCs
exports has increased at the expense of the EU and the US.
However, in the light of the results, it is necessary to realize that China was already experiencing a
period of rapid economic growth at the time of the introduction of its preferential plan. It cannot
therefore be ruled out that China, as the main export market for LDCs, has overtaken the EU and the US
partly because of its increasing domestic demand or because it has begun to be seen by LDCs as a more
prospective trading partner for the years to come.
In conclusion, it should be noted that although the effects of the preferential systems of the European
Union, the United States and China failed to meet the expectations of changes in the pattern of exports
of LDCs, it is not excluded that they largely affected the volume of these exports. However, this will be
the subject of further research.
References
[1] AGOA. (2019). About African Growth and Opportunity Act. TRALAC. Retrieved from:
https://agoa.info/about-agoa.html
[2] Aiello, F. (2010). Evaluating the impact of nonreciprocal trade preferences using gravity models.
Applied Economics. Taylor & Francis Journals, 42(29), pp. 3745-3760.
[3] Aiello, F. (2010). Do Trade Preferential Agreements Enhance The Exports Of Developing
Countries? Evidence From The EU GSP, Working Papers 201002, Università della Calabria,
Dipartimento di Economia.
[4] Bijnen, E. J. (1973). Cluster Analysis: Survey and Evaluation of Techniques. ISBN 978-94-011-
6782-6
[5] Brenton, P. (2003). Integrating the Least Developed Countries into the World Trade System: The
Current Impact of EU Preferences Under Everything But Arms. Journal of World Trade, 37(3),
pp. 623-46.
[6] Cernat L. (2004). The EU Everything But Arms Initiative and the LDCs. In: Guha-Khasnobis B.
(eds) The WTO, Developing Countries and the Doha Development Agenda. Studies in
Development Economics and Policy. Palgrave Macmillan, London
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
64
[7] Cipollina, M. R. (2013). Do Preferential Trade Policies (Actually) Increase Exports? An analysis
of EU trade policies. Agricultural and Applied Economics Association.
[8] Coulibaly, S. (2017). Differentiated Impact of AGOA and EBA on Western African Countries.
Africa Chief Economist Office, the World Bank.
[9] Eurostat. (2019). Glossary:Standard international trade classification (SITC). Europa. Retriewed
from:
https://ec.europa.eu/eurostat/statisticsexplained/index.php/Glossary:Standard_international_trade
_classification_(SITC)
[10] Fayissa, B. and Tadesse, B. (2008). The impact of African Growth and Opportunity Act
(AGOA) on U.S. imports from Sub-Saharan Africa. Journal of International Development. 20. pp.
920-941.
[11] Fernandes, A. M. (2019). Are Trade Preferences a Panacea? The African Growth and
Opportunity Act and African Exports. CESifo Working Paper No. 7672. Available at SSRN:
https://ssrn.com/abstract=3422254
[12] Field, A. (2000). Cluster Analysis. Aims and Objectives. Postgraduate Statistics: Cluster
Analysis. Retrieved from: http://www.discoveringstatistics.com/docs/cluster.pdf
[13] Gil-Pareja S., Llorca-Vivero R., Martínez-Serrano JA. (2019). Reciprocal vs nonreciprocal trade
agreements: Which have been best to promote exports? PLoS ONE 14(2): e0210446.
https://doi.org/10.1371/journal.pone.0210446
[14] Gradeva, K. and Martínez-Zarzoso, I. (2010). The Role of the Everything But Arms Trade
Preferences Regime in the EU Development Strategy. Research Committee Development
Economics, Proceedings of the German Development Economics Conference, Hannover 2010.
[15] Grossman, G. M. and Helpman, E. (2015). Globalization and growth. American Economic
Review, 105(5):100– 104.
[16] Herz, B. and Wagner, M. (2011). The Dark Side of the Generalized System of Preferences.
Review of International Economics, 19. pp. 763-775. doi:10.1111/j.1467-9396.2011.00980.x
[17] Kassa, W. and Coulibaly, S. (2019). Revisiting the Trade Impact of the African Growth and
Opportunity Act: A Synthetic Control Approach. World Bank Working Paper. 1.
[18] Klasen, S. (2016). Trade preferences for least developed countries. are they effective?
preliminary econometric evidence. United Nations. CDP Policy Review No. 4
[19] Minson A. (2007). Will Chinese Trade Preferences Aid African LDCs? Trade Policy Report
No. 19. Johannesburg: South African Institute of International Affairs.
[20] Moral Del, R. (1980). On Selecting Indirect Ordination Methods. Plant Ecology - PLANT
ECOL (Vegetatio). 42. pp.75-84. 10.1007/BF00048873.
[21] Ornelas, E. and Ritel, M. (2018), The not-so-generalized effects of the Generalized System of
Preferences, CEPR Discussion Paper 13208.
[22] Persson, Maria & Wilhelmsson, Fredrik, 2013. "EU Trade Preferences and Export
Diversification," Working Paper Series 991, Research Institute of Industrial Economics. [23] Sakthivel, E. (2015). Clustering Algorithms using Different Distance Measures. Retrieved from:
https://shodhganga.inflibnet.ac.in/bitstream/10603/90817/12/12_chapter8.pdf
[24] Sinharay, S. (2010) An Overview of Statistics in Education. In: Peterson, P., et al., Eds.,
International Encyclopedia of Education, 3rd Edition, Elsevier Ltd., Amsterdam, pp.1-11.
[25] Thelle, M. (2015). Assessment of Economic Benefits Generated by the EU Trade Regimes
Towards Developing Countries. European Union, 2015. In: Belgium. ISBN: 978-92-79-48088-1
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
65
[26] Wamisho, K. (2015). The impact of the african growth and opportunity act (AGOA): An
empirical analysis of sub-saharan african agricultural exports to the United States. Journal of
International Agricultural Trade and Development. 9 (2).
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
66
IMPLEMENTATION OF INDUSTRY 4.0: A RESEARCH BASED ON THE EFFECTIVE
TRAINING OF HRM
Meri Duduci1
1Faculty of Management and Economics – Tomas Bata University in Zlin
Masaryka 5555, 760 01, Zlin Czech Republic
Abstract Industry 4.0 is considered as the new innovation and way of doing business due to the digitization of the
manufacturing sector.
This has brought several changes in the processes of a company. The changes have been tightly connected to the
Human Resources department.
This research aims to analyze and identify the importance of the effective training by analysing factors that
influence and have a crucial impact in the implementation of Industry 4.0.
The research embraces a quantitative methodological approach through an online questionnaire as a mechanism
to gather data.
The results are beneficial and valuable for HRM and companies managers to comprehend the relevance of training
of employees in order to successfully implement Industry 4.0 in a company.
Keywords
Industry 4.0, employees, training, influential factors, questionnaire, HRM.
JEL Classification
M21, M54, N30, 031.
Introduction
Industry 4.0 has brought numerous changes in the way of doing business, especially when it comes to
inner management approach. Training has been one of the crucial points of this innovation. Training
signifies not only having employees with the prerequisite and fundamental skills to perform a job but
also to advance the set of skills required in the eras of change (Armstrong, 2006).
In Industry 4.0 it is not only about preparing to gain some new skills in order to be up to date with their
working position but especially to prepare themselves professionally to be able to maintain their stand
against the hard competition that they have to face with robotics (Arnold, 2016).
Digitalization and automatization of working processes is crucial for the current labor force of an
enterprise to prepare and provide themselves with a series of abilities to sustain and withstand their
current working position.
It is critical and very important for the Human Resources Department to cautiously select the labor force.
After this set of skills being recognised by the HR it is then attainable though the effective direction to
teach the workers to follow their path in this advanced and contemporary working system.
It is the Human Resources Department who plays the essential and decisive role of designing the training
where the company will be oriented into (Armstrong, 2006). The main objective of the whole training
process is to guide the working force into their everyday process and to be able to gain the maximum
production through a favourable environment by eliminating apprehension and nervousness situations
that might affect them (Arnold, 2016).
The training process selected by the HRM should be equivalent and match the strategic objectives of
the company. It is crucial for all enterprises, SMEs and big companies as well to deeply focus and
concentrate toward the correct type of training in order for the implementation of Industry 4.0 to be
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
67
smoothly implemented avoiding in this way the complications and different disputes that might surface
during this process.
Historical Background The Industrial Revolutions are investigated as the ground of modernization, alteration, revolutions in
the economic world. The economic revolution that commenced in England in the year 1760 (Arnold,
2016) expanded to other European countries. Prior to the Industrial Revolution, the economic basis were
focused on agriculture and animal husbandry. With the invention of the steamer it leaded to the
mechanization of agricultural production (Baena et. al. 2016).
The mechanization process induced the growth of production. The capital in use increased production
by the usage of new machineries and the creation of big companies into the economic environment of
countries (Baena et. al. 2016). . This process caused as a chain factor the creation of new working
possibilities and job positions with a noticeable growth of population. As a final result the living standard
improved by the growth and recovery of the economic situation of the countries (Armstrong 2006).
The First Industrial Revolution consists in the beginning of 1760s and going through the 1830s. In this
revolution, production has developed from physical strength to machinery usage. The main shift in this
process was the usage of coal and steam replacing in this way wood, by increasing the power of the
machines. This mechanization process has caused the formation of big factories by displacing the small
family companies and small enterprises (Armstrong 2006).
The first Industrial Revolution commenced and set the roots firstly in England by fastly spreading all
through Europe and America (Arnold, 2016). The usage of steam, iron and coal was considered crucial
as a result of railway development. This set the path to a whole new innovation and modernization of
the economic life due to the facilitation of movement, not only of people but of trading as well
establishing and evolving in this way the economic situation of countries involved. The changes due to
the revolutions affected not only the economic sector but at a very high range also the social
environment. The lifespan was extended and lengthened as well as the population increased. The quality
of everyday life was improved due to the alleviate circumstances of mechanization (Baena et. al. 2016).
The Second Industrial Revolution includes the time period between the years 1840–1870. The
foundations of the Second Industrial Revolution started with Henry Ford’s mass production. The Second
Industrial Revolution has emerged with changes in basic raw materials and energy sources by using in
the production process steel, petroleum and also chemical elements (Armstrong 2006). The usage of
different raw materials has played a key role in the production system, as well as in the first industrial
revolution also in this second one the railway transport was deeply improved. Easier transportation
system for products to reach further markets and the adjoining and approaching of communication
systems have shaped the transmission process. With the electric technology development (Faller, 2015).
In the Second Industrial Revolution, electric technology was developed and started to be used in
production lines (Baena et. al. 2016). This has allowed us to develop the machines, increase the amount
of production, and meet the concept of mass production (Armstrong 2006). The main actors of the
Second Industrial Revolution were England, Germany, USA, and Japan. This revolution is defined as
the massification of production.
During the time period of the first half of the 20th century, the innovation and modernization of
technology was abated and down sized due to the political issued happening in the world. Economic
crisis was deeply felt also in the alteration and novelty of industrialization. Around the year 1970 (Faller,
2015) the third industrial revolution was initiated by the programmable technologies switching in this
way to the well known and big turn of digital technology. The production process was in this way highly
affected by the computers, new machines and the colossal innovation of using renewable energy
(Arnold, 2016). It leaded to a powerful alteration with the production and manufacturing process at the
time.
And finally the last industrial revolution is the fourth one also known as “The Internet of Things”. The
innovation considered the most major one due to the whole production system concealed by the
digitization of the working system and data analysis (Baena et. al. 2016). The leading edge of this new
revolution is the usage and function of machines without the requirement for human force. The first
steps into the 4.0 happened in Germany in the year 2011 (Faller, 2015) and during the whole process
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
68
Germany was the leading country of spreading and extending this new way of doing business in the
whole world.
This colossal change will cause the adjustment and development of different working departments of an
enterprise. This is why even if it was widely embraced by different countries of the world there are still
adjustments to be done and by the year 2020 there are still investments happening in this new industrial
revolution. With the creation of Industry 4.0 there signifies the establishment of new businesses and
new working positions. Also, with the change of the production system was created the possibility to
generate new products and working processes. The transformations have affected not only products but
also the economic, technological and social spheres (Faller, 2015).
Literature Review
The preeminent component that diversifies a successful company form a less successive one is the
Human Resource Department (Arnold, 2016). The investment that should be subjected to the preparation
of the labor force is crucial in order to boost and stimulate the human capital to accomplish the objectives
and goals of an enterprise. The management should be assured that the training and preparation program
are correctly put into action in a prosperous approach to obtain the maximum compensation from it.
The implementation of Industry 4.0 has caused numerous changes in the working process of the business
world due to the different technology used to process and manufacture the products. With the change of
the technology and robotics comprising and involving more processes caused the change of the working
system overall (Çekmecelioğlu, 2013). The activities related to the labor force have been broaden but
still considered slow compared to the overall development that the working mechanism has faced.
As mentioned previously in this research paper this is revolution has been deliberated as the most
impactive one in the economic world (Arnold, 2016). With the high level of automatisation has
declined the demand for a numerous working force. Companies through the world are facing the
problem of adapting the employees to the advanced and contemporary developments. It is crucial to
state the fact that this is not a time period where to withdraw the employees and replace them with
technology on the contrary, managerial approach have changed and altered in a way to prioritise the
working force by preparing them for the change (Faller, 2015) . Vice versa this is change that employees
should recognise and appraise as a favourable opportunity to develop and enhance themselves
professionally.
According to a study of Rhisiart (Rhisiart et al., 2014) the business environment has been changed and
shaped due to three main factors, the first one is the artificial intelligence and robotics, continuously and
repeatedly updates and improvements of the internet services and finally the time and location where
the training takes place.
The working positions are changing in a direction where around 65 % of children in the primary school
will be working in the future in working positions that have not yet been created (WEF, 2016). There
are several factors that affect the evolution of the working positions such as technology based on the
cloud, big data analysis, developed robotics, artificial intelligence and many other develops of
technologies that have affected and deeply changed the process of doing economy. These alteration
deeply affect everyday life in different extents of the economic and social aspects through the growth
and progress of enterprises but also the creation of new categories and varieties of jobs (Rhisiart et al.,
2014).
The creation of new working positions will lead in the necessity and prerequisite of new assemblage of
skills in new working positions as well as in the existing ones.
The management approach will also be crucial to coordinate the employees with the new method and
technique of industry 4.0.
Industry 4.0 will deeply alter the everyday lifestyles of associations and especially societies by
modifying the manners and habits of economy and business life as well. This modernization will have
to balance and sustain with societies changes and companies natures. It is imperative to keep pace with
the changes of technology and all the innovatives of Industry 4.0 to preserve the enterprises from the
risk of those who cannot catch up with change will face the risk of economic vanishing and fading with
the time.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
69
These numerous changes, that have come into existence in different countries, enterprises and
employees culture will mainly affect the human resource in radical manners. Industry 4.0 will alter and
innovate the whole production process and also profoundly affecting the mechanism of marketing and
distribution. The centrality of all these innovations will be the preparation of employees with the correct
methods of preparing and developing their abilities into the regeneration and recondition equalized with
the modernization of new revolution.
Analysis of the influential factors
There are different factors that influence the correct and successful implementation of Industry 4.0 in
the enterprises. Several challenges affect not only Small and Medium Enterprises but also the big
companies when a new way of doing business is implemented. It signifies a new approach and new
methods of management.
In a study conducted by Mrugalska and Wyrwicka (Mrugalska et. al. 2017), it resulted that factors such
as the ¨Working Environment¨ deeply affects the correct implementation of industry 4.0 not only in the
SMEs but also for big companies. According to their study the high level of technology will strongly
affect their working method that will be more driven toward individual processes. It is expected in the
near future for a new approach toward the working hours that will change and develop toward a more
productive one (Mrugalska et. al. 2017).
The research continues with the stressing toward another important factor considered one of the most
influential factors toward the fast and precise adaptation of business culture in the new and innovative
processes of the 4.0, and that is exactly the correct training of employees and management (Mrugalska
et. al. 2017). As one of the most decisive and imperative elements it is tightly connected with the
qualified and prepared workforce to be taught how to use machines. Also the education system will face
an updated version of itself in order to prepare the new labor force at an earlier stage making possible
the faster adaptation toward the new requirements of the different enterprises updated in the 4.0.
Also in another study conducted by Torna and Vaneker (2019), where the main research is to analyse
the main and crucial factors that affect and determine the profitable and prosperous implementation of
Industry 4.0 in the enterprises is Data Based Management (Torna et. al. 2019).
Data-Based Management is the most decisive and essential decision taken proces by the enterprises in
relation to employees (Torna et. al. 2019).. It is crucial to take into accountability the data and important
information of the labor force that is necessary for the company decision making.
These data are meaningful and imperative for the future decisions or for any type of future organization
and development of the labor force system (Lee, 2014).. Commentaries of the labor force are crucial as
the real force and strong point of each company is comprised by the employees and updates and
improvements made by them are deliberated the most ponderous ones (Torna et. al. 2019).
In another study from Meissner (Meissner, 2017) it is acutely analyzed the Performance Management
as another key factor that influences the effective training toward the correct implementation of Industry
4.0. The approach that the management will embrace will deeply shape the business life and the
environment of an enterprise. The management approach is switching from the classical method to an
approach based on young labor force motivated by rewards.
The key to success in an economy that is moving fast toward the change is the manager role that should
be empowering and an inspiring leader (Meissner, 2017). A manager that faces the future and embraces
the changes and innovation. Nowadays it is crucial that the manager and the whole human resource
department should be trained through a digital system in order to keep track of the data to quickly
develop and preserve the leading role in the market.
Based on the literature review analyzed the following hypothesis will be analyzed by this paper in a
quantitative method. H1: The working environment is highly connected with the correct implementation of Industry 4.0 in a company.
H2: Training of Employees is positively correlated with the optimal implementation of Industry 4.0 in a company.
H3: Implementing the appropriate Data Based Management in a company induces to a successful implementation
of Industry 4.0 in the company`s processes.
H4: A well coordinated performance management affirmatively complements the correct and effective
implementation of industry 4.0 in an enterprise.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
70
Methodology and Data
In this paper the methodology used is a mix of approaches that combines qualitative and quantitative
approach to verify the hypothesis covered by this research. The qualitative approach implicates a
comprehensive analysis of the literatures in regards to industry 4.0. The analysis of previous research is
crucial to be able to analyze the hypothesis and objectives of this research in order to study the previous
examinations of the topic. The literature review used are both from the industry and also from the
academic field. In regards to the academic sphere the articles used are from indexed journals as Web of
Science and Scopus and the analysis from the industry are related to companies leaders in the market
such as Siemens, Amazon etc. The analysis of all the articles were done by attentively analysing the key
words similar to the ones of this research. The literature research analysis was essential and imperative
for evolving and developing the concept of training related to industry 4.0 implementation.
The other approach used to analyse and affirm the hypothesis is the Quantitative research. The data
collected were from over a two-month period from an online questionnaire utilizing Google Forms. The
questionnaire was sent to enterprises based in different countries. The respondents were from the
managerial level in order to have full knowledge in regards to Industry 4.0 concept. The answers were
a total of 200 all developed in English language, not considered a barrier due to previous study and
communication conducted with the companies. The data were analysed through a confirmatory factor
analysis to analyze and study if the proposed factors were persistent with the model created. This
analysis is crucial to show through a ranking level the importance of the factors that affect the
implementation of industry 4.0 and where the training factor stands toward the level of importance that
the literature review has proven for it to have.
Empirical Results
The four hypothesis were analyzed through the Confirmatory Factor Analysis through all the data
gathered by the questionnaire. By this analysis is possible to rank them through the level of
importance.
(H1) Working Environment – (env)
(H2) Training of Employees – (train)
(H4) Data Based Management – (dbm)
(H4)Performance Management – (perf)
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
71
Figure 1: Result of statistical analysis of proposed conceptual framework on Industry 4.0
env
0.254
0.875 0.354
train perf
0.421
dbm
After the statistical analysis it is possible that through the gained results to identify the validity of the
hypothesis and to determine which of the factors has the major impact in the successful
implementation of industry 4.0. During the literature review the main factor was considered the
training of the employees and now it is possible to check it statistically.
From all the factors that have been researched and analysed, only one factor or as it is statistically
knows one construct meets the criteria, and it is the construct Training of Employees. So it is
confirmed and statistically demonstrated through both quantitative and qualitative approaches the
elevated relevance and preponderance that the training process of employees has in the correct and
successful implementation of Industry 4.0.
Here is the rank from the most important to the less important factor, and also hypothesis, archived
through the statistical review.
H2: Training of Employees (train)
H3. Data Based Management (dbm)
H4. Performance Management (perf)
H1. Working environment (env)
Table 1: Quality Criteria
AVE
Composite
Reliability
R
Square
Cronbach`s
Alpha
Communalit
y
Redundanc
y
Env 0.9152 0.9864 0 0.9186 0.9142 0
Train 0.9096 0.9697 0 0.9615 0.9124 0
Dbm 0.8908 0.9542 0 0.9684 0.7541 0
Perf 0.886 0.9621 0 0.9514 0.862 0
4.0
imp 0.954 0.99845 0.8112 0.9845 0.962 0.1358
In regards to this table of quality criteria it is possible to check more in details the results. All factors
show their importance toward the successful implementation of Industry 4.0
0.00
0
Ind
4.0-
0.811
0.00
0
0.000 0.00
0
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
72
With reference of Goodness of fit R-square it determines that only Implementation of Industry 4.0 is a
valid factor that expounds the connectivity artificially dependance of this hypothesis towards each
other..
In the results given by Cronbachs Alpha, incomentable is evident that all the factors and hypothesis
have an internal correlation with one another. Regarding the results of Communality it is crucial to be
higher than 0.5. It is a clear demonstration of the emphasis that the factors have toward the general
outcome of 4.0 processes implementation.. The values of communalities of the variables are higher than
0.5, determines and indicates that the validity of the equation is very good and there is no problem with
this specific test.
For the Redundancy test, it is repeatedly confirming as in the R Square results that only the 4.0
implementation factor is significant when compared to the others that have established the value 0 of
validity.
Table 2: Total Effect
Original
Sample
Sample
Mean
Standard
Deviation
Standard
Error
T
Statistics
env -> 4.0 imp 0.087 0.02418 0.155 0.164 0.1524
train-> 4.0
imp 0.5841 0.5879 0.1405 0.1421 4.5018
dbm -> 4.0
imp 0.1124 0.11 0.1254 0.1876 0.7652
perf -> 4.0
imp 0.1198 0.1012 0.1248 0.1431 0.7233
From the results of Table 2 where are shown the results of the total effect generalized. Through the
different tests that have been conducted in this table once again the results are confirmed that the training
of the employees are more relevant compared to the others. These tests show us a low standard error
and a normal value of the T Statistics. Between these four factors exists low standard error and an
average T statistics. Also the factors show a low correlation and dependency between other factors.
Conclusion
Through this research paper it was possible to analyse in different methods and in complex analysis the
factors that have a high impact and a crucial importance toward the correct implementation of Industry
4.0 into a company.
Due to the fact that Industry 4.0 is still relatively a new term in the economic (Vaneker, 2019),
technological and especially in the social world it still requires a lot of research to fill gaps of knowledge
and previous research and papers are very helpful to enrich our information toward this new way of
doing business.
According to the literature review it was possible to understand that the process of training employees
is the key factor for a successful implementation of Industry 4.0. It was further on confirmed by the
qualitative research of the paper the same result. Through the questionnaire used it was possible to
identify the rank and importance of different factors when compared to the implementation of the new
industry (Vaneker, 2019). The fact of the questionnaire being distributed in the managerial departments
and in different countries of the world served as a powerful movement to gather more data and to have
more qualified data when analyzing industry 4.0 and the factors that have affected it through time.
Through the Confirmatory Factor analysis and through a statistical program was possible to analyze the
data and the results confirmed the literature review that the most important factor when switching to this
new innovation is the training process of employees.
Companies should attentively analyze the importance of the training process by meticulously enrich it
to offer all the adequate elements in order to prepare the labor force to be able to adapt and profitably
switch to this new way of doing business, the innovative industry 4.0 (Lee, 2014).
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
73
Labor force have always been considered the crucial point of all enterprises and this new revolution
proved it once again that if the labor force is not prepared , even if it is a revolution based on robotics,
still the labor force will be crucial to keep the company in the success line.
Acknowledgement
This research was supported by Tomas Bata University.
References
[1] Armstrong, M. (2006). A handbook of human resource management practice (10th ed). London,
Philadelphia: Kogan Page.
[2] Arnold, C., Kiel, D., & Voigt, K. I. (2016). How the industrial internet of things changes business
models in diferent manufacturing industries. International Journal of Innovation Management, 20(08),
1640015. https://doi.org/10.1142/S1363919616400156
[3] Baena, F., Guarin, A., Mora, J., Sauza, J., & Retat, S. (2017). Learning Factory: The Path to Industry
4.0. Procedia Manufacturing, 9, 73–80. https://doi.org/10.1016/j.promfg.2017.04.022
[4] Çekmecelioğlu, H. G., & Günsel, A. (2013). The effects of individual creativity and organizational
climate on firm innovativeness. Procedia - Social and Behavioral Sciences, 99, 257-264.
https://doi.org/10.1016/j.sbspro.2013.10.493
[5] Faller, C., & Feldmúller, D. (2015). Industry 4.0 learning factory for regional SMEs. In Procedia
CIRP (Vol. 32, pp. 88–91). https://doi.org/10.1016/j.procir.2015.02.117
[6] Hecklau, F., Galeitzke, M., Flachs, S., & Kohl, H. (2016). Holistic Approach for Human Resource
Management in Industry 4.0. Procedia CIRP, 54, 1-6. https://doi.org/10.1016/j.procir.2016.05.102
[7] Lee, J., Kao, H. A., & Yang, S. (2014). Service innovation and smart analytics for Industry 4.0 and
big data environment. In Procedia CIRP (Vol. 16, pp. 3–8). https://doi.org/10.1016/j.procir.2014.02.001
[8] Meissner, H., Ilsen, R., & Aurich, J. C. (2017). ScienceDirect Analysis of control architectures in
the context of Industry 4.0. Procedia CIRP, 62, 165–169. https://doi.org/10.1016/j.procir.2016.06.113
[9] Mrugalska, B., & Wyrwicka, M. K. (2017). Towards Lean Production in Industry 4.0. In Procedia
Engineering (Vol. 182, pp. 466–473). https://doi.org/10.1016/j.proeng.2017.03.135
[10] PWC (2016). Industry 4.0: Building the Digital Enterprise, retrieved 27.06.2018,
fromhttps://www.pwc.com/gx/en/industries/industries-4.0/landing-page/ industry-4.0-building-your-
digital-enterprise-april-2016.pdf.
[11] Qin, J., Liu, Y., & Grosvenor, R. (2016). A categorical framework of manufacturing for Industry
4.0 and beyond. Procedia CIRP, 52, 173-178. https://doi.org/10.1016/j.procir.2016.08.005
[12] Rennung, F., Luminosu, C. T., & Draghici, A. (2016). Service provision in the framework of
Industry 4.0. Procedia - Social and Behavioral Sciences, 221, 372-377.
https://doi.org/10.1016/j.sbspro.2016.05.127
[13] Rhisiart M., Glover P., Beck H., (2014). The Future of Work Jobs and Skills in 2030, retrieved
23.05.2017, from www.ukces.org.uk/thefutureofwork
[14] Schwab K. (2015). The Fourth Industrial Revolution What It Means and How to Respond, retrieved
23.05.2017, from https://www.foreignaffairs. com: https://www.foreignaffairs.com/articles/2015-12-
12/ fourthindustrialrevolution.
[15] Torna, I.A.R., Vaneker, T.H.J., (2019). Mass Personalization with Industry 4.0 by SMEs: a Concept
for Collaborative Networks. International Conference on Changeable, Agile, Reconfigurable and Virtual
Production, Procedia Manufacturing 28, 135–141, DOI: 10.1016/j.promfg.2018.12.022
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
74
[15] WEF (2016). The Future of Jobs. World Economic Forum. Geneva, 2016.
www.innovarobotik.com: Yatay, dikey entegrasyon, retrieved 05.24.2017, from
https://www.innovarobotik.com/yatay-dikey-entegrasyon.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
75
EVALUATION OF THE EFFICIENCY OF THE SYSTEM OF SELECTED RESIDENTIAL
SOCIAL SERVICES FOR SENIORS IN THE CZECH REPUBLIC
Izabela Ertingerová1
1Department of Public Economics, VŠB – Technical University of Ostrava,
Sokolská tř. 33, Ostrava 70200, Czech Republic
e-mail: [email protected]
Abstract
The contribution deals with the evaluation of the efficiency of the system of selected residential social
services for seniors in the Czech Republic for the period 2006-2018. The efficiency of social services
for seniors of the residential character is evaluated by means of input and output oriented basic models
of CCR method Data Envelopment Analysis from the perspective of selected key seven aggregated
annual parameters. Input parameters are represented by the total number of employees in selected
residential social facilities, the number of employees in direct care, the size of bed capacity, the number
of residential social facilities and the total expenditure. The output parameters represent the amount of
total income and the number of clients using selected residential social services. The monitored
parameters were analyzed within three selected models: Model A, Model B and Model C. Efficiency
results from the point of view of input and output oriented basic CCR models show that the system of
supply of residential social services for seniors cannot be considered
100 % efficient. In particular, ensuring the availability of bed capacity in direct care workers (Model B)
is considerably inefficient in relation to total income.
Keywords
Residential social services, seniors, efficiency, Data Envelopment Analysis, Czech republic.
JEL Classification
C67, J14
Introduction
The need for long-term care in the last decade has been seen not only in the Czech Republic but also in
other European countries as a new social risk that needs to be addressed. All member states of the
European Union strive to ensure access to social care for seniors in all forms - outpatient, field and
residential services. (Greve, 2016) Generally, it is possible to state that the largest volume of provided
social care is provided to persons of post-productive age in the Czech Republic in their home social
environment in the form of nursing services. However, due to the increasing number of people with
dementia and the minimum willingness of families to look after their relatives, there has been a
significant increase in demand for seniors' social services in recent years, or more precisely after their
placement in residential social facilities. These are mainly residential social services in the form of
homes for the elderly, or homes for the elderly with a special regime. (Hrozenská, Dvořáčková, 2013)
According to the available data from the register of Social Service Providers (2019), the number of
social services facilities for the elderly has increased from 541 to 866 (as of 2018) since 2006, ie by 325
facilities. Given the demographic aging of the population and the published projections of demographic
trends over the coming decades, the area of social services for elderly faces high pressure to ensure
universal access to social services of the required level and quality.
The requirement for providing a certain minimum level of quality of social care and the development of
a system of residential social services for seniors is determined by the economic and legal environment.
An important role here is played not only by assessing and evaluating the efficiency of the activities and
processes of individual residential facilities, but also by managing financial resources, the majority of
which is paid to individual providers of residential social services for seniors from public budgets. At
the same time, individual providers of residential social services for seniors emphasize the need to cope
with rising operating costs, which are growing every year. Increased efficiency can be achieved by
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
76
reducing individual costs, which has a positive impact on the activities of all residential social facilities
in the country. (Víšek, Průša, 2012; Greve, 2016; Horecký, 2010)
The aim of the paper is to evaluate the efficiency of the system of selected residential social services for
seniors (retirement homes and homes for people with special regime) in the Czech Republic and to
describe its trend for the period 2006-2018.
Two hypotheses have been defined in relation to the defined objective:
- H1: “In all monitored years, the system of supply of selected residential social services for
seniors within the three models reaches the value of min. 0.85”;
- H2: “The year 2018 shows, in Model A, B and C, input- and output-oriented, the full value of
efficiency, ie 1”.
The mathematical method of Data Envelopment Analysis was chosen to achieve this goal, on which the
subsequent comparative analysis of each Model A, B and C was based. Furthermore, the Super
Efficiency DEA models were used, which can then be used to organize a set of resulting efficient units
and determine which one is the best.
Literature Review
A significant transformation of the social services system in the Czech Republic was recorded in 1989
due to the reaction to the new conditions of the changing social order. The reform of the social services
system, which was based primarily on the concept of a safety net, began in the early 1990s and lasted
for several years. The key moment came only on January 1, 2007, when the new legislative regulation
of the social services sector came into force - the Social Services Act No. 108/2006 Coll., As amended,
which comprehensively and independently regulates the entire system of social services. (Krebs a kol.,
2015)
At the same time, significant economic, social and demographic changes have occurred since 2007,
which have had a direct impact not only on the development of the system of supplying social services
for seniors, but also on their overall cost. These included, for example, the development of civilization
diseases, worsened health of seniors, increased chances of life expectancy, improvement of the level of
medical diagnostics, etc. (Matoušek, 2011; Horecký, 2010)
The system of financing social services also underwent a change in 2007 in the form of a greater
emphasis on multi-source funding and the introduction of a new social benefit - care allowance that was
transformed from social security benefits that were paid by 2006. The aim was to strengthen the ability
to ensure the optimal form of meeting the needs of people in difficult life situations and to significantly
increase the emphasis on the efficiency of the whole social system. The requirement to assess efficiency
is based on the need to use appropriately available resources, which are available to a limited extent. It
was assumed that the introduction of the new funding system, similar to foreign experience, will reduce
the demand for placement in residential social facilities, due to the increasing use of field services.
(Průša, 2008)
The topic of residential social services for seniors, especially addressing the accessibility of these
services, is a very topical theoretical problem, which is given a significant space in both domestic and
foreign studies. Cornea (2017) analyzed the different types of social services for the elderly and points
out the responsibility of public authorities to ensure the availability of these services. Proenca, Proenca
and Costa (2018) define the main factors - social service providers, sources of funding and activities
that have an impact on the emergence and development of a system of social services that is provided
not only by public but also by private, profitable entities. Langhamrová, Šimková and Sixta (2018)
examined their national economic costs to meet the need to provide social services in nursing homes.
The research shows that the system of offering social services for seniors with a residential character is
different in individual regions of the Czech Republic. At the same time, they state that the amount of
public funds paid to ensure the activities of individual providers of residential social services for seniors
is insufficient.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
77
Scientific approaches that assess the efficiency of social services are not regularly implemented in the
Czech Republic, such as Průša (2007), Průša (2008), Horecký (2012). Matoušek (2011) also points out
that the Standards of Quality of Social Services contained in the implementing decree to the Act on
Social Services do not deal with the issue of evaluating the effect of the service.
The Data Envelopment Analysis (DEA) method is considered a suitable non-parametric method for
evaluating efficiency because of its ability to handle multiple input and output variables. Modeling and
evaluation of the efficiency of the system of residential social facilities for seniors according to the DEA
method is of interest to many authors, especially from abroad. From American studies, mention may be
made of publications from the 1990s, such as Nyman et al. (1990), Fizel, Nunnikhoven (1992) or
Kleinsorge, Karney (1992), whose analyzes were aimed at comparing the efficiency of the system
between profitable and non-profit organizations providing residential social services for seniors. Ozcan
et al. (1998) used the DEA method to measure the efficiency of the offer of services for seniors,
respectively of registered US residential facilities. The analysis was carried out on a representative
sample of 10 % of the total of 324 registered residential facilities. The results have shown that the legal
form of providers of residential services (private, public sector) significantly affects the resulting level
of productivity. However, according to Garavaglia, Lettieri, Agasisti, Lopez (2011) and Chang, Cheng
(2013), it can be anticipated that competition between the providers of residential services will positively
lead to a gradual increase in efficiency. The level of efficiency is also closely linked to the occupancy
rate of accommodation facilities, which is also identified by Christensen (2003) in his expert study.
Methodology and Data
Analyzing and evaluating the efficiency of production units and identifying the sources of their
inefficiency is an important prerequisite for improving the behavior, activities and processes of these
units across the market. Čechura (2009) states that among the main, and in favor of, other methods stand
out those methods that are based on an estimation of the production function when evaluating efficiency.
These methods include, for example, the Data Envelopment Analysis (DEA) method, whose basic
models are among the most commonly applied to assess efficiency.
Given that the essence of the basic models of the DEA method is only to identify efficient and inefficient
units from the monitored set, a number of approaches have been developed that deal with the subsequent
arrangement of the set of efficient units. Among the approaches for organizing efficient units are, for
example, cross-effectiveness, optimistic and pessimistic efficient, the AHP model or the Super-
efficiency models. (Jablonský, Dlouhý, 2015) Super-efficiency models will be used to organize the
evaluated units, or the resulting efficient units from the modeling.
DEA CCR basic model with input and output orientation
The CCR model belongs to a group of basic models and is the first ever DEA model (created in 1978).
The input-oriented primary CCR DEA model is based on the assumption of a constant return to scale
(CRS).
The primary CCR model of the DEA method maximizes the rate of the rated unit of Uq, expressed as
the ratio of the weighted inputs to the weighted outputs, subject to the basic conditions where: (i) the
weights must not be negative; (ii) the efficiency rates of all other production units are less than or equal
to one, ie z ≤ 1. For each production unit monitored, the input weights v i = 1,2,…,m shall be obtained,
the virtual input and the output weights ui = 1,2,….,r, virtual output, expressed as:
virtual input (input) = v1x1q + v2x2q +……+ vmxmq,
virtual output (output) = u1y1q + u2y2q +…..+ uryrq.
The entire input-oriented model can be converted from a linear angle programming to a standard
programming problem into a mathematical formulation for the Uq unit by Charnes-Cooper transform:
maximize z =∑ uiri y
iq, (1)
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
78
under conditions ∑ uiri y
ik ≤ ∑ vj
mj xjk, k = 1,2,…,n,
∑ vjmj xjq = 1,
ui≥ ε i = 1, 2, …,r,
vj≥ ε, j = 1, 2, …,m.
If the resulting value of the coefficient z is equal to one, the production unit Uq is evaluated as efficient.
For inefficient units, their efficiency is less than one, ie z < 1. The amount of the coefficient signals the
amount of input reduction needed to make the unit efficient. (Fiala et al., 2010; Cooper, Seiford, Tone,
2007)
The output-oriented primary CCR DEA model also assumes constant scale returns.
The output-oriented DEA CCR model is based on the same assumptions as the input-oriented model
above. The value of the technical efficiency coefficient is given by the ratio of the weighted sum of
inputs to the weighted sum of outputs, however, weights are sought compared to the previous model so
that the value of the efficiency coefficient is greater than or equal to one, ie g ≥ 1.
The primary output-oriented CCR model can be formulated as follows:
minimalize g =∑ vjmj xjq, (2)
under conditions ∑ uiri y
ik ≤ ∑ vj
mj xjk, k = 1,2,… , n
∑ uiri y
iq = 1,
ui ≥ ε i = 1, 2, …,r,
vj ≥ ε, j = 1, 2, …,m.
If the efficiency coefficient g is equal to one, the production unit of interest is considered to be efficient.
However, if higher efficiency coefficient values were found, the unit can be described as inefficient.
The output-oriented CCR model (2) allows you to determine the number of outputs that make an
inefficient unit efficient. (Cooper, Seiford, Tone, 2007; Jablonský, Dlouhý, 2015).
Super-efficiency models
The essence of Super-efficiency models is based on the fact that when calculating the super-efficiency
rate, the weight of the original efficient unit is set equal to zero (the evaluated unit is thus removed from
the set of monitored production units), resulting in a change of the original efficient limit. The model
then measures the distance between the inputs and outputs of the rated unit from the new efficient
boundary (Fiala et al., 2010; Jablonský, Dlouhý 2015).
In input-oriented DEA models, the original efficient units monitored receive a degree of super-efficiency
of more than one or less than one for output-oriented models. Thanks to this, it is possible to classify the
monitored efficient production units and determine which unit is the most efficient of the given set.
The first model from the category of super-efficiency models, which was published in 1993, was the
Andersen and Petersen models (hereinafter the AP model). The mathematical formulation of the input-
oriented model and assuming a constant returns to scale (CRS) can be expressed as:
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
79
minimalize Ѳ𝑞𝐴𝑃
under conditions Ʃ𝑗=1𝑛 𝑥𝑖𝑗𝜆𝑗 + 𝑠𝑖
− = Ѳ𝑞𝐴𝑃 𝑥𝑖𝑞, i = 1,2,…,m, (3)
Ʃ𝑗=1𝑛 𝑦𝑘𝑗𝜆𝑗 − 𝑠𝑘
+ = 𝑦𝑘𝑞, k = 1,2,…,r,
𝜆𝑗 ≥ 0, j = 1,2,…,n, j ≠ q,
𝜆𝑞 = 0.
The higher the value reaches the degree of Super-efficiency, the more stable the evaluated unit is in
efficiency and will be in a higher position within the overall arrangement.
The formulation of an output-oriented AP model can be filled in analogously as follows:
maximalize 𝜙𝑞𝐴𝑃
under conditions Ʃ𝑗=1𝑛 𝑥𝑖𝑗𝜆𝑗 + 𝑠𝑖
− = 𝑥𝑖𝑞, i = 1,2,…,m, (4)
Ʃ𝑗=1𝑛 𝑦𝑘𝑗𝜆𝑗 − 𝑠𝑘
+ = 𝜙𝑞𝐴𝑃𝑦𝑘𝑞, k = 1,2,…,r,
𝜆𝑗 ≥ 0, j = 1,2,…,n, j ≠ q,
𝜆𝑞 = 0.
Data
The subject of efficiency evaluation according to mathematical formulations (2), (3) of the basic model
of CCR DEA method and Super efficiency model (3) is the system of selected residential social services
for seniors of the Czech Republic. The analysis was carried out for the period 2006-2018. Five input
variables (x1 - x5) and two output variables (y1, y2) were selected for modeling the efficiency of the
system of selected social services for seniors, which combined three models: Model A (x1; y1, y2), Model
B (x2, x3; y2) and Model C (x4, x5; y2). All three models consider input (IO) and output orientation (OO)
assumptions. Table 1. shows the scheme of monitored models.
Table 1. Structure Model A, Model B and Model C
Parameters Model A Model B Model C
Total employees (x1) ✔
Employees in direct care (x2) ✔
Number of beds (x3) ✔
Total costs (x4) ✔
Number of accommodation facilities (x5) ✔
Number of clients (y1) ✔
Total revenue (y2) ✔ ✔ ✔
Source: Own calculation.
For an input-oriented model (IO), the observed unit (year) can be considered efficient if the efficiency
value is equal to 1, but inefficient at less than 1. Also in the case of the output-oriented model (OO) the
monitored unit is efficient if the efficiency value is equal to 1, while the inefficient value is greater than
1.
The data for efficiency modeling were drawn from statistical yearbooks of the Ministry of Labor and
Social Affairs of the Czech Republic. The system of selected residential social services for seniors was
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
80
evaluated for the period 2006-2018, respectively for individual years; for the purposes of the research,
the monitored years were designated as DMU_2006 - DMU_2018. Efficiency modeling was carried out
according to the selected methodology (see 3.1, 3.2), through the DEAFrontier Add-In for Microsoft
Excel.
Empirical Results
Results of the evaluation of efficiency according to the basic CCR DEA model
The results of modeling the efficiency of the system of selected residential social services for seniors
contain and compare input and output variables (parameters), and only from the perspective of
individual models (A, B, C) in total for individual years 2006-2018 (n = 13). The observed efficiency
results by model are shown in Table 2. The results show that the number of efficient and inefficient
homogeneous production units (DMUs - years) is, in terms of both input and output oriented, basic CCR
DEA models completely identical.
It is clear from the table that the best results were achieved in Model A and Model C - 3 efficient units
(years). The results of average values and standard deviations show that the assessed system of selected
residential social services for seniors in the Czech Republic is less efficient in Model B, compared to
Model A and C - the efficiency rate for both models improves over time; the minimum values of
efficiency measure show the location of the monitored DMUs in relative proximity to the efficient limit
1. The standard deviation also confirms the mutual differences in model values.
Table 2. Efficiency results according to models (A - C)
Model A; n=13 Model B; n=13 Model C; n=13
input output input output input output
Number of efficient DMUs 3 3 2 2 3 3
Number of inefficient DMUs 10 10 11 11 10 10
Minimum efficiency rate 0,8734 1,1449 0,6818 1,4666 0,9072 1,1023
Average efficiency rate 0,9301 1,0783 0,8336 1,2305 0,9434 1,0612
Standard deviation 0,0511 0,0583 0,1323 0,1962 0,0329 0,0360
Source: Own calculation.
When the results of the efficiency evaluation are entered into the network graphs, the variance of the
result values can be observed, see Figure 3. The 100 % efficient values (years) (θ (Uq) = 1 and φ (Uq)
= 1) lie on the outer circle and find units (years) showing increasingly inefficiency. Because of the
different values of the output-oriented models, these graphs were adapted to the format of graphs of the
input-oriented models for comparability. Identical values of efficient units in the models are only of a
random nature.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
81
Figure 3. Efficiency results according to models assuming input (IO) and output orientation
(OO)
Source: Own processing.
The figure shows that Model C achieved better results in the evaluation of efficiency. While Model A
achieved the same number as the Model C in terms of efficient and inefficient units, the results of the
measure of efficiency were slightly worse. However, these were very good overall results. The biggest
fluctuation of the results was recorded for the Model B. The most significant fluctuations were recorded
in 2006-2011, especially in the case of the output oriented model (OO). The units (years) in this model
were close to 0.6 (60 %). In these years, the amount of total revenue paid from public budgets was
insufficient (inefficient) in relation to bed capacity and direct service workers. Altogether, the number
of workers providing direct social care was not sufficient in individual years to be able to effectively
and quality serve all beds (placed by clients in residential social facilities).
Homogeneous production units (years) that have been shown to be inefficient in the monitored models
assuming input-oriented assumption are recommended, which in the case of Model A means reducing
the number of employees in residential social facilities (x1); in Model B the number of employees in
direct care care (x2) and size of bed capacity (x3) and in Model C a reduction in total expenditure (x4)
and number of social facilities of a residential nature (x5). On the other hand, units that are inefficient in
output-oriented models are recommended to increase the output variables, ie the number of clients (y1)
and total revenue (y2). A possible solution is also to make changes on both sides at the same time, by
reducing inputs accordingly and increasing outputs. Nevertheless, the reduction of the number of
employees can have an negative impact on the quality of provided care and the increase the number of
clients is not so flexible in the social services system.
In terms of efficient units (years), the system of selected residential social services for seniors in the
Czech Republic was efficient in Model A in 2006, 2015 and 2018, in the case of Model B in 2014 and
2018 and in Model C in 2011, 2017 and 2018. On the contrary, the worst results were achieved in 2008
0,60,70,80,91,0
20062007
2008
2009
2010
201120122013
2014
2015
2016
2017
2018
Model A
IO OO
0,60,70,80,91,0
20062007
2008
2009
2010
201120122013
2014
2015
2016
2017
2018
Model B
IO OO
0,60,70,80,91,0
20062007
2008
2009
2010
201120122013
2014
2015
2016
2017
2018
Model C
IO OO
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
82
and 2009 in both Model A and Model B, while in the case of Model C the years 2006 and 2008 were
concerned.
Results of the evaluation of efficiency according to the Super-efficiency models
Using the Super-efficiency models, it is possible to organize the evaluated efficient units and thus
determine their resulting order. Super-efficiency models allow the classification of efficient units by
assigning a value greater than 1 to the efficient units.
Table 4 shows the results of the Super-efficiency analysis in relation to the evaluation of the system of
selected residential social services for seniors in the Czech Republic in the monitored period 2006-2018.
The above results are limited to units (years) when the selected residential care system for seniors in the
Czech Republic in the modeling models A, B and C was totally efficient in the modeling of efficiency,
ie the rate of efficiency was equal to 1.
Table 4. Results of Super efficiency analysis
Model A
Model B
Model C
Rank DMU Score Rank DMU Score Rank DMU Score
1. DMU_2006 1,1069 1. DMU_2018 1,0655 1. DMU_2018 1,0576
2. DMU_2018 1,0445 2. DMU_2014 1,0187 2. DMU_2017 1,0356
3. DMU_2015 1,0270 3. - - 3. DMU_2011 1,0153
Source: Own calculation.
From the table it is clear that in terms of individual models and the total number of efficient units (years),
it ranks among the most efficient unit - the year 2018, except in Model A, where it occupies the second
place. In the given year, the number of workers provided direct social care in bed most effectively in
relation to the total income managed by providers of residential social services. At the same time, in
2018 the most efficient expenditure was spent on securing activities in terms of the total number of
residential social facilities in the Czech Republic. In Model A, the unit ranked first - 2006. This was the
year in which the social services sector had not yet been comprehensively addressed by legislation.
Nevertheless, in the given year, the system showed the most efficient personnel security in relation to
the total number of clients and received funds (income).
Although these results point to the observed efficiency, it is necessary to take into account that it is the
efficiency of the whole system of residential social care for seniors, which is defined by selected input
and output variables. The units are therefore (in) efficient just in the combination of given inputs and
outputs. The specific selection of parameters, the respective inputs, must be supported by the relevant
arguments relating to the outputs and vice versa. The results of the efficiency modeling when looking at
individual residential social facilities for seniors may differ significantly in the monitored years.
Conclusion
The paper focuses on the evaluation of the efficiency of the system of selected residential social services
for seniors in the Czech Republic for the period 2006–2018 from the perspective of aggregated
parameters - annual results such as staffing, financial statements, bed capacity and number of facilities
systems of residential services for seniors of a social character. The system of residential social services
for seniors consists of state, regional, municipal and non-profit (other) facilities, respectively registered
providers.
The Data Envelopment Analysis (DEA) method, more precisely the CCR model with the assumption of
input and output orientation, was chosen as the key method for evaluating the efficiency of the system
of selected residential social services. The efficiency rating is complemented by the arrangement of
efficient units in each model through the Super Efficiency method.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
83
Although the new law on social services came into effect in 2007, which is in line with the content of a
number of contemporary European systems and principles, the previous problems have not been
completely eliminated and many others have paradoxically deepened. One of the problem areas is the
provision of care for the elderly in residential facilities. (Hrozenská, Dvořáčková, 2013; Průša, 2008)
Based on the findings of the evaluation of the efficiency of the system of social services, which provides
residential social care for seniors in the Czech Republic and in the context of the research hypotheses
H1 and H2, it can be stated that the system can be considered inefficient. However, efficiency is
improving.
Verification of hypothesis H1 as follows: “In all monitored years, the system of supply of selected
residential social services for seniors within the framework of all three models reaches the value of min.
0.85” showed that the system rated within the selected input and output parameters in Model B achieved
efficiency levels between 0.6 and 0.7 between 2006 and 2011 for the input-oriented model and between
0.5 and 0.6 for an output-oriented model. This situation was due to the lack of granted and paid funds
from public budgets needed to ensure the operation of residential facilities. Although the total amount
of money spent on residential social services is growing every year, many funds are not returned to the
system, providers of residential social services are increasingly dependent on the provision of subsidies
from the state budget, health insurance rehabilitation care. At the same time, there was a significant
shortage of workers in direct nursing care in the given years, while the total bed capacity increased due
to the response to the growing demand for seniors' placement in residential facilities. The results of the
efficiency measure for Model A and Model C were above the minimum threshold of 0.85. The
hypothesis H1 is therefore refuted.
The results confirm the second research hypothesis H2, stating that the year 2018 shows the same input
and output orientations in Model A, B and C, the full value of efficiency, ie 1 ranked among the most
efficient this year. It is obvious that the system in terms of monitored basic annual parameters (input and
output variables) was in the given year in optimal values, which led to achieving efficiency in terms of
economic and allocation security of provided social care. In view of the expected demographic growth
in the number of people of post-productive age, it is necessary to recommend that the offer of residential
social services for seniors continues to develop in the coming years, not only in terms of productivity
but also in terms of scale.
However, the results of efficiency or inefficiency, whether from a technical, economic or other point of
view, do not constitute a single or final incentive to decide on the further functioning of the whole
system, respectively of individual social facilities. Effective does not always mean socially desirable.
An important role is also played by the added value created by the area of social services with a focus
on the elderly. These include, for example, improving the quality of life not only for seniors themselves,
but also for their families, social inclusion and other services difficult to replace for society.
References
[1] Cooper, W. W., L. M. Seiford and K. Tone. (2007). Data Envelopment Analysis: A Comprehensive
Text with Models, Applications, References and DEA-Solver Software. New York: Springer.
[2] Cornea, V. (2017). Institutional and Administrative Answers to the Phenomenon of Demographic
Aging: (Re)configuration of the Social Services Infrastructure. Public Administration and
Regional Studies, 19 (1), pp. 71-84.
[3] Čechura, L. (2009). Zdroje a limity růstu agrárního sektoru: analýza efektivnosti a productivity
českého argárního sektoru – aplikace SFA. Praha: Wolters Kluwer.
[4] Fiala, P. et al. (2010). Operační výzkum – nové trendy. Praha: Professional Publishing.
[5] Fizel, L. J. and T. S. Nunnikhoven. (1992). Technical efficiency of for-profit and non-profit
nursing homes. Managerial Decision Economics, 13(5), pp. 429–439.
[6] Garavaglia, G., E. Lettieri, T. Agasisti and S. Lopez. (2011). Efficiency and quality of care in
nursing homes: an Italian case study. Health Care Management Science, 14(1), pp. 22–35.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
84
[7] Greve, B. (2016). Long-Term Care for the Elderly in Europe: Development and Prospects. New
York: Routledge.
[8] Horecký, J. (2012). Efektivní financování sociálních služeb v České republice. Praha: Vysoká
škola finanční a správní.
[9] Horecký, J. (2010). Kapacita, dostupnost, struktura a kvalita sociálních služeb. Tábor: Sociální
služby.
[10] Hrozenská, M. and D. Dvořáčková. (2013). Sociální péče o seniory. Praha: Grada Publishing.
[11] Chang, S. J. and M. Cheng. (2013). The impact of nursing quality on nursing home efficiency:
evidence from Taiwan. Review of Accounting and Finance, Emerald Group Publishing, 12(4), pp.
369-386.
[12] Christensen, E. W. (2003). Scale and Scope Economies in Nursing Homes: A Quantile Regression
Approach. Health Economics, 13(4), pp. 363-377.
[13] Jablonský, J. and M. Dlouhý. (2015). Modely hodnocení efektivnosti a alokace zdrojů. Praha:
Professional Publishing.
[14] Kleinsorge, K. I. and D. F. Karney. (1992). Management of nursing homes using data envelopment
analysis. Socio-Economic Planning Science, 26(1), pp. 57–71.
[15] Krebs, V. et al. (2015). Sociální politika. Praha: Wolters Kluwer.
[16] Langhamrová, J., M. Šimková and J. Sixta. (2018). Makroekonomické dopady rozšiřování
sociálních služeb pro stárnoucí populaci České republiky. Politická ekonomie, 66 (2), pp. 240-259.
[17] Matoušek, O. (2011). Sociální služby: Legislativa, ekonomika, plánování, hodnocení. Praha:
Portál.
[18] Nyman, J. A., D. L. Bricker and D. Link. (1990). Technical efficiency in nursing homes. Medical
Care, 28(6), pp. 541–551.
[19] Ozcan, Y. A., S. E. Wogen and L. W. Mau. (1998). Efficiency evaluation of skilled nursing
facilities. Journal of Medical Systems, 22(4), pp. 211–224.
[20] Proenca, T., J. Proenca and C. Costa. (2018). Enabling factors for developing a social services
network. The Service Industries Journal, 38 (5-6), pp. 321-342.
[21] Průša, L. (2008). Efektivnost financování sociálních služeb pro seniory. Praha: VÚPSV, v.v.i.
[22] Průša, L. (2007). Efektivnost sociálních služeb: vybrané prvky a aspekty. Praha: VÚPSV, v.v.i.
[23] Register of Social Service Providers. (2019). Selected residential social services for seniors.
[online database]. Prague Ministry of Labour and Social Affairs. Available at:
<http://iregistr.mpsv.cz/socreg/hledani_sluzby.do?SUBSESSION_ID=1579012964422_1>.
[24] Víšek, P. and L. Průša. (2012). Optimalizace sociálních služeb. Praha: VÚPSV, v.v.i.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
85
ANALYSIS OF THE SPILLOVER EFFECT OF STOCK MARKET RISK: BASED ON EVT-
COPULA-CVAR MODEL
Lun Gao1
1Department of Finance, VŠB – Technical University of Ostrava,
Sokolská tř. 33, Ostrava 70200, Czech Republic
e-mail: [email protected]
Abstract
This paper integarate the analysis characteristics of EVT copula model and CVaR model, construct the
EVT-copula-CVaR model to study the risk spillover effect of American stock market to the U.K. stock
market. The results show that the U.S. stock market has significant risks to the U.K market. Model
diagnosis and post test show that the model can effectively measure the risk spillover of a single
financial institution (or financial market), which is conducive to the financial regulatory authorities to
track the changes of systemic risk in a timely manner..
Keywords
Conditional value at Risk, Extreme value theory, Copula,Spillover Effect, systematic risk.
JEL Classification
G24 G28
Introduction
Financial globalization makes the economic relations between different countries and regions become
closer. With the increasing level of market opening, while improving the market transaction efficiency,
the risk effect is no longer limited to the domestic financial market, but interacts with each other in
different markets. The risk of a single financial institution (or financial market) can spread to other
market systems through open market channels, resulting in risk spillover and systemic risk and financial
crisis. In the 2008 financial crisis, countries around the world initially underestimated the level of Risk
Spillover in the US financial market. The traditional value at risk (VaR) method lacks effective
estimation and measurement of the risk spillover between institutions and markets, which shows some
limitations.
In order to effectively prevent the occurrence of systemic crisis, it is urgent for us to consider the
Financial Risk Spillover in the extreme situation of the market, the economic dependence of financial
markets between different countries and regions under the condition of open economy, and the potential
losses caused by it. Based on this background, this paper uses extreme value theory (EVT), Copula and
conditional value at risk (CVaR) to study the risk spillover effect, and comprehensively measures the
risk contagion spillover effect between different markets and the measurement of conditional value at
risk.
The structure of this paper is as follows: the second part is literature review; the third part introduces
related models; the fourth part is data selection and empirical analysis; the fifth part is the conclusion of
this paper.
Literature Review
Mcaleer and Da Veiga (2005) systematically used VaR method to study the market volatility spillover
effect, and found that VaR method underestimated the market volatility spillover. In order to solve the
undervaluation of market by VaR method, Adian and Brunnermeier (2008) put forward CVaR method.
This method can fully consider the dynamic change of systemic risk in financial market, and also
effectively improve the risk prediction problem of financial market, so it is widely used in the research
field of Risk Spillover (Gideon and Paul, 2017). Gropp et al. (2009) studied the cross-border Risk
Spillover Effect among European banks; Bee and Miorelli (2010) analyzed the market risk during the
financial crisis by using the POT (pecks over threshold) method of extreme value theory and dynamic
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
86
VaR method, and found that this method was very effective in the risk spillover measurement of high-
risk period; Girardi and Erguny (2010) used GARCH method to analyze the dependence of CVaR
between international financial markets, and found that when the vulnerability of financial markets
increased, the risk spillover effect increased the risk contagion; Adams et al. (2010) quantitatively
analyzed the scale and duration of Risk Spillover using copula-VAR model, and found that investment
banks and arbitrage funds play a leading role in the transmission of risk spillover. Most of the above-
mentioned related studies use the independent single GARCH model family method, CVaR method and
EVT to study the spillover effect of the stock market, and most of them focus on the single domestic
market. The financial crisis triggered by the United States fully shows that although we can effectively
manage the normal market fluctuations through the daily institutional arrangements, however, although
the probability of market volatility is low in extreme cases, it often leads to great market risk. In an open
economy, this risk spreads more through the interconnection of different economies, due to the lack of
comprehensive consideration of the risk spillover effect under the extreme conditions of the market, it
eventually leads to serious financial crisis. Therefore, under the condition of open economy, we should
not only measure the risk spillover between different stock markets, but also pay attention to the possible
extreme situation of the market.
