-
Mat-2.108 Independent Research Project in Applied Mathematics
November 20, 2005
Application of Robust Portfolio Modelling to
the Management of Intellectual Property Rights
Helsinki University of Technology Systems Analysis Laboratory
Antti Malava 64705M Department of Engineering Physics and
Mathematics
-
Contents
CONTENTS
.....................................................................................................................2
1.
INTRODUCTION..........................................................................................................3
2. PATENTS AS INTELLECTUAL
PROPERTY...................................................................5
2.1. Background of Patenting
................................................................................5
2.2. Patent Portfolio Valuation
Approaches..........................................................6
3. ROBUST PORTFOLIO MODELLING – A THEORETICAL
FRAMEWORK......................8
3.1. Representation of Additive
Value....................................................................9
3.2. Incomplete
Information.................................................................................10
3.3. Additional Constraints
..................................................................................11
3.4. Non-dominated
Portfolios.............................................................................11
3.5. Additional Information
.................................................................................12
3.6
Robustness......................................................................................................12
4. CASE STUDY ON PATENT PORTFOLIO
....................................................................14
4.1. Problem Statement
........................................................................................14
4.2