Because financial assets generally have "leverage effect", the yield often presents an asymmetric
distribution of "peak and fat tail". The traditional statistical analysis can not fit the distribution of the
yield sequence well, nor predict the change characteristics of financial assets in extreme cases. However,
the extreme value theory has a good goodness of fit for the tail of income, and does not need to model
the whole distribution. It can overcome the shortcomings of other measurement methods in solving fat
tail distribution. It is an effective method to measure the extreme situation of market risk. Copula method
can flexibly select the specific form of asset edge distribution, consider the asset edge distribution and
their correlation structure separately, and capture the nonlinear and asymmetric correlation between
variables. When the profit and loss distribution of traditional VaR method is non normal distribution, it
can't satisfy all the properties of consistent risk measurement. As a result, the local optimal solution is
not necessarily the global optimal solution, which can not deal with the extreme price changes in
financial markets in a timely and effective manner. CVaR can satisfy all the properties and convexity of
consistent risk measurement. It measures the average value of tail loss when the loss exceeds VaR,
which represents the average level of excess loss, and can measure tail loss adequately
Based on the analysis advantages of the above methods and the actual demand of stock market risk
spillover effect measurement, In this paper, EVT copula CVaR model is constructed to analyze the risk
spillover effect between U.S. and U.K. stock markets.
Methodology
This part first introduces the definition and principle of CVaR, and then introduces the specific process
of CVaR calculation using EVT copula model. There are two key steps to measure the relevant structure
of financial market by copula method. The first step is to select appropriate marginal distribution to
combine the sequences respectively, considering that the financial times series generally have the
characteristics of peak and fat tail, this paper uses the semi parametric method to fit the upper and lower
tail of the time series with the generalized Pareto distribution (GPD) of the extreme value theory, while
the middle part of the series uses the empirical distribution. After the marginal distribution is established,
the best fitting function is found from the common copula function family in order to describe the
correlation.
Conditional Value at Risk (CVaR)
Due to the non-linear characteristics such as thick-tailedness and asymmetry, the volatility of financial
asset returns generally cannot be effectively examined by traditional VaR. Huge loss scenarios with
extreme probability of extreme price changes (such as stock market crashes and financial crises) are
often underestimated, leading to inadequate measurements of VaR tail losses. CVaR satisfies all the
properties and convexity of consistent risk measures, and reflects that the tail loss exceeds the average
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
87
value of VaR. Through methods such as sample quantile estimation, a sufficient measure of tail loss can
be achieved without relying on VaR calculations. According to Adrian and Brunnermeier (2008), the
mathematical expression of CVaR from the perspective of risk spillover effects:
Pr(𝑋𝑖 ≤ 𝐶𝑉𝑎𝑅𝑞𝑖𝑗|𝑋𝑗 = 𝐶𝑉𝑎𝑅𝑞
𝑗) = 𝑞 (1)
Among them, 𝑋 represents the level of risk loss, 𝑞 represents a significant level, and 𝑖 and 𝑗represent
financial institutions (or financial markets), 𝐶𝑉𝑎𝑅𝑞𝑖𝑗
indicates that when 𝑗 is in an extremely unfavorable
condition, the level of risk 𝑖 faces is the conditional risk value of 𝑖 with respect to 𝑗, which includes the
value of unconditional risk and the value of spillover risk. The risk spillover effect of 𝑗 on 𝑖 is described
by the numerical relationship between 𝐶𝑉𝑎𝑅𝑞𝑖𝑗
and 𝐶𝑉𝑎𝑅𝑞𝑗. In order to reflect the risk spillover of the
risk event of 𝑗 to 𝑖 more accurately, we define the spillover risk value as 𝛥𝐶𝑉𝑎𝑅𝑞𝑖𝑗
, and the expression
is:
𝛥𝐶𝑉𝑎𝑅𝑞𝑖𝑗= 𝐶𝑉𝑎𝑅𝑞
𝑖𝑗− 𝑉𝑎𝑅𝑞
𝑖 (2)
Due to the large differences in the scale of risk spillovers between different markets and financial
institutions, in order to facilitate comparison, further standardization is required, such as:
%𝐶𝑉𝑎𝑅𝑞𝑖𝑗= (
𝛥𝐶𝑉𝑎𝑅𝑞𝑖𝑗
𝑉𝑎𝑅𝑞𝑖 ) ∗ 100% (3)
%𝐶𝑉𝑎𝑅𝑞𝑖𝑗
removes the influence of dimension, which can more accurately reflect the degree of risk
spillover to 𝑖 when a risk event occurs for 𝑗, and discover the change of system risk when 𝑗 occurs a risk
event timely. 𝛥𝐶𝑉𝑎𝑅𝑞𝑖𝑗
technology combines the risk spillover effect with traditional VaR, which can
more accurately reflect the true level of risk, which is of great significance to the regulatory authorities
concerned about the risk of the entire financial system. Supervisory authorities can therefore accurately
and effectively discover the level of contribution of individual financial institutions (or financial
markets) to systemic risks, and quickly take targeted regulatory measures for the stability of the entire
financial system.
Modeling the marginal distribution using EVT
Extreme value theory mainly deals with extreme cases of risk. It has the ability to estimate beyond
sample data and can accurately describe the tail distribution. Extreme value theory mainly includes two
types of models, the traditional BMM (block maxima method) model and the POT model developed in
recent years. The BMM model often requires a large amount of sample data to model the maximum
value after blocking. This method is difficult to apply in practice due to the limited availability of tail
data. The POT model sets a threshold in advance and models all sample data that exceeds the threshold.
It overcomes the statistical problem of insufficient tail data to a certain extent. It has a clear advantage
over BMM when dealing with tail data in extreme cases. According to the definition of Viviana (2003),
the mathematical expression of EVT is as follows:
Let 𝑋𝑖 , 𝑖 = 1, … , 𝑛 be independent and identically distributed random variables, their common
distribution is 𝐹(𝑥) = Pr (𝑋𝑖 ≤ 𝑥). Let 𝑋 be an arbitrary 𝑋𝑖 . Choose a threshold 𝑢 . We define the
conditional probability distribution of the excess 𝑦 that exceeds the threshold 𝑢 as:
𝐹𝑢(𝑦) = Pr(𝑋 − 𝑢 ≤ 𝑦|𝑋 > 𝑢) =𝐹(𝑦+𝑢)−𝐹(𝑢)
1−𝐹(𝑢), 𝑦 > 0
(4)
The 𝐹𝑢(𝑦) is an over-threshold distribution. Since the generalized Pareto distribution (GPD )can fit the
tail of the yield series well, this article chooses GPD to model the up and lower tail of the yield series
and uses the empirical distribution to fit the middle part of the yield sequence. Finally, the marginal
distribution of the yield sequence 𝑋 is:
𝐹(𝑥) =
{
𝑁𝑢𝐿
𝑁(1 − 𝜉
𝑥−𝑢
𝛽(𝑢))−1
𝜉, 𝑥 < 𝑢𝐿
𝐸𝑐𝑑𝑓(𝑥), 𝑢𝐿 ≤ 𝑥 ≤ 𝑢𝑅
1 −𝑁𝑢𝑅
𝑁(1 − 𝜉
𝑥−𝑢
𝛽(𝑢))−1
𝜉, 𝑥 > 𝑢𝑅
(5)
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
88
Among them, 𝐸𝑐𝑑𝑓(𝑥) is the empirical distribution function on the interval of yield rate 𝑢𝐿 ≤ 𝑥 ≤ 𝑢𝑅.
𝛽(𝑢) is expressed as a positive function scale parameter related to 𝑢, ξ ∈ R is the shape parameter of
the distribution, 𝑢 is the lower-tail threshold, 𝑢𝑅 is the upper-tail threshold of 𝑢, and 𝑢𝑅 is the lower-tail
threshold of 𝑢, 𝑁𝑢 represents the number of observations in the sample that are less than the threshold
𝑢 . The determination of the threshold 𝑢 is a prerequisite for correct estimation of ξ and 𝛽(𝑢). An
excessively high threshold 𝑢 will cause too little excess data, so that the variance of the estimated
parameters will be high, while a too small threshold 𝑢 will produce a biased estimate. In actual
operation, it is usually estimated by mean excess plot and Hill plot, but there is no consistent accurate
estimation method for the selection of the threshold 𝑢. At present, most of paper use the principle
provided by Du Mouchel (1983), that is, selecting the number of samples exceeding the threshold value
to account for 10% of the total number of samples to determine the threshold value. This article uses
this principle to determine the upper and lower tail thresholds of the yield series.
Choose the appropriate Copula function
Copula is actually a class of functions that connect joint distribution functions with their respective
marginal distribution functions. It was first proposed by Sklar (1959). With the development of current
information technology, it began to be used in the financial field in the late 1990s. This article will
choose the corresponding Copula function according to its fit with the actual return series. According to
the definition of Di Clemente (2003), the N-dimensional Copula function refers to a set of functions that
satisfy the following properties C: [0,1]𝑁 → [0,1]. The specific mathematical definition features are
expressed as follows:
1. 𝐷𝑜𝑚𝐶 = 𝐼𝑁 = [0,1]𝑁(DomC represents the domain of function set C);
2. 𝐶 is an N-dimensional increasing function with a grounded surface;
3. The marginal distribution function 𝐶𝑛 of 𝐶 satisfies: Cn (u) = C (1,… 1, u, 1… , 1) = u, where
u ∈ [0,1]. According to the above definition, the Copula function is a conection function that associates a multi-
dimensional joint distribution with a one-dimensional edge distribution. In fact, it is a multivariate
distribution function cluster with [0,1] uniform marginal distribution in N-dimensional [0,1] space.
If 𝐹1 , ... 𝐹𝑁 are univariate distribution functions, then 𝐶(𝐹1(𝑥1),… , (𝐹1(𝑥𝑛),… , 𝐹𝑁(𝑥𝑛)) is a
multivariate distribution function with marginal distribution 𝐹1, ... 𝐹𝑁. According to Sklar's theorem, if
F is an N-dimensional joint distribution function with marginal distributions 𝐹1, ... 𝐹𝑁, there must be a
Copula function C: [0,1]𝑁 → [0,1], so that:
𝐶(𝑢1, … , 𝑢𝑛) = 𝐹(𝐹1(−1)(𝑥1), … , 𝐹𝑁
(−1)(𝑥𝑁)) (7)
The above formula fully illustrates that the Copula function actually reflects a relationship between the
multivariate marginal distribution and its joint distribution that contains all the information between the
variables. Therefore, applying the Copula function can easily obtain the related structure of the
multivariate distribution, and it is not necessary that the edge distribution functions 𝐹1, ... 𝐹𝑁 have the
same distribution form. According to Sklar's theorem, we can get the density function of the joint
distribution function 𝐹:
𝑓(𝑥1, … , 𝑥𝑛 , … 𝑥𝑁) = 𝑐(𝐹1(𝑥1),… , 𝐹𝑛(𝑥𝑛), …𝐹𝑁(𝑥𝑁))∏ 𝑓𝑛(𝑥𝑛)𝑁𝑛=1 (8)
Where𝑐(𝑢1, … , 𝑢𝑛,… 𝑢𝑁) =𝜕C(𝑢1,…,𝑢𝑛,…𝑢𝑁)
𝜕𝑢1…𝜕𝑢𝑛…𝜕𝑢𝑁 is the density function of the Copula function, 𝑓𝑁(𝑥𝑁)is the
density function of the marginal distribution 𝐹𝑛(𝑥𝑛). From Equation (3), we can split a joint distribution
function into a univariate marginal distribution and a dependent structure represented by the copula
function. It provides a method to analyze the multivariate distribution dependent structure without
considering the edge distribution, It makes the solution of multiple univariate joint distribution functions
more convenient.
In this article, we will use t-copula for calculations. That is, consider two risk factors(𝑋1, 𝑋2)𝑇The joint
distribution 𝐹 is unknown. For a certain Copula C, their marginal distributions are 𝐹1 and 𝐹2 stisfy:
F (x1, x2) = C(F1(x1), F2(x2))
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
89
Calculation of CVaR
According to the definition of Gropp et al. (2009), for the yield series 𝑋𝑖 and 𝑋𝑗 , it is assumed that their
joint distribution density function and edge distribution density function are 𝑓 (𝑥𝑖 , 𝑥𝑗 ), 𝑓𝑖 (𝑥𝑖), 𝑓𝑗(𝑥
𝑗).
The conditional distribution density function of the sequence 𝑋𝑖 under the given conditions is:
𝑓𝑖|𝑗(𝑥𝑖|𝑥𝑗) =
𝑓 (𝑥𝑖 ,𝑥𝑗 )
𝑓𝑗 ( 𝑥𝑗 ) (9)
Combining the previous formula (8) of the Copula function, we can further derive the following
formula: 𝑓𝑖|𝑗(𝑥𝑖|𝑥𝑗) = 𝑐 (𝐹𝑖(𝑥
𝑖),… , 𝐹𝑗(𝑥𝑗)) 𝑓𝑖(𝑥
𝑖).Therefore, the conditional distribution function of
the yield sequence 𝑋𝑖 under the given 𝑋𝑗 can be obtained by the following formula:
𝑓𝑖|𝑗(𝑥𝑖|𝑥𝑗) ∫ 𝑐(
𝑥𝑖
−∞(𝐹𝑖(𝑥
𝑖),… , 𝐹𝑗(𝑥𝑗)) 𝑓𝑖(𝑥
𝑖)𝑑𝑥𝑖 (10)
Among them, 𝐹𝑖 and 𝐹𝑗are the edge distributions of the Copula function, which can be obtained by the
extreme value theory which introduced earlier. The derivative of 𝐹𝑖 is 𝑓𝑖, and 𝑐 is the density function
of the selected optimal Copula function. According to the definition of 𝐶𝑉𝑎𝑅𝑞𝑖𝑗, 𝐶𝑉𝑎𝑅𝑞
𝑖𝑗 is the value at
risk of 𝑋𝑖 under 𝑋𝑗 = Va𝑅𝑞𝑗.
𝐶𝑉𝑎𝑅𝑞𝑖𝑗= 𝐹𝑖|𝑗
−1(𝑞| Va𝑅𝑞𝑗) (11)
Among them, 𝐹𝑖|𝑗−1j is an inverse function of 𝐹𝑖|𝑗, or a conditional quantile function. Sometimes it is
difficult to calculate the explicit expression of 𝐹𝑖|𝑗−1, therefore, in the actual solution process, we usually
calculate equation (12), and the solution of 𝑥𝑖 is 𝐶𝑉𝑎𝑅𝑞𝑖𝑗
.
∫ 𝑐(𝑥𝑖
−∞(𝐹𝑖(𝑥
𝑖), 𝐹𝑗(𝑉𝑎𝑅𝑞𝑗)) 𝑓𝑖(𝑥
𝑖)𝑑𝑥𝑖 = 𝑞 (12)
Data selection and empirical analysis
As the world's largest economy, the United States has a dominant financial market in the global financial
system. The turbulence in the US financial market can easily spill over to financial markets in other
countries (regions) through various channels. Therefore, using the stock market as a representative of
the financial market, study the risk spillover effects of US financial markets on other major financial
markets have great practical significance. This article selects the daily closing prices of Standard &
Poor's Index (S&P500) and London Index (FTSE) as raw data. Considering the time difference between
the U.S. stock market and the stock markets of other countries. In the analysis process, 𝑡 − 1 is used as
the US stock market trading day, and 𝑡 is used as the corresponding British stock market trading day.
Take the logarithmic first-order difference of the index closing price to calculate the daily index return.
In order to reduce the calculation error, we multiply all calculation results by 100.
The description of the basic statistical characteristics in Table 1 shows that although the skewness of the
stock index returns are close to 0 corresponding to the normal distribution, the kurtosis are greater than
3 corresponding to the normal distribution, and the probability value of the Jarque-Bera test result is 0.
That is, at a significant level of 1%, each stock index return series is significantly different from the
normal distribution. Based on this, it can be initially judged that each stock index return series does not
obey the normal distribution.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
90
Table 1 Basic statistical description of data
Mean Max Min S.D Skewness Kurtosis J-B
S&P500 0.040 4.840 -6.896 0.938 -0.491 4.525 0.0
FTSE100 0.011 5.032 -6.199 0.936 -0.279 3.124 0.0
In order to further confirm the non-normality of each stock index return series, we make a Q-Q plot
corresponding to each series. Taking the FTSE100 yield sequence as an example, the Q-Q plot is shown
in Figure 1. It can be seen that the upper and lower tails of the FTSE100 yield series deviate significantly
from the normal distribution, and have significant fat tail characteristics.
Figure 1 Q-Q plot of S&P500 and FTSE100
The test of other stock index yield series Q-Q charts also reached the same conclusion, so each stock
index yield series showed a significant "peak and fat tail" phenomenon. Since the GPD distribution in
the extreme value theory can fit the tail data of the yield series well, we use the GPD distribution to fit
the upper and lower tails of each stock index yield series. The data in the middle of the upper and lower
tails of the stock index return series is fitted using an empirical distribution.
After determining the upper and lower tail thresholds of each stock index's return series according to the
Du Mouchel 10% principle, the maximum likelihood method is used to estimate the scale parameter
𝛽 (𝑢) and shape parameter 𝜉 corresponding to the GPD distribution. Based on the estimated results, we
make a GPD distribution fitting diagnostic chart for each series. Taking the FTSE income series as an
example, the upper tail GPD distribution fitting diagnostic chart is shown in Figure 2. From the fitted
diagnostic graph, we found that all points are concentrated near the distribution curve (including the
over-threshold distribution curve and the tail distribution curve), which fits the data well, and the test of
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
91
the fitting effect of S&P500 sequences also gives the same in conclusion. Substitute the parameter
estimates into formula (9) to get the marginal distribution function of each stock index return. After
establishing the marginal distribution of each stock index return series, we use the t-Copula dependency
structure function to capture the correlation structure between the FTSE100 index return series and the
S&P500 return series. The results are shown in Table 2. According to the Kendall τ correlation
coefficient and the upper and lower tail correlation coefficient (mainly focusing on the lower tail
correlation coefficient). From the perspective of risk spillovers, because the bottom-end correlation
coefficients are all positive numbers, S & P500 has a positive risk spillover effect on other stock indexes,
that is, when the S & P500 yield is at its risk level, the probability of potential loss of other stock index
returns will be increase.
Figure 2 upper-tail fitting of ftse100
Table 2 Basic statistical description of data
Parameter estimates Kendall τ lower tail Upper tail
ftse100 θ = 1.153 0.39327202 0.138 0.086 δ = 0.244
So far, we have established the marginal distribution function of each stock index return series and the
Copula function of the S&P500 return series and the UK stock market index return series. In order to
examine the strength of the spillover effect of US stock market risk on the British stock market, we use
the previous method to calculate the CVaR, ΔCVaR, and % CVaR of the other stock index return series
under S&P500 at risk. The results are shown in Table 3.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
92
Table 2 Basic statistical description of data
VaR Cvar △CoVaR %CoVaR( %)
FTSE100 2.5559 3.3081 0.7522 29%
It can be seen that, at a significant level of 5%, the conditional risk value (CVaR) of the UK stock index
return series is greater than the corresponding unconditional value at risk (VaR), that is, the S & P500
risk event has a positive spillover effect on other stock indexes, and the risk spillover intensity (
(Expressed as% CVaR) is 29%. The above analysis shows that, using the EVT-Copula-CVaR model to
better fit the relevant structure between different stock markets, financial institutions and regulators can
apply the model method to risk spillovers when other financial institutions (or financial markets) have
risk events Effective evaluation of direction and intensity will further improve risk management
decision-making capabilities.
Conclusion
The definition of CVaR is based on VaR, which measures the average of tail losses with losses exceeding
VaR and represents the average level of excess losses.
CVaR can effectively measure the tail loss and overcome the inadequacy of traditional VaR for tail loss
measurement. The EVT-Copula model can effectively fit the relevant structure between financial
markets under extreme market conditions. This paper builds the EVT-Copula-CVaR model by
combining the analysis characteristics of the two models. The generalized Pareto distribution is used to
fit the upper and lower tails of each stock index yield series, and the data in the middle of the upper and
lower tails of the stock index yield series is fitted using an empirical distribution. According to the
Kendallτ correlation coefficient and the upper and lower tail correlation coefficients (mainly focusing
on the lower tail correlation coefficients), the risk spillover effects of S & P500 on other stock indexes
are qualitatively analyzed.
Analysis based on this model shows that the US stock market has a strong positive risk spillover effect
on the UK. The intensity of risk spillover of the US stock market to other stock markets is also related
to the size of the US stock market's own risk. The greater the risk of the US stock market, the higher the
risk spillover of other stock markets. This model method can effectively measure the risk spillover of a
single financial institution (or financial market), and it is helpful for financial regulatory authorities to
track the changes in systemic risks in a timely manner. The financial supervision department can carry
out differentiated management according to the contribution of various financial institutions to the
system risk ΔCVaR, focusing on strengthening the supervision of financial institutions with relatively
high ΔCVaR values.
.
References
[1] Adams, Zeno; Füss, Roland and Gropp, Reint.(2010).Modeling Spillover Effects among Financial
Institutions: A State-Dependent Sensitivity Value-at-Risk( SDSVaR).ApproachEuropean
Business School ( EBS) working paper,( 5) ,2010.
[2] Adrian,T.and Brunnermeier, M.(2008). CoVaR. Federal Reserve Bank of New York Staff Reports,
no. 348,2008.
[3] Bee, Marco and Miorelli, Fabrizio. (2010). Dynamic VaR Models and the Peaks over Threshold
Method for MarketRisk Measurement: an Empirical Investigation during a Financial Crisis.
Elenco dei working paper,2010.
[4] Di Clemente, A. and Romano, C. A copula-Extreme Value Theory Approach for modeling
Operational Risk. http: //www.gloriamundi. org, 2003.
[5] Giulio Girardi, A. Tolga Ergün. (2013).Systemic Risk Measurement:Multivariate GARCH
Estimation of CVaR. Journal of Banking & Finance,37(8).
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
93
[6] Hamao,Y. ; Masulis, R. W. and Ng, V. (1990). Correlations in Price Changes and Volatility across
International Stock Markets. Review of Financial Studies, 1990,3( 2) , pp. 281-307.
[7] Hartman, P. ;Straetmans, S and de Vries, C. G. (2004). Asset Market Linkages in Crisis Periods.
Review of FinancialStudies. 2004,86(1) , pp. 313-326.
[8] McAleer, M. and da Veiga, B. (2005). Spillover Effects in Forecasting Volatility and VaR. School
of Economics and Commerce University of Western Australia. 2005.
[9] Samit Paul, Prateek Sharma. (2017). Improved VaR Forecasts Using Extreme Value Theory with
the Realized GARCH model .Studies in Economics and Finance, 2017, 34(2).
[10] Sklar,A. Fonctions de repartition àn dimensions et leurs marges. Publication de l Institut de
Statistique de l Université de Paris, 1959, 8, pp. 229-231.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
94
THE IDENTIFICATION OF FACTORS INFLUENCING HUMAN RESOURCES
MANAGEMENT AND THE EVALUATION OF THEIR INTENSITY: A CASE STUDY ON
HUMAN RESOURCES MANAGEMENT (HRM)
Daniela Kharroubi1
1Department of Management, VŠB – Technical University of Ostrava,
Sokolská tř. 33, Ostrava 70200, Czech Republic
e-mail: [email protected]
Abstract
All experts admit that human resources management is the most important part of any business structure.
The company may have the best technology, output capacity and equipment, but may not achieve the
required profits because its staff is poorly managed. When human resources are managed in the best
possible way, positive results are achieved. For this reason, the HRM must count with factors that
influence its efficiency and have to learn how to deal with these factors. The main objective of this case
study is to identify the main factors that influence HRM in the workplace and to determine whether
these factors have a significant impact on HRM. The research tools used in this paper was a structured
undisguised questionnaire, which was administrated to 25 employees, 18 women and 7 men in the HRM
field. Two independent factors, i.e. external and internal were assessed. The descriptive analysis and the
analysis of Variance (ANOVA) were carried out to derive conclusions about the features of the
mentioned factors. In order to measure the internal consistency of the survey and the reliability of the
scale, the reliability Cronbach’s alphas were calculated. The value of Cronbach’s alpha for each of the
independent factors ranged above 0.9, where the overall C-alpha was 0,943 for the internal factors and
0.991 for the external factors, showing a consistency of the acquired data. In addition to that, the analysis
of Variance was tested to explain the relationship between the independent factors and HRM. The results
of the hypothesis testing showed that more than 50% the mentioned internal and external factors have
an influence on HRM.
Keywords
Human Resources Management, internal factors, external factors, Cronbach’s alpha, analysis of Variance.
JEL Classification
M12 Personnel Management, O15 Human Resources, C46 Specific Statistics, C12 Hypothesis Testing: General.
Introduction
Human Resources Management has changed a lot over the past years. A century ago, most of the people
worked in manufacturing companies and were watched by supervisors. However, companies started of
thinking for ways on how to improve the productivity and efficiency of employees, which was the
approach of scientific management. All of sudden, companies started to study performance standards,
i.e. how much is made by a certain time, job satisfaction, human relations, financial rewards, etc. This
was the beginning of a new approach on how to gain an advantage over competitors and economies of
scale.
As a result, it is renowned that Human Resources Management is the only living factor of production to
control the other factors. In fact, imagine leading companies with impressive buildings and lofty offices
without well-talented employees, definitely they will collapse (Dessler, 2008).
For this reason, there are many aspects that affect the implementation of Human Resources Management
practices. For instance, Budhwar and Baruch (2003) studied HR practices in developing countries and
they found out that it’s associated with organizational and social aspects. In this regard, Oinas Paivi and
Van Gils (2001) identified contextual resources that enhance HR competencies. These aspects included
elements in the external and the internal environment of the company.
As a result, this research study is focused on to explore the factors (external and internal) that influence
the performance of HRM in the Czech Republic;
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
95
To identify the main factors that influence HRM in the workplace
To identify whether the mentioned factors have a significant influence on HRM
Literature Review
In today’ s modest business environment, company’s workforce is in a continual state of flux- skill sets
and job requirement, in addition to regulatory environmental changes at such a rapid pace that the staff
needs had significantly changed.
In the past decades, the HR manager has evolved considerably. Their previous functional approach has
been substituted for a strategic one (Wakely & Point, 2003). Human resource management is mostly
concentrated on leadership (as getting ready for tomorrow) agenda and thoroughly incorporated with
the business (Mooney, 2001). HR manager now has a much deeper understanding of key organizational
challenges, plays a proactive and strategic role and is no more condemned to a reactive and
administrative role (Nasiriour, Afshar Kazemi & Izadi, 2012). Ulrich (1995) even goes so far as to
suggest that HR department should be purged if they fail to become more strategic. For this reason,
HRM is the fundamental strength of organizations in facing the challenges/factors of business today.
Some of the definitions of Human Resources Management: “Human Resources Management is
regarded as a philosophy about the ways in which people are managed at work that is underpinned by
a number of theories relating to the behavior of people and organizations” (Armstrong & Taylor, 2020).
“Human Resources Management is the aspect of managing people in the broad areas of resourcing-
varieties of recruitment and selection, rewarding – forms of pay, developing – forms of training and
assessment, and the building and sustaining of relationships, primarily here, employment relations”
(Rowley & Jackson, 2010).“Human Resource Management is the function within an organization that
focuses on recruitment of, management of, and providing direction for the people who work in the
organization” (Maalderink, 2014).
When analyzing the role of HRM, many challenges exist either internally or externally which adversely
affect its delivery to quality services. In urbanized countries, the HR managers have distinguished the
challenges they face and have developed different strategies to overwhelm these challenges. Now the
question arises as, which internal and external factors impact the role of HR in an organization and how
these factors affect it. In today’s intensive competition and global marketplace, there are a lot of internal
and external factors that affect the role of HR department.
External factors
External factors (Pitra, 2008) have an impact on the internal environment of the organization. They
create an environment in which opportunities and threats arise for the realization of the business plans
of an organization. In general, they can be divided into economic ones that affect the economic
conditions of organizations; political, which are the source of legislation and restrictions; social, which
characterize the lifestyle of society, environmental and technological, which represent the possibility of
applying certain technologies during the implementation of activities in the organization. From the
external aspect, the following external factors are particularly important for the management of human
resource: the situation of workforce in the labor market, such as the level and type of human resources
qualifications, average incomes, labor movements…etc. Labor law (legislations), which affects
activities related to the closure and termination of employment, social security, remuneration…etc.
Socio-cultural environment, such as the average time spent commuting to work, labor norms in a given
region or country, interpersonal relationships, life values and cultural traditions. Competitors (Porter,
2008) affects the behavior of the organization. Organizations focus on other capabilities that the
customer appreciates, gaining a competitive advantage. State regulations that influence the
organization's capabilities through legal norms that impose on them various obligations to reduce
externalities. E.g. the obligation to build a sewage treatment plant, to have a catalyst in the vehicles, to
declare a certain area to be protected…etc. Demographical that affects the overall state of the workforce
and thus the overall level of labor supply. They may result in a shortage of people with the necessary
professional and qualification requirements, the proportion of the working population, changes in age
structure and the quality of the workforce. Globalization is a global process based on the
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
96
internationalization of the economy, i.e. the interconnection of the markets of different countries
through the trade of goods and services and the free movement of capital. Companies operating in the
world markets are merging to form multinational corporations, aiming to dominate as much of the world
market as possible.
The mentioned external factors are summarized in the table (2.1) below.
Table 2.1: External Factors
External Factors
Legislative regulations for business activities
Provided support in selected areas of business
Amount of tax burden and method of payment of taxes
Labor law
State regulation through legal regulations imposing on
them various obligations to reduce externalities
Development of population employment
Lifestyle and consumer habits of population groups
Demographic composition
Average income and savings rate
Gross domestic product
Climate conditions
Infrastructure of transport, energy and telecommunication
networks
The level of production facilities and the state of
development of science and technology
Globalization (linking markets of different countries
through trade in goods and services and free movement of
capital)
Source: own elaboration.
Internal factors
Internal factors that significantly influence the organization's management concept and human resource
management goals are the following (Koubek, 2006): the size of the organization, where the number of
employees is most often used as an indicator of the size of the organization. In a smaller organization,
communication works more easily in the context of direct personal relationships, the organizational
structure is clear, with fewer hierarchical levels and decision-making is seen by specific people. In a
larger organization it is necessary to create mechanisms for communication within the organization, the
organizational structure has more hierarchical levels, it is necessary to formalize decision-making
processes and delegate powers; technologies used are mainly information and communication systems.
New technologies increase productivity and speed up communication. It brings a change in working
practices, restructure of jobs where staff retraining is needed, increased needs for new types of training,
different skills and abilities are required; organizational structure, when choosing such criteria it is
necessary to take into account criteria such as geographical location, functionality or market segment.
The chosen structure affects the number and specialization of recruited employees, types of training,
remuneration system, motivation, job planning, the need to delegate decision-making powers; The
corporate culture that results from a previous development of an organization affecting the
organization's innovative capabilities and internal relationships. It enables employees to share values,
standards and goals together; economic outcomes where an organization can stimulate the performance
of its employees by directly contributing to profits or by preferentially purchasing primary shares whose
value increases, etc. Due to this, employees more easily share and identify with common values,
standards and objectives. The mentioned internal factors are summarized in table (2.2).
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
97
Table 2.2: Internal Factors
Internal Factors
Proper and effective organizational structure affects
human resource planning
Size of the organization
Human resources management policy and strategic goals
of the organization
Training
Economic outcome can stimulate the performance of its
employees by directly contributing to profits
Devotion of employees
Evaluation systems in the area of human resources
management
Organizational culture
Wage policy
HRM style- leadership
Used technologies especially information and
communication systems
Organizational climate (rigid / flexible, friendly / hostile
climate)
Process of adaptation of new employees Motivation
Source: own elaboration.
Methodology and Data
Research design
The research had adopted the descriptive research design. Descriptive research design aims at launching
the current occurrences, just the way it is as the researcher has no control over the variables. This
research type had been established to be ideal when data are collected to describe persons &
organizations (Creswell, 1994).
Research Instrument
The research instrument used for the survey was a structured disguised questionnaire. As it was the
primary tool for the data collection. Secondary tool for deciding about what factors to mention in the
survey was literatures, websites and annual reports.
The questionnaire contained two sections. The first section contained background information and
personal details of the respondents. The second part of the questionnaire identified various factors that
could have an impact on HRM and evaluated the intensity of this impact. This part contained 31
questions related to external and internal factors, as described earlier. The respondents of the study were
HR professionals or their equivalent. The author preferred these respondents since they are directly
involved in the HRM. The respondents were contacted by e-mail and were asked to choose the most
appropriate answer for each question. The Likert-type scale with 1- 5 items was used. As 1 represented
a very small range of impact, 2 small range of impact, 3 slight range of impact, 4 big range of impact
and 5 as very considerable one.
Reliability Test of The Questionnaire
To measure the internal consistency and the reliability of the survey, we used the Cronbach’s alpha.
Cronbach’s alpha (Andrew, Pederson & McEvoy,2011) measures how well a set of variables measures
a single, unidimensional latent construct. It is essentially a correlation between the item responses in a
questionnaire; assuming the statistic is directed toward a group of items intended to measure the same
construct. Cronbach’s alpha values will be high when the correlations between the respective
questionnaire items are high. C-alpha values range from 0 to 1 and values at above 0.7 are desirable.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
98
The formula (Andrew, Pederson & McEvoy,2011) for Cronbach’s alpha is:
=𝑁. c̄
v̄+(𝑁 − 1). c̄
(1)
Where,
• N = the number of items,
• c ̄= average covariance between item-pairs.
• v̄ = average variance.
Data analysis
For the empirical investigation, the researcher had applied the statistical techniques to analyze the data
collected from the survey. The descriptive statistics was used to draw out the basic features of the data.
As it is known that Likert scale types are classified among ordinal data. However, Pecáková (2011)
defended in her book, that the total score obtained by adding the point expression of individual stimuli,
then represents a one-dimensional scale evaluation of the observed phenomenon. So, based on the
procedure used, the scale obtained can be considered as cardinal. For this reason, in this study, the basic
features will be the mean and the standard deviation. Mean, x̄, is the sum of the values in a data set
divided by the number of values (Jacques, 2013) and it is calculated as follows,
x̄ = 1
𝑛∑𝑥 (2)
Standard deviation is calculated as the square root of variance, where variance measures the spread of
data about the mean (Jacques, 2013),
𝑆𝑡. 𝑑𝑒𝑣 = √1
𝑛∑(𝑥 − x̄) (3)
Also, to investigate the researched hypothesis, a parametric hypothesis testing was employed – oneway
ANOVA or the analysis of Variance. The analysis of variance (ANOVA) (Sahai & Ageel, 2000) models
have become one of the most widely used tools of modern statistics for analyzing multifactor data. The
ANOVA models provide versatile statistical tools for studying the relationship between a dependent
variable and one or more independent variables. The results of the statistic test were calculated by the following ways:
Sum of squares within groups
Sum of squares between groups
Sum of squares between-groups examines the differences among the group means by calculating the
variation of each mean (Yj) around the grand mean (Y) (PDX.edu)
𝑆𝑆𝑎 = n∑ (𝑌𝑗. 𝑌)
2
(4)
Sum of squares within-groups (PDX.EDU) examines error variation or variation of individual scores
around each group mean. This is variation in the scores that is not due to the treatment (or independent
variable):
𝑆𝑆𝑠/𝐴 = ∑(𝑌𝑖𝑗 . 𝑌j)2 (5)
Empirical Results
For the data collection, a sample size of 25 HR employees were selected and the socio-demographical
characteristics of the respondents are represented in the tables below (table 4.1, 4.2, 4.3 & 4.4).
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
99
Table 4.1 Gender of the Respondents
Gender Frequency Percentage
Female 18 72%
Male 7 28%
Total 25 100%
Source: own elaboration.
The results obtained shows that 18 were women, while 7 were men, thus showing a high diversity of the
gender. Sands (2019) has noted in his study, that more than 86% of women in the US work as HR
generalists. Table 4.2 Operating Period
Operating years Frequency Percentage
1 year 5 20%
1 – 5 years 11 44%
6 – 9 years 4 16%
10 years 5 20%
Source: own elaboration.
Most of the HR employees – 44% had been working in the organization from one to five years. Table 4.3 Number of Employees in the Company
No. of
employees
Frequency Percentage
1 – 10 3 12%
11 – 50 3 12%
51 – 250 11 44%
250 8 32%
Source: own elaboration.
Majority of the respondents work for companies that have 51-250 employees. Table 4.4 Business Category
Business
Category
Frequency Percentage
Manufacturing 11 44%
Providing
services
11 44%
Others 3 12%
Source: own elaboration.
The response rate was the same in the manufacturing category as the ones providing services – 44%.
Basic features of the data collected are represented in the tables below and the meaningful features for
each factor (external and internal) are highlighted in grey (tables 4.5 & 4.6).
Table 4.5 Basic features of the external factors
1 2 3 4 5 Mean St.dev Legislative
regulations 1 2 11 8 3 3,4
0,191
Provided support
2 3 11 7 2 3,16
0,205
Amount of tax
burden 2 2 9 7 5 3,44
0,231
Labor law 1 4 2 6 12 3,96
0,254
State
regulations 1 7 5 9 3 3,24
0,225
Employment 2 3 6 5 9 3,64
0,263
Lifestyle and consumer
habits
1 - 8 10 6 3,8
0,191
Demographic composition
1 2 7 7 8 3,76 0,225
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
100
Income & saving rate
- 5 7 10 3 3,44
0,192
GDP 3 5 14 3 - 2,68
0,170
Climate conditions
5 3 8 7 2 2,92
0,251
Infrastructure 3 1 3 12 6 3,68
0,249
Level of production
1 4 7 8 5 3,48
0,224
Globalization 1 2 10 7 5 3,52
0,209
Source: own elaboration.
Table 4.6 Basic features of the internal factors
1 2 3 4 5 Mean St.dev Organizational structure
- 1 3 17 4 3,96
0,135
Size of the organization
2 4 6 10 3 3,32
0,228
HRM policy & strategic goals
1 4 1 13 6 3,76
0,226
Training 1 2 4 9 9 3,92
0,223
Economic outcome
4 6 4 7 4 3,04
0,273
Devotion of employees
- 2 5 8 10 4,04
0,195
Evaluation systems
1 4 5 6 9 3,72
0,248
Organizational culture
- 5 4 9 7 3,72
0,22
Wage policy 1 1 4 8 11 4,08
0,215
HRM style management
1 3 2 13 6 3,8
0,216
Communication systems & IS
1 1 7 11 5 3,72
0,195
Organizational
climate - 2 6 8 9 3,96
0,195
Adaptation 2 4 3 7 9 3,68
0,269
Motivation - 2 3 4 16 4,36
0,198
Source: own elaboration.
The results show us that e.g. the researched factor legislative regulations was evaluated by a mean of
3.4 with standard deviation 0.191. While on the other hand, the basic features for the internal factors are
described in table (4.6). From the table, we can notice that the researched factor e.g. Training was
evaluated by a mean of 3.92 and with standard deviation 0.223. Also, we can notice that the mean in
both cases range between 3 and 4 and that the standard deviation is below 1, which interprets the non-
heterogeneity of the opinions. Also, the mean in most cases is in normal distribution.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
101
In order to test the internal consistency of the survey and the reliability of the scale, the reliability
Cronbach’s alphas were calculated in Excel. The value of Cronbach’s alpha for each of the independent
factors are represented in table (4.7).
Table 4.7 Cronbach’s alphas
Cronbach’s alpha
External Factors 0.991
Internal Factors 0.943
Source: own elaboration.
The value of Cronbach’s alpha for each of the independent factors ranged above 0.9, where the overall
C-alpha was 0,943 for the internal factors and 0.991 for the external factors, showing a consistency of
the acquired data.
For further analysis of the researched data, the analysis of Variance was applied for further evaluation
of the mentioned external and internal factors that have an impact on HRM. While hypothesis testing,
the significance level of the study was on =0.05. Also, the external and internal factors were set as
two independent groups. To reach the partial objective of this research two hypothesis were formulated
as following:
H0: Each evaluated factor within the group has no impact on HRM.
H1: Each evaluated factor within the group has an impact on HRM.
Based on the hypothesis formed, the mentioned external and internal factors were tested separately.
After performing calculations using Excel, the following results were deduced (Table 4.8 & 4.9). Table 4.8 Impact of external factors on HRM
Source: own elaboration.
Table 4.9 Impact of internal factors on HRM
Source: own elaboration.
For the validation of the hypothesis, we compared the p-value with the significance level. According to
the p-values in tables 4.8 & 4.9 are lower than 0.05, this means that there is a statistically significant
difference between groups. Therefore, the H0 hypothesis is rejected with 95% probability. This means
that all external and internal factors have an impact on HRM.
ANOVA Source of Variation SS df MS F P-value F crit
Between Groups 41,0714286 13 3,15934066 3,07808681 0,00025347 1,7506351 Within Groups 330,5 322 1,02639752
Total 371,571429 335
ANOVA Source of Variation SS df MS F P-value F crit
Between Groups 35,1085714 13 2,70065934 2,23458811 0,00818186 1,74935988
Within Groups 406,08 336 1,20857143
Total 441,188571 349
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
102
Discussion
In figure 4.1, we can see the external factors that influence the organization as a whole. It illustrates
various types of external factor and displays Labor law as the biggest influencer. Changes of legislations
enacted by governments will continue to have a dramatic impact on the HRM. In fact, these changes
complicate the life of HR employees by levying detailed and demanding liabilities on employers. Also,
the macroeconomic indicator, employment plays an important role in HRM and expresses the current
situation on the labor market. Meanwhile, the development of indicators like inflation rate or the average
salary have an influence on wage policy, because it must be taken into attention while working on the
yearly salary tariffs. Economic situation such as the fluctuation of GDP levels affect the number of
employees necessitated. In addition, it is essential for the HR manager to know the demographical
development of the country. For instance, the age structure of men and women, the percentage of women
and men in productive age. Demographics can also refer to workforce diversity, where various
promoting programs are to be applied to motivate older and even women workforce. For this reason,
policies and practices of the HR must be adapted to embrace the diversity of the workforce.
Technology’s impact on today’s HR can’t be ignored. Thus, the organizations’ structures have been
redesigned and new programs were instituted for the selection and the evaluation of the employees.
Figure 4.1 External factors influencing HRM
Source: own elaboration.
On the other hand, the internal factors (Figure 4.2) impact the run of the company as whole. Developing
the abilities of the employees through trainings can increase their performance and thus help in reaching
the organization’s strategic goals. Motivation also plays an important role in stimulating the individual’s
performance, i.e. through shares from the profit, benefit systems and other motivating programs.
Interpersonal relationships in the workplace have a great influence on the employee’s psychology.
Building friendly interconnections can stimulate the work satisfaction factor and hence lead to better
outcomes. In addition, the management style that the HR department addresses through the leadership
role can have a big impact on to day-to-day operations. Thus, the organizational climate through friendly
and flexible environment is one of the necessities in the workplace. When talking about hiring new
employees, the HR managers should take into mind the adaptation process of these workforces.
Adaptation plays a big role in the devotion of the employees. A devoted employee is usually loyal to
their organizations and work from their heart, which is great in bringing awesome yields. The HR
strategic policy should be transparent and understood by the employees and should work onto reaching
and fulfilling the organizational goals. When the HR plans its procedures, they have to take into
consideration the size of the company. As bigger the company is, the more are sophisticated the
processes. For this reason, a periodical monitoring of the processes can help figure out obstacles and
work on their improvement. Also, the HR management should take have clearly defined performance
measures of the processes. And should update these Key Performance Indicators to adequate the
circumstances. Observing the outcomes of these KPIs reflects shortcomings and would help onto the
optimization of the daily procedures of the HRM.
1,00
2,00
3,00
4,00
5,00legislative regulations
State regulations
Provided support
Lifestyle and…
Level of production
Labor law
Infrastructure
Income & saving rate
Globalization
GDP
Employment
Demographic…
Climate conditions
Amount of tax burden
External factors influencing HRM
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
103
Figure 4.2 Internal factors influencing HRM
Source: own elaboration.
Conclusion
The main objective of this paper was to identify the factors, i.e. external and internal that influence HRM
and to determine whether they have a significant impact on it.
To know the external and internal factors affecting the HR daily procedures, a structured undisguised
questionnaire was sent to a number of HR employees. The mentioned factors in the questionnaire were
generated based on the literatures, websites and annual reports. HR employees were asked to evaluate
the given factors based on a Likert scale from 1 to 5. Where 1 reflects a very small range of impact, 2
small range of impact, 3 slight range of impact, 4 big range of impact and 5 as very considerable one.
After a thorough comparative analysis, the researcher had to realize the most influential factors through
descriptive analysis features. These features were represented in tables 4.5 & 4.6. Among the external
factors ranged legislative regulations, labor law, demographic characteristics, employment rates, amount
of taxes burden and the way of its payment, lifestyle and the consumer’s habits, macroeconomic
indicators, technological advances, infrastructure of transport, energy and telecommunication networks,
and climate changes. On the other hand, among the internal factors ranged the organizational structure,
culture, climate, HR management style or leadership, the organizational size, trainings, process of
adaptation of new employees, motivation, evaluation systems, HR policies and strategic goals, the
devotion of employees, wage policies, and economic outcomes where the employee has a share from
the profits. Based on the descriptive analysis, the most influential external factors are labor laws,
legislative regulations, demographic changes, tax rates, macroeconomic indicators & technological
changes. The most influential internal factors are trainings, wage policies, motivation, evaluation
systems, organizational climate and the devotion of employees. In order to measure the internal
consistency of the survey and the reliability of the scale, the reliability Cronbach’s alphas were
calculated. The value of Cronbach’s alpha for each of the independent factors ranged above 0.9, where
the overall C-alpha was 0,943 for the internal factors and 0.991 for the external factors, showing a
consistency of the acquired data. In addition to that, the analysis of Variance was tested to explain the
relationship between the independent factors and HRM. The results of the hypothesis testing showed
that more than 50% the mentioned internal and external factors have an influence on HRM.
Taking into consideration all the above mentioned external and internal factors, the organization can
then support the organizations development in a desired direction and to stimulate the performance
measures of the organization.
1,00
2,00
3,00
4,00
5,00Proper and effective…
Size of the organization
Training
Human resources…
Economic outcome…
Devotion of employees
Evaluation systems in…
Organizational culture
Wage policy
HRM style- leadership
Used technologies…
Organizational climate…
Process of adaptation…
Motivation
Internal factors influencing HRM
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
104
Acknowledgement
This research was financially supported within the SGS 2020/33 project.
References
[1] ANDREW, D., PERDERSEN, P., MCEVOY, C. (2011) Research Methods and Design in Sport
Management. Illinois (USA): Human Kinetics. ISBN 9780736073851
[2] ARMSTRONG, M., TAYLOR, S. (2020). Armstrong's Handbook of Human Resource
Management Practice. UK: Kogan Page Publishers. ISBN 9780749498283.
[3] BUDWAR, P., YEHUDA, B. (2003). Career Management Practices: an empirical study,
International Journal of Manpower, 24.
[4] CRESWELL, J. W. (1994). Research Design: Qualitative and Quantitative Approaches. USA:
Sage Publishing. ISBN 9780803952546.
[5] DESSLER, G. (2008). Human Resources Management. New Jersey: Pearson Prentice Hall. ISBN
978-0131746176.
[6] JACAQUES, I. (2013). Quantitative Methods. UK: Pearson Educated Limited. ISBN 978-273
776161.
[7] KOUBEK, J. (2006). Řízení lidských zdrojů: Základy moderní personalistiky. 3.vyd. Praha:
Management Press. IBSN 8072610333.
[8] MAALDERINK, Y. (2005). Human Resource Management: Functions, Applications, Skill
Development. USA: CreateSpace Independent Publishing Platform. ISBN 9781503300118.
[9] MOONEY, P. (2001). Turbo-Charging the HR Function. UK: CIPD Publishing. ISBN
9780852928967.
[10] NASIRIPOUR, A., AFSHAR KAZEMI M., and IZADI, A. (2012). Effect of Different HRM
Policies on Potential of employee Productivity, Research Journal of Recent Sciences, 1(16), pp.
45-54.
[11] OINAS, P., VAN GILS, H. (2001). Identifying Contexts of Learning in Firms and Regions. In:
Paivi Oinas and Hein van Gils, The organizational and Industrial Space. ISBN 978-13151868496.
[12] PDX.edu. Notation and Computation of One-Way ANOVA [online]. [cit. 2020-01-11]. Available
at : < http://web.pdx.edu/~newsomj/uvclass/ho_ANOVA.pdf>.
[13] PECÁKOVÁ, I. (2011). Statistika v terénních průzkumech. Praha: Professional Publishing. ISBN
978-8074310393.
[14] PITRA, Z. (2008). Dovednosti a image managera. 2.vyd. Praha: BIVS. ISBN 978- 8072651306.
[15] PORTER, M. (2008). On Competition. USA: Harvard Business Press. ISBN 9781422126967.
[16] ROWLEY, C., JACKSON, K. (2010). Human Resource Management: The Key Concepts. UK:
Routledge. ISBN 9781136901355.
[17] SAND, B. Why is the HR Profession Dominated by Women [online]. [cit. 2020-01-11. Available
at: https://study.com/blog/why-is-the-hr-profession-dominated-by-women.html
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
105
EVALUATION EFFICIENT PRICE OF COMPENSATION OF SELECTED PUBLIC
TRANSPORT IN OLOMOUC REGION AND MORAVIAN – SILESIAN REGION
Natálie Konečná1
1Department of Public Economics, VŠB – Technical University of Ostrava,
Sokolská tř. 33, Ostrava 70200, Czech Republic
e-mail: [email protected]
Abstract
This paper deals with an evaluation efficient price of compensation of selected public transport in 24
selected areas of Moravian-Silesian Region and Olomouc Region, specifically suburban bus transport.
All used data from MSR are accessible in the National Electronic Tool (NEN), data from Olomouc
Region (OR) are from the Tenderarena electronic marketplace. Tenders have been opened within 2018
– 2019. Contract has been concluded with the chosen carrier in these areas. To estimate technical
efficiency, the output-oriented and input-oriented DEA model was chosen with constant and variable
returns to scale working with one input variable (competitive price for 1 vehicle-kilometer) and two
output variables (number of connections and number of vehicle-kilometers). Based on the performed
analysis, it was found that more than 50 % of contracted compensations was inefficient in the all models.
The degree of inefficiency is very dispersed in the all models (output-oriented and input-oriented
constant returns to scale and variable returns to scale models). Only the outputs that have been used in
the model can be objectively evaluated.
Keywords
bus transport, Data Envelopment Analysis, efficiency, compensation
JEL Classification
C21, C67, R48
Introduction
The paper focuses on the issue of securing transport services in the regions in the context of the
efficiency of public expenditure as price compensation and public services. From the point of view of
goods, public transport belongs to mixed public goods, both in terms of security and financing, and in
terms of consumption. Both public administration (state, regions and municipalities) and the private
sector, which is usually a service provider, are involved in securing public transport. Specifically, the
article focuses on suburban bus transport in the Moravian-Silesian and Olomouc regions.
The obligation to provide transport services at the regional level results from Act No. 194/2010 Coll.,
On public passenger transport services and on amendments to other acts. Transport service is defined
according to § 2 of this Act:„ Transport service means ensuring transport on all days of the week,
primarily to schools and school facilities, to public authorities, to work, to health establishments
providing basic health care and to meeting cultural, recreational and social needs, including transport
back, contributing to sustainable territorial district development.“
The regions are divided into individual areas for the purpose of ensuring transport services. For each
given area, the contracting authority, in this case the region, announces a public contract for the provision
of transport services, thus selecting a specific public service provider. Contracts are concluded for a
period of 10 years, mainly because the longer term of these contracts brings some stability to public
transport and also allows carriers to invest more in the fleet. (Transport plan of the MSK territory for
the period 2017–2021, Transport plan of the Olomouc region)
The regions and municipalities have the right to determine the scope of this service, whether it concerns
the required number of connections or the number of vehicle kilometers traveled, or also by determining
whether the service will be provided by public rail passenger transport or public regular transport or by
their interconnection. This will fulfill the condition under the Public Transport Services Act.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
106
Furthermore, the regions are obliged to create a road transport service plan for at least 5 years, which
should in particular map the provision of public passenger transport services in the given territory, the
extent of the expected compensation and the related issue of public service contracts. Regions may
provide public service on their own or conclude public service contracts for the carriage of passengers
with carriers. If the region, respectively. the contracting authority decides for the tendering procedure
on the basis of which the contract will be concluded; in addition to the aforementioned Act No. 194/2010
Coll., on public passenger transport services and amending other acts, it is governed by Act No.
134/2016 Coll., on public procurement, which further defines the procedure for the award procedure.
The contract with the winning bidder must be concluded in accordance with these laws and with the
European Union legislation. This process is supervised by the Office for the Protection of Competition.
Within the framework of the concluded contracts, the range of transport services of individual parts of
the region is set to meet the condition of transport security pursuant to Act No. 194/2010 Coll., On
Public Services. Due to the market environment, the price of transport performance should be lowered
and thus financial savings will be achieved throughout the region. The paper focuses on 24 areas from
the Moravian-Silesian and Olomouc regions, where the tendering procedure was initiated and at the
same time the public service contract was fulfilled in the period 2018–2019. Individual areas are called
DMUs for the purposes of efficiency evaluation. Specifically, these areas are MSK – Karvinsko,
Orlovsko, Frydlant Region, Novojicin East, Novojicin West, Krnov, Bruntal, Rymarov, Opavsko,
Vitkov, Frydek-Mistek, Bilovecko. From the Olomouc Region are the areas – Olomouc Northeast,
Olomouc Southwest, Prerov North and Lipnicko, Hranicko, Sternberg and Uničovsko, Prerov South,
Litovelsko, Prostejov Northwest, Sumperne North, Sumperne South, Mohelnicko, Zabreh. (Transport
plan of the MSK territory for the period 2017–2021, Transport plan of the Olomouc region)
The aim of the paper is to evaluate the technical efficiency of 10-year compensation of suburban bus
transport in 24 selected service areas of the Moravian-Silesian and Olomouc regions according to
selected inputs and outputs.
Technical efficiency is estimated by an input-output oriented Data Envelopment Analysis (DEA) model.
Two research questions (RQs) are verified for evaluation purposes:
RQ1: Is more than 50% of contracted compensation effective in selected regions?
RQ2: Do contracts in the Moravian-Silesian and Olomouc regions achieve comparable average
efficiency values?
Literature Review
Public transport has been the subject of an evaluation and examination of a number of works which have
received considerable attention in recent years. The problems of transport, especially the efficiency of
spending and its utilization, are dealt with by the authors both in the territory of modern European and
non-European countries.
The price of compensation is a public expenditure, in this case it is an expenditure from the regional
budget. The assessment of technical efficiency in bus transport is also addressed by Hanauerová (2019),
who also uses the DEA model to assess efficiency. Beck and Walter (2013) deal with the factors that
influence the bid price in Germany. He also deals with adequate price compensation in his study
Dementiev (2018). Rosell (2017) addresses cost-effectiveness based on examples of municipalities in
the province of Barcelona, concluding that the smaller a municipality has to provide transport services,
the less efficient it is. Vigren (2018) deals with factors influencing those interested in public transport
services. The study uses the Poisson model and concludes that in Sweden the technical safety
requirements, in particular the bus requirements, are a limiting factor. Mathisen (2016) examines, for
example, Norway whether it is really necessary to select carriers on the basis of public contracts, which
bears some uncertainty for the tenderers with regard to the outcome and conclusion of the contract.
There is also the question of how long public transport will be to the extent required by law. Increasing
the number of cars in practice means less use of this service and with it increasing pressure for efficiency,
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
107
resp. on the use of public transport. In his study Zhang et al. Authors (2019) address the example of six
cities in China whether a policy restricting the purchase and use of cars will affect the development of
public transport. Migliore et al. authors (2013) point out in their study the availability of public transport,
which would increase the efficiency and effectiveness of this public service.
Compensation in bus transport
The selection of the carrier is authorized by the contracting authority, Act No. 194/2010 Coll., On public
passenger transport services and amending other laws. This law implies the possibility to conclude a
contract either by tender or by direct award. When choosing the second option, the conditions arising
from Regulation (EC) No 1370/2007 of the European Parliament and of the Council must be fulfilled.
In the Czech Republic, this regulation is followed by Decree No. 296/2010 Coll., On the procedures for
establishing the financial model and determining the maximum amount of compensation. This decree
regulates the method of construction of the financial model. (IODA, 2015)
Regulation (EC) No 1370/2007 of the European Parliament and of the Council on public passenger
transport services by rail and by road and repealing Council Regulations (EEC) Nos 119/69 and 1107/70
implies that public service compensation means:
"Any advantage, in particular financial, granted directly or indirectly by the competent authority from
public sources during or in respect of the period of implementation of the public service obligation".
Region, resp. the client and also the contracting authority commits in the contract to a certain price
compensation, ie payment for ensuring the transport serviceability of the territory by the selected carrier.
In this case, it is a financial cost that the contracting authority (region) has to pay from its budget to the
carrier for the provision of this public service on the basis of a concluded contract. The price entered
into in the contract arises as a bid price of a particular carrier in the public contract. In the Tender
Documentation, the Client determines in advance the number of connections and the number of vehicle
kilometers required for the given location. offer price. By concluding such a contract, the selected
tenderer undertakes to perform within the given scope for the competitively priced price. The winner is
the carrier whose price is the lowest, but not unreasonably low. In this case, the participant would be
excluded. The selected carrier provides a public service through its technical equipment and its staff.
Based on the fulfillment of the contracting authority's requirement, ie the required number of
connections and the number of vehicle kilometers traveled, the area will be managed by public transport.
In this way, the region guarantees to ensure transport accessibility in the given locality and thus fulfills
the obligation arising from the law. In general, price compensation can be understood as a subsidy from
public budgets, ie from the contracting authority. (Hanauerova, 2018).
Act No. 194/2010 Coll., On public passenger transport services and amending other acts, stipulates that
the amount of compensation must be reasonable, otherwise the client may not conclude the contract.
Should such a contract still be concluded, the contract shall be null and void. Prior to the conclusion of
the contract, in the case of direct award, the selected carrier is required to submit a financial model of
costs, revenues and net income. Similarly, the selected carrier shall submit the financial model for the
tender before signing the contract, unless otherwise specified in the tender dossier. (IODA, 2015)
Methodology and Data
The essence of the DEA method lies in the division of surveyed objects into efficient and inefficient
according to the size of consumed resources and the amount of outputs (production). The solution of
DEA models defines empirical production function. Jablonský and Dlouhý (2015) identify output-
oriented models as output oriented models. In the case of minimizing the value of inputs, again while
maintaining the value of the outputs, we are talking about input oriented models. The combination of
these two options creates an additive models, slack – based models.
In output-oriented models, output variables can be used to determine efficiency. Such models calculate
the technical efficiency coefficient, which is determined by the ratio of the weighted sum of inputs to
the weighted sum of outputs, but weights are sought so that the value of the coefficient g is greater than
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
108
or equal to one. Thus, for the effective unit Uq the coefficient g = 1 and for the inefficient unit g> 1. In
order to make the inefficient units effective, it is necessary to increase some or all of the outputs
(Klieštik, 2009).
In the input-oriented models, the effective units within the comparison group are those with a coefficient
value equal to one (g = 1). Within this homogeneous group, units that are less than one (g <1) are
ineffective. This value is then a feedback to improve inputs so that the inefficient unit becomes an
effective unit (Klieštik, 2009).
Reaching the effective threshold for these models is possible in the following ways:
• increasing the value of output consumed while maintaining current input levels – output oriented
models;
• reducing the value of input consumed while maintaining current output levels – input oriented
models;
• a combination of both approaches – additive models, slack-based models. (Jablonsky, Dlouhy,
2015; Vrabkova, Vankova, 2015)
The estimation of the technical efficiency of contracted compensation in suburban bus transport in the
conditions of 24 regions of the Moravian-Silesian and Olomouc regions was carried out according to
the following procedure:
- definition of one input and two outputs for estimating technical efficiency, statistical description
of selected variables (see Table 1),
- calculating an output-oriented efficiency model according to the DEA model, which takes into
account the constant range yield (CCR), according to formula (1);
- calculating an output-oriented efficiency model according to the DEA model, which takes into
account the variable range yield (BCC) formula (2);
- calculating an input-oriented efficiency model according to the DEA model, which takes into
account the constant range yield (CCR), according to formula (3);
- calculating an input-oriented efficiency model according to the DEA model, which takes into
account the variable range yield (BCC), formula (4);
The output-oriented CCR model can be formulated as follows:
Minimalize: g =∑ vjmj xjq, (1)
Under conditions ∑ uiri y
ik ≤ ∑ vj
mj xjk, k = 1, 2, …, n
∑ uiri y
iq = 1,
ui ≥ ε i = 1, 2, …, r,
vj≥ ε, j = 1, 2, …, m.
The output-oriented BCC model can be formulated as follows:
Minimalize: g =∑ vjmi xjq+ v, (2)
Under conditions ∑ uiri y
ik ≤ ∑ vj
mj xjk + v, k = 1, 2, …, n,
∑ 𝑢i𝑟i xiq = 1,
ui ≥ ε, i = 1, 2, …, r,
vj≥ ε, j = 1, 2, …, m,
v – free.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
109
The input-oriented CCR model can be formulated as follows:
Maximalize: 𝑧 =∑ 𝑢iri y
iq, (3)
Under conditions ∑ uiri y
ik ≤ ∑ vj
mj xjk, k = 1, 2, …, n,
∑ vjmj xjq, = 1,
ui ≥ ε i = 1, 2, …, r
vj≥ ε, j = 1, 2, …, m.
The input-oriented BCC model can be formulated as follows:
Maximalize: 𝑧 =∑ 𝑢iri y
iq+ 𝜇 , (4)
Under coditions ∑ u𝑖𝑟𝑖 𝑦𝑖𝑘 + 𝜇 ≤ Σ𝑗
𝑚 𝑣𝑗𝑥𝑗𝑘 , k = 1, 2, …, n,
∑ vjmj xjq, = 1,
ui ≥ ε i = 1, 2, …, r,
vj≥ ε, j = 1, 2, …, m,
𝜇 − any
In order to achieve the goal, the input and output oriented DEA model is chosen in the paper. In both
models, the competitive price is chosen as the input variable, the output side is formed by two variables,
namely the number of connections and the estimated number of vehicle kilometers traveled.
Input
X1 – competitive price (CZK / vehicle). The competitive prize is a price compensation for carriers for
1 mileage covered by the contract. This data is available in the National Electronic Instrument (NEN)
for MSK procurement. The Olomouc Region publishes public tenders in the electronic marketplace
TENDERARENA – electronic tool eGordion.
Outputs
Y1 – number of connections requested by the contracting authority (Moravian-Silesian and Olomouc
regions). Each contracting authority determines more detailed specification of connections for the
competition area itself. Requirements for the number of connections are specified in the Tender
documentation, which is always available in the electronic tools of the contracting authorities.
Y2 – the estimated number of vehicle kilometers traveled in 10 years at the location. This is the
anticipated number of vehicle kilometers traveled, which is needed to fulfill the obligation to serve the
given territory. The Contracting Authority also specifies this requirement in the Tender Documentation
for a specific area. The documents are available in electronic tools for public procurement of the
respective regions.
Table 1. Statistical characteristics of input and output
X1 – competitive
price (CZK/vkm)
Y1 – number of
connections
Y2 – vkm/10 years
Min. 30.46 9 8 350 812
Max. 38.14 43 39 383 743
Mean 35.37 18,08 18 061 082
Median 35.55 16 15 851 699
SD 1.95 8 7 900 933.6
Source: Custom processing.
From Tab. 1, it is evident that the lowest competition price in the compared regions was CZK 30.46 /
1change for the Sternberg and Uničov regions of the Olomouc Region. On the other hand, the contract
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
110
with the highest price (CZK 38.14 / 1vkm) was concluded in MSK for the Frydlant Region. The average
value of the contested price from 24 regions in these regions is CZK 35.37 / vkm. The median or mean
value of the input variable is 35.55 CZK / vkm The standard deviation of MSK input is 1.95.
On the output side, two variables are selected, namely the number of connections and the estimated
number of vehicle kilometers traveled within the time horizon of 10 years. From Tab. 1 shows that the
lowest number of required connections is 9, this number is the same for both regions. In the Moravian-
Silesian Region it is the Karviná region, in the Olomouc region it is the Přerov South region. The highest
number of connections (43) is requested by the contracting authority for the Novojičínsko East in the
Moravian-Silesian Region. The average number of connections is 18.08 connections per area. The
median of this selected output is 16. The standard deviation of the output (Y1) is 8.
The second selected output is defined as the number of vehicle kilometers traveled by the client in the
given location. The lowest number of vehicle kilometers is required for the Frydlant Region (8 350
thousand vehicle-kilometers / 10-years) in the Moravian-Silesian Region. The average mileage is 18
061 ths. Vkm/ 10 years for the site. The mean value for the number of driven kilometers is 15 851 ths.
vkm/10years. The standard deviation of the second variable on the output side (Y2) is 7 900 thous.
vkm/10years.
Empirical Results
Results of output oriented models
The results of the calculation of the efficiency of the Output Oriented Constant Revenue Model (CRS)
model show that out of 24 DMUs, only one procurement is effective – Novojičínsko východ (DMU15).
Contract concluded between the carrier – ČSAD Vsetín, a.s. and the region guarantees the provision of
43 connections and 39 383 thousand vehicle-kilometers / 10-years with the agreed price compensation
of CZK 37.86 /vkm.
In terms of efficiency, the contract for the Opava region (DMU18) is also well based, where it is
contractually agreed to secure 34 connections and 35 559 thousand vehicle-kilometers / 10-years with
compensation of CZK 37.01 / vkm. ČSAD Vsetín, a.s.
On the other hand, a public contract for the Frydlant Region (DMU22) is based on a very ineffective
model, where 12 connections and 8 350 thousand vehicle-kilometers / 10-years are needed for the
management of the area. The price compensation amounts to CZK 38.14 / vkm, and this contract was
re-announced because the price for 1chokokm was exceeded in the first contract for passenger transport
in this locality. This may be one of the reasons why the public service contract in the area is awarded
with the highest price compensation, being the smallest number of vehicle kilometers required and the
second smallest number of connections (12). In both cases, only one participant (ČSAD Frýdek-Místek
a.s.) entered the public contract, which also became the winning carrier for the given territory.
In the field of inefficiency, in addition to the already mentioned Frydlant Region, where the tender was
announced repeatedly, there are also contracts for the Olomouc Southwest (DMU2), Lithuania (DMU7),
Šumperk South (DMU10) in the Olomouc Region and Krnovsko (DMU19) in MSK. Vojtila Trans s.r.o.
won the public contract for the Olomouc region, which committed itself to the management of the
territory with 12 connections and 10 369 thousand vehicles / 10 years for the price of CZK 35.76 / vkm
Three companies entered the tender for the Litovel region, the winner was the carrier ARRIVA
MORAVA a.s. In this locality there are 13 connections, 10 783 thousand driven km / 10 years at an
agreed price of 36.20 CZK / 1 km. As already mentioned, for the operation of the Šumperk South region,
the Olomouc Region had to announce a public contract repeatedly. Three carriers entered this tender,
the winning bidder was ARRIVA MORAVA a.s. The carrier manages the territory on the basis of a
request from the Tender Documentation 13 links, 11 278 thousand vehicle-kilometers / 10-years for the
agreed price of CZK 34.70 / 1-carriage (Figure 1).
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
111
Figure 1. Results of the output-oriented model with constant and variable returns to
scale
Source: own elaboration.
A comparative view of the results-oriented CRS and VRS models is shown in Tab. 2. According to the
output-oriented model with constant scale returns, only the public service contract for the Novojičínská
East (DMU15) is effectively awarded. In contrast, in the output-oriented model with variable returns
from the range, three public contracts are effectively awarded, namely from the Olomouc Region, a
transport service order from Šternberk and Uničovsko (DMU5) and Prostějov Northwest (DMU8). ).
This unit achieves efficiency in both selected models. On the other side of the selected scale there are
those orders that were awarded inefficiently. Based on the results of both output-oriented models, public
tenders for the provision of transport services to the Olomouc South-West (DMU2) and Frýdlant
(DMU22) are awarded inefficiently. According to the output-oriented model with constant yields from
the range, public contracts for the provision of transport services in the Litovelsko (DMU7), Šumperk
South (DMU10) areas from the Olomouc region are also ineffective.
Table 2. Summarized results of efficiency modeling of output-oriented model with CRS
and VRS
CRS VRS
Table g number DMU(s) number DMU(s)
[1] 1 D15 3 D5, D8, D15
[1,001 – 1,499] 3 D8, D18, D23 3 D14, D18, D23
[1,500 – 1,999] 5 D5, D12, D14, D16, D24 8 D1, D3, D4, D6, D12, D13,
D16, D24
[2,000 – 2,499] 8 D1, D3, D4, D6, D9, D13,
D20, D21
5 D9, D11, D17, D20, D21
[2,500 – 2,999] 2 D11, D17 3 D7, D10, D19
[3,000 +] 5 D2, D7, D10, D19, D22 2 D2, D22
Source: own elaboration.
Results of input-oriented models
The DMU15 (Novojičínsko East) is based on all 24 comparison units in the input-oriented model with
a constant return on scale. This area was also most effective in the output-oriented model, both with
constant and variable scale returns. Other units achieve different inefficiencies. Public tenders for the
Frydlant Region (DMU22) in the Moravian-Silesian Region and also the Olomouc Southwest (DMU2)
in the Olomouc Region are ineffective. Also in these units, inefficiencies in the previous output oriented
Olomoucko JZ; 3,385 Litovelsko; 3,163
Šumpersko J II; 3,032 Krnovsko; 3,144
Frýdlantsko II; 3,610
Olomoucko JZ; 3,063
Šternbersko; Uničovsko; 1,000
Prostějovsko SZ; 1,000Novojičínsko východ;
1,000
Frýdlantsko II; 3,583
0,000
1,000
2,000
3,000
4,000
0 5 1 0 1 5 2 0 2 5 3 0
CRS VRS
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
112
model (CRS and VRS). The average value of the efficiency of the input-oriented model with constant
returns from the range is 0.515. The standard deviation is 0.189.
Similarly, the results of an input-oriented model with variable yields from a range can be described. In
this model, as well as the output-oriented variable yield model, DMU5 – the Sternberg and Uničovsko
regions, DMU8 – the Prostějov northwest region and DMU15 – the Novojičínsko east region are
effective units. It can be stated that this model shows efficiency less dispersed compared to the entry-
oriented CRS model. Yet the least effective one is DMU22 – the Frydlant Region. This unit comes out
inefficiently in all the mentioned models. The average efficiency of units in the variable yield model
range is 0.892. The standard deviation of this model is 0.057 (Figure 2).
Figure 2. Results of the input-oriented model with constant and variable returns to scale
Source: own elaboration.
The comparison of the input-oriented model with the constant and variable yields of the range proves
that in the model with the constant yields of the range the efficiency of individual units is significantly
dispersed. The largest number of units, 13 in total, is in the CRS model on a range of values 0.49 – 0.3.
Only the DMU15 unit – the Novojičínsko East area – comes out effectively. In contrast, in the variable
yield model, the largest number of units (20) is on a scale in the range of 0.99 – 0.8 (see Table 3).
Olomoucko JZ; 0,295 Frýdlantsko II; 0,277
Šternbersko; Uničovsko; 1,000 Prostějovsko SZ; 1,000
Novojičínsko východ; 1,000
Frýdlantsko II; 0,799
0,000
0,200
0,400
0,600
0,800
1,000
1,200
0 5 1 0 1 5 2 0 2 5 3 0
CRS VRS
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
113
Table 3. Summarized results of efficiency modeling of input-oriented model with CRS
and VRS
CRS VRS
Table g number DMU(s) number DMU(s)
[1] 1 D15 3 D5, D8, D15
[0,99 – 0,8] 2 D18, D23 20 D1, D2, D3, D4, D6, D7, D9,
D10, D11, D12, D13, D14,
D16, D17, D18, D19, D20,
D21, D23, D24
[0,79 – 0,7] 1 D8 1 D22
[0,69 – 0,5] 5 D5, D12, D14, D16, D24 0
[0,49 – 0,3] 13 D1, D3, D4, D6, D7, D9,
D10, D11, D13, D17, D19,
D20, D21
0
[0,29 – 0] 2 D2, D22 0
Source: own elaboration.
Conclusion
It is worth noting that the number of passenger cars is increasing and people are using this mode more
often than in the past. This is one of the reasons why there is less and less interest in public transport.
However, this mode of transport is also justified.
As for the region, as a contracting authority, the Act No. 194/2010 Coll., On public passenger transport
services and amending other laws, requires the provision of basic transport accessibility, it is also no
less important factor in spending funds to provide this public service.
The aim of the paper was to evaluate the technical efficiency of 10-year compensation of suburban bus
transport in 24 selected service areas of the Moravian-Silesian and Olomouc regions according to
selected inputs and outputs.
The Data Envelopment Analysis model was used to achieve the goal. In the verification of the research
question RQ1: “Is more than 50% of contracted offsets effective in selected regions?” The efficiency
calculation showed that in both cases the input / output oriented CRS and VRS models are more than 50
% of contracted offsets ineffective. For research question RQ2: “Do contracts in the Moravian-Silesian
and Olomouc regions achieve comparable average efficiency values?” According to the efficiency
calculation in the output-oriented DEA model, with both constant and variable yields, contracts
concluded in the Moravian-Silesian Region achieve higher efficiencies. The results of the input-oriented
model with constant returns from the scale show that even in this case contracts in MSK have a higher
efficiency. In contrast, in an input-oriented model with variable returns from scale, the efficiency in both
regions is comparable.
Objectively, it is not recommended to increase the number of joints. to require more vehicle kilometers,
as this could be inefficient in terms of unused line capacity. For example, the Frydlant Region itself is
a mountainous area with a large number of small municipalities where the inhabitants are forced to
commute to work in larger neighboring towns. In order to make inefficient public procurement effective,
the output would have to be increased. The required number of connections and the number of vehicle
kilometers is determined by the client. This implies that the contracting authority itself should know
how many connections and vehicle kilometers are needed to ensure the serviceability of the territory.
Thus, increasing outputs could be inefficient in terms of the use of this public service by citizens and it
would be completely unnecessary for such an increase to occur. The price also reflects the mountainous
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
114
terrain, population, respectively. potential passengers, distance of individual stops as well as the size of
the managed area.
The selection of a particular carrier is carried out in accordance with Act No. 134/2016 Coll., On Public
Procurement, based on which a contract is concluded with the selected (winning) tenderer. Tender is
concluded with the carrier who submitted the most advantageous offer from the perspective of the
region, ie the contracting authority. The best bid is one that brings the lowest price compensation for the
contracting authority's requirements in the Tender Documentation, so-called competition for the lowest
price occurs. As the payment for the public service comes from the region's budget, it is therefore
possible to consider as effective the price compensation that is as low as possible but not unreasonably
low for servicing the region's territory. Should the region still conclude a contract with such a carrier,
there would be a violation of Act No. 134/2016 Coll., On public procurement. In this case, the Office
for the Protection of Competition (ÚOHS) would invalidate the public contract and the concluded
contract.
The DEA model was used to estimate the technical efficiency. The input side consists of one variable –
the competitive price, on the output side two variables are selected – the required number of connections
and the expected number of vehicle kilometers traveled. The estimation of technical efficiency was made
on the basis of 24 homogeneous units, which in fact represent individual areas that form the territory of
the entire region. At these locations, the carrier was currently selected for the upcoming 10-year period.
In both regions it is the same number of 12 localities where the contract was concluded and the obligation
was fulfilled.
When estimating the technical efficiency of both input-output and output-oriented models with constant
yields, a public contract for the Novojičínsko East (DMU15) region is based on an efficient contract. In
both variable yield scale models, in addition to the already mentioned DMU15, tenders were estimated
for the Sternberg and Uničov regions (DMU5) and Prostějov Northwest (DMU8). When estimating the
technical efficiency of output-oriented CRS and VRS models, efficiency is very scattered. In the input-
oriented constant range yield (CRS) model, the efficiency values are also very dispersed. In contrast, in
the input-oriented variable yield model, 20 units are on the g scale in the range of 0.99 – 0.8. The least
effective unit of this model (DMU22 – Frydlant Region II) is 0.799. However, when estimating technical
efficiency, the public contract for the Frýdlant II area is completely ineffective in all models. The reason
for concluding a contract with such price compensation for certain requirements may be the fact that the
contract was awarded for the second time.
All efficiency results achieved through the DEA model are limited both by input and output selection
and by the number of units to be compared (DMU). Furthermore, the results affect the effective and
inefficient units within the model.
References
[1] Act No. 194/2010 Coll. on public passenger transport service, as amended.
[2] Act No. 111/1994 Coll., on road transport.
[3] Act No. 134/2016 Coll., on public procurement, as amended.
[4] Beck, A., Walter, M. (2013). Factors affecting tender prices in local bus transport: Evidence from
Germany. Journal of Transport Economics and Policy, 47 (PART2), pp. 265-278.
[5] Coelli, T., J., Prasada Rao, D. S., O´donell, CH. J., Battese, G. E. (2005). An Introduction to
Efficiency and Productivity Analysis. New York: Springer Science.
[6] Cooper, W. W, L. M Seiford, Tone, K. (2007). Data Envelopement Analysis, Springer, New York
[7] Dementiev, A. (2018). Contracting out public transport services to vertical partnerships. Research
in Transportation Economics, Vol. 69, pp. 126-134.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
115
[8] Hanauerová, Eliška. (2018). Optimalizace kritérií veřejných soutěží v hromadné dopravě osob.
Disertační práce. 2018. MASARYKOVA UNIVERZITA, Ekonomicko-správní fakulta, Brno.
[9] Hanauerová, E. (2019), Assessing the technical efficiency of public procurements in the bus
transportation sector in the Czech Republic. Socio-Economic Planning Sciences.Vol. 66, pp. 105-
111.
[10] IODA (2015). Veřejná doprava v České republice. IODA Informace pro dopravní analýzy.
Fakulta dopravní ČVUT.
[11] Ivan, I. (2010). Advantage of caropooling in comparison with individual and public transport case
study of the Czech Republic. Geographia Technica [online]. 2010, 9(1), pp. 36-46 [cit. 2019-12-
20]. ISSN 18425135.
[12] Jablonský, J., Dlouhý, M. (2015). Modely hodnocení efektivnosti a alokace zdrojů. Praha:
Professional Publishing.
[13] Kliestik, T. (2009). Kvantifikácia efektivity činností dopravných podnikov pomocou data
envelopment analysis. E a M: Ekonomie a Management. 12. 133-145.
[14] Kleprlík, J. (2013). Analýza plánů dopravní obslužnosti krajů a návrhy změn plánů dopravní
obslužnosti kraje. Perner´s contacts.
[15] Mathisen, T.A. (2016). Competitive tendering and cross-shareholding in public passenger
transport. Transport Policy, 48, pp. 45-48.
[16] Migliore, M., Lo Burgio, A., Maritano, L., Catalano, M., Zangara, A. (2013). Modelling the
accessibility to public local transport to increase the efficiency and effectiveness of the service:
The case study of the Roccella area in Palermo. WIT Transactions on the Built Environment, Vol.
130, pp. 163-173.
[17] MINISTRY OF TRANSPORT. Yearbook 2017. [online]. [cit.2019-12-15]. Available from
https://www.sydos.cz/cs/rocenka-2017/index.html
[18] MORAVSKOSLEZSKÝ KRAJ (2019). MSR Plan of transport services for 2017 – 2021
[online].[cit.2019-07-05]. Available from https://www.msk.cz/cz/doprava/plan-dopravni-
obsluznosti-uzemi-moravskoslezskeho-kraje-40792/
[19] NATIONAL ELECTRONIC TOOL. Moravskoslezský kraj – Seznam uzavřených zadávacích
postupů. [online]. [cit.2019-12-20]. Available from
https://nen.nipez.cz/SeznamPlatnychProfiluZadavatelu/MultiprofilZakladniUdajeOZadavateliM-
20523824/SeznamUzavrenychZadavacichPostupu-20523824
[20] OLOMOUCKÝ KRAJ (2019). Olomouc Region Plan of transport services. [online].[cit.2019-12-
05]. https://www.kr-olomoucky.cz/plan-dopravni-obsluznosti-uzemi-olomouckeho-kraje-
aktuality-632.html
[21] Rosell, J. (2017). Urban bus contractual regimes in small- and medium-sized municipalities:
Competitive tendering or negotiation? Transport Policy, 60, pp. 54-62.
[22] TENDERARENA. Olomoucký kraj. [online]. [cit.2019-12-15]. Available from
https://www.tenderarena.cz/profil/detail.jsf?identifikator=Olomouckykraj
[23] Vigren, A. (2018). How many want to drive the bus? Analyzing the number of bids for public
transport bus contracts. Transport Policy, 72, pp. 138-147
[24] Vrabková, I., Vaňková, I. (2015). Evaluation Models of Efficiency and Quality of Bed Care in
Hospitals. Ostrava: VŠB-TUO.
[25] Zhang, L., Long, R., Chen, H.). (2019 Do car restriction policies effectively promote the
development of public transport? World Development, 119, pp. 100-110.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
116
EVALUATION OF CSR DISCLOSURE OF THE BIGGEST COMPANIES IN CZECH
REPUBLIC WITH MCDM METHODS
František Konečný1
1Department of Economics, VŠB – Technical University of Ostrava,
Sokolská tř. 33, Ostrava 70200, Czech Republic
e-mail: [email protected]
Abstract
Corporate Social Responsibility (CSR) is a concept focusing on the future of sustainable business
environment. Sustainability is a very popular keyword and CSR is theory following this umbrella term.
As a practical concept, CSR is however limited in several ways. The predominantly qualitative
conception of the topic makes it difficult to comprehensively assess the level of CSR in a specific
company. The aim of this paper is to evaluate the social responsibility of selected companies using
multi-criteria decision-making methods AHP and TOPSIS. In addition, the evaluation results are used
for comparing selected companies and ranking.
Keywords
Corporate social responsibility, Corporate social performance, Analytic Hierarchy Process, TOPSIS method
JEL Classification
M14, M40, Q56, C38
Introduction
In the last two decades, sustainability or sustainable development has been a frequently used term in
various fields in the academic environment, but also in the private business sector, both nationally and
transnationally. In 2019, the European Commission explicitly sets out its strategies, objectives and
guidelines for sustainable future development. Although sustainability and corporate social
responsibility (CSR) are not necessarily the same, both approaches are strongly forward-looking. There
are several studies that contribute to understanding the long-term benefits of CSR. Although CSR is
partially current trend, the term itself has undergone a long evolution since its major expansion in the
1960s and 1970s. Later, the focus shifted from defining CSR to more practical approaches such as
corporate social performance (CSP) and corporate financial performance, stakeholder theory, business
ethics or other alternative frameworks (Carroll, 1999; Carroll, Schmidt, Rynes, 2016).
The aim of this work is a comprehensive comparison of the social responsibility activities of the five
largest companies in the Czech Republic by turnover. Corporate Social Responsibility (CSR) reports or
other available information from the official sources of selected companies are used for this evaluation.
The method for the evaluation of individual CSR programs and the subsequent comparison is the multi-
criteria decision-making AHP - Analytical hierarchical process and the TOPSIS method.
Modern Corporate Social Responsibility
Recent developments in the theory of CSR are taking place especially on the political field. CSR
standards for businesses are not only created by governments and intergovernmental organizations (EU),
but companies themselves are more involved in political processes by providing public goods. This
makes companies the political actors who shape their institutional environment (Rasche, 2015, Scherer
and Palazzo, 2011). Involvement in CSR activities is essential in multinational corporations with respect
to supply chains that can spread across many countries. These companies usually include standards,
codes and norms with appropriate auditing, while experts are still striving for large corporations to take
a more proactive approach by involving their stakeholders and partners (Quarshie et al., 2015). On the
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
117
other hand, small and medium-sized enterprises (SMEs) make up a large part of the world economy, but
their involvement in CSR activities is currently not well understood. (Scherer et al., 2016).
Campbell et al. (2007) claim that businesses are inclined to integrate CSR activities when certain
economic conditions of the company are met. Simply, companies will act socially responsibly when
their financial and economic health is in good shape. However, there is a need for an institutional
environment that promotes socially responsible behaviour.
Multicriteria decision making methods in CSR
In the following part will be described selected methods of multi-criteria evaluation, which are used in
the practical part for analysis and evaluation of companies according to their social responsibility
(performance). Both methods are described by Saaty (2013), Tryantaphyllou, (2000) and Hwang, Yoon
(1981).
Analytical hierarchy process - AHP
The procedure of this method is based on the distribution of a multi-criteria problem into a system of
levels. The problem is therefore solved at multiple levels, creating a hierarchy. The main step of AHP
is dealing with the structure of the 𝑚 × 𝑛 matrix (where 𝑚 is the number of alternatives and 𝑛 is the
number of criteria). The matrix is created using the relative importance of the variants for each criterion.
The vector (𝑎𝑖1, 𝑎𝑖2, 𝑎𝑖3,…; 𝑎𝑖𝑛) for each i represents its own vector 𝑛 × 𝑛 of the reciprocal matrix,
which is determined by pairwise comparison of the impact of 𝑚 variants on the i-th criterion. An
important feature of this method is therefore also the use of the Saaty paired comparison method for the
determination of criteria weights.
AHP hierarchy for the specific case of five factors and five chosen companies for this paper is in the
Figure 1 below.
Figure 3 Corporate social responsibility in AHP model
Source: Own creation
TOPSIS
TOPSIS was developed in 1980 (Yoon and Hwang) as an alternative to other multi-criteria decision-
making methods. The basic concept of this method is that the chosen alternative should have the shortest
distance from the ideal solution and the greatest distance from the negative-ideal (basal) solution. The
TOPSIS method assumes that each criterion tends to increase or decrease monotonously the usefulness.
For this reason, it should be easy to define the ideal and negative ideal solution. The Euclidean distance
approach was determined to evaluate the relative distance of variants to the ideal solution. Thus, the
order of preference consists of a series of comparisons of these relative distances.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
118
Evaluation method
To objectively asses Corporate Social Performance with MCDM methods, it was needed to code
qualitative factors into the transparent quantitative solution. Companies CSR programmes were
evaluated according to the five factors (criteria) and scored with number (0 – worst, 4 – best). To
understand, what factors were used, there is a Table 1 below with text-based explanation of the factors.
Table 1 Evaluation criteria for AHP and TOPSIS method
Criteria Evaluation
Structure (K1)
There is a coherent report on corporate social responsibility activities. The report
contains a clear and logical summary of all parts of the entire report (Executive
summary), the overall report is logically organized and contains relevant
information. The structure of the report does not lack essential data and important
parts (e.g. introductory word, overview, content, three pillars of activities, GRI
methodology, quantitative data and indicators, or explanatory notes and
comments). The report is clear and readable by any interested party, with
graphical or other data facilitating legibility.
Stakeholders
(K2)
The report contains CSR activities towards all stakeholders. There should also be
activities towards employees, suppliers, customers, local communities or the
wider environment, the state, or the transnational community. The data are
expressed in a clear way and the documented data can be largely verified (e.g. due
to quantitative data).
Quantitative
indicators (K3)
The report contains a large amount of quantitative data and indicators to measure
specific CSR activities. The quantitative data is organized into a clear and simple
structure. The report includes specific social performance activities with
measurable data and indicators of their measurement, as well as an explanation of
the calculation procedure. Data can be verified or traced by documented source.
The data was also verified by an independent evaluator and added comments.
Triple bottom
line (K4)
The report is structured and contains information on the economic, social and
environmental pillar of corporate social responsibility. The report should not omit
activities towards stakeholders in any pillar. There is a large amount of data and
information on all three pillars. All categories are clearly and clearly structured
for any stakeholder group.
GRI
methodology
The report is prepared according to GRI methodology and contains all its
significant parts.
Source: Own creation
Selected companies for comparison
Within the practical application of the AHP and TOPSIS methods, five largest companies were selected
according to their turnover in 2018 in the Czech Republic. Companies are: ŠKODA, ČEZ, EPH,
AGROFERT, UNIPETROL. A thorough analysis of CSR reports (sometimes also referred to as
sustainability report) was carried out for these companies. If a separate CSR report was not found for
the selected company, information from the company's website or annual report was used for the
evaluation. More detailed information about selected companies, evaluation of separate reports,
subsequent determination of weights by the Saaty method and determination of the overall ranking in
the quality of analyzed reports by the AHP and TOPSIS methods are given in this chapter below.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
119
Results of MCDM methods
In this section specific results of previously described process are presented. It is important to firstly
present computed weight priorities in the selected criteria.
Computed criteria weights
According to the Saaty method, weights of the selected criteria were determined. A scale from 1 (least
important) to 3 (most important) was used to assess the importance of specific criteria for the resulting
weights. After the evaluation of the criteria, the first criterion Structure (K1) was rated grade 2,
Stakeholders (K2) also grade 2. Quantitative indicators (K3) were awarded the highest grade 3, Triple
bottom line (K4) were rated 1 and GRI Methodology (K5) 3 points. After pairing with preferences with
respect to the selected degrees of importance, the following table was created, which also contains the
resulting weighting criteria.
Table 2 Computed criteria
Criteria Priority Weights (in %)
Structure (K1) 13,3
Stakeholders (K2) 13.3
Quantitative indicators (K3) 35,4
Triple bottom line (K4) 6,2
GRI methodology (K5) 31,9
Source: Own creation
The K4 (Triple bottom line) criterion came out as the least important from a paired comparison with a
weight of 6.2%. The first (Structure) and the second criterion (Stakeholders) are divided equally with
13.3% importance. The K5 criterion (GRI methodology), according to the results of the Saaty method,
is the second most important criterion with 32.9%. The highest weight is assigned to criterion K3
(Quantitative indicators) with a value of 35.4%.
Reporting quality evaluation according to the chosen methods
After the weights were determined, the CSR reports of selected companies or other available information
related to reporting on CSR (e.g. partial reports, information from the websites of companies) were
evaluated according to the criteria. Two methods of multi-criteria decision making, AHP and TOPSIS
were used for comprehensive evaluation of this information. The application of methods with partial
end results are presented in this chapter
Analytical hierarchy process - AHP
CSR reports of selected companies and other additional information were evaluated for each company
separately by the corresponding quality level from the 0 to 4 scale. Each criterion is presented separately
with partial results, then the overall result of the used method and the resulting order of variants
(companies) are analyzed. Five criteria (designated K1, K2, ..., K5) and 5 variants (designated as V1,
V2, ..., V5) were chosen for the hierarchical system of the AHP method.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
120
All five variants were evaluated according to selected criteria in individual criteria. Following a
comprehensive evaluation of the quality of the CSR reports, the order of the options was presented as
shown in Table 3 below.
Table 3 Final results of the AHP method
Rank Company Final score (in %)
1. ČEZ (V2) 33,8
2. ŠKODA (V1) 28,8
3. EPH (V4) 27,1
4. AGROFERT (V3) 5,4
5. UNIPETROL (V5) 4,9
Source: Own creation
According to the analytical hierarchical process, the second ranking (CEZ) came out of the final ranking,
with a final value of 33.8%. With a minimum difference of 1.7%, V1 (ŠKODA) with 28.8% and V4
(EPH) with 27.1% rank second and third. In all three variants, important quality factors in the individual
criteria were met. Only minor differences determined the resulting order of variants. The remaining
variants V3 (AGROFERT) and V5 (Unipetrol) showed significant shortcomings in several criteria and
stay behind overall with 5.4% V3 in the fourth and 4.9% V5 in the last place.
TOPSIS The overall evaluation of the quality of reporting activities of selected companies was also carried out
using the TOPSIS method. The TOPSIS multi-criteria decision-making method looks for the solution
closest to the ideal variant. The same starting weights from the Saaty method were used to compare the
method. The criteria are the same as the previous method (labeled K1, K2, ..., K5) and the 5 variants
(labeled V1, V2, ..., V5). The values based on the calculation are given in Table 4.
Table 4 Final results of the TOPSIS method
Rank Company Preference (in %)
1. ČEZ (V2) 92,4
2. ŠKODA (V1) 80,3
3. EPH (V4) 80
4. AGROFERT (V3) 14,3
5. UNIPETROL (V5) 0
Source: Own creation CEZ Group (V1) ranked first in a relative ranking of 92.4% with its high-quality sustainability report in
all areas under review. Equally good reports were also evaluated in the case of ŠKODA (V1) and EPH
(V4), where with only a small difference of 3 tenths of percent ŠKODA finished second and EPH third.
AGROFERT (V3) with its report fell to the penultimate 4th place in relative appreciation, mainly due
to low values in criteria K3 (Quantitative indicators) and K5 (GRI methodology). The criteria were also
problematic in the way of reporting on the CSR activities of Unipetrol (V5), but the resulting evaluation
was reduced by the fact that there is no single CSR. Unipetrol is ranked 5th in the overall ranking.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
121
Comparison of the results
By comparing both applied methods of multi-criteria decision-making it is possible to determine the
definite order of the specified variants. Table 6 below was created for easier display and comparison of
results from AHP and TOPSIS methods.
Table 6 Comparison of results from both selected methods
Company Rank with AHP method Rank with TOPSIS method
ČEZ (V2) 1. 1.
ŠKODA (V1) 2. 2.
EPH (V4) 3. 3.
AGROFERT (V3) 4. 4.
UNIPETROL (V5) 5. 5.
Source: Own creation
The order of the variants does not differ for both methods used. The first variant of V1 (ŠKODA)
finished second in both cases, while the difference between V1 and V4 in third place (EPH) was
minimal. CEZ Group (V2) ranked the highest in the ranking by both methods. The penultimate place is
the AGROFERT Group (V3) and the last place in the AHP and TOPSIS evaluation is the last place.
From the point of view of the two selected methods no significant difference was found, and it can be
stated that the choice of the method did not have a significant influence on the final ranking.
Conclusion
A comprehensive evaluation of the social responsibility activities of one or more companies is very
difficult due to often qualitative data and insufficient information. The evaluation should include not
only traditional and audited data from annual reports, but also selected activities to all stakeholders in
the period under the CSR concept. Despite the fact that the principle of volunteering is an important
feature of corporate social responsibility and only a small percentage of companies have to issue a CSR
report, it is expected that in terms of sphere influences of the most important and largest companies
these activities are recorded, measured and subsequently passed to stakeholders in a single report. For
this reason, a deeper analysis of CSR reports and other information regarding CSR activities of the five
largest companies in the Czech Republic according to their turnover for 2018 was carried out.
The aim of this work was a complex evaluation of these data and comparison of individual companies
according to mathematical methods of multi-criteria decision-making AHP and TOPSIS. The qualitative
data was categorized into five cumulative quality factors for a given CSR report and evaluated on a five-
step scale (0 - worst; 4 - best). Based on this evaluation, it was possible to use the AHP and TOPSIS
methodology. There was no significant difference between the two methods, which would influence the
final order of the selected variants. ČEZ, ŠKODA and EPH report their corporate social responsibility
activities using high-quality reports, consistent with international standards. Minor differences in the
assessment of individual factors, however, determined ČEZ in the first, ŠKODA in the second and EPH
in the third place in relative order. In the case of AGROFERT, the report was evaluated, which complied
only with some selected criteria and, thanks to the lack of quantitative data and non-existent GRI
methodology, finished only in fourth place. The last place was placed by Unipetrol, where the low rating
was influenced mainly by the non-existent whole CSR report. Information about CSR activities was
thus very difficult or impossible to find in official sources.
Both mathematical methods fulfilled the basic prerequisite, i.e. the determination of the order based on
the chosen evaluation and comparison. It is therefore possible for a comprehensive evaluation of the
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
122
company's social responsibility activities and subsequent comparison, but it is also necessary to mention
the considerable limitations of this procedure. The methodology of the chosen evaluation in this work
is not standardized and factors for quality measurement can be determined differently by different
authors and with different weights. It should also be noted that the evaluation is relative and therefore
only relevant to the options chosen. To include other companies, it would be necessary to expand the
computational matrix, which complicates the calculations in a larger number of variants.
Finally, it is necessary to state that the data is obtained from reports, which in many cases are optional
and thus the data need not be audited or otherwise checked. Therefore, the resulting ranking of
companies is only appropriate to the extent that the published data of selected companies are authentic
and factually correct
Acknowledgement This article was prepared as a part of the SGS project at the Faculty of Economics, VŠB-TU Ostrava,
project number: SP2019 / 7.
References
[1] CAMPBELL, John L., Jeremy MOON a Sara L. RYNES, 2007. Why would corporations
behave in socially responsible ways? an institutional theory of corporate social responsibility:
A Conceptual Framework for a Comparative Understanding of Corporate Social
Responsibility. Academy of Management Review. 32(3), 946-967. DOI:
10.5465/amr.2007.25275684. ISSN 0363-7425.
[2] CARROLL, Archie, 1999. Corporate social responsibility: Evolution of a definitional
construct [online]. Business and Society
[3] CARROLL, Marc, Frank L. SCHMIDT a Sara L. RYNES, 2016. Corporate Social and
Financial Performance: A Meta-Analysis. Organization Studies. 24(3), 403-441. DOI:
10.1177/0170840603024003910. ISSN 0170-8406.
[4] Data from selected companies (ŠKODA, ČEZ, EPH, UNIPETROL, AGROFERT) were acquired
from their CSR reports, Annual reports and other openly accessible sources (official websites)
[5] HWANG, Ching-Lai a Kwangsun YOON, 1981. Methods for Multiple Attribute Decision
Making. Multiple Attribute Decision Making. Berlin, Heidelberg: Springer Berlin Heidelberg,
1981, 58-191. Lecture Notes in Economics and Mathematical Systems. DOI: 10.1007/978-3-
642-48318-9_3. ISBN 978-3-540-10558-9.
[6] QUARSHIE, Anne M., Asta SALMI a Rudolf LEUSCHNER, 2016. Sustainability and
corporate social responsibility in supply chains: The state of research in supply chain
management and business ethics journals. Journal of Purchasing and Supply Management.
22(2), 82-97. DOI: 10.1016/j.pursup.2015.11.001. ISSN 14784092
[7] RASCHE, Andreas, The Corporation as a Political Actor – European and North American
Perspectives (January 23, 2015). Rasche, A. (2015). The Corporation as a Political Actor:
European and North American Perspectives, European Management Journal, 33(1): 4-8.
[8] SAATY, Thomas L., 2013. The Modern Science of Multicriteria Decision Making and Its
Practical Applications: The AHP/ANP Approach. Operations Research [online]. 61(5), 1101-
1118. DOI: 10.1287/opre.2013.1197. ISSN 0030-364X
[9] SCHERER, A. G., & PALAZZO, G. (2011). The new political role of business in a globalized
world: A review of a new perspective on CSR and its implications for the firm, governance,
and democracy. Journal of Management Studies, 48(4), 899–931.
[10] TRIANTAPHYLLOU, Evangelos, Multi-criteria Decision-Making Methods: A Comparative
Study. 2000. DOI: 10.1007/978-1-4757-3157-6.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
123
MEASURING THE FINANCIAL PERFORMANCE OF A COMPANY BASED ON
SELECTED APPROACH
Filip, Lessl1
1Department of Finance, VŠB – Technical University of Ostrava,
Sokolská tř. 33, Ostrava 70200, Czech Republic
e-mail: [email protected]
Abstract
The paper focuses on the financial performance of a company evaluated on the basis of selected
approaches. At present, the problem of the impact of International Financial Reporting Standards (IFRS)
on the reporting of the financial level of the particular enterprise, in particular of the stock companies
listed on the stock exchange, is at the forefront. The problem arises when choosing the right criteria for
measuring financial performance with respect to international reporting. The starting point is an
understanding of the elementary differences between international accounting standards and Czech
standards. This differentiation distorts the input data for the calculations of individual financial
indicators and thus the overall financial performance. The article will address one of the most widely
used and most comprehensive indicators of economic added value (EVATM). Comparison of financial
performance results for input data according to Czech Accounting Standards and IFRS will be
performed.
Keywords
Value Added (EVA), financial performance, Czech Accounting Standards (CAS), International
Financial Reporting Standards (IFRS), Cost of capital, Risk premium, The Fama French Model (TFF)
JEL Classification
G20, G30, G00.
Introduction
Financial performance is an important and key concept in the financial management of an enterprise and
increasing it is generally considered to be the main objective of financial management of each business.
The term performance is most often defined as the ability to evaluate the capital invested by individual
owners compared to the possibilities of alternative use of that capital for other purposes. However, it is
important to note that the performance of an enterprise can be viewed from a variety of perspectives,
from the point of view of the owners or managers, since the two groups are pursuing slightly different
interests and goals. The world economy has been decimating national borders for decades. In Europe,
together with economic globalization, political unification takes place within the European Union. As a
consequence of these processes, there is a growing need for accounting harmonization. Accounting
information is necessary not only for the implementation of qualified business decisions, but also for
the provision of subsidies, etc. There are currently three significant lines of international accounting
harmonization. These are International Financial Reporting Standards (IFRS), the Accounting
Directives of the European Union, and the US GAAP (General Accepted Accounting Principles) also
play an important role. This paper will only take into IFRS and, of course, Czech Accounting Standards.
The aim of this paper is to determine the cost of capital and to calculate the economic value added based
on the Value Spread.
Literature Review
There are currently many approaches and methods evaluating financial performance. This issue is
addressed by a number of authors, such as Young (2001), EHRBAR (1998). Some approach to measuring
financial performance from a management point of view, for example, with the Balanced Scorecard,
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
124
Damodaran (2011). Economic value added (EVA) is based on concept of an economical profit, which
is part of the financial theory for a long time. Economic value added is measures of firm’s performance,
which have been created with the aim to motivate managers on shareholder”s value increase. EVA is
brand of Stern Steward and company, who has been popularised this approach to measure financial
performance in United State, where this method has been implemented into management system of
many firms. Its role is growing both in transition economies and, above all, in market economies.
According to Dluhošová (2010), the fact that the indicator shows a strong correlation to the development
of the shares of the companies, which is important especially for the shareholders, can be considered.
EVA is a comprehensive instrument in business management. EVA can be used to evaluate an enterprise
or as a tool for managing and motivating workers, especially managers, etc. This way, performance can
be characterized by the 4M (measurement, management, motivation and mindset) rule, as Young (2001)
states. The indicator is relatively new and is becoming increasingly used in the area of performance
measurement indicators. The resulting value indicates whether the value for owners has increased or
decreased. When considering financial performance, there are problems with the quality of input data,
variations in methodological approaches, the risk and uncertainty of future financial flows, etc.
Moreover, another problem arises in the difference between Czech accounting regulations and
International Financial Reporting Standards. (IFRS). Unlike national accounting systems, IFRS do not
provided guidance on accouting procedures.
Definition of International Financial Reporting Standards
The International Financial Reporting Standards are a summary of the best accounting procedures and
experience of the accounting profession and of users´ requirements upon the range of publicly
information. Their purpose is to increase comparability of reporting on financial effectiveness and
financial position of different companies, operating under different national conditions. Financial
Reporting Standards are a summary of the best accounting procedures and experience of the accounting
profession and of users´ requirements upon the range of publicly (IFRS), the original name was
International Accounting Standards, (IAS), are currently one of the three basic regulations in the context
of international accounting harmonization. IFRS used mainly in Europe. Priority objectives of published
standards are not the methodological accounting procedures, but the main emphasis is placed on the
interpretation of accounting data in the form of financial statements. The financial statements prepared
according to these standards provide high-quality, transparent and comparable information, which can
help the users to make economic decisions. (Procházka, 2015).
Czech companies and IFRS
As Dvořáková (2017) states, since 2005 the IFRS application has been compulsory for companies
operating on EU regulated markets. According to the Regulation No. 1606/2002 of the European
Parliament and the Council of 19 July 2002 on the IAS, the accounting entities that are trading
companies and that are issuers of securities registered at a regulated securities market in the EU member
countries, have to apply the IAS, adapted by the European Union law, for accounting and drawing a
financial statement. A key problem of accounting based on IFRS is the tax basis which is obtained from
the accounting profit in the Czech Republic. For this reason, the accounting entities which account and
report according to IFRS by law have, for the purposes of calculation of the profit tax payable, to
transform the business result to such a result which they would have if they accounted and reported
according to the Czech regulations
For income tax purposes, these companies have to rely on the economic result expressed in accordance
with Czech accounting regulations. To solve this situation, there is a two-fold approach:
• To create a high-quality bridge for these operations, which will be displayed
differently in both accounting systems and then to convert it according to IFRS based
on the Czech regulations, or
• to account and prepare financial statements in two accounting systems, i.e. according
to IFRS, and according to Czech accounting regulations.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
125
Application of IFRS in Czech companies requires high demands on the professional knowledge of
accountants and all other employees.
Basic differences between CAS and IFRS
It should be noted that IFRS do not specify a specific form of financial statements, they do not require
any accounting framework. IFRS also do not report standard account balances, as is the case with Czech
accounting. So, IFRS defines the minimum amount of information that an enterprise must publish. The
primary goal of the IFRS financial statements is to provide high quality information for decision-makers.
Conversely, Czech accounting is very closely linked to tax laws. IFRS thus requires the transactions to
be traded consistently according to their economic substance and not in accordance with the legal
standard. Máče (2013).
Calculation of Economic Value Added
General concept of EVA, as a measure of financial performance, expresses the difference between profit
and cost of capital, which reflects a minimal rate of return of capital invested. The calculation of the
EVA is determined on the one hand by the input data and the way of the cost of capital calculation,
variations in methodological approaches, the risk and uncertainty of future financial flows. Moreover,
it is also important, if me want to calculate an absolute or a relative value. According to Dluhošová
(2004), are two basic concepts of calculation: operating profit concept and value spread concept. EVA
calculation on base of operating profit is general defined as:
EVA = NOPAT – WACC · C, (1)
where NOPAT is net operating profit after taxation, WACC is weighted cost of capital and C is value of
total capital invested.
NOPAT is subject to the same adjustment principles as the corrected economic result for DCF. It thus
includes only those revenues and expenses related to the core business of the company. C consists of
assets that are used for operating activities or for the main operation of the company (in the EVA concept
it is replaced by the term NOA, which are the so-called net operating assets). In the value C are assets
that are necessary for operating profit are tied. Thus, there must be some symmetry where NOPAT
should include those revenues and costs related to the assets that are part of the NOA. The calculation
of economic value added can be calculated in two ways, using the cost-of-capital formula or the Value
Spread. As stated in Mařík (2018), the first calculation method described above looks like this:
𝐸𝑉𝐴𝑡 = 𝑁𝑂𝑃𝐴𝑇𝑡 − 𝑁𝑂𝐴𝑡−1 ∙ 𝑊𝐴𝐶𝐶𝑡. (2)
The second aforementioned approaches is being used in this paper, namely the Value Spread concept.
Specifically, this is EVA based on a narrowed concept of value spread, which is defined as follows:
𝐸𝑉𝐴 = (𝑅𝑂𝐸 − 𝑅𝐸) ⋅ 𝐸, (3)
where RE is market cost of equity, E is equity and ROE is return on equity. For the owner, it is important
that the (ROE – RE) spread to be as large as possible or at least positive. Only in this case investment to
the firm brings more than an alternative investment.
Another the decisive factor in EVA calculation is cost of capital, which is one of the key issues due to
their sensitivity to EVA. As Dluhošová (2010) states, Cost of capital represents minimal rate of return,
which must be achieved by firm no do decrease wealth of investors. There are three basic types of capital
costs. The first type is Weighted Average Cost of Capital (WACC), which is a combination of different
forms of capital:
,ED
ERDRWACC ED
+
+= (4)
where D is debt, RE is cost of equity, E is equity, RD is cost of debt, D + E is total capital invested.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
126
The second type is cost of equity. Generally, cost of equity can be calculated using capital assets models
of construction models.
Calculation Cost of capital using rating model (INFA)
In this paper, the cost of equity was determined using the construction model, specifically a rating model
(INFA) used by the Ministry of Industry and Trade of the Czech Republic. Cost of equity can be
expressed as a sum of return of a risk-free assets and risk premiums. The INFA model calculation is as
follows:
𝑊𝐴𝐶𝐶𝑈 = 𝑅𝐸𝑈 = 𝑅𝐹 + 𝑅𝑒𝑛𝑡𝑟𝑒𝑝𝑟𝑒𝑛𝑒𝑢𝑟𝑖𝑎𝑙 + 𝑅𝑓𝑖𝑛𝑠𝑡𝑎𝑏 + 𝑅𝑠𝑖𝑧𝑒, (5)
where Rsize is risk premium for share liquidity, Rentrepreneurial is risk premium for trade risk, RF is risk
free rate, Rfinstab is risk premium resulting for financial stability and WACCU are the weighted average
cost of capital of the non-indebted entity.
Because EBIT · CZ/Z = WACCU. UZ, then the cost of equity can be determined as:
,
A
VK
A
VK
A
UZUM
Z
CZ
A
UZWACC
R
U
E
−−
= (6)
where UZ financial (paid) resources, A are total assets, VK is equity, Z is gross profit, UM is interest
rate, when UM = 𝐼
𝐵+𝑂, where B are bank credits, O are bonds and I are interests.
The cost of equity can then be determined using these risk premiums as follows:
𝑅𝐸 = 𝑊𝐴𝐶𝐶𝑈 + 𝑅𝑓𝑖𝑛𝑠𝑡𝑟 = 𝑅𝐹 + 𝑅𝑒𝑛𝑡𝑟𝑒𝑝𝑟𝑒𝑛𝑒𝑢𝑟𝑖𝑎𝑙 + 𝑅𝑓𝑖𝑛𝑠𝑡𝑎𝑏 + 𝑅𝑠𝑖𝑧𝑒, (7)
where Rfinstr is risk premium resulting for the capital strukture and can be expressed as follows: Rfinstr =
RE - WACCU If 𝑅𝐸 = 𝑊𝐴𝐶𝐶𝑈, after Rfinstr = 0%. In case, when RE – WACCU > 10%, then Rfinstr = 10%.
Risk free rate RF correspond to the yield of the T-bonds with the time to maturity from five to ten years,
most often with a maturity of 10 years.
Determination of risk premium characterizing enterprise size, Rsize, which is the function of the size of
a fimr´s equity. If the value of the financial resources (UZ) > 3 mld. CZK, then Rsize = 0%. If UZ < 100
mil. CZK, after is Rsize = 5.0%. If i tis UZ > 0.1 and UZ < 3 mld. CZK, then Rsiz = (3 mld. Kč – UZ)2 /
168,2.
Rentrepreneurial is risk premium reflecting the production power of the enterprise. This risk premium is
dependent on indicator EBIT/A, which is compared to the indicator XI expressing the replacement of
external paid capital with equity. The indicator is calculated as: ,1 UMA
UZX = Consequently, if
EBIT/A >X1, then Rentrepreneuril = min Rentrepreneuril. If EBIT/A < 0, then Rentrepreneuril = 10.0%. If 0 ≤ EBIT/A
≤ X1, then ,1,01
/12
−=
X
AEBITXR sképodnikatel
Risk premium Rfinstab is a function of gross liquidity (L3), current assets/(short term liabilities + short
term loans). If L3 > XL2, then Rfinstab = 0%. If L3 ≤ XL1, then is 10%. And finally if XL1 < L3 <XL2,
then is Rfinstab= ((XL2 – L3 )/( XL2 – XL1)) 2·0.1. X1 and X2 are the recommended liquidity limits in the
industry, when X1 = 1 and X2 = 2.5.
Calculation Cost of capital using The Fama-French Model
Specifically, a Three-Factor Fama-French model (FF3F) was chosen for the purpose of this paper. The
FF3F model is an asset pricing model developed by University of Chicago professors Eugene Fama and
Kenneth French due to a reaction to the poor results of Sharpe’s and Linter’s in explaining average
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
127
cross-sectional stock returns in the U.S. stock market. FAMA Eugene a Kenneth FRENCH (1992). The
Fama-French Model is an extension of the Capital Asset Pricing Model (CAPM), as stated in the paper
FAMA Eugene and Kenneth FRENCH (1995; 1996a; 2012). Specifically, this model extends the
(CAPM) about the size factor and book-to-market factor in order to capture the cross-sectional variation
in average returns, which would be viewed as certain anomaly in the (CAPM). According to Zmeškal
(2018), the formula is as follows:
E (RI) = RF + β ·(E(RM) – RF) + β · E(RSMB) + βi,HML · E(RHML), (8) where RF is the risk-free return rate, β factor’s coefficient, E (Ri) portfolio’s expected rate of return, (E
(RM) – RF) is market risk premium, RM is the return of the market portfolio.
Factors of The Fama-French model:
• Market risk premium - gives the investor returns above the risk-free rate.
• Small Minus Big, (SMB) - is a size effect based on the market capitalization of a
company. It reflects the difference between the average returns of portfolios of small
and large firms that have a similar ratio of book value to market value. Specifically, it
is taken as the difference between the decile yields of the largest shares according to
market capitalization and the decile of the smallest shares. This factor’s coefficient is
calculated linear regression and it can have negative and positive values.
• High Minus Low, (HML) - it reflects the spread between the average returns of
portfolios with high and low book value ratios. Specifically, it is again taken as the
difference in the yields of the portfolios containing the shares with the highest
proportion of book value and market value (9th decile) of the shares with the lowest
proportion of this ratio (1st decile). As a previous beta coefficient, this risk factor can
be calculated using linear regression, with both positive and negative values.
These parameters βi can be expressed using the matrix notation as:
βi = Var (f) -1 Cov (f, RI – RF), (9)
where f is a vector of risk factors, Var(f) is the variance-covariance matrix of f and Cov (f, RI − RF) is a
vector containing the covariances of risk factors with excess asset return.
Empirical Results
The following key part of this paper will assess the effect of IFRS on the financial performance of the
selected enterprise, measured on the basis of the EVA indicator. The following Table 1. lists the input
data for the calculation of the individual risk premium, the calculation of the cost of capital determined
on the basis of the INFA model (construction model) and for calculating the EVA. It is necessary to
recall that the calculation of the EVA will be based on the accounting model, because the INFA model
is based on the accounting principle.
The input data for the purposes of this article were used by XY. The name of XY is to keep business
secrets. XY is a joint-stock company specializing in the production of heavy castings. The history of the
company dates back to the 1960s. The company is focused on the aerospace industry, where it has a
number of certifications and supplies subcontractors for world-class aircraft brands. In addition to this,
XY is also active in power engineering and electrotechnics. The company's export accounts for 60% of
the company's turnover. According to CZ-NACE the company falls into 25 section - manufacture of
metal structures and fabricated metal products. The input data were obtained on the basis of the
company's annual reports.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
128
Table 1. Input data for the calculation of the risk premium, RE and EVA
(T. CZK)
CAS 2014 2015 2016 2017 2018
Total assets 100 316 107 052 105 727 108 747 106 328
Property, plant and equipment 45 431 46 756 42 732 51 836 43 432
Current assets 54 416 60 072 62 720 56 203 62 391
Total liabilities and equity 100 316 107 052 105 727 108 747 106 328
Equity 44 816 53 513 61 126 69 111 76 567
Non-current liabilities 27 722 25 481 12 841 11 533 8 714
Bank credits and loans 22 722 20 481 7 841 6 533 3 714
Current liabilities 27 778 28 058 31 760 28 103 21 047
Trade and other liabilities 8 000 5 000 5 000 5 000 5 000
Bank credits and loans 11 150 13 676 19 502 16 288 7 219
EBIT 8924 9350 8067 8340 7761
EAT 7373 8319 7613 7985 7456
Interest expense 863 706 206 153 99
IFRS 2014 2015 2016 2017 2018
Total assets 179 339 182 454 174 933 171 865 160 626
Property, plant and equipment 125 342 122 743 112 466 115 465 98 282
Current assets 53 997 59 711 62 467 56 400 62 344
Total liabilities and equity 179 339 182 454 174 933 171 865 160 626
Equity 75 435 83 341 83 462 89 230 92 353
Non-current liabilities 73 025 68 009 57 005 51 448 45 056
Bank credits and loans 22 722 20 481 7 841 6 533 3 714
Current liabilities 30 879 31 104 34 466 31 187 23 217
Trade and other liabilities 18 024 21 636 28 286 23 191 15 379
Bank credits and loans 8 000 5 000 2 000 3 168 2 770
EBIT 7158 8000 4560 7533 6890
EAT 2442 6961 3127 5646 5148
Interest expense 5 505 4 136 2 272 2 882 1 923
Source: Own creation
In the calculation of risk premiums, a risk-free rate was first established. The value-free rates RF were
derived from data published on the Czech National Bank website. These values ranged from 2.26% in
2014 to 0.98% in 2018. Subsequently, it was necessary to calculate the value of the financial (paid)
resources for the calculation of the risk premium characterizing the size of the enterprise Rsize. In the
case of XY, the financial resources are the sum of equity and bank loans. Considering that the value of
the financial sources was less than 100 million crowns for the whole time, then the value of this risk
premium was set at 5% over the whole period. These are CAS values. Then, was determined the value
of the risk premium of Rentrepreneurial, which characterizes the productive power. First, it was necessary to
calculate the value of the indicator X1 and the value of the interest rate. The size of the interest rate was
as follows: UM = I/(B + O) The values of the X1 indicator are then determined as: .
This value was then compared to the return on assets. Given to, that the return on assets is greater than
the X1 indicator, then Rentrepreneurial is equal to the minimum Rentrepreneurial for the sectors available in the
financial analyzes of the Ministry of Industry and Trade published on www.mpo.cz. In the first four
analyzed periods, the overall liquidity was greater than XL2, then the risk premium was calculated as:
In the last year, the indicator L3 was larger than XL2, then the value of the risk premium was 0. Of
,1 UMA
UZX =
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
129
course, the values of individual risk premium were different for IFRSs. Based on the calculation of
individual risk margins, then WACCU and subsequently RE were calculated. WACCU were calculated
based on formula (1) and RE then by (5) and (6), respectively.
Table 2. Risk premium, cost of equity and value of spread (%)
CAS 2014 2015 2016 2017 2018
RF 2,26% 1,58% 0,58% 0,48% 0,98%
Rentepre.min. 4,05% 3,50% 3,00% 3,49% 3,76%
Rsize 5,00% 5,00% 5,00% 5,00% 5,00%
Rfinstab 1,30% 0,57% 1,23% 1,11% 0%
WACCU 12,61% 10,65% 9,81% 10,08% 9,74%
RE 14,86% 14,37% 12,14% 11,34% 9,61%
Rfinstr 2,25% 3,72% 2,33% 1,26% -0,13%
ROE 16,45% 15,55% 12,45% 11,55% 9,74%
Value of spread 1,59% 1,17% 0,32% 0,21% 0,12%
IFRS 2014 2015 2016 2017 2018
Rentepre.min. 3,89% 2,99% 6,21% 5,53% 5,85%
Rsize 4,98% 4,97% 5,00% 5,00% 5,00%
Rfinstab 2,51% 1,50% 2,10% 2,13% 0,00%
RF 2,26% 1,58% 0,58% 0,48% 0,98%
WACCU 13,64% 11,04% 13,90% 13,14% 11,83%
RE 16,70% 10,10% 13,67% 12,14% 11,11%
Rfinstr 3,06% 0,94% 0,23% 0,99% 0,73%
ROE 3,24% 8,35% 3,75% 6,33% 5,57%
Value of spread 2,49% 4,32% 1,87% 2,42% 1,56%
Source: Own creation
All variables are now known to determine the indicator EVA. It was calculated on the basis of the narrow
range according to formula (2.7). The following Figure 1shows EVA values under IFRS and CAS for
the period from 2014 to 2018.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
130
Figure 1. Evolution of EVA over the period from 2014 to 2018
Source: Own creation
As can be seen from the chart, the resulting values of EVA are quite different. It is clear from the analysis
that no positive EVA based on IFRS was achieved at any time.These different results are due to the fact
that the EVA based on IFRS data has a significantly negative value of spread (ROE – RE) against EVA
based on CAS. The negative value of the spread is mainly due to the lower return on equity. Under
IFRS, less profits were due in particular to higher depreciation of new assets, namely machinery and
equipment leases and leased premises. Decreasing return on equity also affected the revaluation of fixed
assets.
The following section will explain and compare the impact of the difference between IFRS and CAS on
EVA. As can be seen, the value of assets, liabilities and profits under IFRS adjustments has changed
significantly. Of course, the names and structure of the individual financial statements have changed.
However, the objective of this paper is not to describe differences in the structure of individual
indicators, but to assess the difference and impact of IFRS financial statements on total financial
performance. The first thing that caused a substantial difference in the resulting value of EVA is the
rental of production premises. According to IFRS, the lease of such premises must be capitalized in the
value of the property as it uses its premises for its economic activity and, at the same time, an increase
in the liability must occur. For CAS, this rental only appears in off-balance sheet records. Therefore, for
realistic display of reality, the item 'Long-term rentals' must be capitalized in the balance sheet. Expert
estimate was the estimated net book of leased premises at 1.1. 2014 in the amount of 52.4 million CZK,
when it is assumed that this property will be used in the form of 25 years. This data is important for
setting up additional costs that have arisen due to the activation of the property (space rent). These
annual depreciations were calculated as the portion of the net book and the economic life of the building,
52.4 mil. CZK / 25 = 2096 thousand CZK. The amount of the annual depreciation is a short-term
liability, the rest of the amortized amount being part of long-term payables. Under IFRS, the original
amount of the lease (4 400 thous. CZK) is deferred to annual depreciation, interest expense and
maintenance and administration. Rental is considered as a form of foreign capital, so it is necessary to
quantify interest. Interest was determined on the basis of the PRIBOR interest rate and the selected risk
margins. The second thing that caused a substantial difference in the resulting value of EVA is financial
leasing. In particular, the Company acquires in the form of financial leasing primarily machinery and
equipment. According to IFRS, the current value of the lease payments must be capitalized in fixed
assets. This will increase valu efor property, plant and equipment. At the same time, there will be an
increase in non-current liabilities and current liabilities, as finance lease liabilities. In addition, the
capture of these assets will also be reflected in the statement of profit and loss. In particular, there will
be an increase in financial costs (interest) and, of course, depreciation. Compared to IFRS, the capture
of the finance leasing was quite different for CAS. According to CAS, financial leasing would only be
reflected in the form of operating costs. Its value is recorded only in off-balance sheet accounts. Another
713,01 628,15 194,41 146,49 95,11
-10157,32
-1453,33
-8279,89
-5189,86 -5109,05
-12000,00
-10000,00
-8000,00
-6000,00
-4000,00
-2000,00
0,00
2000,00
EVA EVA(IFRS)
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
131
very significant difference with CAS is the revaluation of fixed assets. Under IFRS, the tangible and
intangible assets may be revalued at fair value at the balance sheet date. In this way, the situation in the
company more corresponds to reality because the property is valued at fair value. At the same time, the
revaluation is shown in equity. The revalued value becomes the basis for determining the new amount
of depreciation. Precise revaluation of long-term assets can not be fully carried out, as the investor needs
a subordinate internal information.
Conclusion
The objective of this paper was to assess the impact of IFRS on the financial performance of an
enterprise. Comparing accounting procedures and reporting of some items according to the Czech
Accounting Standards and according to IFRS results in differences in the accounting data reporting.
According to one reporting frame a company can reach a profit, while according to another it can show
a loss. Of big difference can also be total balance sums, asset value, and value of other items of property
or liabilities. Financial analyses indices and comprehensive conclusions about performance could differ
considerably. Compilation of the financial statements according to IFRS will change the assessment of
financial stability and financial performance, both in positive and in negative direction. On the contrary,
according to CAS, XY created an economic added value of CZK 0.713 million. Consequently, the
conclusion can be deduced from the results, the impact of IFRS on financial performance is well founded
and more relevant to reality because assets and liabilities are measured at fair value, and all transactions
related to the economic activity of the enterprise are recorded in the financial statements. It can be argued
that EVA according to IFRS is so close to the market method and is not a mere approximation.
References
[1] DAMODARAN, Aswath. Study guide for Damodaran on valuation, security analysis for
investment and corporate finance. New York: Wiley, 1994. 220 s. ISBN 0-471-10897-9.
[2] DAMODARAN, Aswath. Applied corporate finance. 3rd ed. Hoboken: John Wiley & Sons, 2011.
ISBN 978-0-470-38464-0.
[3] DLUHOŠOVÁ, Dana. Finanční řízení a rozhodování podniku. 3. vyd. Praha: Ekopress, 2010. 225
s. ISBN 978-80-86929-68-2.
[4] DLUHOŠOVÁ, Dana a kolektiv. Nové přístupy a finanční nástroje ve finančním rozhodování. 1.
vyd. Ostrava: 2004. 640 s. ISBN 80-248-0669-X.
[5] DVOŘÁKOVÁ, Dana. Finanční účetnictví a výkaznictví podle mezinárodních standardů IFRS. 5.
aktualizované a přepracované vydání. Brno: BizBooks, 2017. 368 s. ISBN 978-80-265-0692-8.
[6] DVOŘÁKOVÁ, Dana. Finanční účetnictví a výkaznictví podle mezinárodních standardů IFRS.
Aktualiz. a rozš. vyd., 4. vyd. Brno: BizBooks, 2014. Daně a účetnictví (BizBooks). ISBN 978-
80-265-0149-7.
[7] EHRBAR, Al. EVA: the real key to creating wealth. New York: Wiley, c1998. ISBN 0-471-
29860-3.
[8] MÁČE, Miroslav. Účetnictví a finanční řízení. 1. vyd. Praha: Grada, 2013, 552 s. ISBN 978-80-
247-4574-9.
[9] MAŘÍK, Miloš a Pavla MAŘÍKOVÁ. Moderní metody hodnocení výkonnosti a oceňování
podniku: ekonomická přidaná hodnota, tržní přidaná hodnota, CF ROI. Přeprac. a rozš. vyd. Praha:
Ekopress, 2005. ISBN 80-86119-61-0.
[10] MAŘÍK, Miloš. Metody oceňování podniku: proces ocenění, základní metody a postupy. 4. vyd.
Praha: Ekopress, 2018. ISBN 978-80-87865-38-5.
[11] PROCHÁZKA, David. Ekonomické dopady implementace IFRS v Evropě. Praha: Oeconomica,
nakladatelství VŠE, 2015. 153 s. ISBN 978-80-245-2097-1.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
132
[12] VERNIMMEN, Pierre. Corporate finance: theory and practice. Chichester: Wiley, c2005. ISBN
0-470-09225-4.
[13] YOUNG, David a Stepfen O'BYRNE. EVA and value-based management: a practical guide to
implementation. New York: McGraw-Hill, c2001. ISBN 0-07-136439-0.
[14] ZMEŠKAL, Zdeněk, Dana DLUHOŠOVÁ a Tomáš TICHÝ. Finanční modely: koncepty, metody,
aplikace. 3. přeprac. a rozš. vyd. Praha: Ekopress, 2013, 267 s. ISBN 978-80-86929-91-0.
[15] ZMEŠKAL, Zdeněk, Miroslav ČULÍK a Tomáš TICHÝ, Finanční řízení a rozhodování: sbírka
řešených příkladů. 5. uprav. a rozš. vyd. Ostrava: VŠB-TU Ostrava, 2018, 204 s. ISBN 978-80-
248-4218-9.
[16] MINISTERSTVO PRŮMYSLU A OBCHODU ČR. Finanční analýza podnikové sféry[online].
MPO. [30.12.2018]. Dostupné z: http://www.mpo.cz/cz/ministr-a-ministerstvo/analyticke-
materialy/#category238
[17] INTERNATIONAL ACCOUNTING STANDARDS BOARD. (2005) International Financial
Reporting Standards (IFRS), London; IASC, 2005. ISBN 1-904230-44-X.
[18] FAMA Eugene a Kenneth FRENCH. "The Cross-Section of Expected Stock Returns". The Journal
of Finance, 1992, 47 (2): 427. doi:10.1111/j.1540-6261.1992.tb04398.
[19] FAMA Eugene a Kenneth FRENCH. „Size and Book-to-Market Factors in Earnings and
Returns.“ The Journal of Finance, 1995, 50(1), pp. 131-155. DOI: 10.1111/j.1540-
6261.1995.tb05169.x.
[20] FAMA Eugene and Kenneth FRENCH. ‘Multifactor Explanations of Asset Pricing Anomalies’.
The Journal of Finance, (1996a), 51 (1), pp. 55–84. doi: 10.1111/j.1540- 6261.1996.tb05202.x.
[21] FAMA Eugene a Kenneth FRENCH. Size, value, and momentum in international stock returns.
Journal of Financial Economics, 2012, 105 (3): 457. doi:10.1016/j.jfineco.2012.05.011
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
133
IDENTIFYING FACTORS OF EMPLOYEE TURNOVER WITH MULTIPLE
CORRESPONDENCE ANALYSIS
Ondřej Mikulec1
1Department of Management, VŠB – Technical University of Ostrava,
Sokolská tř. 33, Ostrava 70200, Czech Republic
e-mail: [email protected]
Abstract
Multiple correspondence analysis of factors of employee turnover represents the practical
implementation of statistical analysis and HR analytics in the area of human resources management. It
is a tool for finding relationships between categories of all categories of variables and the ability to
display these associations graphically in the positional map view of row and column profiles. From this
perspective, you can see what categories of variables you might notice or see if multiple analyses can
be used, such as a quick score tool for data, individual categories, and their associations. This paper
aims to present a new approach to human resource management data visualization and display
associations among factors of employee turnover based on real data from large production company
from Moravia-Silesia region.
Keywords
HR Analytics, Employee Turnover, Multiple Correspondence Analysis, Data Visualization.
JEL Classification
C53, M1, M54.
Introduction
The success of organizations in today's world depends largely on people, and therefore their
management can decide not only whether an organization is successful, but also whether it can survive
in today's competitive market environment (Horváthová et al., 2014). This work deals with the
quantitative approach of HR analytics to solving the problem of human resource management with
employee turnover using multivariate statistical analysis. This study is using multiple correspondence
analysis as another tool of multiple statistical analysis with high potential for possible use in the field of
human resources management. MCA presents strong visualization technique detecting and representing
underlying structures in a data set by expressing each group of categorical variables as a point in a low-
dimensional Euclidean space. Nowadays it is possible to grasp enormous amounts of visualization and
use more data and process it with visual pattern recognition as the basis of exploratory analysis.
Visualization plays important role in human resource management for the need of expressing the right
things the right way as described in Few (2015) or Sinar (2018). Interdependences or associations among
individual factors significant to undesirable employee turnover will be described and displayed based
on multiple correspondence analysis. The output of the application of MCA will be a position map of
row and column profiles visualizing in a graphical form with all categories associated with undesirable
employee turnover.
Methodology and Data
Correspondence analysis (CA) represents graphical and numerical tool expressing hidden inner
dependence among observed variables. Correspondence analysis focus according to Benzécri J. P.
(1992) and Greenacre, M. J. (1993) in Mikulec (2017) is on two-dimensional table of frequencies usually
called contingency table and shows inner associations. Contingency tables are defined by n row
categories and m column categories. Diagram of correspondence analysis is expressed as “subjective
map” with two groups of points: group of n points corresponding to row categories and group of m
points corresponding to column categories as in Mikulec (2017).
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
134
Principles of correspondence analysis
Principles of correspondence analysis are based on contingency tables with n rows and m columns. We
can define matrix U as n × m with elements 𝑈𝑖𝑗 which correspond to elements of contingency table. We
define row sums 𝑁𝑗+ and column sums 𝑁+𝑖 and total sum 𝑁𝑇 as following
𝑁𝑇 =∑𝑁𝑗+
𝑛
𝑗=1
+∑𝑁+𝑖
𝑚
𝑖=1
, (1)
where 𝑁𝑗+ = ∑ 𝑈𝑖𝑗𝑚𝑖=1 and 𝑁+𝑖 = ∑ 𝑈𝑖𝑗
𝑛𝑗=1 .
Chi-quadrat statistics 𝜒2 is testing a null hypothesis of non-existence of associations between rows and
columns is calculated as
𝜒2 = 𝑁𝑇∑∑(𝑝𝑖𝑗 − 𝑟𝑖𝑐𝑗)2/𝑟𝑖𝑐𝑗
𝑚
𝑗=1
𝑛
𝑖=1
, (2)
where 𝑟𝑗 = 𝑁𝑗+/𝑁𝑇, 𝑐𝑖 = 𝑁+𝑖/𝑁𝑇 and 𝑝𝑖𝑗 = 𝑈𝑖𝑗/𝑁𝑇 represent elements of frequency matrix P.
Pearson mean quadratic contingency coefficient 𝑡 = 𝜒2/𝑁𝑇 measures whether statistical properties of
any one part of an overall dataset are the same as any other part or in other words homogeneity
characterized with low t value or heteroscedasticity characterized by large t value. t value can be
expressed as following
𝑡 =∑𝑟𝑖
𝑛
𝑖=1
∑[(𝑝𝑖𝑗/𝑟𝑖 − 𝑐𝑗)2/𝑐𝑗]
𝑚
𝑗=1
, (3)
which is equal to weighted Euclidean distance between vector of relative frequencies and average row
profile.
Let us denote 𝑟 = 𝑃 𝐼 and 𝑐 = 𝑃𝑇𝐼 where I are vectors containing only ones. Then we can denote matrix
J with elements proportional to standardized residues of contingency table U. Matrix J can be defined
as
𝐽 = 𝐷𝑟−1/2(𝑃 − 𝑟𝑐𝑇)𝐷𝑐
−1/2. (4)
Generally rectangular matrix E can be decomposed by singular values decomposition technique into
three matrixes 𝐸 = 𝑈 𝑆 𝑉𝑇 where S is matrix of singular numbers and U and V are left and right
eigenvectors respectively. Row profile components of contingency table 𝑓𝑖 are rows of matrix
𝐹 = 𝐷𝑟−1/2
𝑈𝑆, (5)
and column profile components of contingency table 𝑔𝑖 are rows of matrix
𝐺 = 𝐷𝑐−1/2
𝑉𝑆 (6)
Pairs of row and column components 𝑓𝑖 , 𝑔𝑖 are elements of orthogonal residue decomposition ordered
hierarchically according to importance. This decomposition is being referred to as correspondence
analysis. Components 𝑓𝑖 and 𝑔𝑖 are uncorrelated with zero mean values connected by following linkages
𝐺 = 𝐷𝑐−1/2
𝑃𝑇𝐹𝑆−1, (7)
𝐹 = 𝐷𝑟−1/2
𝑃𝐺𝑆−1. (8)
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
135
Diagonal elements of S matrix are identified as principal inertias. Biplot chart is used as display method
of both profiles as well as for principal components. Main goal is to identify sources of heterogeneity in
contingency tables. Correspondence analysis allows decomposition of 𝜒2 statistics to assess structures
in matrix N.
Multiple correspondence analysis and its eigenvalue correction
MCA is an extension of correspondence analysis which allows one to analyse the pattern of relationships
of several categorical dependent variables. MCA codes data by creating several binary columns for each
variable with the constraint that one and only one of the columns gets the value 1 as discussed in Abdi,
Valentin (2007). This coding schema creates artificial additional dimensions because one categorical
variable is coded with several columns. Therefore, the inertia of the solution space is artificially inflated
and therefore the percentage of inertia explained by the first dimension is severely underestimated.
Correction formula according to Greenacre (1993) can be expressed as
𝜔𝐽 =
{
[(𝐾
𝐾 − 1) (𝜔𝑖 −
1
𝐾)]2
𝑖𝑓𝜔𝑖 >1
𝐾
0 𝑖𝑓𝜔𝑖 ≤1
𝐾
, (9)
where K is a number of nominal variables and 𝜔𝑖 represents proportionally redistributed eigenvalues for
each pattern of relationship. Using this formula gives a better estimate of the inertia, extracted by each
eigenvalue.
MCA applies CA algorithms to each set of categorical variables formed in a Burt table which is analogy
of contingency table for more than two displayed variables. The Burt table is the symmetric matrix of
all two-way cross-tabulations between the categorical variables and has one row and one column for
each level of each categorical variable. It has an analogy to the covariance matrix of continuous
variables. Analysing the Burt table is a more natural generalization of simple correspondence analysis.
Correspondence and multiple correspondence analysis procedure
According to Meloun M., Militký J. and Hill M. (2017) we define the procedure of correspondence
analysis in six following steps:
1) define the objectives of correspondence analysis to assess associations among row and column
categories and rows and columns themselves,
2) task formulation and creation of squared non-negative data matrix,
3) fulfilling assumptions of compositional techniques (completeness of the input characters),
4) presentation of row, column or both categories in common chart where we look for suit-able
number of chart dimensions,
5) interpretation of the results by defining associated categories and comparing row and column
categories,
6) verifying the results.
Data description
The model of MCA for undesirable employee turnover is consisted of fluctuation factors, variables
which were evaluated as significant by the multiple logistic regression model of employee turnover
prediction described in Mikulec (2019) and ongoing research. Categories included in MCA to identify
associations with undesirable employee turnover are Organization, Category, Type of contract, Average
performance, Wage tariff, Age, Gender, Education, Distance to work based on Vnoučková (213),
Horváthová et al. (2014), Rubenstein, Eberly, Lee, Mitchell (2017), Bednář (2018) and analysed
company’s management.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
136
MCA is based on data from 5041 employees including 399 leavers individually described with
elementary categories associated with employee turnover. Stayers, own employees active in the analysed
production company at September 2019, and Leavers, employees who left the company with undesirable
turnover between 2015 and 2018.
Organization (ORG) identifies an employee's job in a particular part of an organization - primary
production, secondary / final production, service and maintenance, engineering production and
organization headquarter with related activities (finance, marketing, human resources, etc.), which were
combined due to the inseparable impact on employee turnover, although these are two distinct parts of
the organization.
Category (CAT) corresponds to the categorization of occupations into blue-collar (mostly manual
labour) and white-collar WC (mostly administration labour). The unlimited / limited contract (U / L)
corresponds to the type of contract between and employee and organization.
Average performance (PERF) corresponds to an average employee performance rating for years 2015-
2018. Employee performance rating ranges from 1 to 5, with 1 being a very low performance rating and
5 a very high performance rating.
Wage tariff (TARIFF) corresponds to the classification of an employee to a certain tariff level. The
organization uses a twelve-degree scale to rank employees with a minimum of 1 and a maximum of 12.
The tariff component of wages is only one of several components of the wage of employees of an
organization and the tariff level is used rather as an indicator of the classification of an employee into a
certain group of professions with similar tariff evaluation.
Age (AGE) variable is a continuous variable that represents the employee's age, with a minimum value
of 19 and a maximum value of 66, which are empirical values that correspond to the actual values under
the conditions of the analysed company.
Gender (GEND) corresponds to the employee's gender as male or female. Major part of analysed
employees are males due to the nature of production company.
Education (EDUC) determines an employee's maximum educational level. For the purposes of the
model and in view of the frequency and possible forms of educational attainment, employees are divided
into three categories with elementary education, high school education, including grammar schools,
secondary vocational schools, secondary vocational schools, and 3 = university education including
tertiary professional schools, bachelor's, master's and doctoral degrees.
Distance to Work (DIST) corresponds to the estimated distance to work. For the sake of simplicity were
created two groups depending whether the employee resides in the same municipality (Short) as the
organization or does not have and therefore a longer commuting time to work is expected (Long).
For continuous variables that were not naturally binary or categorical - namely Average performance
and Age - categories were created so the categories can be included in the multiple correspondence
analysis. A total of three categories were created for average performance: Low_perf for employees
with a low average rating (Perf <2.5), an average rating Avg_perf (2.5 ≤ Perf <3.5), and a high rating
(Perf ≥ 3.5). Age is distributed into age groups regularly used as following: up to 29, 30-39, 40-49, 50-
59, and 60+. The Tariff variable was for the sake of reduced from levels 1 to 12 to three, distributed
into: unqualified professions (Tariff ≤ 6), qualified professions (7 ≤ Tariff ≤ 9) and specialized
professions and sub management (Tariff ≥ 10).
Multiple correspondence analysis of employee turnover
Multiple correspondence analysis of fluctuation factors represents a practical implementation of
statistical analysis and HR analytics in the area of human resources management. It is a tool for analysing
relationships between individual categories of all categorical variables and allows to display these
associations graphically in the form of positional map of row and column profiles. Based on this
approach, it is possible to see which categories of variables are similar or related to each other (Sucháček,
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
137
Sed'a, Friedrich, Koutský, 2014), which makes multiple correspondence analysis an ideal tool for getting
a quick overview of data, individual categories and their associations.
By means of multiple correspondence analysis, a position map is compiled in accordance with the
mutual associations of the individual employee turnover factors described as an important visualization
tool for quick orientation and insight into the fluctuation issue as in Mori, Kuroda, Makino (2016).
Table 1. Outputs of multiple correspondence model of fluctuation factors
dimension % inertia % inertia
cummulative
1 59,32 59,32
2 15,28 74,60
3 2,93 77,53
4 0,23 77,76
5 0,10 77,86
6 0,05 77,91
7 0,01 77,92
8 0 77,92
Source: Own research
Inertia is the total amount of information displayed in a given number of dimensions. The total
advertisement recorded by the multiple correspondence analysis model is shown in Tab. 4.10,
which contains aggregate model indicators, reaching 77.92%, of which 59.32% in the first
dimension, demonstrating the appropriateness of using the method to explain internal
associations in the data.
Table 2. Outputs of multiple correspondence model of fluctuation factors
Factor Category Mass Quality Intertia % Coord x Coord y
LEAVER Stayer 0,092 0,883 0,3% 0,162 0,284
Leaver 0,008 0,883 3,4% -1,887 -3,306
ORG Primary p. 0,022 0,862 1,0% 0,637 -0,654
Secondary p. 0,013 0,557 1,9% 0,821 0,25
Service 0,039 0,655 0,9% 1,031 -1,052
HQ+Engin. 0,027 0,955 4,8% 0,355 0,706
CAT BC 0,071 0,758 6,4% 1,056 0,325
WC 0,029 0,758 15,8% -2,618 0,807
U/L Unlimited 0,082 0,689 1,5% 0,191 0,835
Limited 0,018 0,689 6,9% 0,865 -3,784
PERF Low_perf 0,003 0,949 1,5% 0,063 -5,396
Avg_perf 0,087 0,067 0,2% 0,01 0,038
High_perf 0,009 0,552 0,6% 0,138 1,532
TARIFF UnQualified 0,020 0,512 3,9% 0,92 -1,778
Qualified 0,056 0,694 4,0% 0,897 0,307
Specialist 0,023 0,753 16,6% -2,979 0,816
AGE to 29 0,009 0,657 5,0% 0,408 -4,958
30-39 0,013 0,740 1,9% -1,023 -1,729
40-49 0,031 0,637 0,4% 0,137 0,642
50-59 0,040 0,608 1,1% 0,206 0,952
60+ 0,007 0,525 0,4% 0,541 0,828
GEND Male 0,088 0,638 0,4% 0,209 0,077
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
138
Female 0,012 0,638 2,7% -1,53 0,561
EDUC Elementary 0,004 0,461 1,3% 1,292 -2,047
High school 0,081 0,817 2,3% 0,622 0,065
University 0,015 0,829 14,4% -3,648 0,156
DIST Short 0,043 0,706 0,3% 0,276 0,134
Long 0,057 0,706 0,2% 0,208 0,101
Source: Own research
In Tab. 2 are the results of multiple correspondence analysis for factors associated to
employee turnover. The Mass indicator denotes the line weight as a percentage of information
from all factors, the higher the frequency of the characters in a given category, the higher the
indicator in that category and the whole. The Quality indicator is the sum of the qualitative
score of the extracted dimensions. The inertia% expresses what percentage of the total
variability is explained by the model by a given variable. For each category, the coordinate
values were calculated, the x-axis (Coord x) based on the first dimension and the y-axis (Coord
y) based on the second dimension, so that they could be graphically depicted to show their
degree of proximity and association.
Figure 4. Position map of row and column profiles of multiple correspondence
Source: Own research
Substituting the coordinate values from Tab. 2 we get a position map of row and column profiles for all
categories important for employee turnover in Fig. 1. On this position map you can see the distribution
of points representing individual categories in space. It can be seen from the chart that the monitored
Leaver category representing employees with undesirable fluctuations is relatively far from all other
points taking into account the fact that it is less than 10% of the entire sample. However, some points
are closer than others and therefore have a higher association with the Leaver category. Categories
displayed in the position map visualizes in a very clear way. The Leaver category is the closest to the
30-39 age category, while the second closest age is under 29, which is consistent with the likelihood of
Stayer
Leaver
Primary p.
Secondary p.
Service
HQ+Engin.
BC
WC Unlimited
Limited
Low_perf
Avg_perf
High_perf
UnQualif
Qualif
Specialist
20-29
30-39
40-49
50-5960+
Male
Female
Elementary
High schoolUniversity
LongShort
-6
-5
-4
-3
-2
-1
0
1
2
3
4
-4 -3 -2 -1 0 1 2
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
139
unwanted fluctuation with an older employee's age in Tab. 2. Another point relatively close to the Leaver
category is the WC point, representing white collar employees with administrative and technical
occupation. White collar employees are relatively closely associated with employees with a higher tariff
level (specialists and sub management), higher education and gender female which are all relatively
close to the Leaver category. On the right side of the position map, there are also points, in relative
proximity to the Leaver category, representing employees with fixed-term contracts and those with low
average performance scores. Other points are visibly accumulated around category Stayer indicating
low associations with category Leaver.
Conclusion
People analytics or HR analytics as data-driven approach to managing people at work are based on
implementation of quantitative models into human resource management decision processing and its
influence on the company. Over the past decade, big data analytics and people analytics has been
revolutionizing the way many companies do business. Common use for these techniques is described
for recruitment, talent management, people retention etc. as well as market performance as in Edwards
and Edwards (2016) or Kapoor and Kabra (2016) and this study is oriented on interdependencies and
associations of personnel data with focus on undesirable employee turnover and visualization of the
results using MCA position map of row and column profiles.
Based on the results of multiple correspondence analysis of factors associated with undesirable
employee turnover, it is possible to visually determine which categories are more associated with it and
who fall into those categories and thus who is at greater risk of undesirable turnover. Based on the
results, recommendations can be made to optimize human resource management policy settings to
reduce risk in the categories. Specific measures include, for example, higher payroll progressiveness at
higher tariff levels so that they are more financially motivated to continue with the organization, a
support program such as training for employees with lower performance ratings to improve it, social
events or benefits targeting employees in lower age categories or analogically similar approaches
It is concluded that MCA represents a suitable statistical tool to uncover interdependencies and
associations for human resource management data and can support decisions based on the results of the
analysis which can be used to various HR problems.
Acknowledgement
This research was financially supported by the Czech Scientific Foundation (Grant SGS SP2019/7).
References
[1] Abdi H., Valentin D. (2007). Multiple Correspondence Analysis. In: Encyclopedia of
Measurement and Statistics. Thousand Oaks (CA): Sage. Available at: < http://www.utdallas.
edu/~herve/Abdi-MCA2007-pretty.pdf >
[2] Bednář, V. (2018). Jak omezit fluktuaci a udržet si zaměstnance manažerskými nástroji. Grada
Publishing, 1st ed.
[3] Benzécri J. P. (1992). Correspondence Analysis Handbook, New York: CRC Press.
[4] Edwards, R. Martin and K. Edwards. (2016). Predictive HR Analytics: mastering the HR metric.
London; Philadelphia: Kogan Page.
[5] Few, S. (2015) Data Visualization for Human Perception. The Encyclopedia of Human-Computer
Interaction, 2nd ed.
[6] Greenacre, M. J. (1993a). Correspondence analysis in practice. London: Academic Press.
[7] Greenacre, M. J. (1993b). Biplots in correspondence analysis. Journal of Applied Statistics, 20(2),
pp. 251-269.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
140
[8] Horváthová, P. et al. (2014). Řízení lidských zdrojů pro pokročilé. SOET, vol. 12. Ostrava: VSB-
TU Ostrava.
[9] Kapoor, B. and Y. Kabra. Current and Future Trends in Human Resources Analytics Adoption.
2016. Journal of Cases on Information Technology. 16(1).
[10] Meloun M., Militký J., Hill M. (2017). Statistická analýza vícerozměrných dat v příkladech, Praha:
Academia, 2 ed.
[11] Mikulec, O. (2017). Use of multiple correspondence analysis to identify influence of risk attitude
on trading behavior. Ekonomická revue: Central European Review of Economic Issues. VŠB – TU
Ostrava, vol XX, nr. 2, pp: 53-61.
[12] Mikulec, O. (2019) Predictive HR Analytics: Case of Employee Turnover. Financial Management
of Firms and Financial Institutions, 12th International Scientific Conference Proceedings. pp:
144-152.
[13] Mori, I., Kuroda, M., Makino, N. (2016). Multiple Correspondence Analysis. SpringerBriefs in
Statistics: Nonlinear Principal Component Analysis and Its Applications. pp. 21-28.
[14] Rubenstein, A. L., Eberly, M. B., Lee, T. W., Mitchell, T. R. (2017) Surveying the forest: A meta‐analysis, moderator investigation, and future‐oriented discussion of the antecedents of voluntary
employee turnover. Personnel Psychology. 71(1).
[15] Sinar, E. F. (2018) Data Visualization: Get Visual to Drive HR’s Impact and Influence. Society
for Human Resource Management and Society for Industrial and Organizational Psychology.
Available at: < http://www.siop.org/SIOP-SHRM/2018_03_SHRM-
SIOP_Data_Visualization.pdf>
[16] Sucháček, J., P. Seďa, V. Friedrich, J. Koutský. (2014). Media portrayals of regions in the Czech
Republic: selected issues. E & M ekonomie a management, Vol 17, Iss. 4.
[17] Vnoučková, L. (2013). Fluktuace a retence zaměstnanců. 3. vyd. Adart s.r.o.: VŠEM Praha.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
141
DATA ANALYSIS AND TESTING WITH RESPECT OF PORTFOLIO SELECTION
PROBLEM
David Neděla1
1Department of Economics, VŠB – Technical University of Ostrava,
Sokolská tř. 33, Ostrava 70200, Czech Republic
e-mail: [email protected]
Abstract
In recent years the problem of asset selection for creating an optimal portfolio is a very questionable
topic, not only for economists and scientists, but also for private and institutional investors. It is advised
to analyze the rate of return over a time period before investing funds. The main goal of this paper is
defined as statistical analyzing and testing of stock returns time series in relation to the portfolio
optimization problem. This is due to different assumptions of the portfolio models regarding time series.
Therefore, in this paper basic data characteristics are defined and analyzed, different kinds of probability
results and the relationships between time series, their lags and stationarity. Subsequently, various
methods are used to estimate parameters of suitable distribution. Finally, various types of tests, graphical
as well as statistical, are provided for verifying the correctness of hypotheses. In conclusion, many
results confirm that many assumptions of basic optimization models simplify the model and it is
appropriate to use a better fitted model.
Keywords
Stationarity, Autocorrelation, Probability distribution, Normal distribution, Variance Gamma
distribution, Stable distribution
JEL Classification
C13, G22
Introduction
During recent years the problem of choosing assets for creating an optimal portfolio is a favourite and
questionable topic for economists, scientists and both private and institutional investors. Before
investing is advisable to analyze the evolution of the time series of prices, particularly returns before
investing funds. One of the possible analysis is by using statistics methods due to inserting data to model
and subsequent prediction of future prediction. From a financial point of view, it is also necessary to
examine the relationships between assets for effective diversification. For these reasons, the main goal
of this paper is to statistically analyze and test time series of stock returns with regards to the portfolio
optimization problem.
In the literature, we can find many portfolio optimizations models and methods which include various
assumptions and limitations of the applied data e.g. the Markowitz model contains the assumption about
static probability belief, meaning that the used time series of returns should be normally distributed, see
Markowitz (1952). The same assumption is contained in the CVaR portfolio optimization model. As
already mentioned, another reason to provide data analysis is to predict future evolution of the various
investment instruments included in the asset portfolio by estimated parameters of distribution functions.
Other reasons for analyzing returns time series is the appropriateness of portfolio diversification which
many investors require. Therefore, the dependency analysis is carried out. In this paper only on different
types of correlations are characterized.
The paper is organized as follows. The first section is brief introducing. In the second section is
characterization of used data, then description and formulation of mathematical methodology, included
descriptive statistics, probability distributions and their tests, definition of stationarity and correlation.
At the third section are data characteristics and the main empirical findings are also presented in this
section. Fourth section is conclusion.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
142
Description and Formulation of Mathematical Methodology and Data
In this section is described mathematical methodology, specifically, general descriptive statistics,
different types of probability distribution, distribution tests, estimation methods, stationarity stationarity,
linear and nonlinear types of correlation and autocorrelation coefficients. For the empirical analysis
daily adjusted close prices of stocks included in Dow Jones Industry Average index are used. Time
period is from Monday January 3, 2006 to Friday June 28, 2018. The main source of individual stock
prices is the Yahoo Finance website.
Probability distribution
The probability distribution of dataset is an important characteristic if someone wants to fit parameters
of a statistical (financial) model to data or if someone use specific model with probability distribution
assumption. Almost all data used in finance have underlying stochasticity (randomness) which feeds
into uncertainties in the fitted statistical model parameters. That is the reason why is important to find
what kind of probability distributions typically underlie the stochasticity in data set.
A traditional assumption in financial study normal distribution of dataset, meaning the simple returns
Rt are independently and identically distributed (i.i.d.) with fixed mean value and variance value. Due
to this assumption, the statistical properties of asset returns can be made tractable. In practise, it is not
so clear as it seems. The problems are that the simple return cannot be lower than −1 but the normal
distribution may assume any value in the real line with no limiting interval and “If Rit is normally
distributed, then the multiperiod simple return Rit [k] is not normally distributed because it is a product
of one-period returns.” Tsay (2010). Many empirical results confirmed conclusion that the normality
assumption is reject for asset returns. The general form of normal probability density function is defined
as
𝑓(𝑥) =
1
𝜎√2𝜋𝑒−
12(𝑧−𝜇2)2
, (1)
where 𝜇 is the mean of the distribution and also median and 𝜎 is standard deviation.
The class of distributions included Gaussian, Cauchy and Lévy distributions is stable distributions, see
Lévy (1924). It is a class that allows skewness and heavy tails. The general definition of stable
distribution is described by parameters: the index of stability 𝛼, the skewness parameter β, the scale
parameter γ and the location parameter δ. The reason why to use stable distribution is derived from the
Generalized Central Limit Theorem which defined that the only possible nontrivial limit of normalized
sums of i.i.d. terms is stable. The empirical observation confirmed that large data sets exhibit heavy tails
and skewness. The most used parametrization of this distribution is Zolotarev parametrization defined
as 𝑋~𝑆(𝛼, 𝛽, 𝛾, 𝛿0; 0) with characteristic function:
𝐸 𝑒𝑥𝑝(𝑖𝑡𝑋) = {𝑒𝑥𝑝 (−𝛾𝛼|𝑡|𝛼 [1 − 𝑖𝛽(𝑡𝑎𝑛
𝜋𝛼
2)(𝑠𝑖𝑔𝑛 𝑡)((𝛾|𝑡|)1−𝛼 − 1)] + 𝑖𝛿0𝑡) 𝛼 ≠ 1
𝑒𝑥𝑝 (−𝛾|𝑡| [1 + 𝑖𝛽2
𝜋(𝑠𝑖𝑔𝑛 𝑡)(𝑙𝑛|𝑡| + 𝑙𝑛|𝛾|)] + 𝑖𝛿0𝑡) 𝛼 = 1.
(2)
The parameters occur at an interval 0 < 𝛼 ≤ 2,−1 ≤ 𝛽 ≤ 1, 𝛾 > 0 and 𝛿0 ∈ 𝑅. Special case of the
stable distribution with 𝛼 = 1 and 𝛽 = 0 is called Cauchy distribution. Several methods for estimating
parameters has been founded and distinguished. One of the most popular and empirically confirmed as
reliable method is maximum likelihood estimation. The log likelihood function of i.i.d. variable 𝑋𝑖 is
defined as
ℓ(𝜃 ) =∑𝑙𝑜𝑔𝑓(𝑋𝑖|𝜃 ) .
𝑛
𝑖=1
(3)
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
143
The main problem of this formula is that the general stable density formula does not exist. Zolotarev
(1986) dealt with and describe in detail the topic of stable density. Therefore, the parameters and
can be estimated from the observations 𝑥1, … , 𝑥𝑛 by maximizing the log likelihood function
∑𝑙𝑜𝑔𝑓𝛼 (𝓏𝑖) = 𝑛 𝑙𝑜𝑔𝛼 −
𝑛
𝑖=1
𝑛 𝑙𝑜𝑔(𝛼 − 1)𝜋 +∑𝑙𝑜𝑔 𝓏𝑖𝛼 − 1
𝑛
𝑖=1
+∑𝑙𝑜𝑔∫ 𝑈𝛼
𝜋/2
0
(𝛾, 0)
𝑛
𝑖=1
𝑒−𝑧𝑖𝛼 𝛼⁄ −1
𝑈𝛼(𝛾,0)𝑑𝛾, (4)
where 𝓏𝑖 = |𝑥𝑖 − 𝜇| ∕ 𝜎. To preclude the discontinuity and nondifferentiability of the symmetric -
stable density function at = 1, is restricted to be greater than one. Disadvantage of this method is
due to highly nonlinear optimization problem where no initialization and convergence are not available.
Other allowable method is McCulloch method. McCulloch (1986) generalized and improved Fama-
Roll's method from the set of Sample quantile method, Fama (1971). Firstly, independent of both and
is defined
𝑣𝛼 =𝑥0.95 − 𝑥0.05𝑥0.75 − 𝑥0.25
, (5)
where �̂�𝛼 is the corresponding sample value and it is a consistent estimator of 𝑣𝛼 . Similarly, 𝑣𝛽 is
calculated as
𝑣𝛽 =𝑥0.95 + 𝑥0.05 − 2𝑥0.50
𝑥0.95 − 𝑥0.05,
(6)
where �̂�𝛽is also the corresponding sample value and it is a consistent estimator of 𝑣𝛽. Both, 𝑣𝛼 and 𝑣𝛽,
are functions of and β. The parameters and β may be express as function of 𝑣𝛼 and 𝑣𝛽
𝛼 = 𝜓1(𝑣𝛼, 𝑣𝛽), 𝛽 = 𝜓2(𝑣𝛼, 𝑣𝛽). (7)
Consequently, the scale parameter is given by
𝑣𝛼 =𝑥0.75 − 𝑥0.25
𝜓3(�̂�, �̂�),
(8)
where 𝜓3(�̂�, �̂�)is given by table in McCulloch (1986). McCulloch also gives an estimate of , however,
this procedure is much more complicated.
The generalized Pareto distribution (GPD) is often used to model the tails of another known
distribution. The probability density function of the generalized Pareto distribution is defined as:
𝑓(𝑥) = (
1
𝜎) (1 + 𝑘
(𝑥 − 𝜃)
𝜎)
−1−1𝑘
, (9)
where k is shape parameter, σ is scale parameter and θ is threshold parameter. If k = 0 and θ = 0, the
GPD is the same as the exponential distribution. The GPD has three basic forms, each corresponding
to a limiting distribution of exceedance data from a different class of underlying distributions.
Distributions whose tails decrease exponentially, lead to a shape parameter of zero. Distributions whose
tails decrease as a polynomial, lead to a positive shape parameter. Distributions whose tails are finite,
lead to a negative shape parameter.
Crisis periods such as the 1987 stock market crash or the collapse of Lehman Brothers have changed the
view that extreme events in the financial markets have negligible probability. The use of the extreme
value (EV) distribution is then adequate. The standardized generalized extreme value distribution,
(GEV) in the defined by a location parameter µ, a scale parameter σ, and a tail shape parameter ξ. The
whole equation is defined as:
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
144
𝐹(𝜉,𝜇,𝜎)(𝑥) = 𝑒𝑥𝑝 (−(1 + 𝜉(𝑥−𝜇)
𝜎)−1∕𝜉
), (10)
where
1 + 𝜉
(𝑥 − 𝜇)
𝜎> 0 , 𝜉 ≠ 0. (11)
The probability density functions are respectively:
𝐹(𝜉,𝜇,𝜎)(𝑥) =
1
𝜎(1 + 𝜉
(𝑥 − 𝜇)
𝜎)
−1−1∕𝜉
. (12)
If ξ = 0 then the GEV distribution belongs to the Gumbel class which includes the normal, exponential,
gamma.
The class of GEV distributions is very flexible with the tail shape parameter ξ. Tail index is defined as
𝛼 = 𝜉−1. If ξ > 0 than the distribution is called Fréchet and fat tailed distributions such as Pareto or
Cauchy.
A gamma distribution is one of the general types of statistic distribution. It is a continuous, positive-
only, unimodal distribution with relation to exponential and normal distributions. A random gamma-
distributed variable x with shape parameter 𝑘 and a scale parameter 𝜃 is denoted by the probability
density function
𝑓(𝑥) =
𝑥𝑘−1𝑒−𝑥𝜃
𝜃𝑘𝛤(𝑘), (13)
where 𝑥, 𝑘, 𝜃 > 0 and 𝛤(𝑘) is gamma function evaluated at k.
The formula for the cumulative distribution function of the gamma distribution is defined as
𝐹(𝑥) = ∫ 𝑓(𝜇, 𝑘, 𝜃)𝑑𝜇
𝑥
0
=𝛾(𝑘,
𝑥𝜃)
𝛤(𝑘), (14)
where 𝛾(𝑘,𝑥
𝜃) is the lower incomplete gamma function denoted by
𝛤𝑥(𝑎) = ∫ 𝑡𝑎−1𝑒−𝑡𝑑𝑡.
𝑥
0
(15)
The variance gamma distribution is a continuous probability distribution which is defined as
the normal variance-mean mixture where the mixing density is the gamma distribution. The probability
density function of a variance gamma distribution is unimodal, with a single peak, and heavy tails. The
probability density function for the variance-gamma distribution is given by
𝑓𝑋(𝑥) =(𝛼2 − 𝛽2)𝜆
√𝜋𝛤(𝜆)(2𝛼)𝜆−1∕2|𝑥 − 𝜇|𝜆−1∕2𝐾𝜆−1∕2(𝛼|𝑥 − 𝜇|) 𝑒𝑥𝑝(𝛽(𝑥 − 𝜇)), (16)
where 𝐾 denotes a modified Bessel function of the second kind and 𝛤 is gamma function. Parameters 𝜇
denote location, 𝛼 denote tail, 𝛽 denote asymmetry and 𝜆 denote scale. Variance gamma distribution is
usually used to model daily returns of assets, as the skewness and kurtosis of the data may be a better fit
with this distribution than a normal distribution.
We can find many methods used for estimating individual parameters of this type of distribution e.g.
based on Nelder-Mead algorithm, see Nelder (1965). It is commonly used numerical direct search
method based on function comparison to find the minimum or maximum of an objective function in a
multidimensional space. The problem may be that the Nelder-Mead algorithm usually requires only one
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
145
or two function evaluations at each step, while different other direct search methods use n or more
function evaluations.
The normal inverse Gaussian distribution (NIG) obviously defined as a variance-mean mixture of a
normal distribution and the inverse Gaussian as the mixing distribution. NIG is also often used in finance
to construct stochastic processes for statistical modelling purposes, see Barndorff-Nielsen (1997). The
NIG distribution on the whole real line having density function defined as
𝑔(𝑥; 𝛼, 𝛽, 𝜇, 𝛿) = 𝑎(𝛼, 𝛽, 𝜇, 𝛿)𝑞 (𝑥 − 𝜇
𝛿)−1
𝐾1 {𝛿𝛼𝑞 (𝑥 − 𝜇
𝛿)} 𝑒𝑥𝑝(𝛽𝑥), (17)
where
𝑎(𝛼, 𝛽, 𝜇, 𝛿) = 𝜋−1𝛼𝑒𝑥𝑝 (𝛿√(𝛼2 − 𝛽2) − 𝛽𝜇, (18)
and
𝑞(𝑥) = √1 + 𝑥2, (19)
where 𝐾1 is the modified Bessel function of third order and index 1. Parameter 𝛼 is tail heaviness
parameter, 𝛽 is asymmetry parameter satisfying 0 ≤ |𝛽| ≤ 𝛼, 𝜇 ∈ 𝑅 define location and 𝛿 > 0 define
scale parameter. The distrigution is symmetric around the value 𝜇.
Distribution tests
This group of tests is used for testing if our data are distributed by supposed distribution or not.
Generally, the null hypothesis is defined that the data set is distributed by given type and the alternative
hypothesis is the data set is not distributed by given distribution, see e.g. Tsay (2010).
The first often applied test is the Kolmogorov-Smirnov test (K-S test) which examines if random
variable are likely to follow given distribution. Usually as given distribution is supposed as normal
distribution but by this test the other kinds of probability distribution can be tested. The K-S test is based
on the distance between the empirical distribution function and given cumulative distribution function
and test statistic is defined as
𝐷 = 𝑚𝑎𝑥1≤𝑖≤𝑁
(𝐹(𝑅𝑖) −𝑖 − 1
𝑁,𝑖
𝑁− 𝐹(𝑅𝑖) ), (20)
where 𝐹 is the theoretical cumulative distribution of the distribution being tested.
The K-S test has several important limitations as only continuous distributions can be applied, the test
is more sensitive near the centre of the distribution than at the tails and the distribution must be fully
specified.
Jarque-Bera test can be also used to test the hypothesis that the data are distributed by given
distribution, generally normal distribution. It is based on another way of testing than K-S test. Test
statistic is defined as
𝐽𝐵 =
𝑛
6[𝑆2 +
1
4(𝐾 − 3)2], (21)
where n is number of observations, S presents the sample skewness, K is the sample kurtosis. If the data
comes from a normal distribution, the JB asymptotically has a chi-squared distribution with two degrees
of freedom.
The Shapiro–Wilk test is basically used for data set with observation 𝑁 < 30. The statistic is defined
as
𝑊 =
(∑ 𝑎𝑖𝑥𝑖𝑛𝑛=2 )2
∑ (𝑥𝑖 − 𝜇𝑥𝑖)𝑛𝑖=1
2, (22)
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
146
where 𝑥𝑖 are ordered random sample values, 𝑎𝑖 are constants generated from
the covariances, variances and means of the normally distributed sample (size n), see Shapiro (1965).
The Lilliefors test is also used in statistics to test hypothesis that whether a given distribution more
precisely exponentially distributed population, The Lilliefors test uses the same equation (20) as the K-
S test, but the critical values are compared with the Lilliefors test table. Since the critical values in this
table are smaller, the Lilliefors test is less likely to show that data is normally distributed.
Finally, the Anderson-Darling test is included into group of distribution tests. It is again based on
the K-S test and gives more weight to the tails than the K-S test. The Anderson-Darling test makes use
of the specific distribution for calculating critical values. The Anderson-Darling test statistic is
defined as follows
𝐴2 = −𝑁 − 𝑆, (23)
where
𝑆 =∑
(2𝑖 − 1)
𝑁
𝑁
𝑖=1
[𝑙𝑛 𝐹(𝑌𝑖) + 𝑙𝑛(1 − 𝐹(𝑌𝑁+1−𝑖))], (24)
where F is the cumulative distribution function of the specified distribution and 𝑌𝑖 are the ordered data.
The test is a one-sided and if the test statistic, A, is greater than the critical value the null hypothesis is
than rejected.
Stationarity
In many studies financial rates (interest rates, foreign exchange rates or price series tend to be
nonstationary. For example, price time series are mainly nonstationary since there is no fixed value to
which prices should return. The nonstationary series is usually called unit-root nonstationary time series,
as mention Tsay (2013).
Stationarity means that the statistical characteristics of a time series are almost same over the time
period. Stationarity is very important characteristic because of many statistical tests and many models
(as well financial models) rely on it. If the joint distribution of time series (𝑟𝑡1 , 𝑟𝑡2, … , 𝑟𝑡𝑘) is invetiant
under time shift it is called as strictly stationarity. We assume a weak stationarity which means that there
exists a covariance between rt and lag values which is called autocovariance. It is possible to decide
about stationarity based on graphical view or statistical tests. The graphical point of view includes
analyzing autocorrelation and partial autocorrelation graphs and statistical approach, using statistical
tests like the Dickey-Fuller Test.
Stationarity and autocorrelation tests
The Dickey-Fuller test is used to test the null hypothesis that a unit root is present in an autoregressive
model of a given time series, and that the process is thus not stationary. The original test treats the case
of a simple lag-1 AR model. Before performing the test is necessary to visually decide if mean of time
series is not equal to zero and the trend is linear or quadratic, Tsay (2010, 2013). The Dickey-Fuller test
is testing if 𝜙 = 0 in model of the data:
𝑦𝑡 = 𝛼 + 𝛽𝑡 + 𝜙𝑦𝑡−1 + 𝑒𝑡, (25)
where 𝑦𝑡 is used data set. If 𝜙 = 0, then it is a random walk process.
If we use Ljung and Box test the null hypothesis is defined that the time series does not show
statistically significant dependency structures up to the order m. The Ljung Box test statistic is calculated
as
𝑄(𝑚) = 𝑇(𝑇 + 2)∑
𝑟𝑖2
𝑇 − 𝑖
𝑚
𝑖=1
, (26)
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
147
where 𝑟𝑖 is accumulated sample autocorrelations and m is time lag.
One of many other acceptable tests is Kwiatkowski–Phillips–Schmidt–Shin tests (KPSS test) with
the same null hypothesis as previous. The test is based on a similar principle as the Dickey-Fuller test
and the main principle of this test is to split a time series to three parts:
𝑥𝑡 = 𝑟𝑡 + 𝛽𝑡 + 휀𝑡, (27)
where 𝑟𝑡 is random walk, 𝛽𝑡 is deterministic trend, 휀𝑡 is stationary error and 𝑢~(0, 𝜎2) and are i.i.d.
Depending to this equation, the null hypothesis is defined as 𝐻0: 𝜎2 = 0. The KPSS test statistic is
𝐾 =∑ 𝑆2𝑁𝑛=1 ,
𝑠2𝑁2 (28)
where N is the sample size 𝑠2 is the Newey-West estimate of the long-run variance and 𝑆2 = 𝑒1 +⋯+𝑒3.
The last characterized test is the Phillips–Perron test used for testing the null hypothesis that a time
series is integrated of order 1. It builds on the previous mentioned Dickey–Fuller test.
Correlation and autocorrelation
The dependency structure of random sources is necessarily known indicator in portfolio theory and in
other financial optimization problems. The first and the most used indicator is the Pearson coefficient
of correlation which is based on linear dependency, see Tsay (2010). The formula of Pearson correlation
coefficient between two random variables X and Y is defined as:
𝜌𝑥,𝑦 =
𝐶𝑜𝑣(𝑋, 𝑌)
𝜎(𝑋) ∙ 𝜎(𝑌)=
∑ [𝑥𝑡 − 𝐸(𝑥)][𝑦𝑡 − 𝐸(𝑦)]𝑇𝑡=1
√∑ [𝑥𝑡 − 𝐸(𝑥)]2𝑇𝑡=1 ∙ ∑ [𝑦𝑡 − 𝐸(𝑦)]
𝑇𝑡=1
2
, (29)
where 𝐸(𝑥) and 𝐸(𝑦) are the sample mean of X and Y, respectively, and it is assumed that the variances
exist.
It works well just with normally distributed data which can be a disadvantage. For financial markets
returns the normal (Gaussian) distribution is in most cases rejected, see Fama (1965).
Another possibility to measure the correlation between random variables is Spearman coefficient of
correlation which is a nonparametric statistic. The Spearman correlation between two variables is equal
to the Pearson correlation between the rank values of those two random variables. The formula is defined
as:
𝜌𝑆;𝑥,𝑦 =𝐶𝑜𝑣(𝑟𝑋, 𝑟𝑌)
𝜎(𝑟𝑋) ∙ 𝜎(𝑟𝑌),
(30)
where 𝑟𝑋 is rank of variable X, 𝑟𝑌 is rank of variable Y, 𝜎(𝑟𝑋) and 𝜎(𝑟𝑌) are the standard
deviations of the rank variables.
As the third in order the Kendall correlation coefficient is good to mention. Kendall correlation is also
a non-parametric test based on rank of two random variable. The Kendall′s Tau usually acquire smaller
values than Spearman′s correlation. The formula of Kendall rank correlation is defined as
𝜏 =𝑛𝑐 − 𝑛𝑑12𝑛(𝑛 − 1)
, (31)
where 𝑛𝑐 is number of concordant pairs, 𝑛𝑑is number of discordant pairs.
When the linear dependence between rt and its past values rt−i is of interest, the concept of correlation
is generalized to autocorrelation. The correlation coefficient between rt and 𝑟𝑡−𝑖 is called the lag-i
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
148
autocorrelation of rt and is commonly denoted by �̂�𝜄, which under the weak stationarity assumption is a
function of only. The lag-i autocorrelation of 𝑟𝑡 is given by
�̂�𝜄 =
∑ [𝑟𝑡 − 𝐸(𝑟)][𝑟𝑡−𝑖 − 𝐸(𝑟)]𝑇𝑡=𝜄+1
∑ [𝑟𝑡 − 𝐸(𝑟)]2𝑇𝑡=1
, (32)
where 0 T-1. If 𝑟𝑡 has statistically significant lag-1 autocorrelation, the value 𝑟𝑡−1 might be useful
to predict 𝑟𝑡. The simple linear regression model is referred to AR model of order 1 or AR(1) model.
The definition of AR(1) model can by generalized to the general AR(p) model which is often used in
finance, see Tsay (2010).
Empirical Results
In this section methodology specified in previous sections is used to analyse and test stock return time
series applicable for portfolio optimization problems. In the first step, the continuously return calculation
over each 3396 daily prices was provided for the reason price time series are not usually stationary.
From daily returns the basic descriptive statistics (mean, Min, Max, std. deviation, skewness and
kurtosis) should be calculated.
All daily returns are included in the interval -0,2323;0,2983. The absolute value of the interval
boundary points are roughly similar. In many cases, the difference between limit values in absolute
terms of the interval in which the returns occur are generally smaller than 2 percentage points. In some
cases, if the absolute value of Min and Max are compared the possible maximum return is higher than
possible maximum loss. It may be a sign of a higher positive extreme values or less likely heavier right
tail over a left tail. The means of daily returns are very small, in tens of thousands. When comparing the
mean value with the Max and Min values, there is a large gap where there will certainly be extreme and
outlying values. The highest average daily return relates to stock V before AAPL and NKE stocks.
Otherwise, the least profitable shares are DOW and GS stocks. From these findings, we cannot confirm
that one industry is more profitable compared to an another. According to financial theory, the higher
returns stocks should relate to the higher rate of standard deviation, but this assumption does not apply
in our data set, e.g. the average return of GS is the second smallest, but the standard deviation is one of
the highest. It is contradicted with the basic economic theory. The same non-economic situation can be
seen in DOW and an opposite situation characterized with relatively higher return against smaller
standard deviation is in MCD stock. A zero-skewness value indicates that the values of the random
variable are symmetrically distributed to the left and right of the mean value. For example, the BA value
of skewness is the nearest to zero which indicates a certain symmetry. The most skewed data set relates
to MMM and WBA stocks. In both cases, left-skewed because of negative statistics value. It means that
there is higher probability of achieving a below average return or negative yield. UNH data set is the
most right-skewed and provides a higher than expected daily return. The values of kurtosis higher than
3 is called leptokurtic and indicate fatter tails. In our data set, only with DOW the leptokurtic is
apparently rejected because of the value 0.738. We can also see extreme kurtosis (higher than 25) in two
asset returns (TRV, UNH) and for this reason, it is advisable to observe a different type of distribution
to normal distribution. That fact confirms the hypothesis that stock returns are not based on normal
distribution.
Correlation and autocorrelation
In Pearson correlation matrix, the lower value than 0.3 can be found only in few correlation coefficients
in DOW, but in many cases, it is not significant at confidence level 5 %. In the remaining time series,
the correlation coefficients mostly occur at an interval (0.30, 0.65). Likewise, low values, we can find
exceptions and higher values, except the value 1 which means the correlation between themselves. The
most linear dependent pairs are CVX and XOM with correlation coefficient 0.860 and AXP and JPM
where the coefficient is 0.712. The first pair of companies is from energy industry and the second pair
is from financial services industry. It supports the assumption of the same returns evolution in individual
industries. The values of Spearman correlation coefficients are lower compared to Pearson correlation.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
149
Similarly, the most dependent pairs of returns time series are again (CVX and XOM) and (AXP and
JPM) and the less dependent returns are again between UNH and VZ or WMT in reverse order. Kendall
correlation values were the lowest as expected.
As we can see from autocorrelation coefficients, in all cases the hypothesis that the time series shows
statistically significant dependencies until the lag 15 is rejected.
Normality tests
In recent models, the return normality assumption is abandoned because the normality of asset returns
has been rejected in many studies. To confirm the findings the normal distribution of data is testing by
statistical tests. The results of tests are depicted in Table 1.
For data set smaller than 30 elements, is appropriate use the Shapiro-Wilk test. In this case, due to 3 395
observation for each asset, the Shapiro-Wilk test is not very suitable to use. In all cases except DOW p-
value does not reach 5 % limit of significance than the null hypothesis for all cases can be rejected and
conclude that daily return series do not come from a normal distribution. As for the DOW, according to
Shapiro-Wilk and Jarque Bera tests, the hypothesis of normally distributed returns is confirmed at 95 %
confidence level. After looking at the results of the Anderson-Darling test we can deduce that the null
hypothesis is rejected but compared to the rest, there is a visible difference in test statistics. Only in this
case is the statistics is lower than zero. The same difference is visible in the case of the Jarque Bera test.
Stationarity and autocorrelation tests
If we look at the Ljung box test (Table 2) the conclusion about stationarity is that only in 10 time series
(e.g. AAPL, BA, CAT, VZ) the hypothesis about stationarity of 1 order is confirmed by this test. In
other words, the similar time series as white noise is confirmed in 10 time series. After running function
for Augmented Dickey-Fuller and Phillips–Perron tests we can see that the significances of the statistics
are identically smaller than 0.05 in all cases which means that the null hypothesis is rejected, and the
time series are not stationary. Finally, the KPSS test for level stationarity is provided and the main
attention is again paid to coefficient of significance. Contrary to the previous test all p-values are higher
than 0.05 than the null hypothesis is rejected.
Parameters estimation and testing of particular distribution
It is permissible to find individual parameters defining other types of distribution. Firstly, the parameter
estimation of stable distribution by using maximum likelihood and McCulloch method is in Table 3.
Table 1. Results of particular tests of normality
Kolmogorov-Smirnov Shapiro-Wilk Jarque Bera Anderson-Darling test
Statistic Sig. Statistic Sig. Statistic Sig. Statistic Sig.
AAPL .469 .000 .940 .000 6123.7 .000 39.125 .000
AXP .464 .000 .836 .000 25315 .000 124.09 .000
BA .472 .000 .945 .000 3201.1 .000 38.064 .000
CAT .470 .000 .937 .000 4473.8 .000 42.741 .000
CSCO .471 .000 .888 .000 19942 .000 64.822 .000
CVX .476 .000 .905 .000 24308 .000 44.559 .000
DIS .474 .000 .902 .000 11920 .000 58.439 .000
DOW .476 .000 .975 .174 1.094 .579 .796 .037
GS .467 .000 .856 .000 37323 .000 79.152 .000
HD .474 .000 .935 .000 4576.9 .000 47.115 .000
IBM .478 .000 .927 .000 5487.8 .000 46.978 .000
INTC .471 .000 .939 .000 4223.7 .000 37.646 .000
JNJ .483 .000 .904 .000 25671 .000 49.043 .000
JPM .463 .000 .815 .000 42219 .000 129.71 .000
KO .482 .000 .903 .000 25370 .000 49.465 .000
MCD .480 .000 .939 .000 4756 .000 34.76 .000
MMM .478 .000 .905 .000 11441 .000 63.704 .000
MRK .475 .000 .893 .000 19055 .000 56.446 .000
MSFT .473 .000 .907 .000 13686 .000 55.397 .000
NKE .079 .000 .907 .000 9348 .000 54.223 .000
PFE .074 .000 .932 .000 7229 .000 40.136 .000
PG .481 .000 .917 .000 9285 .000 51.99 .000
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
150
TRV .473 .000 .816 .000 100420 .000 108.25 .000
UNH .469 .000 .852 .000 98849 .000 73.984 .000
UTX .477 .000 .928 .000 7189 .000 45.729 .000
V .471 000 .909 .000 8621.6 .000 48.723 .000
VZ .479 .000 .933 .000 9359 .000 33.505 .000
WBA .474 .000 .911 .000 18076 .000 43.292 .000
WMT .480 .000 .905 .000 18037 .000 46.067 .000
XOM .477 .000 .891 .000 29847 .000 51.242 .000
Source: Neděla (2020)
Table 2. Results of particular stationarity tests
Ljung-Box test Dickey-Fuller test KPSS test Phillips-Perron test
Statistic Sig. Statistic Sig. Statistic Sig. Statistic Sig.
AAPL .037 .847 -42.263 .010 .060 .100 -15.944 .010
AXP 26.770 .000 -45.127 .010 .127 .100 -63.906 .010
BA .262 .609 -41.971 .010 .278 .100 -57.764 .010
CAT .328 .567 -41.254 .010 .050 .100 -57.670 .010
CSCO 4.632 .031 -43.828 .010 .151 .100 -60.686 .010
CVX 22.233 .000 -45.875 .010 .028 .100 -63.487 .010
DIS 11.669 .001 -45.356 .010 .067 .100 -62.100 .010
DOW .015 .904 -5.466 .010 .165 .100 -8.070 .010
GS 5.696 .015 -42.995 .010 .044 .100 -61.054 .010
HD .439 .508 -43.053 .010 .307 .100 -57.708 .010
IBM .911 .340 -41.775 .010 .170 .100 -59.362 .010
INTC 17.673 .000 -43.648 .010 .119 .100 -62.763 .010
JNJ .019 .003 -45.384 .010 .090 .100 -61.519 .010
JPM 29.692 .000 -44.410 .010 .072 .100 -65.417 .010
KO 10.196 .001 -44.646 .010 .076 .100 -62.086 .010
MCD 11.382 .001 -45.848 .010 .056 .100 -62.473 .010
MMM 7.413 .006 -43.573 .010 .089 .100 -61.144 .010
MRK 1.877 .171 -43.237 .010 .067 .100 -59.831 .010
MSFT 16.021 .000 -45.304 .010 .364 .093 -63.370 .010
NKE 9.373 .002 -43.993 .010 .032 .100 -62.133 .010
PFE 7.698 .006 -44.363 .010 .144 .100 -61.534 .010
PG 21.600 .000 -45.884 .010 .077 .100 -63.556 .010
TRV 114.890 .000 -48.591 .010 .052 .100 -71.905 .010
UNH 1.366 .242 -42.180 .010 .391 .081 -59.469 .010
UTX 12.992 .000 -44.229 .010 .030 .100 -62.150 .010
V 25.467 .000 -42.462 .010 .039 .100 -59.864 .010
VZ .384 .535 -44.926 .010 .040 .100 -59.148 .010
WBA 3.030 .082 -42.674 .010 .125 .100 -60.006 .010
WMT 12.150 .000 -44.945 .010 .070 .100 -62.385 .010
XOM 49.594 .000 -48.367 .010 .088 .100 -66.954 .010
Source: Neděla (2020)
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
151
Table 3. Parameter estimation of stable distribution
Maximum Likelihood Estimation McCulloch Estimation
AAPL 1.6198 -.0439 .0108 .0012 1.5100 .0300 .0101 .0008
AXP 1.3796 -.0832 .0085 .0010 1.3260 -.0880 .0083 .0008
BA 1.6199 -.1073 .0094 .0012 1.5450 -.1040 .0093 .0011
CAT 1.6082 -.0953 .0106 .0001 1.4670 -.0490 .0098 .0007
CSCO 1.5748 -.0886 .0087 .0009 1.4920 -.0610 .0082 .0007
CVX 1.6533 -.1709 .0086 .0010 1.5560 -.1810 .0082 .0012
DIS 1.5878 -.0502 .0080 .0008 1.4770 -.0670 .0076 .0009
DOW 1.9990 -.1964 .0158 .0001 1.4240 -.2460 .0115 .0035
GS 1.5654 -.0765 .0106 .0007 1.4750 -.0910 .0102 .0007
HD 1.5649 -.0356 .0082 .0007 1.3900 -.0130 .0078 .0005
IBM 1.6972 -.1501 .0073 .0007 1.5200 -.0950 .0068 .0006
INTC 1.6437 -.1158 .0098 .0009 1.5430 -.0870 .0094 .0009
JNJ 1.6363 -.0921 .0053 .0007 1.5160 -.0480 .0050 .0004
JPM 1.4031 -.0185 .0093 .0004 1.3420 -.0800 .0090 .0005
KO 1.6237 -.0944 .0057 .0007 1.5260 -.0770 .0054 .0007
MCD 1.6735 -.1104 .0064 .0009 1.5540 -.0920 .0060 .0010
MMM 1.5429 -.1281 .0066 .0011 1.4530 -.0860 .0063 .0009
MRK 1.6328 -.0248 .0077 .0006 1.5870 -.0090 .0075 .0004
MSFT 1.5805 .0013 .0084 .0006 1.4960 -.0240 .0080 .0005
NKE 1.6082 -.0406 .0085 .0008 1.4610 -.0250 .0082 .0008
PFE 1.6339 .0002 .0072 .0004 1.5020 .0036 .0068 .0001
PG 1.5844 -.0163 .0054 .0004 1.4640 -.0050 .0052 .0003
TRV 1.4923 -.0701 .0070 .0008 1.4280 -.0570 .0070 .0008
UNH 1.5755 -.0706 .0093 .0008 1.5250 -.0240 .0091 .0006
UTX 1.6025 -.0670 .0075 .0007 1.5000 -.0600 .0072 .0006
V 1.5538 -.0775 .0091 .0013 1.4720 -.0610 .0086 .0014
VZ 1.6917 -.0129 .0072 .0008 1.6320 -.0880 .0071 .0008
WBA 1.6593 -.0427 .0089 .0004 1.5970 -.0340 .0088 .0003
WMT 1.6564 -.1030 .0064 .0006 1.5550 -.0750 .0062 .0007
XOM 1.6500 -.1015 .0076 .0006 1.5550 -.0690 .0073 .0004
Source: Neděla (2020)
In the first method of estimation the average values about 1.6 of the parameter suggest the stronger
presence of heavy tails, since the skewness values are mostly negative, and average is about -0.07
meaning the presence of more values to the left of the mean. By the second method the estimated
parameters are slightly different, especially in the first two parameters.
Estimation of variance gamma distribution by used method based on Nelder-Mead algorithm follows in
Table 4.
Table 4. Parameter estimation of variance gamma distribution
AAP
L .0000
.042
7 .0017
2.483
0 IBM
-
.000
8
.024
0
.001
3
2.645
2 PFE
.000
5
.012
6
-
.000
2
.2269
AXP -
.0002
.025
4 .0004
2.070
9 INTC
.001
4
.017
7
-
.001
1
.8408 PG .000
4
.010
3
-
.000
1
.9396
BA .0027 .016
5
-
.0021 .2005 JNJ
.000
0
.016
1
.000
6
2.546
0 TRV
.001
2
.014
0
-
.000
7
.3073
CAT .0009 .040
7
-
.0002
2.506
4 JPM
-
.000
6
.051
3
.008
7
2.824
5 UNH
.000
0
.038
3
.000
7
2.715
0
CSC
O .0017
.015
9
-
.0013 .2949 KO
.000
9
.010
4
-
.000
5
0.883
5 UTX
.001
5
.013
4
-
.001
2
.2075
CVX .0016 .015
5
-
.0012 .8344 MCD
.001
2
.014
4
-
.000
6
1.721
5 V
.001
6
.017
0
-
.000
6
.1704
DIS .0009 .015
3
-
.0003 .9987
MM
M
.000
5
.019
7
-
.000
4
2.370
7 VZ
.001
4
.021
4
-
.000
7
2.484
3
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
152
DOW .0088 .023
4
-
.0080
1.271
0 MRK
.000
4
.014
5
.000
1 .9082 WBA
.001
8
.015
4
-
.001
7
.2137
GS .0012 .031
8
-
.0007
2.131
2
MSF
T
.000
9
.015
3
-
.000
4
.2020 WM
T
.000
7
.022
5
-
.000
5
2.520
2
HD .0004 .015
6 .0002 .9957 NKE
.000
0
-
.015
4
.000
8 .2019 XOM
.001
7
.013
5
-
.001
6
.1334
Source: Neděla (2020)
It can be seen in Table 4 large differences between γ parameter estimated values which is scale parameter
and small disparity is also of tail index. The remaining parameters do not show any significant
changes.
Other parameters estimation is applied for generalized extreme value distribution by using maximum
likelihood method. This distribution is chosen because some financial time series contain many extreme
values.
The location of the GEV distribution changed from -0.100 to -0.0041, and the scale is from 0.0131 to
0.0325. From the estimated shape parameters can be calculated tail index and it is advisable to perform
statistical tests if the distribution is related to dataset. Results of particular tests in Table 5 indicate
conclusion: in all cases the hypothesis (time series are distributed by given distribution) is rejected.
Table 5. Particular tests for estimated distribution
Source: Neděla (2020)
Conclusion
The main goal of this paper was to analyze and test time series of daily stock returns during a chosen
time period for using in portfolio optimization models and strategies.
It was found out that the distances between the mean values and the maximum or minimum values were
large, indicating that the presence of extreme values. If density curves and their logarithm were created,
the fluctuations that indicated heavy tails of data set, caused by founded extreme values, should be seen.
After testing normal distribution by four statistical tests was detected that we should consider
distribution with heavier tails. In this paper, Pearson, Spearman and Kendall correlations were calculated
and compared. Based on autocorrelation results, in all cases the hypothesis that the time series is
statistically significant dependent until the lag 15 was rejected. Through Ljung-Box autocorrelation test
in 10 time series, the hypothesis about dependency of first order was confirmed. Due to Augmented
Dickey-Fuller and Phillips-Perron tests the null hypothesis was rejected, but by the KPSS test the null
hypothesis about level stationarity was failed to reject.
K- S test for stable distribution Lilliefors test for generalized extreme values
distribution
Stat. Sig. Stat. Sig. Stat. Sig. Stat. Sig. AAPL .479 .000 MCD .491 .000 AAPL .149 .001 MCD .174 .001
AXP .478 .000 MMM .489 .000 AXP .241 .001 MMM .174 .001
BA .485 .000 MRK .480 .000 BA .189 .001 MRK .210 .001
CAT .481 .000 MSFT .479 .000 CAT .166 .001 MSFT .237 .001
CSCO .483 .000 NKE .482 .000 CSCO .210 .001 NKE .197 .001
CVX .491 .000 PFE .483 .000 CVX .253 .001 PFE .167 .001
DIS .482 .000 PG .486 .000 DIS .224 .001 PG .219 .001
DOW .483 .000 TRV .483 .000 DOW .163 .001 TRV .309 .001
GS .478 .000 UNH .481 .000 GS .261 .001 UNH .311 .001
HD .483 .000 UTX .484 .000 HD .195 .001 UTX .212 .001
IBM .489 .000 V .483 .000 IBM .167 .001 V .200 .001
INTC .484 .000 VZ .484 .000 INTC .155 .001 VZ .227 .001
JNJ .491 .000 WBA .481 .000 JNJ .255 .001 WBA .212 .001
JPM .472 .000 WMT .471 .000 JPM .261 .001 WMT .221 .001
KO .491 .000 XOM .488 .000 KO .262 .001 XOM .251 .001
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
153
At the end of the application part, the parameters of the stable, variance gamma and generalized extreme
value distribution functions were estimated by using several methods. Then the parameters of the
different distribution functions were tested. KS test for stable distribution or Lilliefors test for
generalized extreme values distribution, identically rejected the null hypotheses. However, if a QQ plot
for AXP or HD is compiled than it may be recommended to use variance gamma distribution.
References
[1] BARNDORFF-NIELSEN, Ole E. Normal Inverse Gaussian Distributions and Stochastic
Volatility Modelling. Scandinavian Journal of Statistics. 1997, 24(1), 1–13.
[2] FAMA, Eugene. The Behavior of Stock-Market Prices. Journal of Business. 1965, 38(1), 34-
105.
[3] FAMA, E. and ROLL, R. Parameter Estimates for Symmetric Stable Distributions. Journal of
the American Statistical Association. 1971, 66(334),331-338.
[4] LÉVY, Paul. Théorie des erreurs. La loi de Gauss et les lois exceptionnelles. Bulletin de la
Société Mathématique de France. 1924, 52, 49-85. ISSN 0037-9484.
[5] MARKOSE, Sheri and Amadeo ALENTORN. The generalized extreme value distribution,
implied tail index, and option pricing. The Journal of Derivatives. 2011, 18(3), 35-60.
[6] MARKOWITZ, Harry. Portfolio Selection. The Journal of Finance. 1952, 7(1), 77-91.
[7] MCCULLOCH, John. Simple consistent estimators of stable distribution
parameters. Communications in Statistics - Simulation and Computation. 1986, 15(4), 1109–
1136.
[8] NELDER, John A. a Roger MEAD. A Simplex Method for Function Minimization. The
Computer Journal. 1965, 7(4), 308–313.
[9] SHAPIRO, S. S. a M. B. WILK. An Analysis of Variance Test for Normality (Complete
Samples). Biometrika. 1965, 52(3/4), 591-611.
[10] TSAY, Ruey S. Analysis of financial time series. 3rd ed. Cambridge, Mass.: Wiley, 2010. ISBN
978-0-470-41435-4.
[11] TSAY, Ruey S. An introduction to analysis of financial data with R. Hoboken, N.J.: Wiley,
2013. ISBN 978-0-470-89081-3.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
154
INSURABLE AND UNINSURABLE RISKS AND THEIR CLASSIFICATION FROM THE
PERSPECTIVE OF A CZECH EXPORTER
Michaela Petrová1
1Department of Law, VŠB – Technical University of Ostrava,
Sokolská tř. 33, Ostrava 70200, Czech Republic
e-mail: [email protected]
Abstract
Nowadays, not only Czech manufacturers are trying to export their products abroad. This is probably
due to the trend of trade liberalization. However, the implementation of international business
transactions involves a higher level of risk compared to domestic transactions. This paper deals with the
issue of risks that may arise in international trade and their division. The aim of the article is to define
and identify risks from the perspective of a Czech exporter based on a summary of domestic theoretical
knowledge related to risks in the international trade. Risks are also associated with insurance. Against
individual risks, exporters can take advantage of insurance offers from commercial insurance companies
or use insurance with state support. However, not all risks can be insured, so individual risks must be
viewed from the perspective of insurability and insurability on the Czech market.
Keywords
Risks, Insurance, International Trade, Insurable and Uninsurable Risks, Export
JEL Classification
F23, G22
Introduction
Every exporter is always exposed to some risk when exporting his products. Generally, risk is defined
as a result that differs somehow from the expected result due to random events. For example, according
to Smejkal (2010), risk is defined as the volatility of a financial variable around the expected value due
to some parameter changes.9 Each area of business is associated with risks, and business on foreign
markets is associated with specific risks. The concept of risk can be viewed from different perspectives.
From the perspective of an ordinary person, this is the possibility that something unfavourable can
happen. From this perspective, risk is perceived as an illness or personal loss of something valuable such
as property, loved ones, employment, etc. From a business perspective, the risk is an economic loss –
financial loss, bankruptcy, loss of market position, loss of credibility, etc. From an insurance
perspective, risk is seen as an event that may adversely affect the ability to achieve our goal. In some
cases, the term risk is confused with the word uncertainty. However, this substitution is not correct.
There is a significant difference between risk and uncertainty. Uncertainty cannot be measured or
estimated against risk. Therefore, we can say that risk is uncertainty, but we have to added that it is
uncertainty that can be measured. 2
This paper deals with the division of risks from the perspective of a Czech exporter, so the article focuses
on risks in the area of international trade. The thesis deals with the issue of using insurance as one of the
possibilities of protection against risks. The insurance transfers the impact of the risk from the
policyholder to the insurer, thus supporting the exporter to carry out his business.
The main aim of the article is to evaluate selected risks in terms of their insurability and uninsurability
on the Czech market. This objective is preceded by the need to define and identify risks from the
perspective of the Czech exporter on the basis of a summary of domestic theoretical knowledge relating
to risks in the area of international trade.
In this article are used the methods of analysis, synthesis and description. In order to write the paper, it
was necessary to search the domestic literature related to risks in the field of international trade. For
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
155
assessing insurability of selected risks, it was necessary to study the insurance conditions of insurance
companies.
Classification of the risks in international trade
The risk needs to be seen more specifically and the risk has to be characterized. Each author divides
risks differently according to various criteria, such as measurability, the effect of risk on the outcome,
causes of origin, areas, etc. There is no consistent nomenclature, but in some areas the individual authors
agree with each other. In the area of international trade, according to Černohlávková (2014), the main
types of risks are market risks, commercial risks, transport risks, territorial risks, currency risks, liability
risks and others.6 Janatka (2011) divides risks into commercial, manufacturing and market risks, risks
from non-fulfillment of obligations by the contracting parties, risks arising from errors in the negotiation
of a commercial operation, transport risks, risks associated with non-payment for delivered goods and
services, risk of liability for damage caused by a product defect.5
Böhm (2009) classify the risks even more narrowly. He declares that in the area of foreign trade practice
it is appropriate to divide the risks into traditional and modern. The traditional risks include business
risks, production risks and risks from erroneous conclusion of business operations. This group of
traditional risks is complemented by other risks like risks, which are connected with the selection of
business partners, with tariff and non-tariff barriers, with patent and trademark protection of goods.
Furthermore, risks that arise from the nature of goods, transport risks, liability and legal risks. Modern
risks include payment risks, credit and investment risks.1
Other authors classify the risks according to the areas in which they occur. For example, Cipra (2015)
characterizes individual types of risks from the perspective of banks and insurance companies.
According to him, these are the following: market risks, credit risks, liquidity risks, model risks,
operational risks and insurance risks.2
No matter how the risks are classified by individual authors, there are links and connections between
these types of risks. Some species may occur together or complement each other.5 In addition, there is a
need to look at risks in a comprehensive way, because it may happen that the restrictions of one risk can
increase the possibility of occurrence of the other risk.5
For insurance and insurability, it makes sense to divide risks into commercial and territorial, as Böhm
(2009) states. In general, commercial risks are normally insurable, while territorial risks are more
difficult or not at all. Commercial and territorial risks are among the financial risks of insurance
companies, in terms of credit insurance.1 In terms of credit risks, Petrusheva (2016) divides risks into
commercial and so-called non-commercial, which includes political and natural disaster risks.7
Commercial risks
In general, commercial risks are those resulting from the nature of the goods. Pursuant to the Convention
International Sales of Goods, “the seller must deliver the goods, hand over any documents relating to
them and transfer the property in the goods as required by the contract and this Convention”10, this
provision is also enshrined in the Czech Civil Code. The seller has a statutory obligation to deliver the
agreed goods and all documents. In order to minimize commercial risks, the contract needs to define the
characteristics of the goods as accurately as possible. However commercial risks are not only due to the
nature of the goods but can be classified according to various criteria.5
In a broader context, we perceive commercial risks as risks that may arise during the preparation of
production, through production financing, production itself, closing of sales, delivery of goods,
takeover, payment to potential claims. Specifically, as stated by Janatka (2011), these are risks related
to the production and nature of goods, risks related to the sale and delivery of goods, risks arising from
errors and deficiencies in the negotiation of sales contracts, business partner risks, territorial risks, risks
connecting with transportation, handling and storage, payment and exchange risks, product liability risks
and other specific (unforeseeable) risks.5
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
156
In terms of credit insurance, commercial risks are related to the economic and financial situation of a
foreign entity - the buyer. These are risks that can be influenced by the buyer - mainly insolvency and
protracted default.1 Insolvency generally means that the debtor is unable to pay its obligations. On the
other hand, protracted default means that the debtor is in breach of the contract by failing to pay the
claim for no legal reason. It is the unwillingness to pay. Černohlávková (2014) adds to the insolvency
and protracted default other forms of commercial risks, such as the withdrawal of a business partner
from the contract, failure to fulfil the contract or defective performance of the contract by the supplier
and unjustified non-acceptance by the customer.6
Territorial risks
Territorial risks constitute the second group of risks in international trade. In the literature we also find
the terms like non-commercial risks or political risks. Territorial risks are associated with the political,
economic, financial and legal environment in the exporting country – the debtor country. They arise
from political and economic events and measures. Unlike commercial risks arising from the debtor's
economic or financial situation. Janatka (2011) states the individual causes that lead to territorial risk.
These are various political events in the country of the debtor, such as war, revolution, rebellion, civil
unrest, strikes. Then it is legislative and political measures or administrative decisions, such as
withdrawal of import or export license, embargo, restriction of the movement of goods, etc. It may also
be caused by situations of expropriation – in the form of nationalization or confiscation. In this list of
causes must be also included the natural disasters.5
This is the most serious type of risk because it is unpredictable. They are so-called force majeure and
that is why are difficult to quantify in advance. They arise from the uncertainty of political and
macroeconomic development of individual countries. Černohlávková (2014) claims that these risks have
a negative impact on the results of individual business transactions and on the implementation of future
business plans. In her view, political risks are the most serious, because they can lead to a reduction or
even severance of relations with the country and thus to great damage.6
Risk insurability and insurability criteria
Risk prevention and their effects are used in a variety of instruments. In the time of loss, the entity may
use its own resources – so-called self-insurance or special financial institutions. The second option is
used more often. Risks are therefore closely connected to the concept of insurance.
Insurance means the transfer of risk to the insurer. The main purpose of the insurance is to provide
compensation for loss incurred as a result of accidental events.6 In addition, it is an optimal possibility
of obtaining funds in the event of damage.4 The insurer in this legal relationship is the insurance
company with which the policyholder concludes the insurance contract. The insurance company has the
right to have a reward for taking over the risk, which is provided in the form of insurance premiums.
The essence of insurance is risk management, taking risks from clients and working with risks.11
Probably everyone knows that not all risks can be insured. In order to consider whether the risk can be
insured, it must meet the requirements of insurability. Therefore, there are criteria to determine whether
the risk is insurable. These are the so-called insurability criteria and the individual criteria are the
contingency criterion, the uniqueness criterion, the estimability criterion, the independence criterion, the
criterion of the size and moral principles. The insurability criteria are summarized and explained in the
following table (Table 1). These are the conditions under which insurance companies determine the
characteristics of a given risk and assist them in deciding whether to take the risk into insurance.3
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
157
Table 1. Insurability criteria
Criterion Explanation
Contingency criterion A random event; uncertain, uncontrollable, unpredictable
and unexpected risk
Uniqueness criterion Clearly descriptive, demonstrable and unmistakable risk
Estimability criterion Identifiable and measurable probability of realization of
risk, valuability of damage
Independence criterion Independence of individual risks*
Criterion of the size The insurer's ability to bear the risk**
Moral principles Do not help to avoid punishment for damages caused by
acting in circumstances that are not considered moral
Notes: * Dependent risks are cumulative risks, contagion risks, risks with fluctuating underlying probability such as storm or hail
** The size of the risk is determined by the amount of damage that may occur in the realization of the risk
Source: Ducháčková (2009)
If any of the above criteria is not met, it is an uninsurable risk.
Insurability is further investigated by the insurer in three terms:3
– in terms of the accidental nature of insured events – the probability of the insured event
not being too high (certainty that the insured event will occur) and the occurrence of
the event must not be influenced by the policyholder,
– in terms of the size of assurance benefit in case of realization of the risk – the
possibility of causing too much damage,
– in terms of the achievement of insurance protection – insurability options in terms of
surface and temporal distribution of risks (objective risk assessment).
Risks in international trade are different from domestic risks. They are difficult to predict and are harder
to remove. According to Černohlávková (2014), the most frequently used types of insurance in
international trade are transport risk insurance, credit and investment risk insurance including payment
instruments, liability insurance and insurance for fairs and exhibitions.6
Discussion – Insurance of selected risks on the Czech market
The division of risks into commercial and territorial is important from the insurance point of view,
namely because the risk can be insured with a regular commercial insurance company or if it is necessary
to use insurance with state support. For the scope and use of state-supported insurance are important
directives issued by the European Council and the European Parliament. According to these directives,
the risks are divided according to whether they are marketable. Marketable risks are commercial risks.
On the other hand, territorial risks are not marketable due to insufficient reinsurance capacity of
commercial insurance companies. Marketable risks include:5
– arbitrary failure to recognize a contract by a foreign debtor,
– arbitrary refusal to take over the goods by the debtor,
– insolvency of the foreign debtor or its guarantor,
– long-term default on obligation of a foreign debtor.
The above risks can be covered by commercial credit insurance, for all the others can use state-supported
insurance. The purpose of state aid is to assume a certain risk and to provide a guarantee, especially
where the guarantees of commercial insurance companies cannot cover the exporter to such an extent
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
158
or are unavailable to the exporter.8 Insurance companies themselves decide about insurability. Insurance
companies therefore decide which risks they will insure and at what price.
In the article I present selected risks that a Czech exporter may encounter when is exporting his goods
or services. Risks are viewed in terms of insurability on the Czech market. To assess the insurability of
selected risks were examined the individual insurance conditions of insurance companies that offer their
insurance products on the Czech market.
Defective performance
This type of risk can be easily insured with commercial insurance companies on the Czech market. A
product defect is a condition where the product does not exhibit specified, notified or agreed
characteristics. Characteristics which can reasonably be expected of it to have regard to all the
circumstances - the intended purpose for which the product is to be used (also taking into account when
the product was launched).
In most cases, insurance is offered under liability insurance. Most often it is the liability insurance for
damage caused by a product defect. The aim of insurance is to minimize the damage that a defective
product can cause to the end consumer. The purpose of the insurance is therefore to cover the damage
to health or property of third parties, which occurred in connection with the use of a product which did
not guarantee the safety characteristics that could be reasonably expected and which the insured persons
is obliged to replace. The idea is not to relieve the policyholder of the liability for the damage caused,
but to limit and assume the negative economic impacts. Liability for damage caused by product defects
is usually agreed separately in the insurance contract.
This is an insurable risk, but there may also be situations where normally insurable risk becomes
uninsurable. These are situations that are contained in exclusions. The insurance does not cover, for
example, damage caused intentionally, damage caused by insufficiently pre-tested product, product that
does not reach the agreed functional parameters, etc.
Insolvency and protracted default
As part of commercial insurance, entrepreneurs have the possibility to insure against situations such as
insolvency and protracted default. This is again an insurable risk. In practice, it is possible to meet with
insurance of debts, where there is a possibility to insure the risk of non-payment of invoices for goods,
such as insolvency and protracted default. Insolvency occurs when the insolvency of the customer is
detected by the competent authority under the insolvency law. Protracted default is defined in the
insurance terms as non-payment of a debt by the due date for reasons other than insolvency.
Even within this insurance, we may encounter situations where the insurance may not occur. Insurance
companies will not insure claims in the event of a fault caused by the policyholder, invalidity of the
contract, breach of the relevant legal regulations or due to breach of the terms of the contract by the
insured person, etc.
Against this risk can be also used state-supported insurance offered by the Export Guarantee and
Insurance Company (EGAP) in the Czech Republic. EGAP offers, for example, the possibility of
insolvency insurance under insurance against the inability to fulfil an export contract. The inability to
fulfil the contract here means the partial or total impossibility of performance, even the economic
impossibility, when the insured person cannot be reasonably required from the insured point of view,
even if the obligation is legally and effectively fulfilled. According to EGAP, the inability to fulfil may
be firstly due to the insolvency of the importer - a bankruptcy decision or the rejection of an insolvency
petition for lack of property. Secondly, breach of the export contract by the importer - refusal of
performance or inaction.
Withdrawal from the contract
The above mentioned risk of inability to fulfil the contract may lead to the suspension or even
cancellation of the export contract. The reasons, in addition to insolvency and protracted default, may
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
159
be the refusal to take delivery of the goods supplied, or may be due to the political, financial or
macroeconomic situation in the importer's country. Withdrawal risk is a commercial risk, so it is
insurable risk.
Situations such as the existence of a dispute over the fulfilment of an export contract, the insured person
has breached the norms or customs of international law, etc. may lead to the risk, that the risk will be
difficult insurable or uninsurable.
On the Czech market, this type of insurance can be insured through the EGAP insurance company. It
offers insurance against cancellation or interruption of the contract during production.
Natural disasters
In the Czech Republic there is a possibility of insurance against natural disasters. The best way to
reduction the financial consequences of disasters is to use commercial insurance. Insurance against
natural disasters does not meet the criterion of independence. Natural disasters affect a high percentage
of the population, causing high damage to property in similar locations and at the same time. The
insurance company solves this problem by grouping several risks into one insurance contract. Another
criterion that is not met is the contingency criterion. The problem is the low level of predictability of
disaster risk. Insurance companies are not able to obtain sufficient information about the economic
impacts, and this uncertainty prevents insurance companies from correctly assessing the risks, which
they undertake and estimate the probabilities of disasters of varying severity.
Nevertheless, these risks are at least partially insured on the Czech market. Modern risk estimation
methods, evaluation tools and risk maps have been developed primarily for earthquakes, tropical
cyclones, hurricanes and floods. In the Czech Republic, insurance companies use flood maps, but on the
other hand, for example, for the risk of a windstorm there is no geographical information system.
Catastrophic risks appear in commercial insurance as part of life insurance, which covers the risk of
death in connection with a catastrophic event and then in property insurance, primarily to cover damage
caused by catastrophic events. Insurance companies to natural risks include fire, explosion, lightning,
windstorm, flood, hail, frost, earthquakes, falling trees and poles, land subsidence, landslip or collapse
avalanches, heavy snow and glaze ice. From the point of view of insurability is important for the
insurance company the area to which the insurance will be apply. In the event of a recurrent natural
disaster in the area, this risk may be uninsurable. In practice, insurance companies do not like to insure
natural disasters, such as earthquakes, tsunamis, volcano eruptions, etc., mainly because of concerns
about serious impacts.
Increasing damage from natural disasters often exceeds the capacity of insurance markets and thus
creates further barriers to insurability of catastrophic risks, therefore the state is also involved in the
coverage of damage caused by natural disasters.
Terrorist attacks
Nowadays terrorism is on the expansion. Recent events all over the world enhance uncertainty and are
merely examples of terrorist acts and violence caused by terrorism. From the policyholder's point of
view, this is an event against which he or she has a need to protect himself and this insurance option is
important to him. For the exporter, this represents the confidence to expand abroad and to areas where
the possibility of terrorism is possible. In recent years, it has been shown that attacks can occur even in
a quiet region.
In insurance a terrorist attack is a violent act or series of acts committed by any person or group of
persons acting alone or in connection with any organization founded for political, religious or
ideological reasons in order to influence any government and/or raise public fear.
This type of insurance is quite new in practice and can be said that is insurable with difficulty. This is a
threat that is very difficult to predict and qualify. The insurance does not usually cover damage caused
by war, invasion, rebellion, uprising, nuclear energy or nuclear radiation and deliberate actions of the
policyholder or the insured person. The high risk and high level of uncertainty of this risk is reflected in
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
160
very high premiums or even in the unwillingness of the private insurance sector to cover these risks.
Most insurance companies today cover at least partially the consequences of terrorist attacks. However,
a frequent limitation of the payment of assurance benefits is if the terrorist attack occurs at a place where
the Ministry of Foreign Affairs has declared that it does not recommend citizens to travel to the area or
state.
A summary table with an overview of risks and insurability rates can be seen below (Table 2). We can
see that common commercial risks are insurable on the Czech market without a problem. Almost all
commercial insurance companies offer insurance against these risks. Insurance against territorial risks
is already more difficult in practice, but it can still be at least partially insured. The only exceptions are
war, rebellion, uprising and other war events that cannot be insured against.
Table 2. Overview of selected risks and their rate of insurability
Risk Type of risk Insurability
Defective performance commercial insurable
Insolvency commercial insurable
Protracted default commercial insurable
Withdrawal from the contract commercial insurable
Natural disasters (in general) territorial partially insurable (uninsurable)
Terrorist attacks territorial insurable with difficulty
War territorial uninsurable
Source: own processing
In case of insurance of some risks it is advantageous to use ART methods. So we can use alternative
transfer of risk than conventional insurance, either due to uninsurability or high premium.
Conclusion
The subject of the paper was to define, identify and allocate risks from the perspective of a Czech
exporter in the field of international trade. The aim of the article was to evaluate selected risks in terms
of insurability and uninsurability on the Czech market. Insurance companies decide on insurability based
on the fulfilment of insurability criteria, which are the criteria of contingency, uniqueness, estimability,
independence, criterion of the size and moral principles.
The research shows that commercial risks are insurable without problems. These are risks that can be
influenced by the buyer and are primarily insolvency and protracted default. On the other hand,
insurance companies are generally not able to insure deliberate actions, unlawful actions, too high
expected damage and too high and unquantifiable risk. Commercial risks are insurable on the Czech
market by regular commercial insurance companies. For some risks it is advisable to use state-supported
insurance, in some cases it may even be the only insurance option. We can also say that today there are
few risks that cannot be at least partially insured against. However, everything is subsequently reflected
in the amount of the premium and the amount of the participation. An exception to insurability is
represented by various war events, such as war, rebellion, uprising, etc., against which cannot be insured.
Insurance companies also report absolute exclusions that cannot be included in any type of insurance.
These are mainly damages caused by deliberate actions of the policyholder or the insured person or by
another person on their orders. Furthermore, the insurance company does not provide assurance benefits
if there is a deliberate negligence, breach of the relevant legal regulations or due to breach of the terms
of the contract, etc.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
161
However, ART methods, such as captive insurance, can be used for insurance of uninsurable risks. The
use of ART methods instead of conventional insurance may be subject to further investigation.
Today, insurers are looking for new approaches to risk insurance, so that the risk remains insurable
while not endanger the existence of the insurance company itself. Most often, insurance companies make
premiums more expensive, change the amount of participation and the upper limit of assurance benefits.
In extreme cases, the insurance company has to withdraw its insurance product and the risk becomes
uninsurable. An example is flood insurance in risk areas.
Acknowledgement
This paper was financially supported within the VŠB–Technical University SGS grant project No.
SP2020/77 (Risk Assessment in International Trade in Selected OECD Countries and Risk Minimization
in the Context of the Czech Exporter).
References
[1] BÖHM, Arnošt. Pojištění pohledávek v mezinárodním obchodě. Praha: Professional Publishing,
2009. ISBN 978-80-7431-004-1.
[2] CIPRA, Tomáš. Riziko ve financích a pojišťovnictví: Basel III a Solvency II. Vydání I. Praha:
Ekopress, 2015. ISBN 978-80-87865-24-8.
[3] DUCHÁČKOVÁ, Eva. Principy pojištění a pojišťovnictví. 3. aktualiz. vyd. Praha: Ekopress,
c2009. ISBN 978-80-86929-51-4.
[4] JANATA, Jiří. Pojištění a management majetkových podnikatelských rizik. Praha: Professional
Publishing, 2004. ISBN 80-86419-64-9.
[5] JANATKA, František. Rizika v komerční praxi. Praha: Wolters Kluwer Česká republika, 2011.
ISBN 978-80-7357-632-5.
[6] MACHKOVÁ, Hana, Eva ČERNOHLÁVKOVÁ a Alexej SATO. Mezinárodní obchodní
operace. 6. aktualiz. a dopl. vyd. Praha: Grada Publishing, 2014. ISBN 978-80-247-4874-0.
[7] Petrusheva, N. (2016). Management of Financial Risks in International Trade Financing. Časopis
za ekonomiju i tržišne komunikacije, Vol. 6, Issue 1. pp. 81-92
[8] ROZEHNALOVÁ, N. (2010). Právo mezinárodního obchodu. 3. Vydání. Praha: Wolters Kluwer,
ČR, a. s.
[9] SMEJKAL, V., RAIS, K. Řízení rizik ve firmách a jiných organizacích. 3. vydání. Praha: Grada
Publishnig, 2010. ISBN 978-80-247-3051-6
[10] Úmluva OSN o smlouvách o mezinárodní koupi zboží
[11] VÁVROVÁ, Eva. Finanční řízení komerčních pojišťoven. Praha: Grada Publishing, 2014. Expert.
ISBN 978-80-247-4662-3
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
162
GOVERNANCE STRUCTURES OF MUNICIPAL ENTERPRISES – EMPIRICAL STUDY
OF EFFICIENCY OF HOSPITALS
Sabrina Lee1
1Department of Economics, VŠB – Technical University of Ostrava,
Sokolská tř. 33, Ostrava 70200, Czech Republic
e-mail: [email protected]
Abstract
This paper deals with the complex of topics of efficiency measurement of municipal hospitals in Baden-
Württemberg, Germany. The municipal economy and in particular the assurance of public health
services has a very high economic and socio-political importance. The aim is to make a scientifically
solid and empirically proven contribution to the efficiency measurement of hospitals and the influence
of their governance structures. The core question is which essential input and output factors exist in
municipal hospitals for measuring efficiency and how these can be supported with regard to responsible
public governance. To answer this question, a modelling was performed using Data Envelopment
Analysis (DEA) and Free Disposable Hull (FDH). It was found that smaller hospitals with facultative
supervisory boards have considerable potential. Larger hospitals could be filtered out as inefficient.
Thus, this work provides a contribution to the science and practice of municipal hospitals.
Keywords
Health services, hospitals, hospitals in public ownership, efficiency, Data Envelopment Analysis
model (DEA), Free Disposable Hull (FDH).
.
JEL Classification
C10, C14, I10, I18
Introduction
The municipal economy is an important part of the overall economy and directly affects all citizens. The
performance of public tasks by different public enterprises is also clearly visible to citizens. Whether in
local public transport, health care, real estate and housing, water supply, energy (electricity and gas) or
swimming pools, museums and theatres, the practical relevance of public enterprises is significant
(Westermann and Cronauge, 2006; Fabry and Appel, 2011). As a representative of the performance of
public tasks, the German health care system in particular has for many years been the subject of
increasing political and economic discussion. Almost in alternation, the efficiency of the health care
system, the quality of the services offered and the level of expenditure and costs are the subject of dispute
(Kuchinke et al., 2004; Helmig, 2005; Augurzky and Schmitz, 2010; Kalb, 2010). This political and
economic discussion on the efficiency and quality of the public health care system has been taking place
for many years, not only in Germany but also at international level (see exemplary Vrabková and
Vaňková, (2015); Kalb, (2010)).
In the discussion on cost saving potentials, of the various groups of service providers such as physicians,
pharmacies and hospitals and their respective associations, the concentration on the hospital sector
appears to be the most appropriate. Especially since the hospital sector is regularly mentioned first when
it comes to cost saving potentials and the question of the cost explosion in the health care system
(Helmig, 2005; Taube, 1988). At first glance, this seems understandable, since in absolute figures,
expenditure on inpatient hospital services makes up the largest share of costs in the German health care
system. As a result, hospitals are subject to particularly strong public observation and are under great
pressure to justify their corporate policy and financial activities. Publicly owned hospitals are especially
in the spotlight because they are considered comparatively inefficient and are financed by public funds
(Helmig, 2005). There are great differences between hospitals in this respect. While some have already
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
163
taken steps to increase efficiency years ago and are now in a good economic position, there are also
hospitals that could not survive without the financial support of their owners (Augurzky and Schmitz,
2010). In this paper, efficiency is understood to mean technical efficiency. This is defined as the ability
to produce a certain amount of output with the least possible amount of input (Augurzky and Schmitz,
2010; Kuchinke et al., 2004). In contrast, inefficiency exists if the production costs are higher than the
costs that can be achieved with the given state of the technology (Kuchinke et al., 2004).
The present study therefore continues at this point. Using econometric methods, it investigates and
measures the efficiency and quality of German hospitals owned by local authorities. For this purpose,
the most recent data from official hospital statistics are used. A special focus is on the differences
between the different management structures in various legal forms of municipal hospitals.
The objective of the paper is to evaluate the technical efficiency of 89 German public hospitals in the
year 2018 based on selected inputs and outputs as well as the input-oriented DEA and FDH.
Relevance of previous research
Also in the literature on efficiency in the public sector, the health care sector has received most attention
in recent decades. There are a large number of studies that examine the technical or cost efficiency and
the respective influencing factors of health care facilities (such as hospitals or even nursing homes) in
different countries. However, the absolute majority of studies refer to the United States of America
(Kalb, 2010; Helmig, 2005).
Early studies go back to Banker et al. (1986), Wilson and Jadlow (1982), Grosskopf and Valdmanis
(1987), Nyman and Bricker (1989) and Valdmanis (1992). These studies use linear programming
techniques to evaluate different aspects of the technical efficiency of hospitals and nursing homes in the
United States of America. The last four studies also examine the influence of the form of ownership on
technical efficiency (Kalb, 2010).
Hospitals are an essential part of the health care system. In the Federal Republic of Germany, the
assurance of health care and thus also of the hospital system is regarded as a public task (Helmig, 2005).
The declared aim of the legislature is to have a variety of providers and operators. In particular - in
addition to the public institutions - the economic sustainability of non-profit and private hospitals must
be guaranteed (Deutscher Bundestag, 2019a).
Only just one in three hospitals is still in public ownership. Between 1991 and 2017, the share of public
hospitals fell from 46.0 % to 28.8 %. In the same period, the share of privately owned hospitals increased
steadily from 14.8 % to 37.1 %. This shows that privatization in the hospital sector is also continuing to
make strong progress. The share of non-profit hospitals, on the other hand, decreased only slightly from
39.1 % to 34.1 %. Figure 1 (appendix) shows the hospitals by ownership and the number of beds by
ownership in 2017. Around two thirds (59.8 %) of public hospitals are organized under private law (e.g.
limited liability companies, in germen “GmbH”). The share of private legal forms has thus doubled since
2002 to 28.3 % within a very short time. By way of comparison, the share of public hospitals that are
operated as legally dependent institutions, such as own operations ("Eigenbetrieb") and controlled
operations ("Regiebetrieb"), will be 15 % in 2017. In 2002, their share was 56.9% (Statistisches
Bundesamt, 2018).
Therefore, the ownership and legal form of hospitals is of enormous importance. Against this
background, special challenges arise with regard to the governance, management and supervision of
public municipal hospitals. On the one hand, they are situated in the area of conflict between market
economy and economic efficiency, but on the other hand, as institutions of the public sector, they have
to perform tasks of public services of general interest at the municipal level and are therefore subject to
a dual system of objectives (Ruter et al., 2005; Schaefer et al., 2008; Hilb et al., 2013). In order to make
clear the connection between the efficiency measurement of public hospitals and the differences between
the respective legal forms and the management structures derived from them, a corresponding study is
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
164
necessary. From the studies on efficiency measurement mentioned so far, it can be deduced that in some
cases they only take into account the ownership as exogenous influencing factors. Only Helmig (2005),
Frohloff (2007), Herr (2008), Herr et al. (2009, 2011), Tiemann and Schreyögg (2011) consider the
differences between private, non-profit and public sponsorship. Public sponsorship with its different
legal forms (legally independent, legally dependent and under private law) is not examined in any of
these studies.
Theoretical approach to analyze the efficiency of municipal hospitals and their
governance structures
The present paper refers to the theoretical approach of the "New Institutional Economics". This looks at
the enterprise as an institution for the combination of production factors and analyses the process of the
production of goods from a legal and economic point of view. The focus is not on production factors
from a technical-economic perspective and their ownership, but on property rights over a factor or a
specific good (Wöhe et al., 2016; Roßberg, 2007). While the external factors of enterprises cannot be
influenced for the most part, internal factors of enterprises, such as structures and processes, can be
designed in many ways (Wöhe et al., 2016).
The essential three sub-areas of the New Institutional Economics are in particular
− The "Property Rights Theory", which forms the theoretical foundation of the new institutional
economics and deals with the rights of disposition of resources and their transactions.
− The "principal-agent theory", which deals with the problems of optimal contract design within
a contractual relationship between principal and agent on the basis of incomplete information
or information asymmetries.
− As well as the "transaction cost theory", which deals with the costs of transferring rights of
disposition and their cost minimization (Roßberg, 2007).
Applied to the field of municipal hospitals and to Public Corporate Governance, this means that aspects
of corporate governance can be explained and solved using the principal-agent approach.
Research contribution and aim of this paper
The main objective of this work is to make a scientifically solid, empirically proven research
contribution to the evaluation of the efficiency of municipal hospitals in Germany and to propose
improvements where necessary. The contribution is based on already existing economic findings and
own empirical studies. The aim of this work is to construct and evaluate the benefits and limitations of
the most widely accepted efficiency model DEA under the conditions existing in Germany. In particular,
the paper focuses on the different legal and governance structures of public hospitals to ensure
responsible Public Corporate Governance. This research aspect represents a novelty, which has not yet
been further considered in any existing study on efficiency measurement of public hospitals.
Research design: methods used and solution process
The methods of multiple-criteria decision-making (MCDM) are among the most frequently used
methods in health care economics today. The models are based on applied efficiency formation and
evaluation. Efficiency is generally achieved when the expenditure/costs of ensuring certain processes
(inputs) do not exceed the profits achieved at the end of the process (outputs) (Vrabková and Vaňková,
2015). Fried et al. (2008) state that efficiency is a component of productivity. They refer to the
comparison between the actual and the optimal amounts of inputs and outputs (Vrabková and Vaňková,
2015). As already mentioned, technical efficiency is defined as the production of goods of a predefined
quality at the lowest possible cost (Kuchinke et al., 2004; Augurzky and Schmitz, 2010). In order to
measure the technical efficiency of hospitals, which are called decision-making units (DMUs, such as
schools, public administrations, waste management enterprises, etc.), it is first important to define a
reasonable set of input and output combinations. The inputs and outputs are then used to construct a
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
165
best-practice frontier - a frontier which then contains the most efficient decision-making units.
Consequently, DMUs that are below this line are not efficient (Kalb, 2010).
Data Envelopment Analysis Modell (DEA)
Among the non-parametric approaches, one of the best-known methods proposed for the construction
of a best-practice frontier is Data Envelopment Analysis (hereinafter DEA) (Kalb, 2010; Helmig, 2005).
With regard to the majority of studies on hospital care and other health services, the DEA method is also
used predominantly (Vrabková and Vaňková, 2015; Helmig, 2005). Originally this approach goes back
to the research of Farrell (1957) and Charnes et al. (1978) (Kalb, 2010). Farrel's model for measuring
the efficiency of units with one input and one output therefore represents the original starting point. The
model was subsequently extended in 1978 by Charnes, Cooper and Rhodes to CCR (both input-oriented
and output-oriented models) and by Banker, Charnes and Cooper to BCC (modified CCR extended by
variable returns to scale) (Vrabková and Vaňková, 2015). Due to its limited scope, this paper focuses
on the CCR model.
The DEA method is used to assess technical efficiency and aims to measure the relationship between
the inputs and outputs of homogeneous units (Vrabková and Vaňková, 2015). In concrete terms, the
efficiency of the decision unit is measured by how efficiently it succeeds in transforming input factors
into output factors in a production process (Helmig, 2005). Since there can be many different types of
inputs and outputs, DEA models belong to the methods of multiple-criteria decision-making (MCDM)
(Vrabková and Vaňková, 2015).
In the case of multiple inputs consumed in the production of multiple outputs, the relative measurement
of the degree of efficiency Uq is used. The latter can be expressed as follows:
𝐸𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑐𝑦 𝑟𝑎𝑡𝑖𝑜 (𝑈𝑞) = 𝑤𝑒𝑖𝑔ℎ𝑡𝑒𝑑 𝑠𝑢𝑚 𝑜𝑓 𝑜𝑢𝑡𝑝𝑢𝑠
𝑤𝑒𝑖𝑔ℎ𝑡𝑒𝑑 𝑠𝑢𝑚 𝑜𝑓 𝑖𝑛𝑝𝑢𝑡𝑠= ∑ 𝑢𝑖𝑦𝑖𝑞𝑖
∑ 𝑣𝑗𝑥𝑗𝑞𝑗
Formula 1: Efficiency ratio
There are different variants of DEA, which concern on the one hand the behavioural objective of the
decision units and on the other hand the economies to scale. If the behavioral goals of the decision units
are considered first, technical efficiency can be identified either as a proportional reduction in input
consumption for a given output or, conversely, as a proportional increase in output production for a
given set of inputs. In the following, efficiency indices that are calculated according to the method of
reducing input consumption are referred to as input-oriented technical efficiency measures. While the
method of increasing the output quantity is called output-oriented efficiency measures. In contrast,
returns to scale deal with the question of how output changes when all inputs increase proportionally
(rate of increase, in production theory called return to scale). In general, DEA can be based on one of
the following three assumptions:
1. constant returns to scale (CRS),
2. variable returns to scale (VRS); or
3. non-increasing returns to scale (NIRS).
The differences between the three DEA frontiers and the differences between input and output
orientation are shown in Figure 3 (appendix) for the special case of one input and one output (Kalb,
2010).
Figure 3 (appendix) illustrates that at constant returns to scale, only decision making unit B is considered
efficient. However, assuming no increase in returns to scale, the best-practice frontier runs through
points B, C and D. Finally, the assumption of variable returns to scale also includes decision-making
unit A as an efficient point. Merely decision-making unit P is regarded as inefficient in all three cases,
since it is always below the frontier (Kalb, 2010).
As for inputs, two basic categories are considered: Labour and capital. Labour inputs is usually measured
in terms of the number of employees (nurses, physicians, specialists, administrative staff and support
staff). Capital inputs usually include the number of beds, the number of operating rooms, bed occupancy
in days and the average length of stay (Vrabková and Vaňková, 2015). .
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
166
Alternative models for evaluating efficiency
The Free Disposable Hull Model and the Malmquist Productivity Index are other widely accepted
models of multiple-criteria decision-making (MCDM) used in health care to evaluate efficiency and
productivity. Another special category is the Health Technology Assessment Model (often also HTA),
which evaluates the efficiency of health care. In the case of health care institutions, this model focuses
on the efficiency assessment and the evaluation of the technical and technological coverage of the
examined institutions (Vrabková and Vaňková, 2015).
Due to the limited scope of this work, the Free Disposable Hull model is described in more detail below
for comparison with the DEA model outlined above.
Free Disposable Hull Modell
The Free Disposable Hull Model (FDH) is one of the so-called dynamic models of the production
function. This FDH model was formulated by Deprins et al. (1984). The basic characteristic of the FDH
model is the non-convex characteristic of the production possibility sets. In contrast to the DEA models,
the production unit can only be evaluated in relation to other existing units and not in their convex
combinations. The advantage of the FDH is the fact that the economies of scale are not affected by any
assumptions. The FDH models analyse both input and output-oriented approaches. The input and output
matrices 𝑋 and 𝑌 represent the structural coefficients of an application. The variables of the model are
𝜆, 𝑠+, 𝑠− vectores, the 𝜃 variable (in case of the input-oriented model) and the Φ variable (in case of the
output-oriented model). To determine the efficiency of all units, it is necessary to solve for each unit
independently, i.e. n times. The value of the target function measures the distance of the unit from the
frontier of production possibility. Depending on the type of model orientation (input- or output-
oriented), it indicates the amount by which outputs must be increased or inputs reduced in order for the
production unit to be judged as efficient (Vrabková and Vaňková, 2015).
Figure 4 (appendix) graphically shows the difference between DEA CRS, DEA VRS and FDH (again
for the input and output case). While the best practice frontier in the case of DEA CRS and DEA VRS
is represented by a straight line or convex curve, the FDH model generates a best practice frontier in the
form of a staircase (Kalb, 2010).
Against this background, the choice of the DEA method chosen in this paper is justified in particular by
the fact that the DEA method is clearly the most frequently used in studies on the efficiency of health
care facilities and hospitals (Vrabková and Vaňková, 2015; Kalb, 2010; Helmig, 2005; Vera, 2010).
In order to address the research problem of measuring the technical efficiency of public hospitals that is
being considered here, it is - as already mentioned - first important to define an appropriate set of input
and output combinations. In the further procedure, these are then used to construct a best-practice
frontier.
The most common input parameters used in health care are those that can be reduced rationally and
sensibly for reasons of efficiency (with regard to maintaining the availability and quality of health care).
For output parameters, those are preferred for which rational and sensible growth is desired (Vrabková
and Vaňková, 2015).
Typically, the input and output parameters listed in Table 2 (appendix) are used in the literature for
measuring the efficiency of health care institutions.
Research application and results
The public hospitals examined in this study (DMU n = 89) have specialized clinics in basic medical
fields such as anesthesia, surgery, gynecology and obstetrics, ear, nose and throat medicine and internal
medicine. The selected data set consists of all public hospitals of the state of Baden-Württemberg in
Germany. The current official statistical number of all hospitals in Baden-Württemberg is 267 from
2018, whereof 89 are in public ownership. The proportionate distribution of every third publicly-owned
hospital in Baden-Württemberg is thereby representative of the nationwide distribution (see section 1.2).
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
167
In order to ensure an unbiased measurement of efficiency, all independent psychiatric centers or clinics
and university hospitals are not included in the sample. These differ too much in their characteristics
from general hospitals (e.g. the number of beds of 0 in many psychiatric day hospitals) and could
therefore influence the result by inhomogeneity of the units.
Three classic input parameters (number of beds, number of physicians and number of nursing staff) and
one classic output parameter (case numbers per year) were defined for the analysis. The models take
into account the approach of input orientation as well as constant and variable returns to scale (CRS and
VRS).
The data material was taken from official statistics (Statistisches Bundesamt, 2018), the German
Hospital Register (Deutsches Krankenhaus Verzeichnis) and the respective annual reports or quality
reports of the hospitals, which the hospitals are legally obliged to provide according to Section 137 I of
the Social Security Code V (Deutscher Bundestag, 2019b).
Distinction by size class
The arithmetic mean of the number of beds in the public hospitals examined is 323,6. Figure 5
(appendix) gives an overview of the DMU distribution of size classes by number of beds. However,
most hospitals only provide 101 to 250 beds. It can be observed that the majority of hospitals belong to
the small and medium size classes.
DEA model M1
The model M1 represents the input-oriented efficiency with constant returns to scale (CRS). The optimal
result of the efficiency measure calculated with the DEA model is equal to one (𝑧 = 1). The results of
the efficiency modelling are expressed in percent, i.e. an efficient decision unit has a value of 100%,
while the value of inefficient decision units is correspondingly less than 100%.
Figure 6 (appendix) shows the percentage distribution of DMUs based on their efficiency measures. Of
the total number of 89 DMUs, however, only 6 DMUs use their input variables efficiently, or better than
the remaining 83 DMUs. Most hospitals (n = 87) use their input variables inefficiently (𝑧 < 1).
It can be seen that the percentage input-oriented efficiency of most public hospitals ranges between 70
% - 79 %. The efficient hospitals are: H15 Hegau-Bodensee-Klinikum Stühlingen, H49 Krankenhaus
Herrenberg, H18 Hohenloher Krankenhaus Öhringen, H54 Krankenhaus Stockach, H75 Rems-Murr-
Klinikum Winnenden, H44 Krankenhaus 14 Nothelfer GmbH.
The following 3 DMUs are the most inefficient: H38 Klinikum Schloß Winnenden, H43 Klinikum am
Weissenhof, H3 Alb-Donau Klinikum Standort Langenau.
The aggregated results of the M1 model are shown in Table 3 (appedix). More detailed results are given
in the appendix.
DEA model M2
The M2 model represents input-oriented efficiency with variable returns to scale (VRS) in all 89 DMUs.
The optimal result of the efficiency measure calculated with the DEA model is also equal to one (𝑧 =1). The results of the efficiency modelling are expressed in percent, i.e. an efficient decision unit has a
value of 100%, while the value of the inefficient decision units is correspondingly less than 100%.
Figure 7 (appendix) shows the percentage distribution of DMUs based on their efficiency measures. Of
the total number 89 DMUs, 12 DMUs use their input variables efficiently, or better than the remaining
77 DMUs. Most hospitals (n = 77) use their input variables inefficiently (𝑧 < 1).
It can be seen that the percentage input-oriented efficiency of most public hospitals ranges between 70
% - 79 %. The efficient hospitals are: H44 Krankenhaus 14 Nothelfer GmbH, H15 Hegau-Bodensee-
Klinikum Stühlingen, H34 Klinikum Mittelbaden Baden-Baden Ebersteinburg, H72 Ostalb-Klinikum
Aalen, H18 Hohenloher Krankenhaus Öhringen, H54 Krankenhaus Stockach, H75 Rems-Murr-
Klinikum Winnenden, H48 Krankenhaus Hardheim, H39 Klinikum Stuttgart – Katharinenhospital und
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
168
Olgahospital / Frauenklinik, H41 Klinikum am Gesundbrunnen, H83 Städtisches Klinikum Karlsruhe,
H49 Krankenhaus Herrenberg.
The following 2 DMUs are the most inefficient: H38 Klinikum Schloß Winnenden und H43 Klinikum
am Steinenberg.
The aggregated results of the M1 model are shown in Table 4 (appendix). More detailed results are
given in the appendix.
FDH Model
The FDH is one of the independent models of a production function. In difference to the DEA models,
a production unit can only be evaluated in relation to other existing units and not in relation to their
convex combinations. The advantage of the FDH is that it is not restricted by any prerequisites regarding
the character of the economies to scale. The FDH model also determines for each production unit a
suitable production unit for benchmarking. The FDH model tests if o DMU is non-dominated, i.e. Pareto
efficient. Units suitable for benchmarking are units that are efficient in the given variant of the model
(Vrabková and Vaňková, 2015).
Figure 8 (appendix) shows the input-oriented efficiency of the evaluated 89 DMU. The optimum level
of the efficiency measure equals one (Φ = 1). The results are given in percentage. An efficient unit thus
has the value of 100 %.
According to the FDH calculation IM1 model, 53 DMU are efficient, which means they cannot improve
one production factor without worsening another and have reached their limits of production
possibilities. The efficient 53 DMU shown in Table 5 (appendix) are therefore suitable for
benchmarking.
All efficient units are suitable benchmark primarily for themselves. Table 6 (appendix) shows the
benchmark of efficiency DMU. The units H22 and H52 have the most numerous benchmark in the input-
oriented model IM1.
Discussion on results
The efficient It is noticeable that the hospitals that are efficient as a result fall either into the category of
the smallest and small size classes with an average number of beds of only 93.85 (7 DMU) or into the
large to largest size classes (1 DMU with 399 and 4 DMU with an average number of beds of 1110.5).
Although the majority of DMUs are in the small to mid-size category based on the number of beds,
many of these DMUs are not efficient as a result.
Of 12 efficient hospitals, the following 9 are organised in the private law form of a limited liability
company (GmbH) with supervisory board and additionally in a cooperative of several hospitals: H15
Hegau-Bodensee-Klinikum Stühlingen, H49 Krankenhaus Herrenberg, H18 Hohenloher Krankenhaus
Öhringen, H75 Rems-Murr-Klinikum Winnenden, H44 Krankenhaus 14 Nothelfer GmbH, H34
Klinikum Mittelbaden Baden-Baden Ebersteinburg, H41 Klinikum am Gesundbrunnen Only H83
Karlsruhe Municipal Hospital and H54 Stockach Hospital are not part of any network. 8 of the 9 efficient
hospitals under private law have a mandatory supervisory board. H 54 has an optional supervisory board.
3 of the 12 efficient hospitals are nevertheless organised under public law. H39 and H72 are organised
as large units in the legal form of a "joint municipal institution under public law" (gkAöR) and H48 as
a very small unit in the form of a public law "Zweckverband". All 3 have a supervisory board as a
controlling organ.
Against this background, it could be deduced that hospitals in associations are much more efficient than
hospitals acting alone. Furthermore, it is important to note that due to the considerable spread of the
private law legal form of the GmbH, it cannot be deduced that these are more efficient than public law
legal forms. Hospitals are also efficient in a public law legal form in combination as an association, in
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
169
this case the legal form of the "joint municipal institution" (gkAöR) or the " Zweckverband". Another
efficiency criterion is the existence of a supervisory board.
It is also interesting to note that all three of the most inefficient hospitals - H38 Klinikum Schloß
Winnenden with 572 beds, H43 Klinikum am Steinenberg with 578 beds and H3 Alb-Donau Klinikum
Standort Langenau 375 beds - are just not in the range of the efficient poles of small and very large
hospitals.
The One-Third Participation Act (Drittelbeteiligungsgesetz) in Germany requires a mandatory
supervisory board for companies in the private law form of a GmbH with more than 500 employees.
Below this number, the supervisory board is considered optional. The management structures of the
efficient hospitals evaluated all have a supervisory board.
Thus, it can be deduced from these results that a clinic group contributes to efficiency, especially for
very small and very large hospitals. Governance structures have a particularly positive effect on
efficiency when the management is additionally controlled by an optional or mandatory supervisory
board.
Conclusion
In this paper, the technical efficiency of hospital services was evaluated using the example of 89 selected
regional hospitals under public ownership in 2018 in Germany. The technical efficiency was evaluated
using input and output indicators using the DEA model and the variant with a focus on the inputs and
the constant and variable returns to scale of the various hospitals. It must also be added that the
evaluation of the efficiency of the units examined in the DEA model can also be approached on the basis
of other rational economic indicators, such as 1 day of stay or 1 hospitalized patient as well.
One model was constructed in the FDH model, which took the input-orientation into account. The FDH
model is not only suitable for the evaluation of input and output variables like the DEA model, but also
additionally for the determination of benchmarking basic data.
However, the application transfer of these derivations also has limitations. In addition, the exact
structures, rights and obligations, communication channels, and instruction rights of the governing
actors of the different hospitals in their respective legal forms must be researched. Only in the following
step concrete recommendations for the design of a governance structure that affects the efficiency of
hospitals can be formulated.
Research on the effects of the legal forms with their respective governance structures on the efficiency
of hospitals, which follows on from these results, must also be verified by means of qualitative research,
such as expert interviews. The extent to which hospital networks can increase efficiency must also be
questioned. The results of this study show that unconnected, middle-sized hospitals are inefficient.
References
[1] Augurzky, B. and Schmitz, H. (2010), Effizienz von Krankenhäusern in Deutschland im
Zeitvergleich: Endbericht - November 2010, RWI Projektberichte, Rheinisch-Westfälisches
Institut für Wirtschaftsforschung (RWI), Essen.
[2] Banker, R.D., Conrad, R.F. and Strauss, R. (1986), “A Comparative Application of Data
Envelopment Analysis and Translog Methods. An Illustrative Study of Hospital Production”,
Management Science, Vol. 32 No. 1, pp. 30–44.
[3] Bremeier, W., Brinckmann, H. and Killian, W. (2006), Public Governance kommunaler
Unternehmen: Vorschläge zur politischen Steuerung ausgegliederter Aufgaben auf der
Grundlage einer empirischen Erhebung, Edition der Hans-Böckler-Stiftung, Vol. 173, Hans-
Böckler-Stiftung, Düsseldorf.
[4] Budäus, D. (Ed.) (2008), Corporate Governance in der öffentlichen Wirtschaft: Referate eines
Symposiums der Gesellschaft für öffentliche Wirtschaft, … am 23./24. November 2006 in Berlin,
Beiträge zur öffentlichen Wirtschaft, Vol. 27, Ges. für öffentliche Wirtschaft, Berlin.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
170
[5] Charnes, A., Cooper, W.W. and Rhodes, E. (1978), “Measuring the efficiency of decision making
units”, European Journal of Operational Research, Vol. 2 No. 6, pp. 429–444.
[6] Deprins, D., Simar, L. and Tulkens, H. (1984), Measuring labor-efficiency in post offices.
[7] Deutscher Bundestag (2019a), Gesetz zur wirtschaftlichen Sicherung der Krankenhäuserund zur
Regelung der Krankenhauspflegesätze(Krankenhausfinanzierungsgesetz - KHG):
Krankenhausfinanzierungsgesetzt.
[8] Deutscher Bundestag (2019b), Sozialgesetzbuch (SGB) Fünftes Buch (V): SGB V.
[9] Deutsches Krankenhaus Verzeichnis, “Deutsches Krankenhaus Verzeichnis”.
[10] Dittrich, G., Karmann, A., Steinmann, L. and Zweifel, P. (2005), “Effiziente Krankenhäuser?
Ein Vergleich sächsischer und schweizerischer Krankenhäuser”, ifo Dresden berichtet, Vol. 12
No. 2, pp. 9–18.
[11] Fabry, B. and Appel, M. (Eds.) (2011), Unternehmen der öffentlichen Hand: Handbuch, Nomos-
Praxis, 2. Aufl., Nomos-Verl.-Ges, Baden-Baden.
[12] Farrell, M.J. (1957), “The Measurement of Productive Efficiency”, Journal of the Royal
Statistical Society. Series A (General), Vol. 120 No. 3, p. 253.
[13] Fried, H.O., Lovell, C.A.K. and Schmidt, S.S. (Eds.) (2008), The measurement of productive
efficiency and productivity growth, Oxford Univ. Press, Oxford.
[14] Frohloff, A. (2007), Cost and Technical Efficiency of German Hospitals – A Stochastic Frontier
Analysis.
[15] Grosskopf, S. and Valdmanis, V. (1987), “Measuring hospital performance. A non-parametric
approach”, Journal of Health Economics, Vol. 6 No. 2, pp. 89–107.
[16] Hammerschmid, G. (Ed.) (2010), Zukunft der öffentlichen Wirtschaft: Referate einer vom
Wissenschaftlichen Beirat des Bundesverbandes Öffentliche Dienstleistungen am 25./26.
Februar 2009 in Eppstein (Taunus) veranstalteten Tagung, Beiträge zur öffentlichen Wirtschaft,
Vol. 31, Berlin.
[17] Helmig, B. (2005), Ökonomischer Erfolg in öffentlichen Krankenhäusern, Teilw. zugl.: Freiburg,
Univ., Habil.-Schr., 2001, Schriften zur öffentlichen Verwaltung und öffentlichen Wirtschaft,
Vol. 185, 1. Aufl., BWV Berliner Wiss.-Verl., Berlin.
[18] Herr, A. (2008), “Cost and technical efficiency of German hospitals. Does ownership matter?”,
Health economics, Vol. 17 No. 9, pp. 1057–1071.
[19] Herr, A., Schmitz, H. and Augurzky, B. (2009), Does higher cost inefficiency imply higher profit
inefficiency?: Evidence on inefficiency and ownership of German hospitals, Ruhr economic
papers, Vol. 132, RUB Dep. of Economics, Bochum.
[20] Herr, A., Schmitz, H. and Augurzky, B. (2011), “Profit efficiency and ownership of German
hospitals”, Health economics, Vol. 20 No. 6, pp. 660–674.
[21] Herwartz, H. and Strumann, C. (2011), On the effect of prospective payment system on hospital
efficiency and competition for patients in Germany, Economics Working Paper, Kiel University,
Department of Economics, Kiel.
[22] Hilb, M., Hösly, B. and Müller, R. (2013), Wirksame Führung und Aufsicht von Öffentlichen
Unternehmen, VR- und GL-Praxis, 1. Aufl., Haupt, Bern.
[23] Kalb, A. (2010), Public Sector Efficiency: Applications to Local Governments in Germany,
Gabler Research, Gabler Verlag / GWV Fachverlage GmbH Wiesbaden, Wiesbaden.
[24] Kuchinke, B., Kallfass, H.H., Schneider, H. and Kirn, S. (2004), Krankenhausdienstleistungen
und Effizienz in Deutschland: Eine industrieökonomische Analyse, Zugl.: Ilmenau, Techn. Univ.,
Diss., 2004, Gesundheitsökonomische Beiträge, Vol. 43, 1. Aufl., Nomos Verl.-Ges, Baden-
Baden.
[25] Meyer, M. and Wohlmannstetter, V. (1985), “Effizienzmessung in Krankenhäusern”, Zeitschrift
für Betriebswirtschaftslehre, No. Band 55 Nummer 3, pp. 262–281.
[26] Nyman, J. and Bricker, D.L. (1989), “Profit Incentives and Technical Efficiency in the
Production of Nursing Home Care”, The Review of Economics and Statistics, Vol. 71 No. 4, pp.
586–594.
[27] Papenfuß, U. (2013), Verantwortungsvolle Steuerung und Leitung öffentlicher Unternehmen:
Empirische Analyse und Handlungsempfehlungen zur Public Corporate Governance, Zugl.:
Hamburg, Helmut-Schmidt-Univ., Diss., 2012, Research, Springer Gabler, Wiesbaden.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
171
[28] Papenfuß, U. and Reichard, C. (Eds.) (2016), Gemischtwirtschaftliche Unternehmen:
Bestandsaufnahme und Perspektiven für Forschung und Praxis, 1. Auflage, Nomos
Verlagsgesellschaft mbH & Co. KG, Baden-Baden.
[29] Roßberg, I. (2007), Die marktorientierte Umstrukturierung kommunaler Kultureinrichtungen:
Besonderheiten und Lösungsansätze, Zugl.: Ostrava, Techn. Univ., Diss., 2006, Tectum-Verl.,
Marburg.
[30] Ruter, R.X., Sahr, K. and Waldersee, G. (2005), Public Corporate Governance: Ein Kodex für
öffentliche Unternehmen, Gabler Verlag, Wiesbaden, s.l.
[31] Schaefer, C., Reichard, C., Schulz-Nieswandt, F., Püttner, G., Grossi, G., Reck, H.-J.,
Mühlenkamp, H., Harms, J., Theuvsen, L., Röber, M., Frentrup, M., Rottmann, O., Lenk, T. and
Papenfuß, U. (2008), Public Corporate Governance: Bestandsaufnahme und Perspektiven, 1.
Auflage, Nomos Verlagsgesellschaft mbH & Co. KG, Baden-Baden.
[32] Schedler, K., Müller, R. and Sonderegger, R.W. (2011), Public corporate governance:
Handbuch für die Praxis, Public Management (PM), 1. Aufl., Haupt, Bern.
[33] Staat, M. (2006), “Efficiency of hospitals in Germany. A DEA-bootstrap approach”, Applied
Economics, Vol. 38 No. 19, pp. 2255–2263.
[34] Štastná, L. and Votápková, J. (2014), Efficiency of hospitals in the Czech Republic: Conditional
efficiency approach, IES Working Paper, Charles University in Prague, Institute of Economic
Studies (IES), Prague.
[35] Statistisches Bundesamt (2018), Grunddaten der Krankenhäuser: Fachserie 12 Reihe 6.1.1 -
2017.
[36] Steinmann, L., Dittrich, G., Karmann, A. and Zweifel, P. (2003), Measuring and comparing the
(in)efficiency of German and Swiss hospitals, Dresden discussion paper in economics, Techn.
Univ., Fak. Wirtschaftswiss, Dresden.
[37] Taube, R. (1988), Möglichkeiten der Effizienzmessung von öffentlichen Verwaltungen: Eine
ökonometrische Untersuchung am Beispiel von Krankenhäusern der Bundesrepublik
Deutschland, Zugl.: Münster, Univ., Diss., 1988, Volkswirtschaftliche Schriften, Vol. 383,
Duncker & Humblot, Berlin.
[38] Tiemann, O. and Schreyögg, J. (2011), Changes in hospital efficiency after privatization, HCHE
Research Paper, University of Hamburg, Hamburg Center for Health Economics (HCHE),
Hamburg.
[39] Vahs, D. and Schäfer-Kunz, J. (2012), Einführung in die Betriebswirtschaftslehre, 6. Aufl.,
Schäffer-Poeschel Verlag, s.l.
[40] Valdmanis, V. (1992), “Sensitivity analysis for DEA models. An empirical example using public
vs. NFP hospitals”, Journal of Public Economics, Vol. 48 No. 2, pp. 185–205.
[41] Vera, A. (2010), Krankenhausmanagement in einem wettbewerbsorientierten Umfeld, Zugl.:
Köln, Univ., Habil.-Schr., 2007, 1. Aufl., Eul, Lohmar.
[42] Vrabková, I. and Vaňková, I. (2015), Evaluation models of efficiency and quality of bed care in
hospitals, Series on advanced economic issues, 2015, vol. 36, 1. vydání, VŠB-TU Ostrava,
Ostrava.
[43] Westermann, G. and Cronauge, U. (2006), Kommunale Unternehmen: Eigenbetriebe -
Kapitalgesellschaften - Zweckverbände, Finanzwesen der Gemeinden, Vol. 3, 5., überarb. Aufl.,
Schmidt, Berlin.
[44] Wilson, G.W. and Jadlow, J.M. (1982), “Competition, Profit Incentives, and Technical
Efficiency in the Provision of Nuclear Medicine Services”, The Bell Journal of Economics,
Vol. 13 No. 2, p. 472.
[45] Wöhe, G., Döring, U. and Brösel, G. (2016), Einführung in die allgemeine
Betriebswirtschaftslehre, Vahlens Handbücher der Wirtschafts- und Sozialwissenschaften, 26.,
überarbeitete und aktualisierte Auflage, Verlag Franz Vahlen, München.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
172
Appendix
Table 3: Overview of empirical studies on the efficiency of hospitals in Germany
Short title of study Author Yea
r
Method Basis of
analysis
Inputs and outputs /
main Efficiency measurement in hospitals
(Meyer and Wohlmannstet
ter, 1985)
1985 DEA 20 German hospitals
-
Possibilities for measuring the
efficiency of public administra.
(Taube, 1988) 1988 Regressions
analyse SFA
613 German
hospitals
Outputs: patients in different
departments Inputs: Costs
Economic success in public
hospitals
(Helmig,
2005)
2005 DEA 418 German
hospitals
Inputs: Number of beds, treatment
cases per year, sponsorship
Efficient hospitals?
A comparison
(Dittrich et
al., 2005)
2005 DEA 105 Saxon and
251 Swiss
hospitals
Inputs: Number of staff, costs, days
of care
Main findings: Swiss hospitals are
less efficient than German hospitals
Efficiency of hospitals in
Germany: a DEA-bootstrap
approach
(Staat, 2006) 2006 DEA 160 German
hospitals
Inputs: daily rates, number of beds
Outputs: Treatment cases per year,
length of stay
Cost and Technical Efficiency of German Hospitals
(Frohloff, 2007)
2007 SFA 1500 German general hospitals
Inputs: e.g. ownership Main findings: private and non-
profit hospitals are on average less
efficient than public hospitals
Cost and Technical Efficiency
of German Hospitals: Does Ownership Matter?
(Herr, 2008) 2008 SFA 1500 German
hospitals
Inputs: Number of beds, treatment
cases per year, sponsorship Main findings: private and non-
profit hospitals are less cost-
effective and technically less
efficient than publicly owned h.
Does Higher Cost Inefficiency
Imply Higher Profit Inefficiency? - Evidence on
Inefficiency and Ownership of
German Hospitals
(Herr et al.,
2009)
2009 SFA 374 German
hospitals
Inputs: Number of beds, treatment
cases per year, ownership Main findings: private (for-profit)
and (private) non-profit hospitals
are less cost-efficient but more
profitable than publicly owned h.
On the effect of prospective
payment system on hospital efficiency and competition for
patients in Germany
(Herwartz and
Strumann, 2011)
2011 DEA 1500 German
general hospitals
Inputs: Material costs, personnel,
number of beds as non-descretionary input
Outputs: Treatment cases per year,
number of trainees
Main findings: Improvement of overall efficiency after DRG introd.
Profit Efficiency and Ownership of German
Hospitals
(Herr et al., 2011)
2011 SFA 541 German hospitals
Main findings: higher profit efficiency of private hospitals
compared to public hospitals
Changes in hospital efficiency
after privatization
(Tiemann and
Schreyögg,
2011)
2011 DEA 1878 German
acute hospitals
Main findings: Conversion from
public to private ownership resulted
in increased efficiency
Source: Own illustration.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
173
Figure 5: Percentage of hospitals by ownership 2017
Source: (Statistisches Bundesamt, 2018).
Figure 6: DEA frontiers due to economies to scale Figure 7: Overview of frontiers
with input and output orientation
Source: Own illustration based on Source: Own illustration based on
(Kalb, 2010; Vrabková a nd Vaňková, 2015)). (Kalb, 2010; Vrabková and Vaňková, 2015).
Table 4: Typically used inputs and outputs for measuring the efficiency of health care institutions
Source: Own illustration based on (Vrabková and Vaňková, 2015; Vera, 2010; Helmig, 2005; Kalb, 2010).
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
174
Figure 8: DMU Distribution to size classes Figure 9: Percentage distribution of
efficiency measurement according to DEA model M1 according to number of
beds
Table 5: Aggregated results of modelling M1 Figure 10: Percentage distribution DEA
model M2
Table 6: Aggregated results of modelling M2 Figure 11: FDH model IM1
Table 7: Aggregated results of modelling IM1 Table 8: Benchmark of Efficiency DMU in input-oriented model
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
175
GENERALIZED LINEAR MODELS IN A MOTOR HULL INSURANCE PORTFOLIO
Adéla Špačková1
1Department of Economics, VŠB – Technical University of Ostrava,
Sokolská tř. 33, Ostrava 70200, Czech Republic
e-mail: [email protected]
Abstract
Actuaries in insurance companies try to find out the best model for an estimation of insurance premium.
It depends on many risk factors, e.g. the car characteristics and the profile of the driver. In this paper,
an analysis of the motor hull insurance portfolio using a generalized linear model (GLM). Our aim is to
predict the relation of claim frequency on given risk factors. The models with different predictor
variables are compared by Likelihood ratio test, analysis of deviance, Akaike information criterion
(AIC) and Bayesian information criterion (BIC). Based on this comparison, the model for the best
estimate of annual claim frequency is chosen. All statistical calculations are computed in STATA
environment.
Keywords
Generalized Linear Models, Claim Frequency, Motor Hull Insurance Portfolio, Risk Factors
JEL Classification
C13, G22
Introduction
The need and the necessity of establishing internal models is still growing. Internal models are important
for the determination and management of risks in each insurance company. Risk management is often
complex, therefore, it is important to establish a model for the estimation of the claims frequency and
severity, which are important for the calculation of insurance premiums. Our aim is to predict the relation
of claim frequency on given risk factors. In this paper will be conducted an estimate of the claim
frequency and the selected models of this frequency will be tested using LR test, the deviance, the
Akaike criterion (AIC) and Bayesian information criterion (BIC).
In statistics, the likelihood-ratio test assesses the goodness of fit of two competing statistical models
based on the ratio of their likelihoods, specifically one found by maximization over the entire parameter
space and another found after imposing some constraint. If the constraint (i.e., the null hypothesis) is
supported by the observed data, the two likelihoods should not differ by more than sampling error. Thus
the likelihood-ratio test tests whether this ratio is significantly different from one, or equivalently
whether its natural logarithm is significantly different from zero.
Deviance measures the discrepancy between the current model and the full model. The full model is the
model that has n parameters, one parameter per observation. The Akaike information criterion (AIC) is
an estimator of out-of-sample prediction error and thereby relative quality of statistical models for a
given set of data. In statistics, the Bayesian information criterion (BIC) is a criterion for model selection
among a finite set of models; the model with the lowest BIC is preferred. It is based, in part, on the
likelihood function and it is closely related to the Akaike information criterion (AIC).
Literature Review
Generalized linear modelling is a methodology of descriptions possibilities of relationships between
variables. This methodology have been find out by great scientists Nelder and Wedderburn in 1972.
Nowadays, these models generally deals with many of the authors as Gray and Pitts (2012), Hardin and
Hilbe (2012), Long and Freese (2008) etc. Practical example using negative binomial distribution is
demonstrated in Hilbe (2011). The above mentioned titles are aimed more generally. The work focused
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
176
on pricing process using non-life insurance data is well described in Ohlsson and Johansson (2010).
General part about the insurance risk is possible to find out in Cipra (2006 and 2012). Pretty well
application paper is written by Valecký (2015).
Methodology and Data
In this section is described the methodology of generalized linear regression models. first, generalized
linear models in general, then it is described the methodology of the estimation of the individual
parameters. At the very end of the methodology are described the tools of comparison of the selected
models, specifically, the deviance and information criteria.
As the data file has been selected a random sample of contracts of insurance accident insurance collected
in 2005-2010 in the Czech Republic territory.
Generalized linear models
Generalized linear models have been formulated by John Nelder and Robert Wedderburn as the way of
others regression statistical models, including linear that permit for independent variable utilize other
than normal distribution. The basic of these models are defined as an extentions of the Gaussian linear
predictor derived from the exponential family. The main purpose of theese models is to estimate random
explanation variable (denoted y), depending on certain number of explanatory variables (Xi).
Generally, GLM includes three main assumptions:
• A probability distribution has to be from an exponential family
• A linear predictor is transformed by link function, such as:𝑛 = 𝑥′𝛽
• A link function 𝑔 (𝜇
𝑛) = 𝑥′𝛽,
where g is a link function, 𝜇 is mean, n is called the exposure.
Link function can be diverse, but for the purposes of this paper the logarithm link function is selected
(Jong, Heller (2008) and Gray and Pits) .
Thus, link function g is log, that becomes:
ln ´ ln ln ´x n x
n
= = = +
(1)
where ln n is called an „offset“. This offset is another variable x , where the coeficient is equal to
one. Offsets are usualy used to correct differing time period of observation.
All probability distribution can be decribed of the general form:
( )( ) ( , ) exp
y af y c y
−=
(2)
where is the canonical parameter and is called the dispersion parameter. ( )a and ( , )c y are
functions determining the actual probability function such as normal, gamma, binomial etc.
For the purposes of this paper the negative binomial distribution is choosen. Description of exponencial
family parameters is shown in Table 1.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
177
Table 9. Parameters distributions in exponential family framework
Distribution ( )a ( )E y ( )
( )Var y
V
=
Negative-
binomial( , )
ln1
+
1ln(1 )e
− − 1 (1 ) +
source: Jong and Heller(2008)
Maximum likelihood estimation
The standard method of estimation parameters is maximum likelihood estimation. This method is
based on selecting parameter estimates and maximize the likelihood of the observed sample:
1
( , ) ln ( ; , )n
i
i
f y =
(3)
where ( )if y is a probability function depends on the canonical parameter and the dispersion
parameter If the maximum likelihood estimation is the exponential family probability function:
1 1
( )( , ) ln ( ; , ) ln ( , )
n ni i i
i i
i i
y af y c y
= =
− = = +
(4)
Maximalization of likelihood called the log-likelihood is a logarithm of the likelihood with respect to
j :
1
ni
ij i j
=
=
(5)
where the parameters are folowing:
( )i i i iy a y
− −= =
(6)
i i i i
ij
j i j i
x
= =
(7)
ijx is a component i of jx .
If the equation (5) is equal to zero, that estimation equations for is:
1
( ) 0 ´ ( ) 0n
iij i i
i i
x y X D y
=
− = − =
(8)
According to the equation (8) it is clear, that parameter is implicit and working throught and D .
Generalized linear models are estimated using Newton-Raphson method, or the method of IRLS
(method of iteratively weighted least squares). Using the algorithm Newthon-Rapson can obtain the
observed information matrix (OIM), on the contrary, the method of IRLS we obtain the expected
information matrix (EIM) see GRAY, Roger J. a Susan M. PITTS (2012).
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
178
Claim frequency models
Claim frequency model (or observed number of claims) is a situation, where the random dependent
variable is discrete and conditioned by vector of explanatory variables ( characteristic of risk based on
individual characteristic of shareholders). According to the purposes of this paper, the Negative-
binomial and Poisson distribution is selected. The Negative-binomial probability of random variable Y
fitting into the exponential family framework (2) is given:
1 ( )ln ( ) ln ln(1 )
1
y af y y
−= − + =
+
(9)
where the dispersion parameter is 1 = and canonical parameter ln1
=
+
Mean and variance function is denoted:
( ) ( )
1
eE y a
e
= = =
− (10)
2( ) ( ) (1 )
(1 )
eVar y a
e
= = = +
− (11)
where ( )a and ( )a are first and second derivates of ( )a with respect to .
The Poisson probability of random variable Y fitting into the exponential family framework (2) is given:
ln ( ) ln ln ! ln !
y af y y y y
−= − + − = − (12)
where the dispersion parameter is 1 = and canonical parameter e = (see table of parameters 1),
see JONG, Piet de a Gillian Z. HELLER (2008).
Mean and variance function is denoted:
( ) ( ) ( ) ( )a e E y a Var y = = = = = (13)
where ( )a and ( )a are first and second derivates of ( )a with respect to .
Models´ goodness of fit
The goodnes of fit of a model to a data is natural question arises with every statistical modelling. The
literature presents many statistic tools, that can be used to select and to assess the performance of count
data models. As discussed in Jong and Heller (2008) one way how to assess the fit of given model is to
compare with the model with the best possible fit.
Likelihood Ratio
Indicator of the credibility (likelihood ratio), the LR is very often used for comparison of models with
different distribution of probability. For this test it is necessary to calculate the parameter estimates for
both the full model and for the model reduced (estimated to best describe the reality). Test statistics is
then given by the expression:
𝐿𝑅 = −2(𝑙𝑜𝑔𝐿𝐼 − 𝑙𝑜𝑔𝐿𝐹)
Likelihood statistic has in large selections division of Pearson 𝜒2 statistics with the number of freedom,
which is then equal to the difference of the number of parameters of the tested models, i.e.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
179
𝐿𝑅~𝜒 𝑘𝑓−𝑘𝑖2 (0,1 − 𝛼)
The value of the LR test is high if the explanatory variable (linear regresor) in the model affect the
explanatory variable, see Hardin, Hilbe (2012). Test ratio assurance is an alternative to the F-test in
linear regression model and is suitable to use if it is considered about adding other explanatory variables
into the model.
Akaike and Bayesian information criteria
The basic of these criteria is the comparison of models among themselves and in the most suitable model
is considered to be such a model, the value of which AIC and BIC is the lowest. Akaike information
criterion is following:
2 2log( )AIC k L= − (15)
where k is number of predictors of a model including constants and log(L) means log-likelihood model.
Bayesian information criterion is:
2log log( )BIC L k n= − + (16)
n is the number of observations.
Deviance residuals
By using residual analysis is possible to find out the information about the suitability of the model. The
deviance residuals can be used for assessing the quality of the model, e.g. for the detection of remote
observation and verification of the assumption about the variance.
The general form of deviance residual is:
( )ˆ( ) (D
i i i i ir sign y d y = − (17)
where ( )iiyd ( denotes the distance function, which represents the remoteness from the estimated
mean values to observed.
Empirical Results
Every person, which applying vehicle insurance is divided into a class, where the risk is homogenous.
One of the criteria used for assigning an individual to a certain class is number of claims. It is a very
important task for insurance companies to model the number of claims in given insurance portfolio.
Our aim is to predict the relation of claim frequency on given risk factors. For the purposes of this paper
a random selection of real data in vehicle insurance is used and collected during the years 2005-2010 in
the Czech Republic teritory. The file contains 18 111 contracts.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
180
Figure 12. Summarize count
The mean number of count is 0,064, Variance is 0,070 it is a bit more than mean, it can mean that data
are overdispersed. Negative-binomial regression can be used for overdispersed count data, that is when
the conditional variance exceeds the conditional mean.
The model is going to be estimated with these seven following predictors, because each vehicle
insurance contract includes this following individual characteristic of the policyholders:
Table 2. Variable description
variable description
agecar Age of car
gender Drivers gender
volumkw Engine power
fuel Type of fuel
ageman Age of driver
price Vehicle price
district District area
Source: Špačková (2020)
The following fig. shows histogram of empirical frequency:
Figure 2. Empirical claim frequency
Source: Špačková (2020)
According to the histogram, it is obvious positive skewness. In the following table is shown observed
and predicted claim frequency by Negative binomial distribution.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
181
Table 3. Observed vs. predicted frequency
Claim frequency Observed Negative-binomial
0 17 045 17 104,88
1 984 897,14
2 76 75,21
3 6 2,66
Source: Špačková (2020)
According to the table it is obvious, that in 17 045 suffered no claim. One claim was occurred in 984
cases, two in 76 cases, three in 6 cases. The table provides, that Negative-binomial fits good to our
insurance data. In following table we can see the β parameters estimated by maximum likelihood
methods.
Table 4. Analysis of parameters
variable Negative-binomial
parameter St. error p-value
agecar -0.129 0.012 0.000
gender 0.267 0.089 0.000
volumkw 0.006 0.004 0.000
fuel
2 0.398 0.089 0.017
3 -10.303 890.294 0.986
4 0.560 1.05 0.575
ageman -0.013 0.002 0.000
price 5.81e-07 1.95e-07 0.000
district
2 -0.421 0.319 0.200
3 -0.428 0.319 0.180
4 0.192 0.2347 0.411
5 -0.296 0.239 0.188
6 0.365 0.246 0.133
7 -0.440 0.320 0.215
8 -0.328 0.295 0.219
9 -0.303 0.327 0.389
10 -0.188 0.221 0.430
11 0.156 0.192 0.556
12 -1.082 0.382 0.005
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
182
13 -0.011 0.263 0.961
14 -0.113 0.204 0.532
Source: Špačková (2020)
P-value of variables fuel and the district was higher than 0.05, which of course signifies the
irrelevance to the variable count. According to LR test procedure, it is necessary to estimate a second
model. Thus, next step contains second model, which is estimated without variables fuel and district.
This model, called model 2 is nested to the model 1 and subsequently it is possible to test it by LR test.
The results of likelihood ratio test are in table 5.
Table 5. Likelihood ratio test
LR test 2 (16) prob. >
2
46,1 0,0001
Source: Špačková (2020)
According to the results it is clear that the more accurate model is model 1 including all variables. In the
following table it is shown criteria for assessing goodness of fit of full and restricted model.
Table 6. Criteria for assessing goodness of fit
Full model Restricted model
Log Likelihood -2704.994906 -2686.450867
AIC 0.4317 0.4482
BIC -114970.7 -115264.6
Source: Špačková (2020)
Preferred model is still full model, because the Log-likelihood is higher. According to the results of AIC
a BIC, the better model is full model again, because the value of AIC and BIC is lower. The Akaike
Information Criterion (AIC) and Bayesian Information Criterion (BIC) provide a method for assessing
the quality the model through comparison of related models. It’s based on the Deviance, but penalizes
you for making the model more complicated. Eventually, it was proved, that the full model is better.
The models are also possible to compare by deviance In the figure is comparison of predicted deviance
count by each model.
Figure 3. Predicted deviance by full (left) and restricted model
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
183
From figure 2 clearly shows that the estimate of the mean value of the variable count model 1 is more
accurate, since the residues for full model are not so much scattered. In restricted model contained a
larger number of extreme values compared to the full model. Based on the results of deviance residual
was again full model evaluated as more appropriate for modelling the number of claims.
From the results of the comparison of the two models, both essential characteristics, so the test of
credibility and a comparison of information criteria, etc., full model appeared to be more appropriate for
modelling the number of claims.
Conclusion
Our aim was to predict the relation of claim frequency on given risk factors from which the premium is
derived. It was found out that the claim frequency depends on many risk factors – agecar, gender,
volumkw, fuel, ageman, price and district. In this paper have been conducted an estimate of the claim
frequency and the selected models of this frequency will be tested using LR test, the deviance, the
Akaike criterion (AIC) and Bayesian information criterion (BIC).
The standard method of estimation is generalized linear models. The GLM models are used for
estimation claim frequency in this paper. In the theoretical part the GLM are introduced, including the
definition of link function and specification of an exponential family of probability density functions.
The standard method of estimation parameters is maximum likelihood estimation. This method is
based on selecting parameter estimates and maximize the likelihood of the observed sample.
The main part of the paper is GLM application in vehicle insurance. For the purposes of this paper a
random selection of real data in vehicle insurance have been used and collected during the years 2005-
2010 in the Czech Republic territory. The file contained 18 111 contracts. The drivers have been divided
into groups on the basic of risk factors. The negative-binomial distribution with log link function is used.
Two different models with various variables are considered. Based on goodness of fit the best model
have been chosen. The best model is a full model which includes all above mentioned variable.
References
[1] CIPRA, Tomáš. Finanční a pojistné vzorce. Praha: Grada Publishing, 2006. ISBN 80-247-1633-
X.
[2] CIPRA, Tomáš. Pojistná matematika: teorie a praxe. 2. aktualiz. vyd. Praha: Ekopress, c2006.
ISBN 80-86929-11-6.
[3] CIPRA, Tomáš. Riziko ve financích a pojišťovnictví: Basel III a Solvency II. Vydání I. Praha:
Ekopress, 2015. ISBN 978-80-87865-24-8.
[4] GRAY, Roger J. a Susan M. PITTS. Risk modelling in general insurance: from principles to
practice. Cambridge: Cambridge University Press, 2012. ISBN 978-0-521-86394-0.
[5] HARDIN, James W. a Joseph HILBE. Generalized linear models and extensions. 3rd ed.
College Station: Stata Press, 2012. ISBN 978-1-59718-105-1.
[6] HILBE, Joseph. Negative binomial regression. 2nd ed. Cambridge: Cambridge University Press,
2011. ISBN 978-0-521-19815-8.
[7] JONG, Piet de a Gillian Z. HELLER. Generalized linear models for insurance data. Cambridge:
Cambridge University Press, 2008. ISBN 978-0-521-87914-9.
[8] LONG, J. Scott a Jeremy FREESE. Regression models for categorical dependent variables using
Stata. Third edition. College Station: Stata Press Publication, 2014. ISBN 978-1-59718-111-2
[9] OHLSSON, Esbjörn a Björn JOHANSSON. Non-life insurance pricing with generalized linear
models. Berlin: Springer, c2010. ISBN 978-3-642-10790-0.
[10] VALECKÝ, Jiří. Modelling claim frequency in vehicle insurance. Acta Universitatis
[11] Agriculturae et Silviculturae Mendelianae Brunensis. 2015. 10 s. ISSN 1211-8516.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
184
OPTIMALIZATION OF DIRECT COSTS OF THE RAILWAYS OF THE SLOVAK
REPUBLIC
Adrián Kuka1, Adrián Šperka2, Michal Petr Hranický3, Jozef Gašparík4
1Department of Railway transport, ŽU – Univerzity of Žilina,
Univerzitná. 1, Žilina , Slovak Republic
e-mail: [email protected]
2 Department of Railway transport, ŽU – Univerzity of Žilina,
Univerzitná. 1, Žilina , Slovak Republic
e-mail: [email protected]
3Department of Railway transport, ŽU – Univerzity of Žilina,
Univerzitná. 1, Žilina , Slovak Republic
e-mail: [email protected]
4Department of Railway transport, ŽU – Univerzity of Žilina,
Univerzitná. 1, Žilina , Slovak Republic
e-mail: [email protected]
Abstract
The article deals with the issue of direct costs in the railway transport environment, specifically in the
environment of the infrastructure manager. It is important for the infrastructure manager to gradually
reduce the direct costs and thus seize the opportunity to save money. One of the largest items of direct
costs is staff costs. This can be seen in operating professions where there is a trend to reduce staff and
implement dispatching centralization in railway transport operating. The issue of reducing direct costs
through the introduction of dispatching centralization is currently topical also due to the lack of
employees caused by high demands and low attractiveness of operating professions in the railway
sector.
Keywords
remote control track, infrastructure manager, costs, human labor
JEL Classification
R4
Introduction
Rail transport is one of the fastest growing transport branches. It is essential that it respond to new
challenges from passenger passengers and freight carriers. A dynamic modernization process, which
brings many positive changes, also facilitates its correct response. Everything from the switches to the
locomotives is being modernized.
Modernization and reconstruction measures also cause direct costs of rail transport. The effort of the
infrastructure manager and individual carriers is to eliminate these costs. One of the popular and
currently inevitable ways to solve this problem is to control railway lines remotely. This will not only
reduce direct personnel costs but also the number of operating staff. The other side of this measure is
where to put workers who have lost their jobs as a result of these measures.
Rail transport costs
Almost all human needs are characterized by movement, change of place. The higher the level of
transport, the better the social division of labor and cooperation, the distribution of means of production
and consumer goods, the change of activities and the change of goods can develop (Řezníček, 1982).
The development of the transport enables (Řezníček, 1982): • closer social relations,
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
185
• stronger material and cultural connections between nations,
• a much richer life for people and its better protection of the participation of transport in
strengthening state defense.
But every development costs something. The transport sector is also one of the most cost-intensive
sectors. Its cost depends on various factors and is also different for different modes of transport.
Classification of costs
The classification of costs serves mainly for a more detailed breakdown of costs in the company, thus it
organizes individual cost items into groups, with each group having its own characteristic feature.
(Dolinayová & Nedeliaková, 2015).
In railway transport, it is necessary to monitor costs from various points of view and their classification
is most often according to (Dolinayová & Nedeliaková, 2015):
• cost types - economic division of costs,
• relation to the production process - purpose-specific cost division,
• calculation formula items,
• cost dependency,
• responsibility for their creation,
• decision making - managerial understanding of costs.
The classification of costs is mainly used for (Dolinayová & Nedeliaková, 2015):
• getting more accurate information,
• finding the level of management of the company and its individual organizational units,
• more precise allocation of costs to individual activities and processes in the company and its
organizational units,
• cost planning.
The division of costs by cost type is the basic one used by each undertaking, given that these costs are
recorded in the cost accounts. It is a value expression of consumption of individual types of production
factors at the input to the production process (Dolinayová & Nedeliaková, 2015).
Costs according to cost types can be divided eg. as follows (Míka, 2005):
1. Material and energy costs
2. External services costs:
• analysis,
• mediation,
• consulting,
• service, reparation.
3. Personnel costs:
• wages,
• rewards,
• educational costs.
4. Taxes and fees
5. Depreciation and provisions
6. Income taxes
The division shows that employees' wages are included in personnel costs. A very important aspect in
rail transport is also the training of employees due to the demanding nature of operating professions.
Calculation formulas are used to calculate the cost part of the cost formula, which allows to assign the
respective part of each cost type to individual products (Kupkovič, 1999).
In terms of calculations, we divide costs into two groups (Kupkovič, 1999):
• direct costs - costs that can be directly identified and calculated by the cost bearer,
• indirect costs - they are incurred jointly by several cost carriers and are allocated to cost carriers
using various methods, most often by using a mark-up surcharge.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
186
The general calculation formula is designed to calculate the full cost (Kupkovič, 1999). The type
calculation formula of the infrastructure manager (hereinafter referred to as ŽSR) is based on the
purpose-specific cost classification. The type calculation formula serves as a basis for calculating the
performance of the individual company units, which may add to the calculation formula other items
related to the calculation of the cost of services and services as shown in Table 1.
Table 1. Calculation formula of railways of the Slovak Republic
Category Cost
1. DIRECT MATERIAL
2. DIRECT WAGES
3. OTHER PERSONNEL COSTS
3. 1. Legal social insurance
3. 2. Other social security
3. 3. Legal and other social costs
4. OTHER DIRECT COSTS
4. 1. Other services
4. 2. Direct depreciation
4. 3. Other direct costs
1. – 4. Operating overhead
5. OPERATING RICE
1. – 5. Own costs of operation
6. CORRECT SURVEY
1. – 6. Total own cost of performance
7. PROFIT OR LOSS
1. – 7. Price without VAT
8. VAT
9. Other price components within the
meaning of the Price Act
Source: (Železnice Slovenskej republiky, 2012)
The table shows that the infrastructure manager's labor costs are direct costs. All items of direct costs
must be related to the valued performance and the cost bearer (in our case the employee) is determined
directly or by technical conversion. (Dolinayová & Nedeliaková, 2015).
Wage costs
Labor costs are one of the largest items that the company seeks to minimize. It is necessary for the
company to be able to control and direct them so that the interests of employees and the economic
objectives of the company are preserved. (Štetka, 2014).
Employer's labor costs per employee (labor price) consist of the following parts (Šutyová, 2019):
1. Gross wage of emloyee
2. Payments to the tax office
3. Contributions to the Social Insurance Agency
4. Payments to the Health Insurance Company
Employers' contributions are 35.2% of the employee's gross salary according to the items listed above.
In relation 1 there is a sample calculation of labor costs. (Šutyová, 2019)
𝐿𝑎𝑏𝑜𝑢𝑟 𝑐𝑜𝑠𝑡 = 𝐺𝑟𝑜𝑠𝑠 𝑤𝑎𝑔𝑒 𝑜𝑠 𝑒𝑚𝑝𝑙𝑜𝑦𝑒𝑒 + 𝑒𝑚𝑝𝑙𝑜𝑦𝑒𝑟 𝑐𝑜𝑛𝑡𝑟𝑖𝑏𝑢𝑡𝑖𝑜𝑛𝑠 (1)
It is true that the more employees a company has, the greater the labor cost of its employees is.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
187
Dispatching centralization on ŽSR network
The high pace of introduction of new technologies in railway traffic management requires a review of
ŽSR's approach in this area and before the actual implementation of traffic management centers and
remote-controlled lines it was necessary to decide strategic issues and define requirements for their
implementation on the ŽSR network.
Main milestones of implementation of dispatching centralization on ŽSR network (Odbor stratégie a
vonkajších vzťahov GR ŽSR, 2019):
• analysis of the current situation,
• a proposal for the future situation,
• implementation of the proposal.
Proposal of strategic distribution of traffic managements centers on ŽSR network
The ŽSR network has a total length of 3,627 km, of which 1,586 km are electrified. Even for a relatively
short length, it is necessary that the traffic management centers (CRDs) are located according to certain
principles and conditions respecting the individual parameters of the lines and stations.
The proposal for the implementation of the CRD stems from the following principles and conditions
(Odbor stratégie a vonkajších vzťahov GR ŽSR, 2019):
• coherence of transport - linking traffic management to the direction of traffic flows,
• link to the modernization of lines according to sections,
• composition of the lines - significant and minor routes, corridor and sub-routes,
• transparency in traffic management,
• traffic intensity and number of dispatchers,
• the location of CRDs at major transport hubs, following the availability of local controlled
workplaces in the event of emergencies,
• acceptance of the current state of long-distance traffic management at the site of ŽSR,
• anticipation of the development of demand and the structure of transport on individual lines,
• possibility of integration into integrated transport systems,
• size of controlled stations and lines,
• the expected number of controlled elements in each station,
• shift making rules,
• Labour Code and othew laws,
• the average wage and employment in the region.
The following are excluded from the draft CRD layout concept (Odbor stratégie a vonkajších vzťahov
GR ŽSR, 2019):
1. Large marshalling yards, which will be locally controlled and only entrances and exits of trains,
ie branches:
• Bratislava východ,
• Žilina-Teplička,
• Čierna nad Tisou,
• Košice nákladná stanica.
2. Large border crossing points on wide-gauge lines to be operated locally:
• Čierna nad Tisou,
• Maťovce.
3. Lines with simplified traffic management and lines with low traffic frequency.
Figure 1 is a map background showing the planned CRD network on the ŽSR network.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
188
Figure 1. Traffic management centers on ŽSR network
Source: (Odbor stratégie a vonkajších vzťahov GR ŽSR, 2019)
Table 2 shows the number of individual CRDs, associated local CRDs, and km of tracks to be controlled
from local CRDs.
Table 2. CRD location on the network ŽSR
CRD Local CRD Length of controlled lines
in km
Bratislava
Jablonica 69
Kúty 93
Bratislava-Nové
Mesto 36
Dunajská Streda 95
Nové Zámky 131
Galanta 94
Lužianky 76
Prievidza 105
Trnava 145
Žilina
Púchov 138
Žilina 70
Kraľovany 139
Zvolen
Lučenec 198
Zvolen 145
Brezno 179
Banská Bystrica 80
Levice 55
Košice
Poprad 94
Kysak 27
Košice 36
Michaľany 105
Moldava nad Bodvou 66
Plaveč 104
Prešov 117
Humenné 91
Trebišov 100
Haniska pri Košiciach 105
Source: (Odbor stratégie a vonkajších vzťahov GR ŽSR, 2019)
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
189
A total of 27 local CRDs should be built on the territory of ŽSR, which will fall below 4 CRDs. No
CRD will serve an area greater than 200 km. On the one hand, it is relatively enough CRD for such a
small railway network, but considering individual catchment areas, operating staff may not move
directly to CRD construction sites, but may commute.
In the Czech Republic, where there are two CRDs (Prague and Prerov), there is a problem with relocating
operational employees to these locations. As the housing issue is not being solved on a long-term basis,
especially in Prague, the operational employees (dispatchers) of the infrastructure manager are not
satisfied.
Reduction of operational staff due to dispatching centralization
Under the ŽSR conditions, mainly railway dispatchers, signalmen and switchmen are involved in the
management of railway transport. Collectively, they can be called traffic management employees.
The work of the switcherman comprises (Majerčák, et al., 2015):
• operation of manually and locally adjusted switches for train movements and for shunting,
• controlling whether the track is free and thus ready for a train movement,
• controlling the train movements,
• inspection and operational service of switches and derails
• operation of railway crossings’ signalling systems, as long as they are within their responsibility.
Contents of the dispatcher's work (Majerčák, et al., 2015):
• control, coordination and management of train traffic within the allocated perimeter of the
railway station, track or remotely operated track sections,
• shunting management and control,
• management and control of train-forming activities,
• managing and coordinating the work of the relevant staff at the railway station and train staff
while in the railway station and adjacent interstation section and in remotely controlled stations.
Due to the introduction of CRD on the ŽSR network, the number of traffic management employees will
decrease. Figure 2 shows the estimated number of employees in each CRD.
Figure 2. Number of current traffic management staff and their reduction after the introduction
CRD
Source: (Odbor stratégie a vonkajších vzťahov GR ŽSR, 2019)
From the graph it is clear that at present, the number of employees involved in traffic management is a
total of 3,846. After the CRD is built, there will be only 1915, which is a saving of 1931 employees.
1258
675850
1063
667
314396
538
0
200
400
600
800
1000
1200
1400
Bratislava Žilina Zvolen Košice
Current number of traffic management staff
The proposed number of traffic management staff
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
190
Dispatching centralization track Banská Bystrica – Zvolen and its economic impact
According to the facts from Chapter 3, one of the sections included in the dispatching centralization is
also the section Banská Bystrica - Zvolen mesto. It is a single track 20 km long, which should be
controlled from the CRD Banská Bystrica after completion of the remote control. For charging purposes,
the track falls into category no. A (Dopravný úrad, 2019).
The section includes the following railway stations:
• Banská Bystrica,
• Radvaň,
• Vlkanová,
• Sliač kúpele.
The following train stops are located on the track section:
• Banská Bystrica mesto,
• Hronsek,
• Veľká Lúka,
• Zvolen mesto.
The number of transport employees (dispatchers and switchmen) depends mainly on the size of the
railway station and the type of signalling system. Figure 3 shows the number of tracks at individual
railway stations on the section in question.
Figure 3. Number and type of tracks
Source: (VLAKY.NET, 2004)
It is clear from the graph that the most rail tracks are in the Banská Bystrica railway station, therefore
the operational need of transport employees will be the greatest at this point of transport. Other railway
stations are intermediate stations with a smaller scope of operational work.
Staff costs in the current state
The costs of employees in the operation of railway transport depend mainly on the number of employees.
Traffic on the section Banská Bystrica - Zvolen (outside) is controlled by dispatchers in each station on
the section. Since the signalling system does not require the interaction of switches, the switches are
switched directly by the dispatcher and therefore the importance of switchmen at individual stations
decreases from year to year. Just in 2019 ŽSR proceeded to abolish switchmen on the track section.
Table 3 shows the number of dispatchers at each transport site.
15
23 3
5
21 1
2
0
2
00
2
4
6
8
10
12
14
16
Banská Bystrica Radvaň Vlkanová Sliač kúpele
Traffic tracks Handling tracks Other tracks
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
191
Table 3. Operational need for transport staff at individual stations
Railway station Number of dispatchers per shift Total need of dispatchers
Banská Bystrica 2 8
Radvaň 1 4
Vlkanová 1 4
Sliač kúpele 1 4
Odb. Zvolen mesto - -
Source: Authors
The amount of wage for each employee depends on several factors, especially on the labor intensity of
individual transport operations. There is a table system of remuneration at ŽSR where individual stations
are grouped according to their operational indicators. According to the groups, individual tariff classes
for operational employees are developed. For the purposes of this article, we will consider a uniform
hourly rate for dispatchers and switches, as shown in Table 4.
Table 4. Data for direct cost calculation
Current status Working hours of
dispatchers
Hourly rate of
dispatchers
Annual rate of
dispatchers
Banská Bystrica
12 17,-€ 148 920,-€ Radvaň
Vlkanová
Sliač kúpele
Odb. Zvolen
mesto -
- -
Source: Authors
Each of the aforementioned traffic is in operation 24 hours a day and 7 days a week. The dispatchers
work for 12 hours, then take turns. The annual rate of dispatchers is calculated according to formula 2
as follows:
𝐴𝑛𝑛𝑢𝑎𝑙 𝑟𝑎𝑡𝑒 = ℎ𝑜𝑑𝑖𝑛𝑜𝑣á 𝑠𝑎𝑑𝑧𝑏𝑎 ∗ 24 ℎ𝑜𝑢𝑟𝑠 ∗ 365 𝑑𝑎𝑦𝑠 (2)
The annual number of dispatchers shall be calculated by taking into account the number of dispatchers
and their annual rate. In our case, we will not take into account holidays, sickdays, overtime, or even
days when dispatchers have time off between changes. The annual cost of transport service staff shall
be calculated according to formula 3 as follows:
𝐴𝑛𝑛𝑢𝑎𝑙 𝑐𝑜𝑠𝑡 = 𝑎𝑛𝑛𝑢𝑎𝑙 𝑟𝑎𝑡𝑒 ∗ 𝑠ℎ𝑖𝑓𝑡 𝑛𝑒𝑒𝑑 (𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑑𝑖𝑠𝑝𝑎𝑡𝑐ℎ𝑒𝑟) (3)
Table 5 shows the annual costs of dispatchers at each railway station on the line.
Table 5. Current status of annual costs of transport service employees
Railway station Annual costs for
dispatchers
Banská Bystrica 1 191 360,-€
Radvaň 595 680,-€
Vlkanová 595 680,-€
Sliač kúpele 595 680,-€
Σ 2 978 400,-€
Source: Authors
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
192
The highest costs for dispatchers are in the railway station Banská Bystrica, due to double occupancy
compared to other railway stations on the track section.
Staff costs following the introduction of a remote-controlled line
A fundamental change is in the occupation of individual transport. While in the previous chapter all the
traffic was occupied by at least one dispatcher, only one centralized traffic, in this case Banská Bystrica,
is occupied by the dispatcher.
Following the introduction of centralized dispatching, railway transport will be organized as follows:
• station Banská Bystrica, station Kostiviarska, station Uľanka – 1 dispatcher,
• station Radvaň, station Vlkanová, station Sliač kúpele – 1 dispatcher.
Emergency dispatchers (with a switchman wage rate) will be available for 24 hours in other stations.
They will take turns after twelve hours. They will be available at railway stations in case of failure of
station systems and line signalling equipment. Table 6 shows the dispatches of the dispatchers and their
turn-over need after the introduction of dispatching centralization.
Table 6. Operational need for transport staff at dispatch control
Railway station Number of dispatchers per shift Total need of dispatchers
Banská Bystrica 2 8
Radvaň 0 0
Vlkanová 0 0
Sliač kúpele 0 0
Odb. Zvolen mesto - -
Source: Authors
At the Radvaň, Vlkanová and Sliač stations there will always be one emergency dispatcher and their
need will be four emergency dispatchers. Table 7 provides the basis for cost calculation.
Table 7. Data for direct cost calculation after optimalization
Railway station Working hours of
dispatchers
Hourly rate of
dispatchers
Annual rate of
dispatchers
Banská Bystrica
12
17,-€ 148 920,-€
Radvaň 14,-€ 122 640,-€
Vlkanová 14,-€ 122 640,-€
Sliač kúpele 14,-€ 122 640,-€
Odb. Zvolen
mesto -
- -
Source: Authors
The calculations are carried out according to the same formulas, but with different rates for remote line
dispatchers and emergency dispatchers. Table 8 calculates staff costs after optimization.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
193
Table 8. Current status of annual costs of transport service employees for optimalization
Railway station Annual costs for
dispatchers
Banská Bystrica 1 191 360,-€
Radvaň 490 560,-€
Vlkanová 490 560,-€
Sliač kúpele 490 560,-€
Σ 2 663 040,-€
Source: Authors
At first glance it is obvious that the costs are lower due to the different station occupancy. Figure 4
shows the amount of savings achieved by cost reduction due to dispatching centralization.
Figure 4. Cost savings
Source: Authors
The total annual saving of labor costs for operating employees is 315 360, - € for the 20 km long track
after the dispatching centralization. As the signaling equipment for the dispatching line control is already
prepared in the railway station Banská Bystrica, the additional costs for the construction of the
dispatching apparatus will not be so extensive.
Conclusion
Dispatching centralization is the way that all developed railway reports are driven worldwide. In
addition to increasing throughput and security, there is also the economic side of the project. Dispatcher
centralization can greatly help reduce operating personnel costs, thus saving considerable money for
railway administrations. The aim of the article was to analyse the current state of dispatching
centralization on the ŽSR network with a link to human resources and to point out possible reductions
in personnel costs. It can be stated that the goal has been met.
References
[1] Dolinayová, A. & Nedeliaková, E., 2015. Controlling v železničnej doprave. Prvý ed. Bratislava:
DOLIS s. r. o..
[2] Dopravný úrad, 2019. Podmienky používania železničnej siete 2020. [Online] Available at:
https://www.zsr.sk/files/dopravcovia/zeleznicna-infrastruktura/podmienky-pouzivania-zel-
infrastruktury/podmienky-pouzivania-zel-siete-2020/priloha3_3_1_1-
kategoriatrati_pre_ucely_spoplatnovania-01_01_2019.pdf [Accessed 24 December 2019].
[3] Kupkovič, M. 1999. Náklady podniku, komplexný pohľad na náklady. Prvý ed. Bratislava:
Sprint.
€2 978 400,00 €2 663 040,00
€-€500 000,00
€1 000 000,00 €1 500 000,00 €2 000 000,00 €2 500 000,00 €3 000 000,00 €3 500 000,00
Staff costs before the introduction ofdispatching centralization
Staff costs following the introduction ofcentralized dispatching
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
194
[4] Majerčák, J. Gašparík, J. & Blaho, P., 2015. Železničná dopravná predvázka: Technológia
železničních stanic. Prvý ed. Žilina: EDIS – Vydavateľstvo Žilinskej univerzity v Žiline.
[5] Míka, V., 2005. Mikroekomómia: Vybrané přednášky z mikroekonómie a podnikovej ekonomiky
pre študentov bezpečnostního manažmentu. [Online] Available at:
http://fsi.uniza.sk/kkm/files/publikacie/mie/mie_07.pdf [Accessed 18 December 2019].
[6] Odbor stratégie a vonkajších vzťahov GR ŽSR, 2019. Postup zavádzania centier riadenia
dopravy a diaľkovo ovládaných tratí na sieti ŽSR. [Online] Available at:
http://www.betamont.sk/userfiles/editor/files/08_ZSR.pdf [Accessed 19 December 2019].
[7] Řezníček, B., 1982. Ekonomika železničnej dopravy. Prvý ed. Bratislava: Vydavateľstvo
technickej a ekonomickej literatúry ALFA.
[8] Štetka, P., 2014. Mzdové náklady. [Online] Available at:
https://peterstetka.wordpress.com/2014/10/03/mzdove-naklady/ [Accessed 18 December 2019].
[9] Šutyová, Z., 2019. Hrubá a čistá mzda od 1. 1. 2020. [Online] Available at:
https://www.podnikajte.sk/socialne-a-zdravotne-odvody/hruba-cista-minimalna-mzda-2020-dan-
odvody [Accessed 18 December 2019].
[10] VLAKY.NET, 2004. ŽSR 170: Vrútky – Zvolen. [Online] Available at:
https://www.vlaky.net/trate/47/zsr-170-vrutky-zvolen/ [Accessed 24 December 2019].
[11] Železnice Slovenskej republiky, 2012. Kalkulácia nákladov a tvorba cien ŽSR. Bratislava:
Odbor 330 GR ŽSR.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
195
DIGITAL TRANSFORMATION AND BUSINESS PROCESS MANAGEMENT
IN CREATIVE INDUSTRIES: THE CASE OF FILM PRODUCTION PROCESS
Tran Van Hai Trieu1
1Faculty of Management and Economics, Tomas Bata University in Zlín
nám. TGM 5555, 760 01 Zlín, Czech Republic
e-mail: [email protected]
Abstract
Business Process Management is an approach to model, analyse, and improve the business processes
that are applied for performance enhancement, cost reduction, and risk management. Also, some of the
business process management systems such as total quality management, business process
reengineering, etc. are the management tools for organizations to increase business competitiveness,
moreover, they help to achieve better system performance, for instance, higher profit, quicker response,
and better services for customer using services or products. Especially, the digitalization and digital
transformation based on the trend of technology 4.0 have contributed the changes all aspects of life,
culture, economy, and social, that is a reason why the purpose of the paper is to analyse the business
process management in the case of film production process, as well as study how the technology 4.0
trend and digital transformation affect the business process management in the creative industry.
Keywords
Industry 4.0, audiovisual, digital transformation, business process management, creative industry.
JEL Classification
M10, O30
Introduction
The explosion of industry 4.0 in the era of 21st brings many innovations of technology with the Artificial
Intelligent (AI), Big Data, Internet of thing (IoT), Cloud, Blockchain (Imran et al., 2018), as well as it
was the play of an important role for the backbone network to integrate physical objects, human actors,
production lines, intelligent machines, and processes across organizational boundaries (Schumacher et
al., 2016). One of the most important of industry 4.0, it is necessary to apply the new technology for
digitalization and digital transformation for any fields such as the creative industry, etc. with some
applications namely social media, mobile devices, and analytics, or embedded devices to improve
business performance through customer experience enhancement, business activities and creating a new
business model (Krizanic et al., 2019). Moreover, in the enterprise perspective, digital transformation
was considered as an organizational transformation to technology platforms, for instance, data analytics,
cloud, mobile, and social media platforms (Nwankpa and Roumani, 2016). Besides, business process
management was a systematic approach to identify, map, document, design, implement, measure and
control business processes, as well as it embraced the increasing IT support to improve, innovate, and
manage processes thoroughly, determining business results and creating customer value, achieving thus
the business goals with greater flexibility (Hitpass and Astudillo, 2019). Therefore, business process
management in the creative industry such as film production is so important, and it needs to digitize and
automate business process workflows to support the transparent interoperations of product or service
vendors, and achieving business results, creating customer value, and the business goals with greater
flexibility.
Theoretical background
Industry 4.0
The phrase "Revolution of industry 4.0" is taking place in many countries around the world and it is
discussed in many social networks. The study of Imran et al. (2018) showed that there were four stages
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
196
for the industrial revolution, the first industrial revolution took place in the 18th century with the steam-
based machine, the second took place in the 19th and 20th century with the electrical energy-based mass
production and the third happened at the end of the 20th century with the computer and internet-based
knowledge. The fourth, in the 21st, has been taken place with an era of the Artificial Intelligent (AI), Big
Data, Internet of thing (IoT), Cloud, Blockchain, and the perspective of Schumacher et al. (2016) also
mentioned industry 4.0 that was the internet and assistive technologies such as embedded systems,
played an important role for the backbone network to integrate physical objects, human actors,
production lines, intelligent machines and processes across organizational boundaries to form a new
kind of intelligent, networked and agile value chain. Besides, Haseeb et al. (2019) found the industry
4.0 features that included cloud computing, augmented reality, multilevel customer interaction,
advanced algorithms with big data, smart sensors, mobile devices, IoT platforms, location detection,
advanced human-machine, and 3D printing.
Digital Transformation
Along with the development of industry 4.0, the term of digital transformation is mentioned so much
and there are many terms and concepts related to digital transformation, and Nwankpa and Roumani
(2016) showed that it was the change and transformation with the technology platforms. In the enterprise
perspective, digital transformation was considered as an organizational transformation to technology
platforms such as data analytics, cloud, mobile, and social media platforms. Krizanic et al. (2019) gave
the definition of digital transformation as the digital technologies’ application namely social media,
mobile devices, and analytics or embedded devices to improve business performance through customer
experience enhancement, business activities and creating new business model, while Schallmo et al.
(2017) in their study proposed the definition of digital transformation that included business and
customer-related elements across all value chain segments and the application of new technologies so
that it could be an extraction, exchange, data analysis and convert information for using, evaluation and
decision making for business operations. Besides, digital transformation was relevant to business
models, processes, relationships and products that helped to increase the performance and company
operation scope.
Creative Industry
In this era, creativity is considered a key factor in the knowledge economy, which leads to creativity and
technology change, thereby it contributes to competitive advantages for business and country, the
transformation of creative ideas has contributed to the increase of tangible products and intangible
services. Martinaitytė and Kregždaitė (2015) said that the creative industry played an extremely
important role in economic development and social security, affecting macroeconomic results
throughout GDP index, technology performance, and employment, personal income, unemployment,
interest rates, and related welfare programs.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
197
Figure 1: Classification of creative industries. Source UNCTAD (2008, p.13)
Particularly, economic growth was influenced by the creative industry throughout the economic, social,
technological, scientific and creative industry was developed by factors related to art, social values,
cultural and local economic welfare, etc. In fact, the creative industry is new in the 20 th century,
UNCTAD (2008, p.13) classified four broad groups as heritage, arts, media, and functional creations,
these groups was divided by nine subgroups that consisted of the traditional culture expressions (arts
and crafts, festivals and celebrations), cultural sites (archaeological sites, museums, libraries,
exhibitions), visual arts (painting, sculpture, photography, and antiques), performing arts (live music,
theatre, dance, opera, circus, puppetry), publishing and printed media (books, press and other
publications), audiovisuals (film, television, radio and other broadcasting), design (interior, graphic,
fashion, jewelry, toys), creative services (architectural, advertising, cultural and recreational, creative
research and development, digital and other related creative services) and new media (software, video
games, and digitalized creative content).
Another opinion of Müller et al. (2009), creative industry was separated by core groups including six
sectors were content (film, games, journalism, authors, music, performing arts, photography and sound
studios), design (arts and crafts, design and fashion, graphic design, engineering design, and web
design), software (programming and computer services (excluding web design and computer games)),
architecture (architecture including landscaping and urban planning), advertising (planning, creating and
putting in place advertising campaigns, public relations management, market research, advertising
services), publishing (publishing of books, newspapers and other printed matter, including printing
services). Besides, from the perspective of Mangematin et al. (2014), music, movies and videos,
publishing, video games and television in the creative industry have been transformed by digitization.
Digital technology affected not only the dissemination, circulation, and storage of content but also it
changed the way viewers choose content to watch. Musicians and artists were determined by the number
of views on social video platforms as Youtube or other video channels that allowed users to participate
heavily in content development before being published.
Business Process Management
Business process management was mentioned by Paschek et al. (2017) as a management concept for
controlling, adapting and optimizing business processes, they defined that it was “a systematic
approach, to capture, shape, execute, document, measure, monitor and steering automatic and non-
automatic processes to reach coordinated and sustainable company targets” based on a concrete
definition from European Association of Business Process Management (EAPM), and they analysed the
digital transformation impacted the business process management while using methods like machine
learning or artificial intelligence. Besides, business process management was emerged by succeeding
concept of the total quality management (TQM) in the 1980s, and the business process reengineering
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
198
(BPR) in the 1990s, following BPR, there were many IT systems such as enterprise resource planning
(ERP) and customer relationship management (CRM) gained organizational focus (Brocke and Sinnl,
2011), while Utama and Ratnapuri (2018, p.1002) gave out seven rules of business process management
that were addressed as “1) major activities should be mapped and well-documented, 2) there are
horizontal linkages between key activities, 3) Business process management should rely on documented
procedure and system, 4) Business process management should measure an activity to assess the
performance, 5) based on continuous approach, 6) Business process management has to represent the
best practice, 7) Business process management is an approach for culture change in organization”. In
the study of Hitpass and Astudillo (2019), they referred business process management that was a
systematic approach to identify, map, document, design, implement, measure and control business
processes, as well as it embraced the increasing IT support to improve, innovate, and manage processes
thoroughly, determining business results and creating customer value, achieving thus the business goals
with greater flexibility.
Business process management of film production in creative industry
In creativity-intensive processes managing, there were two main perspectives as task-level, or activity-
level analysis and process-level analysis that were distinguished, in which, the task-level perspective
pertained to the questions of how pockets of creativity were characterized and how they could be
supported while the process-level perspective took a view of the overall business process, consequently
the existence of creative tasks within a business process significantly affected the process as a whole
(Seidel and Rosemann, 2008), for instance, in visual production process model emerged in subsequent
data analysis by Becker et al. (2011), as follow:
Figure 2: The format of production process. Source Becker et al. (2011, p.4)
In this process, the first phase was the idea generation for future visual formats that were generated by
producers, scriptwriters or directors. Moreover, commissioning broadcaster sometimes developed new
format and requirement of the production companies deploying the new concepts for the visual format,
for example, game shows or serials, and the output of this phase generally described an exposé for the
rough cut of the idea in few sentences. And then, when the broadcast networks accepted the exposé,
producers, and scriptwriters would develop more detailed concepts in terms of a script and screencast
including budget planning for the format production, and resent the exposé until receiving the feedback
of the second approval from the broadcasting network for the script and calculation, the visual
production was conducted by an established production team with the phase had to be strictly organized
so that there was the right staff and cast accordant with the right equipment at the right place to the right
time that was planned in pre-production, whilst the actual shooting of footage took place in the
production phase, besides, the footage was modified by the directors and producers expectations in post-
production. After finishing the visual production, the broadcasting network was received the format
including market research, TV program planning, marketing activities and the actual broadcasting of the
format (Becker et al., 2011). Especially, the prepared film for the edit process model was referred by
Seidel et al. (2007) that was a predetermined structure, as follow:
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
199
Figure 3: The prepared film for edit process. Source Seidel et al. (2007, p.524)
Although the prepared film for the edit process model was an example of a static model which was
relatively simple, in which, many points in the process where decisions or choices must still be made on
which branch of the model to execute based on the conditions of a particular instance, that was any
flexibility had to be incorporated into the process control flow as explicit conditional branches to decide
the control flow mixing with business process logic (Seidel et al., 2007).
At last, from two above illustrated examples related to the business process management of film
production, it can be seen that there were many challenges arise from the existence of creative tasks
within business processes such as allocating resources (task-level, process level), enhancing creativity
(task-level), managing creative risks (task-level, process-level), and enhancing process performance
(process-level), in which, the allocating resource with the special creative tasks that were resource and
time-intensive, as well as the process owner had to decide what budget, equipment while creative
individuals had to be allocated to what ask. For enhancing creativity, the process owner wanted to
enhance the quality of the creative output as the core output of that task through generating a new idea,
the evaluation of alternative proposals, or a selection process. Moreover, managing creative risks related
to creative tasks were inherently connected to high variance of possible outcomes because the fact of
creative was original and came up with novel ideas, and solutions that was reason why happened
unwanted consequences, for instance losing control of process (losing control of time and budget), low
product quality (which would lead to customer dissatisfaction), and lack of external compliance (which
could lead to a loss of reputation or even to lawsuits), especially it related to the film industry when the
customer was often unable to specify the requirements and the visual effects studio such as providing a
set of iterative solutions to get closer to the actual requirements. Besides, the time and budget had to be
controlled by the company, as well as complying with external requirements such as governmental
policies and legal requirements, hence the identification of creative tasks and their attributes within a
process was the prerequisite to successfully implementing risk management strategies. For enhancing
process performance, creativity-intensive processes were characterized by a high demand for flexibility
that was conventional process automation approaches such as workflow management or even more
sophisticated approaches such as exception handling or evolutionary workflow solutions would not be
appropriate, in fact, processes included both well-structured parts and pockets of creativity that did not
have any obvious structure at all, thus, identifying and better understanding these pockets of creativity
allowed for designing an IT solution providing a maximum level of automation where it was suitable
(Seidel and Rosemann, 2008).
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
200
How digital transformation impacts business process management in creative
industries
Digital transformation and business process management 4th wave
Each organization had its business processes that could be specified in terms of the goals they wanted to achieve but letting the process efficiency, quality and agility were the keys to business success, every organization had to invest in the quality of business process management. Digital transformation replied on the creation of integrated, stable and reliable processes and structured data that were further drivers of flexibility and agility, as well as the integration of various technological innovations into business, previous automation and processes managed by the software within the business process management. Furthermore, the technological impacted on process changes, two other impacts could be considered, such as (1) the impact of data and (2) the impact of human factors. During the digital transformation of organizations, named two impacts on business processes also had to be taken into the consideration since they generated three main areas of business process management trends in digitalization: (i) Business process management influenced by data, (ii) Business process management influenced by social factors and (iii) Business process management based on process cases (Vugec et al., 2018). Besides, the study of Pihir (2019) analyzed and gave out three waves of process evolution and adding the last 4th wave as a digital transformation wave of business process management (refer to Table 1).
Table 1: Business process management and digital transformation across time:
Digital transformation as a new business process management wave.
Phase Time Focus Business Technology Tools/Enablers
Industrial Age
Industrial Age 1750 -
1960s
- Specialization of Labour - Task Productivity - Cost Reduction
- Functional - Hierarchies - Command - Control - Assembly Line
- Mechanization - Standardization - Record-keeping
- Scientific Management - PDCA Improvement Cycle; Financial Modeling
Informational Age
1st Wave
Process
Improvement
70s - 80s
- Quality
Management
- Continuous
Flow
- Task Efficiency
- Multi-
Industry
Enterprises
- Line of
Business
- Organization
Mergers &
Acquisitions
- Computerized
Automation
- Management
Information
Systems
- MRP
- TQM
- Statistical
Process Control
- Process Improvement Methods
2nd Wave -
Process
Reengineering
1990s - Process
Innovation
- “Best
Practices”
- Better, Faster,
Cheaper
- Business via the Internet
- Flat
Organization
- End-to-end
Processes
- Value
Propositions –
Speed to
Market,
Customer
Intimacy,
Operational
Excellence
- Enterprise
Architecture
- ERP - CRM - Supply Chain Management
- Activity Based
Costing
- Six Sigma
- Buy vs. build -
Process
Redesign/ Reengineering Methods
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
201
3rd Wave -
Business
Process
Management
2000 -
2015
- Assessment,
Adaptability, &
Agility
- 24x7 Global
Business
- Continual Transformation
- Networked
Organization
- Hyper
Competition
- Market
Growth Driven
- Process
Effectiveness
over Resource
Efficiency
- Organizational Effectiveness Over Operation
- Enterprise
Application
Integration
- Service
Oriented
Architecture
- Performance
Management
software
- BPM Systems
- Balanced
Scorecard
- Self Service &
Personalization
- Outsourcing,
Co-Sourcing, In-
sourcing
- BPM Methods
4th Wave -
Digital Transformation
2015 +
- Process/
Product
Innovation by
Creative Use of
New
Technology,
- Using Disruptions as New Possibilities not Problems
- Added Value
to Old
Customers/
Products,
- New Value
through New
Business
Models
- Radical change driven by Technology and Shift in Mind
- AI; Big Data;
Cloud
Computing;
- Data Analytics,
− Implantable
technologies
- IoT; Smart
cities; 3D print;
Driverless cars
- Robotics − Block chain − Sharing economy
- Process
Oriented
Applications
(POA)
- Intelligent
BPMS Systems
- Software as a
Service
- New digital
transformation
methods
- New digital transformation tools
Source: Pihir (2019, p.358)
Business process management capabilities in the digital age
The penetration of economy and society based on digital technologies referred to digitalization, a global
phenomenon leading to an opportunity-rich, hyper-connected, fast-moving, and highly competitive
environment, for instance, the social and mobile technologies (e.g., social media and social collaboration
platforms) helped the people to communicate and emancipate work from time and location, as well as
equipping physical objects with sensors, actuators, computing power, and connectivity, the internet of
things boosted the fusion of the physical and digital world. Besides, when combined with the potential
of blockchain-empowered solutions, it enabled novel value exchanges among individuals, businesses,
and smart things, reducing the distance between customers and companies, and grants access to so far
unexplored data sources. Further, data analytics, including the latest advances in cognitive technologies
enabled capitalizing on data in a diagnostic, predictive, and prescriptive manner, building the foundation
of data-driven business models, the automation of unstructured tasks, and natural interaction between
humans and machines (e.g., social robotics), and 3D/4D printing disrupted supply chains and value
networks by enabling highly decentral, delayed production facilities. Digital technologies enabled so far
unimaginable business processes, digitalization was a true game-changer for business process
management, posing manifold challenges and opportunities. In an environment characterized by
advanced process automation capabilities and new digital process design opportunities, a purely reactive
and problem-driven approach to business process management is no longer sufficient. Instead, business
process management needed to become ambidextrous, i.e., it must leverage digital technologies for both
streamlining and innovating business processes. With the uptake of digitalization, business process
management had to continue emphasizing the people's perspective to ensure an optimized augmentation
for employees and customers in the future of work and consumption (Kerpedzhiev et al., 2017).
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
202
Furthermore, Kerpedzhiev et al. (2017) showed business process management capabilities including 30
capabilities structured along with the established core elements such as strategic alignment, governance,
methods or IT, people, and culture in the digital age (refer to Table 2), in which, the strategic alignment
referred the continual alignment of organizational priorities and processes, enabling the achievement of
business goals, and governance established relevant and transparent accountability and decision-making
processes to align rewards and guide actions, while methods were the approaches and techniques
supporting and enabling consistent process actions and outcomes through using information technology
application such as software, hardware, and information systems that enable and support business
processes. For the people in the table were the individuals and groups who continually enhanced and
applied their process-related expertise and knowledge, and culture included the collective values and
beliefs that shape process-related attitudes and behaviors.
Table 2: Framework of business process management capabilities in the digital age.
Strategic
Alignment
Governance Methods/ Information
technology
People Culture
Strategic
Business
Process
Management
Alignment
Contextual
Business
Process
Management
Governance
Process
Context
Management
Multi-purpose
Process Design
Business
Process
Management
and Process
Literacy
Process
Centricity
Strategic
Process
Alignment
Contextual
Process
Governance
Process
Compliance
Management
Advanced
Process
Automation
Data
Literacy
Evidence
Centricity
Process
Positioning
Process
Architecture
Governance
Process
Architecture
Management
Adaptive Process
Automation
Innovation
Literacy
Change
Centricity
Process
Customer
and
Stakeholder
Alignment
Process Data
Governance
Process Data
Analytics
Agile Process
Improvement
Customer
Literacy
Customer
Centricity
Process
Portfolio
Management
Roles and
Responsibilities
Business
Process
Management
Platform
Integration
Transformational
Process
Improvement
Digital
Literacy
Employee
Centricity
Source: Kerpedzhiev et al. (2017, p.2)
In the digital age, the strategic alignment had to ensure the transparency, and benefits associated with
business processes and business process management that were arranged with the expectations of re-
conditioned, digitally savvy customers, and other stakeholders. Besides, business process management
and process governance had to be highly contextual with the methods and tools were chosen and
customized that was associated with organizational contexts, and process design fitted multiple purpose
such as customer-centric, risk-aware, or flexibility-aware processes, and mass-personalized processes
that was the same to the process data analytics and business process management platform integration,
as well as process automation forced to tackle unstructured tasks and enable new forms of human-
machine interaction by leveraging opportunities of digital technologies such as cognitive automation,
social robotics, and smart devices. Especially, related to the people in the digital age was knowledge
requirement about digital technology that was so important for the people, for instance, data analytics,
data privacy, data security techniques, innovation techniques, digital economy, and digital business
models along with the substantial knowledge about business process management methods and tools.
Moreover, new process values and beliefs were required, and business process management had to
actively involve the people in process decisions, the same time forecasted the effects of these decisions
on their work lives, took customer feedback seriously and granted them the sovereignty to make self-
dependent decisions although they encountered unprecedented challenges.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
203
Conclusion
The research topic studies the overall issues of technology 4.0 with the concepts related to the artificial
intelligent, big data, internet of thing, blockchain, cloud computing, augmented reality, multilevel
customer interaction, advanced algorithms with big data, smart sensors, mobile devices, internet of thing
platforms, location detection, advanced human-machine, and 3D printing, as well as the role of digital
technology, and digital transformation affects the business models, process business, business process
management that helps to increase the performance and company operation scope, especially companies
in the creative industry. Besides, the works are specified by business process management, for instance,
total quality management, business process reengineering, and enterprise resource planning that is an
innovation through the systematization method to identify, map, document, design, implement, measure
and control business processes, and it embraces the increasing IT support to improve, innovate, and
manage processes thoroughly, determining business results and creating customer value, achieving the
business goals with greater flexibility.
Moreover, this paper also analyzes the creativity-intensive processes managing through two main
perspectives that are task-level (activity-level) analysis and process-level analysis, and it is illustrated
by two examples that are the visual production process model and the prepared film for edit process
related to the business process management of film production with many challenges arise from the
existence of creative tasks within business processes such as allocating resources (task-level, process
level), enhancing creativity (task-level), managing creative risks (task-level, process-level), and
enhancing process performance (process-level). Thus, each organization wants to achieve the process
efficiency, quality and agility for business success that they must invest in the quality of business process
management through using digital technology, and digital transformation that is reason why there are
four waves of process evolution, in which, the last 4th wave as a digital transformation wave of business
process management, as well as the business process management capabilities including 30 capabilities
structured along with the established core elements such as strategic alignment, governance, methods or
IT, people, and culture in the digital age, have been referred and analyzed, especially, the role of digital
technology is so important for business process management through the methods and tools that are
associated with organizational contexts, and process automation forces to tackle unstructured tasks and
enable new forms of human-machine interaction by leveraging opportunities of digital technologies such
as cognitive automation, social robotics, and smart devices.
To sum up, the development of economy and society based on digital technologies along with
globalization trend has facilitated an opportunity-rich, hyper-connected, fast-moving, and highly
competitive environment such as the social and mobile technologies, equipping physical objects with
sensors, actuators, computing power, etc. The people play a central role in the digital age who is required
a good knowledge of digital technology such as data analytics, data privacy, data security techniques,
innovation techniques, digital economy, and digital business models along with the substantial
knowledge about business process management.
References
[1] Becker, J., Bergener, K., Schwehm, M. O., and Voigt, M. (2011). CONFIRMING BPM THEORY
IN CREATIVE INDUSTRY CONTEXT – A CASE STUDY IN THE GERMAN TV INDUSTRY. 19th
European Conference on Information Systems, ECIS 2011.
[2] Brocke, J. vom, and Sinnl, T. (2011). Culture in business process management: a literature review.
Business Process Management Journal, 17(2), pp. 357–377. Available at <https://doi.org/10.1108/14637151111122383>.
[3] Haseeb, M., Hussain, H. I., ´Slusarczyk, B., and Jermsittiparsert, K. (2019). Industry 4.0: A
solution towards technology challenges of sustainable business performance. Social Sciences,
8(5). Available at <https://doi.org/10.3390/socsci8050154>.
[4] Hitpass, B., and Astudillo, H. (2019). Editorial: Industry 4.0 Challenges for Business Process
Management and Electronic-Commerce. Journal of Theoretical and Applied Electronic
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
204
Commerce Research, 14(1), pp. I–III. Available at <https://doi.org/10.4067/S0718-
18762019000100101>.
[5] Imran, M., ul Hameed, W., and ul Haque, A. (2018). Influence of Industry 4.0 on the production
and service sectors in Pakistan: Evidence from textile and logistics industries. Social Sciences,
7(12), pp. 0–21. Available at <https://doi.org/10.3390/socsci7120246>.
[6] Kerpedzhiev, G., König, U., Röglinger, M., and Rosemann, M. (2017). Business Process
Management in the Digital Age. BPTrends, (July), pp. 1–6. Available at
<https://doi.org/10.13140/RG.2.2.12087.42408>.
[7] Krizanic, S., Sestanj-peric, T., and Tomicic-pupek, K. (2019). THE CHANGING ROLE OF ERP
AND CRM IN DIGITALTRANSFORMATION. (May), pp. 23–24.
[8] Mangematin, V., Sapsed, J., and Schüßler, E. (2014). Disassembly and reassembly: An
introduction to the Special Issue on digital technology and creative industries. Technological
Forecasting and Social Change, 83(1), pp. 1–9. Available at
<https://doi.org/10.1016/j.techfore.2014.01.002>.
[9] Martinaitytė, E., and Kregždaitė, R. (2015). THE FACTORS OF CREATIVE INDUSTRIES
DEVELOPMENT IN NOWADAYS STAGE. Economics and Sociology, 8(1), pp. 55–70.
[10] Müller, K., Rammer, C., and Trüby, J. (2009). The role of creative industries in industrial
innovation. Innovation: Management, Policy and Practice, 11(2), pp. 148–168. Available at
<https://doi.org/10.5172/impp.11.2.148>.
[11] Nwankpa, J. K., and Roumani, Y. (2016). IT capability and digital transformation: A firm
performance perspective. 2016 International Conference on Information Systems, ICIS 2016, pp.
1–16.
[12] Paschek, D., Luminosu, C. T., and Draghici, A. (2017). Automated business process management-
in times of digital transformation using machine learning or artificial intelligence. MATEC Web
of Conferences, 121, pp. 1–8. Available at <https://doi.org/10.1051/matecconf/201712104007>.
[13] Pihir, I. (2019). BUSINESS PROCESS MANAGEMENT AND DIGITAL TRANSFORMATION.
41st International Scientific Conference on Economic and Social Development, (May), pp. 353–
360.
[14] Schallmo, D., Williams, C. A., and Boardman, L. (2017). Digital transformation of business
models-best practice, enablers, and roadmap. International Journal of Innovation Management,
21(8), pp. 1–17. Available at <https://doi.org/10.1142/S136391961740014X>.
[15] Schumacher, A., Erol, S., and Sihn, W. (2016). A Maturity Model for Assessing Industry 4.0
Readiness and Maturity of Manufacturing Enterprises. Procedia CIRP, 52, pp. 161–166. Available
at <https://doi.org/10.1016/j.procir.2016.07.040>.
[16] Seidel, S., Adams, M., ter Hofstede, A., and Rosemann, M. (2007). MODELLING AND
SUPPORTING PROCESSES IN CREATIVE ENVIRONMENTS. Proceedings 15th European
Conference on Information Systems, pp. 516–527.
[17] Seidel, S., and Rosemann, M. (2008). Creativity Management–The New Challenge for BPM.
BPTrends, (May), pp. 1–8.
[18] UNCTAD. (2008). Creative Economy Report 2008. The Challenge of Assessing the Creative
Economy: towards Informed Policy-making. In Harvard Business Review (Vol. 8).
[19] Utama, I. D., and Ratnapuri, C. I. (2018). Comparing the Business Process in Creative Industry
at Bandung. Proceedings of the International Conference on Industrial Engineering and Operations
Management, 2018-March, pp. 999–1004.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
205
[20] Vugec, D. S., Stjepić, A., and Vidović, D. I. (2018). The Role of Business Process Management
in Driving Digital Transformation: Insurance Company Case Study. International Scholarly and
Scientific Research & Innovation, 12(9), pp. 730–736.
Contact information
Trieu Tran Van Hai, Ph.D student
University: Tomas Bata University in Zlín
Faculty: Management and Economics
E-mail: [email protected] or [email protected]
ORCID: 0000-0002-2532-8016
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
206
AVOIDANCE OF COST INCREASES DURING CHANGE MANAGEMENT Rijad Trumic1
1Department of Economics, VŠB – Technical University of Ostrava,
Sokolská tř. 33, Ostrava 70200, Czech Republic
e-mail: [email protected]
Abstract
This work deals with the problem of cost increases in purchasing after a supplier nomination. The first
part of the paper explains the problem and the AHP method used to find a solution for this issue.
Furthermore, the TOPSIS method is performed to substantiate the AHP result. The expectation is that
the subject of overhead is ranked as the first priority and the catalogue of changes is ranked second
because it means a large output with a manageable time schedule. In the last part a summarized
conclusion and an outlook for the dissertation is also given.
Keywords
Analytical Hierarchy Process (AHP), TOPSIS, change catalogue, overhead, technical requirements.
Introduction
For emphasizing a word or phrase, please, use italic. Do not use boldface typing or capital letters except
for section headings (cf. remarks on section headings, below). Achieving the best costs and best prices
for the nomination of a supplier over the serial running time to the aftersales is one of our goals.
Unfortunately, in reality, good nomination results are often forfeited because of changes in the
component development process until the "start of production" (SOP).
Often, we commit ourselves to a supplier over a period of up to 25 years at an early stage and with
relatively low maturity of the component! Thus, it is undisputed: The nomination has the largest impact
on costs during the whole process
But not the changes after the nomination: They are a significant cost driver for OEMs and they often
have to be negotiated in a single-source environment.
Change requests are never entirely avoidable in a vehicle development process, certainly the efforts in
concept plausibility must be intensified and the quality specifications must be improved.
All car manufacturers nominate suppliers for vehicle components many years before the SOP. This
means that the technical procurement status of these components does not have the same technical status
as for the SOP because the components face technical changes due to the development phase after a
nomination. However, since the suppliers have been defined in this phase and the supply relationship
has been contractually defined, the changes are in many cases implemented followed by higher prices
by the suppliers. Car manufacturers very often have no option to negotiate real costs or to withdraw
from the contract, as the development has already been performed and the security of supply is the first
priority. At the end, the high costs are very often accepted without any negotiating leverage.
Nevertheless, changes are an integral part of the development process, in order to continuously improve
vehicles to market maturity, to develop them further and to keep them competitive through necessary
changes, and also to be competitive on the cost side despite the changes. The objective of this paper is
to prioritize and focus on cost reduction measures and is a sub-aspect and integral part of the dissertation,
which also deals with the tools from the measures found.
The main goal is defined as the cost savings in change management which is in level one. The cost
savings are examined for further 6 aspects and decision criteria in level two. These are briefly explained:
The speed of implementation is a very important factor in the selection of the tools. It is very important
how fast each topic can be implemented in practice, how complex the topics are in the preparation and
how much capacity must be used in terms of man power and time.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
207
Furthermore, it is also very important whether premises can be set for the respective topic. For example,
if premises are kept too coarse and generous in a change catalogue, the costs cannot be precisely defined.
A precise and detailed definition of the premises also enables a detailed statement of costs for a specific
measure.
It is also important to ask whether the know-how is available internally. The employees and their
experience is essential. Employees from development and purchasing can bring the topics into the tools
from Lessons Learned. These topics have to be evaluated by the supplier.
Finally, output is the last also very important criteria. It may be that everything can be implemented very
fast, with low capacity and high know-how, but if the output is small or it brings little savings, the focus
is usually placed on another topic. These decision criteria are speed of implementation, complexity,
capacity effort, setting of premises, internal Know-How and output.
In the context of this paper, three alternatives of tools were selected, which are evaluated in quality and
in relation to the decision criteria. These 3 subject areas are change catalogue or pre-negotiation of
possible changes in the future, Improvement of the technical requirements and specifications.
A decrease of the overhead and profit surcharge or a question whether the used „Surcharge calculation”
by many OEMs, is future-oriented.
The use of a cost catalogue after a nomination can be useful in many cases also to be able to negotiate
changes better and more effectively. Increasing product requirements are an important part of a
nomination. A good change catalogue is developed in close coordination between the purchasing and
developing colleagues. In the second step, the contribution of a cost calculator is of course very
important in order to calculate the measures or changes requested in the change catalogue. This leads to
better negotiation results.
The aim of the Dissertation is to develop a consistent method so that purchasing can pre-negotiate
possible changes that may occur in the future before a supplier nomination. If these changes occur in
reality, the conditions are already established and so they then have their validity. The conditions that
have been agreed with a supplier prior to being awarded are significantly better than conditions that can
be achieved after a suppliers’ nomination. The cost differences in a product line are in the 6-digit range.
The method should also represent the relationship between different component families. To a certain
extent, changes are pre-negotiable. A change catalogue contains technical increases and reductions as
well as commercial issues, such as changes in the volume of the components during the project, premises
for the nomination, relocations, raw materials or currency issues. A detailed elaboration of the change
catalogue will be developed and presented in the context of the dissertation.
Improving the quality of the specifications can avoid many changes in the series development, so the
catalogue of changes can be superfluous or greatly reduced. However, this improvement is very often
not possible in the early phase or at the time of nomination because the requirements for the component
and the development goals are often unknown. For example, the requirements and premises are often
kept very general at the time of nomination, so they are not defined more precisely. The concept and
development maturity is simply not given at the time of nomination, so changes have to occur in most
cases. On the other hand, changes are also deliberately decided so that OEMs stay competitive or to get
the innovation and technological leadership. The requirement specifications describe the requirements
for the component in a minimal form, which is required as a function in the vehicle (in order to also
achieve the lowest costs during a nomination) and the change catalogue describes the possible extensions
with the additional costs. Too generously designed specifications lead to higher costs during the
nomination. A reduction of the technical requirements in the specifications after a nomination does not
lead to the desired material cost savings as opposed to before a nomination. This means a mature
specification sheet in combination with a good and detailed change catalogue are the optimal solution.
As the third main aspect in avoiding costs in series development, but even more importantly in series
delivery, is a decrease of overhead and profit surcharges. Many OEMs use a surcharge calculation as a
calculation base. The calculation uses the botton-up approach to calculate the cost components and then
add the overhead and profit surcharges as a percentage of the material and production costs.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
208
This is primarily determined during the nomination and agreed with the supplier. This means that the
supplier confirms that his business case, with a certain volume, a certain price and the stated overhead
and profit, has been accepted. This also results in the profit and overhead as a sum in EUR (not in %)
over the duration of a project. However, there are often volume changes in the project duration compared
to the nomination status. Nevertheless, the percentage surcharges are still added to every single
component that the OEM buys, even though the supplier agreed to the lower sum x during nomination.
This gives the supplier additional pure profit, which he did not expect and which a supplier does not
have to make any additional effort for (especially since most OEMs pay the investment separately). It
is therefore sufficient if the supplier gets an additional profit by the increase of volume, which may also
finance an additional investment required to produce the higher volume. Currently, this approach is
given little attention and it is quite easy and fast to implement. This type of surcharge has to be frozen
during the nomination. This means the supplier needs to point out the overhead and profit in their offer
as an absolute number in EUR which is then paid over the nominated quantity as a surcharge. As soon
as the volume changes, the term of payment of the apportionments also changes, in other words, the
apportionment is canceled earlier than at the EOP (End of Production). Instead, many suppliers are
overpaid till the EOP and very often the OEMs have not even noticed this phenomenon. Figure 2 shows
the problem of costs increasing hierarchically. The goal is noted on the top level. The second level lists
the criteria that are important. The third level shows the individual alternatives or tools that can be
considered. The AHP is also well suited to solving the problem because the assessment is quantitative
in nature.
Methodology and Data
The Analytical Hierarchy Process (AHP) presented by SAATY is a method for solving multi-criteria
decision problems3. In this method, a problem is broken down hierarchically into sub-problems, thus
reducing complexity. The sub-problems are solved step by step.
The evaluation of the alternatives in the AHP procedure takes place with regard to their importance to a
superordinate element or criterion and is compared in pairs4. The criteria are compared at one level in
the hierarchy of the problem5. If a problem is divided into several levels, the pair comparisons are
continued first on the criteria level and then successively for the other criteria levels6. The results of all
comparisons are presented in an evaluation matrix. The priorities are calculated through an iterative
process with the so-called eigen value method. There are three steps. In the first step, the evaluation
matrix is squared7. In a second step, the matrix is normalized and the local weights are determined. In
the third step, this process of squaring and normalization is repeated in an iterative process until the
calculated weights deviate very small from the values determined from the previously calculated
matrix8.
𝐴 =
(
𝑎11 … 𝑎1𝑗 … 𝑎1𝑛… … … … …𝑎𝑖1 … 𝑎𝑖𝑗 … 𝑎𝑖𝑛… … … … …𝑎𝑛1 … 𝑎𝑛𝑗 … 𝑎ₙₙ)
with
∀𝑖 = 1,… , 𝑛 ∀𝑗 = 1,… , 𝑛: 𝑎𝑖𝑗 > 0 ∀𝑖 = 𝑗: 𝑎𝑖𝑗 = 1 ∀𝑖 = 1, … , 𝑛
3 See Saaty (2000); Saaty (2001) 4 See Saaty (1994b), p.22; Saaty (2000), p. 105 5 See Saaty (1994b), p.22; Saaty (2000), p. 105 6 See Saaty (1994b), p.22; Saaty (2000), p. 105 7 Dolan JG, Isselhardt BJ, Cappuccio JD, Med Decis Mak. 1989;9 (1):40–50 8 Saaty TL. The analytic hierarchy process: planning, priority setting, resource allocation. 2. Aufl. New York:
McGraw-Hill; 1980. S. XIII, 287.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
209
∀𝑗 = 1,… , 𝑛: 𝑎𝑖𝑗 = 𝑎𝑗𝑖−1
Each element of the comparison shows how much an element is more significant with respect to the
element of the overlying level9.
Table 1: Relative meaning of the criteria for the superior element
AHP Scale of Importance
for comparison pair (aij)
Numeric Rating Reciprocal (decimal)
Extreme Importance 9 1/9 (0,111)
Very strong to extremely 8 1/8 (0,125)
Very strong Importance 7 1/7 (0,143)
Strongly to very strong 6 1/6 (0,167)
Strong Importance 5 1/5 (0,2)
Moderately to Strong 4 1/4 (0,25)
Moderate Importance 3 1/3 (0,333)
Equally to Moderately 2 1/2 (0,5)
Equal Importance 1 1 (1,0) In order to determine the inconsistency between the pairs, SAATY has developed a consistency index
(C.I. = Consistency Index) and a consistency value (C.R. = Consitency Ratio)10. The largest eigenvalue
𝜆𝑚𝑎𝑥 of the evaluation matrix is equal to the dimension n of the matrix for the consistency of the pairs.
The consistency of the matrix can be checked using the largest eigenvalue 𝜆𝑚𝑎𝑥 .
𝐶. 𝐼. = 𝜆𝑚𝑎𝑥 − 𝑛
𝑛 − 1
The necessity of revising the evaluation matrix can be checked with the consistency value (C.R.). A C.R
value <0.1 is permissible. SAATY indicates that for .C.R.≥0,1 a revision of the pair comparisons in the
evaluation matrix should be made11.
𝐶. 𝑅. = 𝐶. 𝐼.
𝑅. 𝐼.
The Random Index (R.I.) is an average consistency index that was randomly generated from reciprocal
matrices. SAATY shows in table 2 the determined values.
Table 2: Random Index12
n 2 3 4 5 6 7 8 9 10 11 12 13 14 15
R.I
.
0,0
0
0,5
2
0,8
9
1,1
1
1,2
5
1,3
5
1,4
0
1,4
5
1,4
9
1,5
1
1,5
4
1,5
6
1,5
7
1,5
8
9 See Saaty (1994b), p.23 10 See Saaty (2000), p. 47; Saaty (2001), p. 80 11 See Saaty (1994b), p. 27; Saaty (2000), p. 84 f.; Saaty/Vargas (2001), p. 9. 12 See Saaty (2000), p. 65 and 84.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
210
Application of the AHP method
Input data and model quantification of criteria
Table 3: Evaluation matrix of all criteria in the second decision level
Speed Complexity Capacity Premises Know
How
Output
Speed 1 0,2 0,2 0,125 0,111 0,111
Complexity 5 1 3 0,167 0,111 0,111
Capacity 5 0,333 1 0,143 0,2 0,2
Premises 8 6 7 1 1 1
Know How 9 9 5 1 1 3
Output 9 9 5 1 0,333 1
The result for the evaluation matrix in table 3 is 𝜆𝑚𝑎𝑥 = 6,35.
Consistency-Index C.I. (N=6)
𝐶. 𝐼. =6,35 − 6
6 − 1= 0,07
Calculation of the consistency rate C.R. (Random-Index R.I.= 1,25)
𝐶. 𝑅. =0,07
1,25= 0,056 < 0,1
Since the C.R. value is 0.056, it is not necessary to revise the evolution matrix.
Calculation of alternatives in terms of speed
Table 4: Evaluation matrix of all alternatives related to the speed of implementation
Speed Change cataloge Technical
Requirements
Overhead
Change cataloge 1 4 0,142
Technical
Requirements
0,25 1 0,142
Overhead 7 7 1 The result for the evaluation matrix in table 4 is 𝜆𝑚𝑎𝑥 = 3,086.
Consistency-Index C.I. (N=3)
𝐶. 𝐼. =3,086 − 3
3 − 1= 0,043
Calculation of the consistency rate C.R. (Random-Index R.I.= 0,52)
𝐶. 𝑅.=0,043
0,52= 0,082 < 0,1
Since the C.R. value is 0.082, it is not necessary to revise the evolution matrix.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
211
Calculation of alternatives in terms of complexity
Table 5: Evaluation matrix of all alternatives based on the complexity of the implementation
Complexity Change cataloge Technical
Requirements
Overhead
Change cataloge 1 0,25 3
Technical
Requirements
4 1 3
Overhead 0,333 0,333 1 The result for the evaluation matrix in table 5 is 𝜆𝑚𝑎𝑥 =3,015.
Consistency-Index C.I. (N=3)
𝐶. 𝐼. =3,015 − 3
3 − 1= 0,0078
Calculation of the consistency rate C.R. (Random-Index R.I.= 0,52)
𝐶. 𝑅. =0,0078
0,52= 0,0088 < 0,1
Since the C.R. value is 0,0088, it is not necessary to revise the evolution matrix.
Calculation of alternatives in terms of capacity
Table 5: Evaluation matrix of all alternatives based on the capacity during implementation
Capacity Change cataloge Technical
Requirements
Overhead
Change cataloge 1 5 0,167
Technical
Requirements
0,2 1 0,167
Overhead 6 6 1 The result for the evaluation matrix in table 6 is 𝜆𝑚𝑎𝑥 = 3,122.
Consistency-Index C.I. (N=3)
𝐶. 𝐼. =3,122 − 3
3 − 1= 0,061
Calculation of the consistency rate C.R. (Random-Index R.I.= 0,52)
𝐶. 𝑅.=0,061
0,52= 0,069 < 0,1
Since the C.R. value is 0,069, it is not necessary to revise the evolution matrix.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
212
Calculation of alternatives in terms of premises
Table 7: Evaluation matrix of all alternatives based on the premises during implementation
Premises Change cataloge Technical
Requirements
Overhead
Change cataloge 1 4 0,333
Technical
Requirements
0,25 1 0,167
Overhead 3 6 1 The result for the evaluation matrix in table 7 is 𝜆𝑚𝑎𝑥 = 3,027.
Consistency-Index C.I. (N=3)
𝐶. 𝐼. =3,027 − 3
3 − 1= 0,0139
Calculation of the consistency rate C.R. (Random-Index R.I.= 0,52)
𝐶. 𝑅. =0,0139
0,52= 0,0156 < 0,1
Since the C.R. value is 0,0156, it is not necessary to revise the evolution matrix.
Calculation of alternatives in terms of know-how
Table 8: Evaluation matrix of all alternatives related to implementation know-how
Know -how Change cataloge Technical
Requirements
Overhead
Change cataloge 1 3 0,333
Technical
Requirements
0,333 1 0,167
Overhead 3 6 1 The result for the evaluation matrix in table 8 is 𝜆𝑚𝑎𝑥 = 3,009.
Consistency-Index C.I. (N=3)
𝐶. 𝐼. =3,009 − 3
3 − 1= 0,0048
Calculation of the consistency rate C.R. (Random-Index R.I.= 0,52)
𝐶. 𝑅. =0,0048
0,52= 0,0054 < 0,1
Since the C.R. value is 0,0054, it is not necessary to revise the evolution matrix.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
213
Calculation of alternatives in terms of output
Table 9: Evaluation matrix of all alternatives related to the output during implementation
Output Change cataloge Technical
Requirements
Overhead
Change cataloge 1 3 0,5
Technical
Requirements
0,333 1 0,5
Overhead 2 2 1 The result for the evaluation matrix in table 9 is 𝜆𝑚𝑎𝑥 = 3,11.
Consistency-Index C.I. (N=3)
𝐶. 𝐼. =3,11 − 3
3 − 1= 0,055
Calculation of the consistency rate C.R. (Random-Index R.I.= 0,52)
𝐶. 𝑅.=0,055
0,52= 0,0622 < 0,1
Since the C.R. value is 0,0622, it is not necessary to revise the evolution matrix.
Summary of results
Table 10: Ranking of results
Weights Ranking
Change cataloge 0,286 2
Technical Requirements 0,111 3
Overhead 0,601 1
In this summary, the importance or priorities are clearly visible. The top priority is overhead, followed
by the change catalogue. In contrast, the subject of specifications is at the last place, which is also
reflected in reality.
Conclusion
From experience, at the beginning of this thesis there was an expectation that this sequence would also
be reflected after the qualitative presentation. The same priorities were presented and confirmed using
the AHP and TOPSIS method. The top priority is clearly the overhead issue, followed by the cost
catalogue on second place. Even after a sensitivity analysis, the overall ranking does not change, which
indicates a very stable result. These two topics will be more detailed in the dissertation.
It will be interesting to see whether this result will be confirmed even after the survey in the various
purchasing departments. The tendency can be derived from this work. But a delimitation in the
application also must be considered. The implementation of the methods can be carried out for a new
project. In the case of series projects, where contracts have already been signed by the suppliers,
implementation and negotiation becomes much more difficult because the framework conditions for the
project have been defined. Subsequent implementation is much more difficult, so the timing for the
application of the methods is essential.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
214
As a perspective, it can be stated that the result of the dissertation should take the form of a contract,
that the premises and specifications are included in the inquiry documents sent to the suppliers, and that
the content must be confirmed by the supplier.
Furthermore, it can be investigated which content, for example from the overhead, is variable and which
is fixed. The fixed portion is independent of the volume of parts and should not be paid during the
change management, and the variable part of the costs should be paid. In this work there is no difference,
the whole overhead is fixed part.
Implementation in a change catalog is partly possible for a new component (innovation parts). There is
no technical basis or experience of possible changes in the future. Only costs regarding volume effects
can be determined.
References
[1] Christian GILLE. Gestaltung von Produktänderungen im Kontext hybrider Produkte. Springer
Gabler, 2013. ISBN 978-3-658-02693-6.
[2] Dolan JG, Isselhardt BJ, Cappuccio JD. The analytic hierarchy process in medical decision
making – a tutorial. Med Decis Mak. 1989;9(1):40–50.
[3] Günther SCHUH, Wolfgang STÖLZLE. Anlaufmanagement in der Automobilindustrie
erfolgreich umsetzen. Springer-Verlag, 2008. ISBN 978-3-540-78406-7.
[4] Martina Carolina WICKEL. Änderungen besser managen – Eine datenbasierte Methode zur
Analyse technischer Änderungen. Technical University Munich (TUM), Dissertation 2017.
[5] Peter MILLING, Jan JÜRGING. Der Serienanlauf in der Automobilindustrie: Technische
Änderungen als Ursache oder Symptom von Anlaufschwierigkeiten? Gabler, 2008. ISBN 978-3-
8350-5583-4.
[6] Ralf REICHWALD, Juan-Ignacio CONRAT. Vermeidung von Änderungskosten durch
Integrationsmaßnahmen im Entwicklungsbereich. Chair of general and Industrial Business
Administration at the technical University Munich (TUM), 1993.
[7] Saaty, T.L. How to Make a Decision: The Analytic Hierarchy Process. In: Interfaces,
[8] Vol. 24 (1994), No. 6, p. 19-43.
[9] Saaty, T.L. Fundamentals of Decision Making and Priority Theory with the Analytic
[10] Hierarchy Process. 2. Aufl., Pittsburgh 2000.
[11] Saaty, T.L.; Vargas, L.G. Decision Making in Economic, Political, Social and
[12] Technological Environments. The Analytic Hierarchy Process Series, Vol. 7. Pittsburgh
[13] 1994.
[14] Saaty TL. The analytic hierarchy process: planning, priority setting, resource allocation. 2.
Aufl. New York: McGraw-Hill; 1980. S. XIII, 287.
[15] Saaty TL. How to make a decision: the analytic hierarchy process. Eur J Oper Res. 1990;48
(1):9–26.
[16] Udo LINDEMANN. Das Änderungsmanagement Report 2015. Competence in Design and
Development, Working Paper Series. 2005. ISSN 1861-079X.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
215
BUSINESS STUDIES IN TIMES OF CHANGE (INDUSTRY 4.0)
Susann Wieczorek1
1Department of Business Sciences, Westsächsische Hochschule Zwickau,
Dr.- Friedrichs- Ring 2, Zwickau 08056, Germany
e-mail: [email protected]
Abstract
The fourth industrial revolution, or Industry 4.0 for short, brings changes for politics, society and
companies. At the same time, digitisation does not stop at the education sector, especially the
universities. For this reason, the business studies program is also undergoing a change that has never
been seen before. Two research phases (I. Document analysis and II. Expert interviews) made it possible
to record the first innovations for the business studies programme. With the help of two sequentially
conducted online surveys of companies and universities, these points from the first two research phases
could be substantiated. As a future-oriented requirement, digital competence was identified as a novelty,
alongside professional competence, methodological competence and personal/social competence.
Keywords
Tertiary education, higher education, further development of business studies, business administration,
teaching, science, digital literacy
JEL Classification
A23, M20, M21
Introduction
Changing market requirements, caused by rapidly growing challenges induced by technology and
automation, promote competition between companies. (Casper- Hehne/Reiffenrath, 2017) A key driver
for companies to be able to hold their own in the market is their innovative ability. This involves
production improvement, new products and networked systems to meet customer needs. Suitable
personnel are necessary for this innovative ability. This high level of market dynamics and flexibility
requires managers who recognize opportunities and use their powers of judgment to find solutions.
(Kirsch/Picot, 2013) Social and economic debates also show that with the increasing use of digital
technologies, the world of work is undergoing and will undergo change. (Hirsch-Kreinsen, 2017)
Education is thus becoming an even stronger decisive driver of innovation.
The inevitable consequence of Industry 4.0 is the elimination of traditional occupations, but also the
emergence of new activities. For this reason, we have to deal with these changes already today. When
teaching in tertiary education, it is of course important to know what skills and abilities are expected of
working people in order to be prepared for the future. A large proportion of companies believe that
universities do not sufficiently prepare students for working life. For this reason, the issue of
employability is constantly being viewed critically. (Haberfellner/Sturm, 2012) It is precisely for this
reason that it is imperative that universities undergo an adaptation induced by Industry 4.0.
The aim of this paper is to identify the competencies that will become increasingly important and
necessary for business studies in the future, especially for science, teaching and practice, and to derive
recommendations for action.
Methodology
Industry 4.0 brings ground-breaking changes. Using the example of business administration studies,
essential requirements with regard to competences were examined more closely. Based on this research
focus, the present paper will examine the research objective with regard to Industry 4.0: Requirements
for university graduates, especially for business administration, to derive a new type of study curriculum.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
216
With the data collected during the qualitative and quantitative research it was possible to use the mixed
method research approach. Due to the limited literature available, the exploratory research approach was
chosen. Document analysis (phase I) was launched (Mayring, 2016), which examined three studies on
the subject of expectations of graduates. The following three studies were used. The Association of
German Chambers of Industry and Commerce e. V. (DIHK) has already addressed the requirements of
Bachelor and Master graduates in 2014. Around 2,000 companies took part in the survey, including
companies from the manufacturing industry, other services and trade (Bauer, 2016). In autumn 2014,
Graz University of Technology launched a two-stage survey with the aim of making an initial
assessment of the current employment situation of university graduates. At this point it is assumed that
the requirements of Austrian and German companies and graduates do not differ (Bauer/Sadei, 2015).
In the study carried out by Lödermann/Scharrer at the University of Augsburg in summer 2009, a total
of 1,789 companies from the Augsburg/Swabia region were surveyed. The response rate was 14 %, i.e.
only 249 questionnaires could be evaluated. The survey focused on employability (Lödermann/Scharrer,
2010)
The findings were put into a category system. Based on these findings, semi-structured expert interviews
(phase II) were conducted with experts from companies and universities in Germany. With the help of
qualitative content analysis according to Mayring and Strauss' Grounded Theory (Mayring, 2016), these
interviews were evaluated. Subsequently, the findings from the document analysis and the expert
interviews contributed to the creation of two online questionnaires which were considered separately.
The two online questionnaires (phase III) were pre-tested after their preparation. The online
questionnaires were sent to German universities and companies. IBM SPSS Statistics Version 25 was
used to verify or falsify the hypotheses made.
Research phases I - III
After completion of the three research phases mentioned above, the following results could be stated.
Using the example of business administration studies, essential aspects of digitisation were examined
more closely. are effective: stronger interdisciplinary cooperation across degree programme borders and
Introduction of digitalization modules, taking into account the level of competence and knowledge
transfer.
Research phases I
Within the framework of the document analysis, the three studies presented in Section 2 (DIHK Business
Survey, Graz University of Technology and Augsburg Study), which form the basis for the category
system, are to be used. In this phase it should be possible to create an overview of the three studies.
Research phases II
The findings of the first research phase are the starting point for the semi-structured expert interviews.
The individual competencies of the graduates were examined more closely and transferred to the
business studies program. The results are to be confirmed or rejected with the help of ten interviews
each with experts from companies and universities. The interview guidelines were divided into five
topics:
1. personal details: position, work experience, number of employees, industry
2. industry 4.0: Terms, future strategies, qualification measures, digitalisation specialists,
previous changes in the company
3rd business graduate: relevance of competences, interest in digital focal points,
recommendations for action, theses
4 CDO and Disruption: Relevance and influence on business studies
5. participation in the working group
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
217
Research phases III
The findings of research phases I and II were to be transferred into a sequential online survey for
companies and universities, so that a fully structured written survey could be conducted. The
questionnaire was to contain points:
1. introduction: topic incl. contact possibilities to the creator, target group
2. sociodemographic data: Activity, professional experience, sectors, degree of digitisation as
well as university name, type of university, name of degree programme incl. degrees, etc.
3. thematic catalogue of questions: Industry 4.0, skills of the business studies graduate, potential
for improvement for business studies in the digital age, etc.
4th degree: Next steps, farewell
The questionnaire was created via the online platform www.umfrageonline.com and distributed to the
two groups of participants, companies and universities. The questionnaire mainly asked closed
questions. The online survey should be answerable within ten minutes in order to keep the dropout rate
low. In the course of a follow-up action, if necessary, the answering of the questionnaire should be
reminded. The lynchpin of this scientific work was to find out what the requirements for business studies
are and what measures can be derived from them. In order to test the functionality of the online survey
for the companies, it was made available to a test person in advance (pretest). As the survey has changed
only slightly for the universities, except for socio-demographic information such as course of study and
position in the university, no further pretest was conducted.
Results of research phases
On the basis of the expert interviews conducted in research phase II, the following hypotheses for the
online survey of enterprises could be established.
No. Hypotheses Statistical methods
H1 The larger the company, the more important digital literacy is. T-Test and Spearman rank
correlation analysis
H2 The degree of digitalisation varies from company size to
company size.
Single factorial analysis of
variance or simple. ANOVA
H3 The more personnel are involved in digitisation, the more
important digitisation is for a company.
Spearman rank correlation
analysis
H4 The basic knowledge of a business master's degree is more
important for professional life than technical specialisation.
Spearman rank correlation
analysis
H5 Comparison of the importance of methodological, technical,
social and digital competence
H5a The methodological competence of business studies graduates is
more important to employers than professional competence.
Spearman rank correlation
analysis
H5b The methodological competence of business studies graduates is
more important to employers than social competence.
Spearman rank correlation
analysis
H5c The methodological competence of business administration
graduates is more important to employers than digital
competence.
Spearman rank correlation
analysis
H6 The degree of digitalization of a company has an impact on the
business requirements profile.
Spearman rank correlation
analysis
H6a There is a correlation between the degree of digitization and the
requirement profile of bachelor's degree graduates.
Spearman rank correlation
analysis
H6b There is a correlation between the degree of digitization and the
requirements profile of master's degree graduates.
Spearman rank correlation
analysis
H6c There is a correlation between the degree of digitisation and the
professional competence.
Spearman rank correlation
analysis
H6d There is a correlation between the degree of digitisation and
methodological competence.
Spearman rank correlation
analysis
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
218
H6e There is a correlation between the degree of digitisation and
social competence.
Spearman rank correlation
analysis
H6f There is a correlation between the level of digitisation and digital
literacy.
Spearman rank correlation
analysis
H7 The higher the level of digitisation in a company, the greater the
interest in exchanging information on digitisation with
representatives of educational institutions.
Spearman rank correlation
analysis
H8 For companies, the modular further training (modular studies,
certificates, part-time studies) of employees is more important
than attendance at a university.
Frequencies
Every single hypothesis was checked and either verified or falsified. With the respective results the
following statements could be derived for this scientific elaboration.
Science
The author refers to a triangular relationship in science, which is characterized by key figures,
methodology and basic IT understanding.
Key figures play an important role in the daily work of the business administration graduate. To
be able to explain the key figures correctly, the necessary expertise must be available.
In addition, the business administration graduate must have a basic understanding of IT. A sound
knowledge of the technologies used in the company is essential. Questions that should be
answered by business administration graduates are, for example: what a technology offers and
what are the chances and risks, etc. In this way, it can be rationally decided in the course of
negotiations whether or not the technology presented will bring added value to the company.
In order to be able to handle the mentioned key figures and the basic understanding of IT
correctly, methodological competence is required. In the context of this work, methodological
competence is understood as the use of cause-effect chains, as well as the ability to solve
problems and to think economically.
The research approaches and developments of Industry 4.0 and digitisation mentioned in this paper are
difficult to foresee from today's perspective. The reasons for this lie in the fast moving and technological
change. Nevertheless, it remains to be noted that the trend is increasingly towards harmonising the
corporate business areas. In other words, all corporate divisions are now subject to digitisation. This is
supported by the use of technologies to make detailed statements for the present and future.
Business administration covers all the original tasks of a company, so that a wide range of research areas
are created. For this reason, it is important to research creative ideas and thus always be one step ahead.
At this point, suggestions for further research studies are e. g. mentioned:
Does science, in terms of research, unite with business?
How will university professors do their research in 2030?
Can the Internet of Services (Facebook, Instagram, Twitter, YouTube) replace university
professors in the future?
Teaching
In this section, the contents to be taught in the business studies programme, learning formats and
methods, etc. are discussed. Following on from this, model curricula for business studies, for the degrees
'Bachelor of Science' and 'Master of Science', are used to provide a forward-looking orientation.
When designing the content of the course of study, great importance was attached to supplementing the
classic business administration modules with digitisation modules. The unchanged business
administration modules include: 'Controlling', 'Financing', 'Cost and Performance Accounting', 'Micro
and Macro Economics', 'Personnel', 'Organization' and 'Taxes'. The surveys helped to identify issues that
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
219
were considered particularly important. Thus, relevant digitisation modules or focal points for business
studies could be worked out: business analytics, AI, data mining, IT law & IT security, IT data protection
& ethics, AR & VR, agile projects, blockchain.
It would be conceivable to successively restructure an existing business studies course, i.e. to include
elective modules in the first step. Subsequently, these modules would be combined to form digitisation
priorities, and, in the final instance, a new course of study adapted to the above modules could be
designed and introduced. If business studies do not yet exist at a university, such a course could be
introduced in such a form.
The desire for greater interdisciplinarity across the boundaries of study programmes was expressed by
all experts in universities and companies. The challenges posed by Industry 4.0 are so diverse and
complex that they can usually only be solved in an interdisciplinary manner and not just within a specific
discipline. This also broadens the knowledge horizon of all those involved.
In order to incorporate the digitisation modules into teaching, it is essential to break up the prevailing
column thinking created by departments or faculties. This means that interdisciplinary complexes of
topics or questions should be posed in an interdisciplinary way and thus lead to success. In this way,
many different perspectives can be incorporated. The students should be given the freedom to find their
own solutions. At the same time, the ability to communicate and work in a team is also strengthened in
this way. Knowledge labs offer a platform for new and agile topics. In order to strengthen or intensify
the handling of new technological developments, such teaching factories, in which several universities
join forces, can be a possible instrument. The testing of new technologies is the core objective in order
to implement a holistic way of thinking. Such cross-cutting modules allow students from other faculties
to work together on entrepreneurial and technological solutions. In this way, a close bond between
universities and companies is created and consolidated. The author sees the introduction of knowledge
labs, especially at universities of applied sciences, as the focus is more on practical application in
companies.
With a revision of the previously known learning formats, such as classroom or frontal teaching, the
business studies programme can be made more sustainable. This is relevant and important both for the
continuing education sector from a business perspective and for the up- and-coming generation, the
digital natives. Hybrid learning formats for teachers and learners are conceivable, regardless of the time
and place component, to enable more flexible learning. Learning nuggets could be a good possibility in
the organisation of the course. These are small, concise learning units, which are generally intended to
impart knowledge for no longer than five minutes. With Learning nuggets is a great learning success
connected. The interviewees see great opportunities for universities to position themselves better,
especially in the modular further education possibilities, such as modular studies, part-time studies,
certificate courses. This continuing education programme follows the mindset of precise knowledge
transfer in small learning units, so that continuing education in enterprises can be further expanded and
thus the attractiveness of universities increased. These findings are accompanied by the answer to
another research question.
With a new curriculum 'BWL - digital management (B. Sc.)', a modern business administration course
of studies is to be offered. The aim of the course is to impart digital skills in addition to business
management skills. The digital influence mentioned at the beginning is new to this course of study. For
this reason, topics such as Business Intelligence, Data Management as well as Artificial Intelligence etc.
should not be missing. These modules are supported by safety-relevant and legal aspects that a business
economist must be familiar with. Using scientific and modern methods, it is possible to record and
evaluate the data correctly and to derive the results logically. The vertical introduction of these digital
modules makes it possible to have a solid foundation of new technologies.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
220
Table 1. Business studies - digital management (B. Sc.)
Business modules Digitization modules Bachelor thesis
Basics of business administration,
Mathematics I, Personnel
management, External accounting
Basics of digitization,
IT management and Digital business
models
Taxation, Mathematics II,
Macroeconomics, Internal
accounting
IT security, Digital marketing and
social media
Controlling, Statistics I,
Microeconomics, Cost and
performance accounting
IT law and data protection,
Enterprise resource planning,
Business intelligence
Financing, Statistics II, Economic
policy, Corporate governance
Data management, Artificial
intelligence, Modern methods
Source: own Source
In addition, the business studies programme covers all important business management modules,
including controlling, financing, human resources, taxation, etc. In addition, the university education
will have a higher level of detail on the modules: mathematics, micro- and macroeconomics, internal
and external accounting and statistics, as shown in Table 1. The acquisition of additional qualifications,
e.g. MS Office and soft skills, should be given at any time during the course of the study. A compulsory
semester abroad could round off the course of study.
The Master's degree programme 'Digital Management - Accounting, Controlling, Finance (M. Sc.)'
provides in-depth knowledge in the areas of 'Accounting', 'Controlling', 'Finance'. A major role is given
to statistical methods, as these will become increasingly important in the future. The standard repertoire
also includes 'Controlling', 'Management', 'Accounting', 'Corporate Governance' and 'Business Law' etc.
In addition, the selection consists of a pool of focus modules to choose from. Possible focus modules
could be, for example: 'international accounting standards', 'financial controlling', 'risk and compliance
management' and 'auditing'. During the course of studies, the focus is on methodological competence,
so that the subject areas of qualitative and quantitative research, such as 'Multivariate Statistics' and
'Predictive Analysis', can also be taught.
Table 2. Digital management – Accounting, Controlling, Finance (M. Sc.)
Business modules Digitization modules Master thesis
Multivariate statistics, Controlling
systems, Strategic management,
Business law
Business intelligence,
Enterprise resource planning
Accounting, Project controlling,
Corporate management and
corporate governance, Technology
and innovation management
Business analytics,
Data mining
Scientific work Multivariate statistics,
predictive modeling
Source: own Source
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
221
The two study profiles presented here have been expanded to include the field of activity of the Chief
Digital Officer (CDO), which is understood to mean the planning and control of digital processes and
their transformation. The author succeeded in fusing business management know-how and IT
understanding. With the presented sample study plans further answers to the research questions could
be found.
Practice
This paper has shown that Industry 4.0 is changing all our lives and has a significant impact on an
economy; from government and business to science and education. One aim of this work was to increase
employability. This means enabling graduates to carry out the required activities in their professional
life more quickly. Conversely, the newly required course contents and the newly created sample
curricula based on new trends and current topics are intended to increase the graduates' know-how for
entrepreneurial use in practice and thus reduce the training period of the newly graduated.
The digital modules support to get a better understanding of Industry 4.0 and to solve modern problems.
Business intelligence and artificial intelligence are particularly important for German and Slovakian
SMEs, as they provide considerable added value over competitors. The reason is that companies can
back up more and more data and want to understand and interpret it. The view is directed towards the
future. On the part of the universities, knowledge labs have been integrated into the model curricula,
which lead to better cooperation between practice and universities. Entrepreneurial topics can thus be
identified more quickly and taken into account in teaching. As a result, universities are becoming more
agile, which was a point of criticism from the experts. In addition, the study profile of business
administration studies was extended to the field of activity of CDOs.
With more flexible and modular further training options, it is possible to increase the range of further
training on offer, from which companies also benefit. Because companies are constantly trying to find
new and interesting further training offers for their staff in order to have a knowledge advantage over
their competitors. It is important that the employees perform their work despite further training. For this
reason, part-time, distance learning, modular studies or certificates are suitable for the workforce.
External lecturers can promote cooperation between practice and universities and provide new impulses.
Conclusion
The elaboration has shown that other authors have already devoted themselves to this topic. Universities
are faced with the decision what is the right way to study business administration. This paper aims to
provide guidelines and contribute to the discussion. However, the author's clear message is that change
is necessary and important.
With the research phases I - III it could be stated that the four competences technical, methodological,
social and digital competence are of central importance, they serve as a starting point for the results. In
the age of digitalisation, politics is increasingly challenged. Education is the pioneer of our future.
There is ample evidence that it is not yet possible to foresee what final changes Industry 4.0 will bring.
But one thing can be said for sure: today and tomorrow it is important to develop an understanding in
order to solve problems, no matter how good the expertise is. In this way, the required problem and
solution orientation, or rather the cause-and-effect principle, is achieved. Knowledge labs are suitable
as a possible tool or as a platform to try out unknown things. These are small laboratories in which new
ideas or solutions are developed with the help of IT; a kind of creative workshop.
Creativity opens up new avenues, so such a skill is essential in the digital age. Personal and social skills,
which include intuition, help in the decision-making process. The results also flow into the design of the
study programme in order to prepare students as well as possible for the labour market.
A central task is to ensure that universities are sensitised to bring about change as quickly as possible.
But there must also be a rethink in politics, so that new positions for professors with a changed profile
can be advertised.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
222
References
[1] BAUER, U. 2016. Employability: Welche Kompetenzen fordern Unternehmen von TU-
Absolventen? 2016. DOI: 10.1007/978-3-658-12097-9_1.
[2] BAUER, U. - SADEI, C. 2015. Studie zur Employability der TU Graz Absolvent/inn/ en - 2015:
Ergebnisse einer Primärerhebung unter Arbeitgeber/inne/n und TU Graz Absolvent/inn/en.
Research Report. Technische Universität Graz. Graz 2015. DOI: 10.3217/978-3-85125-386-3.
[3] Casper-Hehne, H. - Reiffenrath, T. (2017): Higher education in global and local contexts: The
Göttingen Model of Curricular Internationalisation. pp. 98.
www.ssoar.info/ssoar/bitstream/handle/document/55106/ssoar-interculturej-2017-27/28- casper-
hehne_et_al-Hochschulbildung_in_globalen_und_lokalen.pdf?sequence=1, [accessed
15.05.2019].
[4] Haberfellner, R. - Storm, R. (2012) "Job opportunities in higher education": Longer-term
employment trends of university graduates.
www.econstor.eu/bitstream/10419/97929/1/787185000.pdf, [accessed 30.07.2019].
[5] Hirsch-Kreinsen, H. (2017). Arbeiten 4.0 – Qualifikationsentwicklung und Gestaltungsoptionen.
pp. 473. www.wiwi.tu-dortmund.de/wiwi/ts/de/forschung/veroeff/soz_arbeitspapiere/AP-SOZ-
38.pdf, [accessed 13.05.2019].
[6] Kirsch, W. - Picot, A. (2013). Business administration in the field of tension between
generalisation and specialisation: Wiesbaden: Gabler Publishing 2013. ISBN 3-663-0690-95.
[7] LÖDERMANN, A.-M. - SCHARRER, K. 2010. Beschäftigungsfähigkeit von
Universitätsabsolventen – Anforderungen und Kompetenzen aus Unternehmenssicht. Universität
Augsburg. Augsburg 2010. ISSN: 0171-645X.
[8] Mayring, P. (2016). Einführung in die qualitative Sozialforschung Eine Anleitung zu qualitativem
Denken. ISBN 978-3-407-25734-5.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
223
RESEARCH ON THE IMPACT OF CHARACTERISTICS OF THE BOARD OF
DIRECTORS OF CHINESE APPLIANCE LISTED COMPANIES ON CORPORATE SOCIAL
RESPONSIBILITY
XIAOJUAN WU1
1Department of Economics, VŠB – Technical University of Ostrava,
Sokolská tř. 33, Ostrava 70200, Czech Republic
e-mail:[email protected]
Abstract
With the increasingly serious environmental and social problems, more and more Chinese companies
are voluntary or required to balance social and environmental interests while pursuing economic
benefits. Sustainable development has been reflected in all aspects of China's economic activity as the
current goal of the Chinese economy. Corporate social responsibility (CSR) is a key indicator for
measuring sustainable development and has become one of the most important references when
investing in companies. Previous empirical research based on data collected from global companies
found that the composition of the board of directors has a significant impact on sustainable development,
but few articles have studied the situation of Chinese companies. This article takes the Chinese home
appliance listed companies as a sample to study the impact of board composition-independence,
diversity and CEO duality on CSR. The linear regression results show that only the diversity of board
members has a statistically significant and negative impact on CSR, and there is no significant
relationship between board independence and CEO duality and CSR.
Keywords
Corporate social responsibility, Corporate governance, Board of directors, Sustainable development
goals.
JEL Classification
M210 Business Economics.
Introduction
With the growing environmental problems, the Chinese government proposed an economic goal for
scientific development in 2003. In 2006, the Shenzhen Stock Exchange issued the “Guidelines for Social
Responsibility of Listed Companies”, encouraging listed companies to establish corresponding social
responsibility institutions by the guidelines, and advocating that listed companies also disclose their self-
assessed social responsibility reports and financial statements. In 2008, the Shanghai Stock Exchange
released the "Guidelines for Environmental Information Disclosure of Listed Companies", which
stipulates that listed companies should promptly disclose incidents related to environmental protection
that may affect stock prices. After more than ten years of development, the overall implementation of
Chinese corporate social responsibility (CSR) has shifted from the initial bystander stage to the starter
stage (Chinese Academy of Social Sciences, 2018). But the general situation is still not ideal. Exploring
which factors may affect the implementation of CSR has become one of the focuses of scholars. Recent
research has found that the characteristics of the board of directors (BOD) have a significant impact on
CSR (Chams and García-Blandon 2019; Naciti 2019). China's home appliance industry is one of the few
internationally competitive industries in China. What is the current status of CSR of Chinese home
appliance companies, and what factors can affect it are issues worth studying? To the best of the author’s
knowledge, few articles examining the relationship between the characteristics of the board of directors
of Chinese household electrical appliances companies and CSR. Therefore, this study attempts to
provide evidence supporting the nexus between BOD characteristics and implementation of CSR of
Chinese home appliance companies. With the help of regression techniques, the research question arises:
do the characteristics of BOD have any effect on CSR of Chinese home appliance companies?
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
224
The objective of the article is to analyze the impact of specific characteristics of board composition
(board independence, board diversity and CEO duality) on CSR of the Chinese home appliance
companies. Our research provides a descriptive knowledge that links the implementation of CSR with
corporate governance.
The remainder of the study is presented as follows. The next section reviews the existing literature about
CSR (or sustainable development) and characteristics of BOD and develops the hypotheses. In the third
section, we present the sample, the data, and the applied methodology. The fourth section lays out the
empirical results. In the last section, we present our conclusion.
Literature Review
In 2014, the UN Member States followed a decision made at the Rio+20 Conference to launch a process
to develop a set of Sustainable Development Goals (SDGs) that will build upon the Millennium
Development Goals (MDGs). The SDGs recognize that companies play a crucial and decisive role as
they are the chief motivators of sustainable development. When companies face such new development
goals, they should satisfy the interests of stakeholders by maintaining balanced economic, social and
environmental development rather than just pursuing economic interests to satisfy the interests of
shareholders. The board of directors, as the highest decision-maker, plays a decisive role in setting and
implementing goals. For this reason, it is important to reconfigure the governance system that has been
tasked with defining and implementing Corporate Sustainability (CS) policies and strategies (Van
Marrewijk, 2003). The following section presents the recent literature about the relationship between
CSR (or sustainability) and three characteristics of the board -- board independence, board diversity and
CEO duality, which are our main research objects.
Board Independence
Based on the Stakeholders Theory, the independence of the BOD is expected to be positively connected
with CSR, since these are less subject to shareholder pressures. A board with a large portion of
independent directors can provide supervision for management and protect all the stakeholders’
interests. Hussain et al. (2018) found that “board with a higher proportion of independent directors
positively impacts environmental and social performance” by researching the US-based companies. As
Nicola Cucari et al. (2018) stated: “that firms with more independent directors lead to an increase in the
information related to ESG disclosure”. Pucheta‐Martínez et al. (2018) used a sample of financial
companies listed in Spain between 2004 and 2015 and found evidence that the presence of independent
directors encouraged financial entities to report CSR matters, which demonstrated the effectiveness of
the corporate governance mechanism. Hence, it is expected that BODs with more independent directors
exhibit then better CSR.
Hypothesis 1. A high number of independent directors on a BOD is positively and significantly
associated with CSR.
Board Diversity
The diversity of BOD has been explained in several ways. Gender and nationality are the main
characteristics of a diverse board. In the Stakeholders Theory framework, the presence of women on a
BOD is expected to be positively connected with CSR. Female directors have certain personality traits,
such as low-risk aversion, transparency, responsiveness, and recognition of social and environmental
issues, which improves sustainable performance (Boulouta, 2013). Hussain et al. (2018) find “that board
diversity enhances the social dimension of sustainability”. Naciti (2019) stated that firms with more
diversity on the board showed higher sustainability performance based on data from 362 large industrial
firms on the 2013—2016 Fortune Global 500 list. Accordingly, the hypothesis is formulated to test the
link between BOD diversity and CSR:
Hypothesis 2. A high proportion of board diversity, in terms of gender, is positively and significantly
associated with CSR.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
225
CEO Duality
CEO duality occurs when the CEO is the company's president and chairman of the board. According to
agency theory, separation of board chairman and CEO roles increases board oversight of management.
Therefore, if the CEO also serves as chairman, it can have a significant impact on board decision-making
processes and company operations. The previous empirical literature has not provided consensus results
on the relationship between CEO duality and CSR. Naciti (2019) found firms with a separation between
chair and CEO roles exhibit higher sustainability performance. Pucheta‐Martínez (2018) found that
CEO duality is negatively correlated with CSR. Chams and García-Blandon (2019) did not find a
significant relationship between CEO duality and sustainability. Thus, this study predicts that the dual
nature of the CEO will have a negative impact on the implementation of CSR.
Hypothesis 3. CEO duality is negatively and significantly associated with CSR.
Methodology and Data
Sample and Data
This study aims to explore the impact of characteristics of BOD of Chinese appliance firms on the level
of CSR. Thus, the study analyzed data from Chinese home appliance listed companies in 2018. The
sample list is from the wencai database and includes 26 companies listed on the mainboard, 25
companies listed on the Small and Medium Enterprise board, and 8 companies listed on the Growth
Enterprise Market (GEM). Chinese authority agency -Rankins CSR Ratings (RKS) will release the CSR
level of listed companies according to the CSR report. However, not every listed company is willing to
disclose its CSR status, so there are only a few listed companies included in its report. This paper adopted
the concept of CSR advocated by Elkington13 – the so-called triple bottom line principle, which assumes
that if an enterprise forms an economic and social system, then its development objectives should
constitute a triple beam, which relates to the profit, the people associated with the company, and care
for the planet. Therefore, according to the triple bottom line principle, the author calculates the CSR
score of appliance listed companies from three aspects: economics (shareholder, consumer, employee,
supply chain and government), environment and society, as the measurement of the dependent variable
of CSR. As for what criteria are included and how to assign the specific weights for each criterion can
be seen in the author’s other paper (Wu, 2019). In order to get the newest result, this paper used the data
of 2018, the related criteria and corresponding weights are the same as the author’s other paper (Wu,
2019), and just the data is updated to 2018. All sample data are collected from wencai database, financial
statements and CSR reports (if any).
Methodology and Applicability Analysis of OLS
When data only has a cross-sectional dimension, a multiple linear regression model is proposed to test
the hypotheses formulated in the former section. This method is considered an appropriate analytical
tool when the outcome variables are measurable and continuous. Thus, the empirical model is provided
as
𝐶𝑆𝑅 = 𝛽0 + 𝛽1𝐵𝐼𝑁𝐷𝐸𝑃 + 𝛽2𝐵𝐷𝐼𝑉𝐸𝑅 + 𝛽3𝐶𝐸𝑂𝐷𝑈𝐴𝐿 + 𝛽4𝑅𝑂𝑆 + 𝛽5𝐹𝑅𝐼𝑆𝐾
+ 𝛽6𝑆𝐼𝑍𝐸 + 휀 (1)
where 𝛽𝑖 represents the partial regression coefficient, all the variables included in the study is described
in detail in Table 1.
13 J. Elkington: Cannibals with Forks: The Triple Bottom Line of 21st Century Business. Capstone Publishing Limited, Oxford 1997.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
226
Table 1. Summary of Variables and Their Measurements in the Study
Name of variables Abbreviation Measurement
CSR CSR CSR score calculated by AHP method
Independent Variables
Board Independence BINDEP Proportion of independent directors on the
board
Board Diversity BDIVER Proportion of female directors on the board
CEO Duality CEODUAL Value of 1 if CEO is both president and
chairman of BOD; 0 otherwise.
Controlling Variables
Return on Sales ROS Operating income to total revenue
Financial Risk FRISK Liabilities divided by total assets.
Company Size SIZE The logarithm of the number of assets
Note: Table 1 presents the description and abbreviation of the variables included in this study.
Table 2 shows the descriptive statistics of input data in columns 4 and 5, while minimum and maximum
scores are shown in columns 2 and 3.
Table 2. Descriptive Statistics of Given Variables
Minimum Maximum Mean Std. Deviation
CSR 0.0585 0.5913 0.349622 0.1229467
BOD Independence 0.3000 0.6000 0.377711 0.0593242
BOD Diversity 0.0000 0.5556 0.165026 0.1476641
CEO Duality 0.0000 1.0000 0.322034 0.4712667
ROS -74.8300 40.0600 2.178475 20.9121352
Financial Risk 12.2800 195.1700 48.699831 27.0375150
Size 8.5617 11.4211 9.682264 0.6092080
Source: SPSS
Testing the data with SPSS software, we found that the data used meets the assumptions of the ordinary
least squares (OLS) method, so this paper uses OLS to estimate model parameters. The test results are
shown in Figure 1 to 4 and Table 3 to 5.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
227
Figure 1.The Linear Relationship between Independent Variables of BOD Independence and
BOD Diversity and Dependent Variable of CSR
Source: SPSS
Figure 1 shows the linear relationship between independent variables of BOD dependence and BOD
diversity and the dependent variable of CSR. Because the CEO duality is a categorical variable, it is not
necessary to examine its linear relationship with the dependent variable of CSR.
Figure 2. Relationship between Independent Variables of BOD Independence and BOD
Diversity and Unstandardized Residuals
Source: SPSS
We can observe the relationship between independent variables of BOD independence and BOD
diversity and unstandardized residuals through Figure 2 and find independent variables are not
correlated with the residuals. Because the CEO duality is a categorical variable, it is not necessary to
examine its relationship with the residuals.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
228
Figure 3. Histogram and P-P plot of Regression Standardized Residual
Source: SPSS
Figure 4. Scatter Plot of Regression Standardized Predicted Value and Regression Standardized
Residual
Source: SPSS
We can get the conclusion that the residuals are normally distributed, and their mean value is zero from
Figure 3. From Figure 4, we can see the scattered point fluctuation range of the standardized residual is
basically stable and does not change with the change of the standardized prediction value. It can be
considered that the homogeneity of the variance is basically satisfied.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
229
Table 3. Model Summary
Model R R Square
Adjusted R
Square
Std. Error of
the Estimate
Durbin-
Watson
1 0.595a 0.354 0.279 0.1044000 2.098
Source: SPSS
From Table 3, we know the value of Durbin-Watson is 2.098, which is very close to 2. Thus, we can
infer that there is no obvious correlation between the residuals; that is, the residuals are independent.
Table 4. Correlations
CSR BOD
independence
BOD
diversit
y
CEO
dualit
y
ROS
Financia
l
risk
Size
Pearson
Correlatio
n
CSR 1
BOD
independence -
0.132 1
BOD diversity -
0.284 -0.049 1
CEO duality 0.012 0.288 0.122 1
ROS 0.146 0.109 0.05 0.08 1
Financial risk 0.158 -0.049 0.024 -0.111 -
0.323 1
Size 0.459 0.021 -0.017 0.193 0.023 0.226 1
Sig.(1-
tailed)
CSR
BOD
independence 0.16
BOD diversity 0.015 0.356
CEO duality 0.465 0.014 0.179
ROS 0.135 0.205 0.355 0.274
Financial risk 0.117 0.355 0.428 0.201 0.006
Size 0 0.437 0.45 0.072 0.431 0.042
Source: SPSS
Table 4 displays the Pearson correlation coefficient between any two variables of all variables and their
corresponding p-values. The results show that all the correlation coefficients between the continuous
independent variables are less than 0.5, and the p-values are larger than 0.05, indicating that the
correlation between the independent variables is weak, and it can be considered that there is no
collinearity. Because the independent variable CEO duality is a categorical variable, it is not appropriate
to use the Pearson correlation coefficient for testing. And we can get the same conclusion from the value
of collinearity statistics—Tolerance and VIF from Table 5. For each variable, the value of Tolerance is
larger than 0.2, and the VIF is less than 10, indicating that there is no collinearity between all variables.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
230
Table 5. Coefficients Estimated by OLS
Model
Unstandardized
Coefficients
Standardize
d
Coefficients t Sig.
Collinearity
Statistics
B Std.
Error Beta
Toleranc
e VIF
1
(Constant) -0.324 0.246 -1.315 0.194
BOD
independence -0.367 0.244 -0.177 -1.505 0.138 0.9 1.111
BOD diversity -0.251 0.094 -0.302 -2.663 0.010 0.968 1.033
CEO duality 0.004 0.032 0.016 0.132 0.895 0.841 1.188
ROS 0.001 0.001 0.212 1.775 0.082 0.873 1.145
Financial risk 0.001 0.001 0.131 1.065 0.292 0.819 1.222
Size 0.085 0.024 0.42 3.546 0.001 0.887 1.127
Source: SPSS
Empirical Results and Discussion
Empirical Results
As previously mentioned, the sample comprises 59 sample companies in the Chinese appliance industry.
Table 4 shows the Pearson correlation results. We find that CSR has a negative and significant
correlation with BOD diversity. The correlation coefficient between CSR and BDIVER is -0.284 at 5%
significance level. We also discover that at a significance level of 5%, CSR was not associated with
BINDP and CEODUAL. In Table 4, we notice that company size is positively correlated with CSR at
1% significance level.
From Table 3, we find the Coefficient of Determination R2 is 0.354, which means 35.4 per cent of the
total variation in CSR can be explained by the multiple regression model provided in this study. Thus,
the goodness of fit of this multiple regression model is acceptable.
In Table 5, we report the results after testing the hypotheses with OLS; that is, the estimated results of
the determinants of CSR. Results show that the particular regression coefficient of BOD independence
is -0.367, and the p-value is 0.138, which is slightly greater than 10% significant level. It represents a
relatively weak linear relationship between CSR and board independence, which is inconsistent with
hypothesis 1. BOD diversity is significantly but negatively associated with CSR at 5% level.
Specifically, an increase in board diversity leads to a decrease in CSR by 0.302 points, which is contrary
to our expectation. The p-value of CEO duality is 0.895, which is much greater than the 5% significant
level. It demonstrates that CEO duality is not associated with CSR at 5% significant level, which does
not support hypothesis 3. In addition, ROS is positively correlated to CSR at 10% significant level, and
company size is positively associated with CSR at 1% significant level.
Results Analysis
Possible explanations to our result about the negative relationship between board diversity and CSR can
be found in Cucari et al. (2018), where it was shown that “being a woman director does not necessarily
mean having a different outlook. Therefore, it is not the gender that determines a positive level of ESG
disclosure”. According to Jain and Jamali (2016) and Rodríguez-Ariza et al. (2017), the relationship
between directors’ gender and CSR behaviour is very complex, because a female director has many
characteristics, except for her genders, such as a woman with specific expertise or experience as an
executive or non-executive director.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
231
Another empirical result on the relationship between board independence and CSR is the same as that
found by Lau et al. (2016), which explained that a small number of outside directors was not sufficient
to make a change in the board. The connection between CSR and CEO duality is the same as the finding
of Chams and García-Blandón (2019). The reason may be the most sample companies are private
companies (45 in 59), where the actual controller is also the chairman and CEO in most cases, so the
agency theory is not appropriate. Thus, there is no relationship between CSR and CEO duality.
Conclusion
This paper examines the impact of the characteristics of the board of directors on CSR. We measure
CSR as the weighted score of seven aspects using the method of AHP to get their corresponding weights.
Based on a sample of 59 Chinese appliance listed companies and applying empirical tests, we find that
only the relationship between board diversity and CSR is significant but negative, and there are no
statistical links of board independence and CEO duality with CSR.
References
[1] Boulouta, I. (2013). Hidden Connections: The Link between Board Gender Diversity and
Corporate Social Performance. Journal of Business Ethics, 113(2), pp. 185–97.
[2] Chinese Academy of Social Sciences. (2018). China Corporate Social Responsibility Research
Report. Social Science Literature Press.
[3] Chams, N. and J. García-Blandón. (2019). Sustainable or Not Sustainable? The Role of the Board
of Directors. Journal of Cleaner Production, 226, pp.1067–1081.
[4] Cucari, N., S. E. De Falco. and B Orlando. (2018). Diversity of Board of Directors and
Environmental Social Governance: Evidence from Italian Listed Companies. Corporate Social
Responsibility and Environmental Management, 25(3), pp. 250–266.
[5] Hussain, N., U. Rigoni. and R. P. Orij. (2018). Corporate Governance and Sustainability
Performance: Analysis of Triple Bottom Line Performance. Journal of Business Ethics, 149(2),
pp. 411–432.
[6] Jain, T. and D. Jamali. (2016). Looking Inside the Black Box: The Effect of Corporate Governance
on Corporate Social Responsibility. Corporate Governance: An International Review, 24(3),
pp.253–273.
[7] Lau, C. M., Y. Lu. and Q. Liang. (2016). Corporate Social Responsibility in China: A Corporate
Governance Approach. Journal of Business Ethics, 136(1), pp. 73–87.
[8] Naciti, V. (2019). Corporate Governance and Board of Directors: The Effect of a Board
Composition on Firm Sustainability Performance. Journal of Cleaner Production, 237: 117727.
https://doi.org/10.1016/j.jclepro.2019.117727.
[9] Pucheta-Martínez, M. C., I. Bel-Oms. and M. Nekhili. (2019). The Contribution of Financial
Entities to the Sustainable Development through the Reporting of Corporate Social Responsibility
Information. Sustainable Development, 27(3), pp. 388–400.
[10] Rodríguez-Ariza, L., B. Cuadrado-Ballesteros., J. Martínez-Ferrero. and I. M. García-Sánchez.
(2017). The Role of Female Directors in Promoting CSR Practices: An International Comparison
between Family and Non-Family Businesses. Business Ethics, 26(2), pp. 162–174.
[11] Van Marrewijk, M. (2003). Concepts and Definitions of CSR and Corporate Sustainability:
Between Agency and Communion. Journal of Business Ethics, 44, pp. 95–105.
[12] Wu, Xiaojuan. (2019). Research on the implementation of CSR of listed companies in the Chinese
household appliance industry. 21st international conference Mekon 2019. pp.255-262
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
232
MULTI-CRITERIA DECISION MAKING USING THE ENTROPHY METHOD APPLIED
ON SELECTED VARIABLES FROM THE AREA OF DIGITALIZATION AND
DEVELOPMENT IN THE CENTRAL EUROPE TERRITORY
Martina Žwaková1
1Department of Economics, VŠB – Technical University of Ostrava,
Sokolská tř. 33, Ostrava 70200, Czech Republic
e-mail: [email protected]
Abstract
A comparison of the regions can be considered as a complex procedure, where multiple criteria are
requested to be evaluated. The aim of this paper is to rank the selected regions based on the variables
from the field of technology, its use and development, used for the comparison and assess the selected
regions from the area of Central Europe at the NUTS2 level. There are 30 NUTS2 regions assessed. The
variables related to human resources, use of the digital technologies, information and communication
technology and science are incorporated in the evaluation. This approach is used to obtain results which
are more complex. The variables were chosen also with regards to an availability of statistical data. The
Entropy method is used in the paper to calculate weights of the variables identifying an importance of
each variable for overall ranking for the selected group of the regions. Results from the analysis are
presented comparing the ranking of the regions and describing main characteristics, potential strengths
and weaknesses in terms of digitalization and development area defined by the selected variables.
Keywords
Entropy method, NUTS 2, multi-criteria decision making, digitalization, Central Europe
JEL Classification
D80, O30, O33, R11
1 Introduction
This paper is focused on application of entropy method used for purposes of assessment NUTS 2 regions
from the area of Central Europe from a perspective of variables related to areas with possible influence
on digitalization in the regions. The aim of the paper is to rank the selected regions based on the
indicators considering their importance for the evaluation expressed by the calculated weights.
The motivation for this ranking was to obtain an overview of the ranking of the regions Czech Republic
in a wider geographical context based on the variables related to the trend of digitalization. This
comparison was aimed to contribute to a description of a business environment of the regions.
The variables have been selected based on their availability, consistency of the data from a point of
missing values and appropriate time range – the restriction has been set not to use data older than five
years in the paper to use the most current data possible. There were seventeen variables selected for the
evaluation from 2016. This year has been selected due to balanced demands on availability and also
suitability of a distance from current time period. The variables can be divided into three separate groups
based on the areas which are covered by them. The first group of indicators is from the field for human
resources in the technological branches and branches with a potential importance for development. In
the second area there are the indicators of use of digital technologies included. The indicators related to
added value of selected branches are included in the third group of variables.
The values for the 17 variables have been collected for 30 regions from the Central Europe area and
Slovenia, in total five countries: Czech Republic, Slovakia, Austria, Slovenia and Hungary. Due to a
change in methodology of NUTS 2 regions it was needed to re-calculate data from two separate regions
in Hungary, Budapest and Pest which were originally merged in the Közép-Magyarország area and
according to the later changes into methodology divided into the Budapest and Pest region. As for the
use-of-technologies related variables, the data was available only for the merged region it was needed
to unify the regions used, the older divides have been used.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
233
To identify the importance of the selected values for the overall ranking the objective weighting method
has been used to get the results which are not dependent on an evaluator but based on the statistical data.
The Entropy method has been selected for this purpose due to its suitability to detect the most various
indicators (Deng et al., 2000). The TOPSIS approach has been then used to combine results from the
seventeen variables. The TOPSIS method has been selected due to the approach considering the position
of the variant compared to the best and to the worst possible result (Brožová, 2014, p. 37), the evaluation
is performed in a context.
The outcomes of the evaluation are briefly described focusing on the results from regions with the
highest and the lowest rankings. The results in specific areas are also compared and the main findings
are presented.
2 Brief overview of the topic
As the multi criteria decision making using Entropy method and TOPSIS method have been applied on
the problematic of assessment of the selected NUTS2 regions from the perspective related to digital
technologies, there is a concise outline of schemes used for evaluation in terms of digitalization. In this
paper the variables have been selected mainly based on their accessibility on the regional level and they
are focused more on the usage of technologies, value created and human resources in knowledge
intensive areas.
As the term of digitalization is used, this term is outlined at first as presented by Vogelsang (2010, p. 3)
where the author emphasize the technical aspects of the phenomenon in terms of use a binary form of
information which creates a background for digitalization. Vogelsang (2010, p. 3) also emphasize the
relation with the wide use of internet and worldwide consequences.
There are several structures of indexes described in a literature including a complex set of areas
indicating level of digitalization or indicators with a high impact on this area.
The indexes as the index used by the World Economic Forum are composed of the wide range of
variables from several areas from the technological to social (Baller Silja, Dutta Soumitra and Bruno
Lanvin, 2017). This index is not the only index used for assessment of area of digitalization. As the other
examples of such indexes the methodology used by Roland Berger Consultants is used as explained by
Soldatos et al. (2016, p. 166) or DESI indicator (European Commission, 2019). Another index with a
wide range of variables focus is the Networked readiness index. The description of this index is provided
for example by Vogelsang (2010, p. 12) or Xu (2014, p. 11-12).
Although variables from similar areas are used the indexes, the scope of variables is more narrow and
focused only on three main areas. The importance of such factors as research and development and the
problematic of human resources in terms of digitalization is described for example by Mařík et al.
(2016).
The topic of digitalization is subject of research of several authors as Vogelsang (2010), Xu (2014),
from the Czech authors as Mařík et al. (2016) focusing of Industy 4.0, Veber et al. (2018) focusing on
impacts of the digitalization and also implications in the economical braches including industry. The
topic is also monitored by institutions as World Economic forum as the Global Economic Forum, as the
Global Information Technology Report was issued, for example social (Baller Silja, Dutta Soumitra and
Bruno Lanvin, 2017).
The area of weighting methods and multi criteria decision making is described for example by the
following authors Zardari et al. (2015) or Brožová et al. (2014).
3 Methodology and Data
The variables taken into the assessment are at first weighted. The approach using the different weights
has been preferred before the same weighting for all variables to emphasize the importance of the
variables, which are suitable for differentiating the regions as the principle of the preference of the
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
234
variables with varied results is also described by Zardari et al. 2015, p. 65). This approach also helps to
identify the areas where different levels of the results are achieved.
As method for calculation of weights the Entropy method has been selected. This method has been used
due to its base in the statistical data and objective approach excluding influence of the evaluator. The
method is focused on identification of uncertainty and the more various variables are ranked with a
higher weight.
After calculation for the weights, then additive method is applied to calculate the overall results for each
region based on which they are ranked. In the methodology part of this article, the symbols in equations
have been unified as multiple literature resources have been used.
3.1 Methods used for assessment of the NUTS2 regions
At first a data matrix marked as Aij is set from the particular values yij where j=1,2,3…n, where n
represents seventeenth variable and i=1,2,3…m, where m is the thirtieth of the assessed regions. (Bao,
2012)
The values needs to be normalized at first as the different measures are used applying the equation (1)
=M
i
ij
ij
ij
y
yx
(1)
where xij is the normalized value, yij is the original value of the jth variable for ith region from M
regions.The normalization by the overall sum is described by Zardari et al. (2015, p. 33) or Brožová et
al. (2014, p, 13).
The assessment of the importance of the criteria follows with calculation of information entropy E j based
on the equation (2) as presented by Zardari et al. (2015, p. 33) or Brožová et al. (2014, p, 13).
( ) ( )MxxE ij
M
i
ijj ln/ln
−= (2)
Where xij represents normalized values and M represents a number of the variants of values for each
variable, in this paper it is the number of the NUTS2 regions. Finally the weight of jth variable
represented by wj is calculated using equation (3)
−
−=
N
j
j
j
j
E
Ew
1
1
(3)
Where Ej is information entropy of jth variable and N is a number of variables included in the assessment.
(Zardari et al., 2015, p. 33) or (Brožová et al., 2014, p, 13).
The second method used for the assessment is the Statistical variance method. Before application of the
method the values are normalized using the equation (1) to eliminate an influence of the different scales
for the variables. Then variances are calculated using the equation (4):
( )
M
xx
V
M
i
ijij
j
=
−
= 1
2
(4)
where a number of variables is represented by M. . (Zardari et al., 2015, p. 35)
The weights wj are calculated by dividing each variance by a sum of variances as in the equation (5) as
described by Zardari et al. (2015, p. 33).
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
235
=N
j
ij
ij
ij
v
vw
(5)
3.2 TOPSIS method
As a method of calculation of overall ranking the TOPSIS method has been selected. At first
normalization of the original data is needed based on the following calculation (6)
=M
i
ij
ij
ij
y
yr
2
(6)
where rij is a normalized value of jth variable for ith region (Brožová, 2014, p. 36). The normalized values
are then multiplied by already calculated weights in equation (4) and the v ij value is obtained. There are
also different approaches used for normalization, for example Bao et al. (2012, p. 112) uses
normalization by dividing each value by the highest value for the variable whereas Chan and Wang
(2013, p. 116) uses the same weighting procedure as used in the equation (6).
For each variable the value with the most and the least preferred variant is selected and then the distances
of the values for the regions are calculated from the most and also for the least preferred value. For the
distance calculation from the most preferred one the equation (7) is used
( )=
+ −=N
j
jiji hvd1
2 (7)
Where d+ i represents the overall distance of the best variant, vij is normalized and weighted value of the
jth variable for ith region and hj is the best variant for the jth variable. Similar procedure is applied for a
calculation of the overall distance from the least preferred variant d- i displayed in the equation (8)
described by (Brožová, 2014, p. 37).
( )=
− −=N
j
jiji lvd1
2 (8)
where lj represents the least preferred value for the jth variable. The overall ranking ci is obtain by
application of equation (9) described by (Brožová, 2014, p. 37).
−+
−
+=
ii
i
idd
dc (9)
3.3 Data preparation
As the main aims of the comparison was to base the assessment on the objective comparison method,
the data from the reliable resource as Eurostat (2019)14 were collected. There were variables from three
areas which were identified as ones with a potential to influence digitalization level or spread selected.
These three categories have been created focusing on the similarity of the area of the basis of the single
indicators. The three main areas are human resources, relationship of users to digital technologies and
vale adding in the selected areas. In the diagram 1 there is a structure of the used indicators visible.
14 DISCLAIMER: The opinions expressed in this publication are those of the individual author alone and do not
necessarily reflect the position of the European Commission.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
236
Diagram 1. Main groups of indicators used for the assessment
Source: Author’s processing
Regarding the indicators from the area of human resources, the structural point of view regarding
employment is described by Mařík et al. (2016, p. 159 - 162) where the importance of employment in
knowledge intensive manufacturing and services, especially information and communication
technologies and science and research, is highlighted, so the variables from such areas were included
into the comparison.
According to Mařík et al. (2016, p. 162 - 165) there is also a high importance of computer usage skills
not only on the side of the employees, but also end users and the importance of the partial areas as
internet access, use of internet for different purposes including communication with government are
mentioned, such areas are also included in DESI indicator (European commission, 2019). This can be
considered as a supportive factor to use the indicators related to the use of technologies for the
comparison.
Also the aspect of level of value added in the sector is mentioned by Mařík et al. (2016, p. 162) in a
context of the ratio of the employment in information and communication technology areas, where it is
noted, that not only the ratio itself, but also a structure, from a value added point of view , is important.
The specific variables divided in the presented three categories with an overview of the used measures
are listed in the Table 1
Selected areas of interest
Human resources
Value addedApproach to technologies
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
237
Table 1. An overview of the used variables
Indicator Measure
Value added ICT to overall % from overall value added
Value added professional and scientific activities to overall % from overall value added
Gross value added mil. EUR for ICT Per 1 employee in a branch
Gross value added mil. EUR of profess., scientific and technical activities Per 1 employee in a branch
Persons employed - computer programming % from economically actives15
Persons employed - information service activities % from economically actives
Persons employed - manufacturing of computers % from economically actives
Persons employed in High and medium high manufacturing % from economically actives
Persons employed in knowledge intensive services % from economically actives
Persons employed - scientists and engineers % from economically actives
Households with broadband access % of population
Individuals who ordered goods/ services over the internet for private use in last 3 months % of population
Individuals who used the internet for interaction with public authorities (in last 12 months) % of population
Individuals who used the internet for interaction with public authorities – submitting forms
(in last 12 months) % of population
Individuals who used the internet, frequency of use and activities – daily use of internet % of population
Individuals who used the internet, frequency of use and activities – internet banking % of population
Individuals who used the internet, frequency of use and activities – sell of goods % of population
Source: Author’s processing based on names and measures from Eurostat (2019)
The data is gathered for 30 NUTS2 regions for five countries: Czech Republic, Slovakia, Hungary,
Slovenia and Austria. These countries were selected on a geographical basis as the focus was on the
Central Europe to compare a situation in Czech Republic with the geographically near regions to identify
a position of each NUTS2 regions within the area. Also the perspective of data availability was needed
to be taken into consideration. For this reasons NUTS2 regions of Poland had to be excluded as for the
variables from the group containing indicators of use of technologies the values weren’t available at the
NUTS2 level and also Slovenia was added to the comparison.
The NUTS2 regions were selected to compare not the countries but to use more detailed view. Due to
the data availability a lower level of detail than NUTS2 couldn’t be used.
The year 2016 was selected as a suitable time period for the assessment where the data availability
perspective for all regions and the requirements on actual data were needed to be equalized. The year
2016 also meets the requirement to not to use the data older than 5 years from the current year.
To obtain data in a comparable format there were additional calculations needed except of the variables
of usage of digital technologies among population in a region which were already available in a form of
percentage. The other variables were re-calculated either to a percentage of the overall value (value
added or employed persons depending on the particular area) or per one unit – this approach has been
used in case of value added related variables, where the re-calculation has been performed.
There were also adjustments needed due to an inconsistency in methodology of data presentation from
Eurostat’s side. Data from two separate regions in Hungary, Budapest and Pest were originally merged
in the Közép-Magyarország area. For the variables related to human resources and for the value added
related variables the data was available according to a newer methodology divided into separate regions,
but for the data for the area of usage of digital technologies the data was available only for a merged
region Közép-Magyarország. To unify the methodology, the older version with merged region was
applied also in case of the variables for areas human resources and value added. The values for Budapest
and Pest regions were aggregated to obtain the value for the former Közép-Magyarország area and this
15 15 – 65 years
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
238
aggregated value has been used for calculations. The methodology of dividing regions is described by
Eurostat (2019).
3.4 Comparison of empirical results from the assessment
The approach using the weighted variables by Entropy method is used combined with the TOPSIS
method. The weight calculation applying the Entropy method also compared to the weighting by
Statistical variance approach to verify, if there is a significant difference in the weights obtained
identified.
After the calculation of the overall values for each NUTS 2 region the overall ranking is completed. The
focus is on the regions with the highest and the lowest overall score. These regions are then described
more in detail.
The analysis for the variables itself is also done assessing the regions not from the multi-criteria point
of view, but by each variable separately using also fundamental descriptive statistics and graphical tools.
Those are also used for the description of the most and the least successful regions in the ranking.
4 Empirical Results
In this section of the paper, main findings from the application of described methods are described. At
first the focus is on the calculation of the weights and comparison of results using two different objective
calculation methods. After the description of results of weights’ calculation, there are described the
results from application of TOPSIS decision making method. As weights used for a calculation the
results from the entropy method were applied. The presented results are based on processing of data
from Eurostat (2019).
4.1 Weights calculated by Entropy method and Statistical variance
As the weights are calculated by objective methods, there is no influence of assessors, the calculation
are based only on statistical parameters. Due to this fact, it is not possible to deduce the assumptions
about an importance of selected variables for digitalization itself as there are not included all the
interdependencies. The weights represent importance for sorting the regions according to these selected
variables by adding higher preference to the variables based on which the regions can be sorted more
sufficiently, in the other words the variables which have the higher measure of uncertainty (Entropy
method) or have a higher value of the average difference from the average values of the variables
(Statistical variance method).
At firsts the results from the Entropy method weight calculation are presented in the Table 2, where the
variables are sorted according to achieved weights.
Table 2 . Weights calculated using Entropy method
Variable Weight Rank
Persons employed - information service activities 0.215756 1
Persons employed - computer programming 0.181908 2
Value added ICT to overall 0.127692 3
Persons employed - manufacturing of computers 0.104176 4
Gross value added mil. EUR of profess. sci-tech. activities per 1 econ. active person 0.087879 5
Gross value added mil. EUR for ICT per 1 economically active person 0.050905 6
Persons employed in High and medium high manufacturing 0.048423 7
Persons employed - scientists and engineers 0.038730 8
Individuals who used the internet for interaction with public authorities – submitting forms 0.038157 9
Value added professional and scientific activities to overall 0.037554 10
Individuals who used the internet, frequency of use and activities – sell of goods 0.020088 11
Individuals who ordered goods/ services over the internet for private use 0.018620 12
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
239
Individuals who used the internet, frequency of use and activities – internet banking 0.011476 13
Individuals who used the internet for interaction with public authorities 0.009051 14
Persons employed in knowledge intensive services 0.007325 15
Individuals who used the internet, frequency of use and activities – daily use of internet 0.001454 16
Households with broadband access 0.000806 17
Source: Authors processing based on data from Eurostat (2019)
It can be seen, that in general the values for variables related to the users’ behaviours can be evaluated
as less varied and are of the more similar digits in all assessed regions. The first three variables chosen
as the ones with the biggest impact on the ranking are Persons employed - information service activities,
Persons employed - computer programming which have significantly higher values and the Value added
ICT to overall where a value close to the fourth one - Persons employed - manufacturing of computers
– is achieved.
The indicator with the lowest assigned importance is the Household with broadband access where the
numbers in all regions are very similar.
The second rank displayed in the Table 3 is calculated using the Statistical variance procedure. The
overall ranking varies very little from the results from the Entropy method, only the ranks of Individuals
who used the internet for interaction with public authorities and Value added professional and scientific
activities to overall are switched.
Table 3. Weights calculated using Statistical variance method
Variable Weight Rank
Persons employed - information service activities 0.251532 1
Persons employed - computer programming 0.207787 2
Value added ICT to overall 0.136985 3
Persons employed - manufacturing of computers 0.089487 4
Gross value added mil. EUR of profess, sci-tech. activities per 1 econ. active person 0.078581 5
Gross value added mil. EUR for ICT per 1 economically active person 0.042422 6
Persons employed in High and medium high manufacturing 0.039308 7
Persons employed - scientists and engineers 0.034667 8
Value added professional and scientific activities to overall 0.033404 9
Individuals who used the internet for interaction with public authorities – submitting forms 0.029891 10
Individuals who used the internet, frequency of use and activities – sell of goods 0.016474 11
Individuals who ordered goods/ services over the internet for private use 0.015235 12
Individuals who used the internet, frequency of use and activities – internet banking 0.008822 13
Individuals who used the internet for interaction with public authorities 0.007190 14
Persons employed in knowledge intensive services 0.006354 15
Individuals who used the internet, frequency of use and activities – daily use of internet 0.001203 16
Households with broadband access 0.000660 17
Source: Authors processing based on data from Eurostat (2019)
The results are more varied in a matter of the distribution of the values of the weights when the ends of
the rank are reviewed. The importance of the first three indicators is higher than in case of Entropy
method and there is a higher difference between the weight on the 3rd and 4th position.
Based on the both applied methods it is visible, that the regions differs mostly in the variables related to
human resources in IT branches and also in case of the ratio of the value added to overall value added
in the region.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
240
4.2 Results from application of TOPSIS method
The weights calculated using Entropy method were combined with the TOPSIS multi-criteria decision
making procedure to obtain overall score in a rank from 0 to 1 base on the distance from the most
preferred variant. As all the variables can be considered as maximizing criteria there was no
transformation needed among criteria type. The original values have been transformed to the same scale
as described in the part of the paper dedicated to methodology.
The final ranking of the regions is presented in the Table 4, where we can see that the first four regions
with evaluation above 0.5 are achieved in case of Praha, Bratislavský kraj, Wien and Közép-
Magyarország, so the highest overall values are achieved in the capital regions of the countries, where
there is a higher usage of technologies, concentration of human resources and value added.
Table 4. Final ranking of the NUTS2 regions combining the Entropy and TOPSIS methods
Region Score Rank Region Score Rank Region Score Rank
Praha 0.782087 1 Steiermark 0.215584 11 Niederösterreich 0.14443 21
Bratislavský kraj 0.690954 2 Közép-Dunántúl 0.201964 12 Střední Morava 0.142244 22
Wien 0.623235 3 Severovýchod 0.198683 13 Burgenland 0.127493 23
Közép-Magyarország 0.549858 4 Vorarlberg 0.184004 14 Jihozápad 0.125833 24
Zahodna Slovenija 0.304428 5 Východné Slovensko 0.174372 15 Dél-Dunántúl 0.117726 25
Jihovýchod 0.285439 6 Tirol 0.171391 16 Střední Čechy 0.107704 26
Salzburg 0.247281 7 Západné Slovensko 0.161791 17 Észak-Alföld 0.106295 27
Kärnten 0.220348 8 Nyugat-Dunántúl 0.151063 18 Vzhodna Slovenija 0.095132 28
Oberösterreich 0.220028 9 Moravskoslezsko 0.149815 19 Dél-Alföld 0.067743 29
Észak-Magyarország 0.216701 10 Stredné Slovensko 0.148306 20 Severozápad 0.061491 30
Source: Authors processing based on data from Eurostat (2019)
The results for the regions, which have been ranked with the highest scores, are described more in details.
For this comparison, the normalized data used for weights’ calculation are used also for the comparison
as it is suitable approach for increasing readability of the graphs presented in the Appendix 1 as they are
scaled to the same average (0.033). The description is focused on the defined three sub-areas.
Praha – In case of value added related variables the values are above the average values for the variables.
In case of Value added ICT to overall the highest value for the whole group of regions is achieved. For
the variable Value added professional and scientific activities to overall the third highest value has been
reached. Except of value for the variable Gross value added mil. EUR of professional and sci-tech.
activities per 1 employee there were high values achieved in general for this area.
In case of area of human resources there are high values achieved, in case of Persons employed -
computer programming, Persons employed in knowledge intensive services and Persons employed -
scientists and engineers the highest values from all regions have been achieved, except of variables
Persons employed - manufacturing of computers and Persons employed in High and medium high
manufacturing where the values are even below the averages for these variables.
When the group of variables related to the approach to use of technologies is assessed, the results are
more various. Except from the variables Individuals who ordered goods/ services over the internet for
private use, Individuals who used the internet for interaction with public authorities and Individuals who
used the internet for interaction with public authorities - submitting forms the achieved values were
above the averages for the values. In case of variable of households with broadband access the value
was the highest from all regions.
Based on these findings it can be derived that the strengths of this region in terms of the assessed
variables are mainly human resources except of the manufacturing and also area of value adding. Good
results are also achieved in case of the approach to digital technologies.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
241
Bratislavský kraj – In Bratislavský kraj in area of value added related variables high values are achieved
in case of the ratios of the value added for ICT and for professional and scientific activities to the overall
value added. Values lower than the average values were achieved in case of the variables of value added
per one employee in the ICT and professional and scientific activities branches.
In case of the human resources are the values for the selected variables are mostly above the average
values. For the variable Persons employed - information service activities the highest value from all
regions has been achieved and for the variable Persons employed - computer programming there was
the second highest valued identified. For the variables related to manufacturing the values are lower –
for Persons employed - manufacturing of computers the value is above the average and for Persons
employed in High and medium high manufacturing the value is close to the average.
When the variables from the area of approach to digital technologies the values for the related variables
are assessed the values are above or close to the average except of Individuals who used the internet for
interaction with public authorities - submitting forms, which is below the average.
For the Bratislavský kraj area we can consider results from the area of the human resources as a strength
except of these related to manufacturing. The results of variables of ratios ICT and professional and
scientific GVA to the overall GVA are also considered as high.
Wien – For Wien there are above average values of value added related variables identified. In case of
variables Value added professional and scientific activities to overall and Gross value added mil. EUR
for ICT per 1 employee the highest values from all regions are achieved.
In the area of human resources related variables high values above the overall average are achieved
except of the variables Persons employed - manufacturing of computers and Persons employed in
High and medium high manufacturing where the values are relatively low. For the variables Persons
employed in knowledge intensive services and Persons employed - scientists and engineers the second
highest values are achieved among all assessed regions.
The variables in the area of attitude to the digital technologies usage are above or close to the averages
for the variables except of the variable of Individuals who used the internet, frequency of use and
activities – sell of goods which is slightly below the average.
For Wien the values can be considered as balanced, mainly above the averages for variables. As the
exception there are variables related to the human resources in manufacturing areas included to this
assessment.
Közép-Magyarország – In this region results for the area of value added related variables the similar
pattern as in case of Praha area is followed. High values are achieved in case of variables Value added
ICT to overall and Value added professional and scientific activities to overall. In case of variables Gross
value added mil. EUR for ICT per 1 employee and Gross value added mil. EUR of professional and sci-
tech. activities per 1 employee the values are below the averages.
In the area of human resources below average values are also identified for variables Persons employed
- manufacturing of computers and Persons employed in High and medium high manufacturing. For the
variable the third highest value is achieved and for Persons employed - computer programming
Persons employed - information service activities the fourth highest value from all regions has
been identified.
In the third assessed area the values for variables are identified as slightly higher or close to the average
values except of the variable Individuals who used the internet for interaction with public authorities -
submitting forms and in case of the variable Individuals who ordered goods/ services over the internet
for private use the value achieved is under the average.
Generally it could be considered that Közép-Magyarország follows similar patterns as Praha in the
assessed areas of variables with lower values with some exceptions as Persons employed -
manufacturing of computers.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
242
The regions with the lowest values of the overall ranking are Dél-Alföld and Severozápad region. The
value under 0.1 is achieved also in the region Vzhodna Slovenija, but the value doesn’t differ so
significantly from regions on the 26th positions.
More details for characteristics for the regions with the lowest rankings are also described.
Dél-Alföld – All values from the area of value adding are below the averages for these variables. The
highest value from these variables for the region are achieved for the variable Value added professional
and scientific activities to overall, the lowest for the Gross value added mil. EUR of professional and
sci-tech. activities per 1 employee, but the values are not significantly highly varied.
In the area of human resources, the values are also relatively lower except of the value of Persons
employed in knowledge intensive services where the score is close to the average value. The lowest
value among the variables for this region is identified for the variable Persons employed - information
service activities.
In case of variables related to the attitude to the usage of digital technologies, three values are
significantly below the average - Individuals who ordered goods/ services over the internet for private
use, Individuals who used the internet, frequency of use and activities - internet banking and
Individuals who used the internet, frequency of use and activities - sell of goods. The variable
Individuals who used the internet for interaction with public authorities - submitting forms is identified
as above average one and the remaining are below the average.
Severozápad region – In this region the values related to the GVA are generally below the average,
except of the variable Gross value added mil. EUR for ICT per 1 employee. The lowest value for the
region in this area is identified for the variable Gross value added mil. EUR of professional and sci-tech.
activities per 1 employee.
In the area of human resources, the values are relatively low with the exception of the variable Persons
employed in High and medium high manufacturing where the score is above the average and also the
value of Persons employed in knowledge intensive services which is higher than the remaining scores
in this group of variables.
In the area of usage of digital technologies there were for values significantly lower than the average
identified - Individuals who ordered goods/ services over the internet for private use, Individuals who
used the internet for interaction with public authorities, Individuals who used the internet for interaction
with public authorities - submitting forms and Individuals who used the internet, frequency of use and
activities - sell of goods. The remaining ones are more approaching to the average value.
There are also groups with the lower differences among them, one with values around the score 0.2 and
around the value 0.15. The biggest differences are between the lowest and highest values and the rest of
the assessed regions.
5 Conclusion
Based on the statistical data gathered for the seventeen selected variables for the thirty regions from the
area of Central Europe the weights for these variables have been calculated using two objective
weighting methods – Entropy method and Statistical variance. The variables with a highest and the
lowest importance for the ranking of the regions were identified and it was found out, that the variables
from the area of usage of digital technologies has the lowest impact on the multi-criteria decision
making.
The results from both methods have been also compared and the differences identified. There were no
significant changes in the rank of impact of the variables, but the weights were distributed differently.
In case of Entropy method the difference between the lowest and highest weights was lower than in case
of the Statistical variance method. The weighting according to Entropy method calculation was used in
combination with a TOPSIS method and the overall scores for the regions were achieved and the final
ranking of the regions based on the selected values presented.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
243
There were four regions identified with the score above the value 0.5 Praha, Bratislavský kraj, Wien and
Közép-Magyarország. The results for these four regions were described more in details by a
summarization of the main findings in each of the three groups of variables.The same procedure has
been applied for the cases of the regions with a low score under the value 0.1. There were three regions
identified - Dél-Alföld and Severozápad region Vzhodna Slovenija. The descriptions of the achieved
results has been done for the regions Dél-Alföld and Severozápad as they have achieved similar scores
and the score for the Vzhodna Slovenija is relatively close to the score of the region on the 27 th position
of the ranking with a value slightly above 0.1.
When application of the selected weighting methods, mainly the Entropy method is evaluated, as the
main advantage of the method is eliminating the need of the subjective evaluation of the importance of
variables and the results are not dependent on the specific evaluator. This method was well applicable
on the data of various measures assigning the high importance to the variables which are suitable to
select the regions as also mentioned by (Zardari et al. 2015, p. 65).
The scope of the research in this paper was significantly predetermined by the data base available on the
level of the level of more detailed geographical areas. This aspect influenced a range of variables used
for the assessment as well as the selection of the regions.
There are also several topics for the further research in the area of the variable selection and areas of
interest (within the limitation of data availability) and also in terms of the relation between the variables
– the aspect of correlation. It would be also beneficial to compare the results within time series to
evaluate the development of positions of the regions in time.
References
[1] Baller, S., Dutta, S. and B. Lanvin. (2016) The global Information Technology Report 2016.
[online]. World Economic Forum and INSEAD. Geneva. Available at:
<www.weforum.org/reports/the-global-information-technology-report-2016, (cit. 12.11. 2019)>.
[2] Bao, Q. et al (2012) [Kahraman, c. ed.] Computational Intelligence Systems in Industrial
Engineering: With Recent Theory and Applications. Paris: Atlantis Press, 2012-11-6. p. 109-130.
[3] Brožová, H. et al. (2014). Modely pro vícekriteriální rozhodování. Praha: Credit.
[4] Deng, H., Yeh, C. H., & Willis, R. J. (2000). Inter-company comparison using modified TOPSIS
with objective weights. Computers and Operations Research, 27(10). p. 963–973.
[5] European commision. (2019). Digital single market: DESI by components. [Online]. European
commision. Available at: <https://digital-agenda-data.eu/charts/desi-
components#chart={%22indicator%22:%22DESI%22,%22breakdown-
group%22:%22DESI%22,%22unit-measure%22:%22pc_DESI%22,%22time-
period%22:%222018%22} , (cit. 28.11.2019)>.
[6] Eurostat (2019). Regional statistics by NUTS classification. [online database]. Luxembourg:
European Commision, Available at: <https://ec.europa.eu/eurostat/data/database, (cit. 25.
5.2019)>.
[7] Chan, H. K. and X. Wang (2013). Fuzzy hierarchical model for risk assessment. New York:
Springer.
[8] Mařík, V. et al. (2016). Průmysl 4.0: výzva pro Českou republiku. Praha: Management Press.
[9] Soldatos J., Gusmeroli S., Malo P. and Di Orio G. (Friess, P, Vermesan O. - Editors) (c2016)
Digitising the Industry – Internet of Things Connecting the Physical, Digital and Virtual Worlds.
Delft: River Publishers. p. 153-181.
[10] Veber, J. et al. (2018) Digitalizace ekonomiky a společnosti: výhody, rizika, příležitosti. Praha:
Management Press.
[11] Vogelsang, M. (2010). Digitalization in Open Economies – Theory and Policy Implications.
Heidelberg: Physica-Verlag Heidelberg.
22nd International Conference
MEKON 2020
February 6, 2020, Ostrava, Czech Republic
Conference Proceedings of MEKON 2020
VŠB-Technical University of Ostrava
Faculty of Economics
244
[12] Xu, J. (2014). Managing Digital Enterprise: Ten Essential Topics. Paris: Atlantis Press
[13] Zardari, N. et al. (2015). Weighting Methods and their Effects on Multi-Criteria Decision Making
Model Outcomes in Water Resources Management. Cham: Springer International Publishing.
245
Appendix 1
Table 1 – Variables related to gross value added
Source: Authors processing based on data from Eurostat (2019)
246
Table 2 – Variables related to human resources
Source: Authors processing based on data from Eurostat (2019)
247
Table 3 – Variables related to approach to digital technologies usage
Source: Authors processing based on data from Eurostat (2019)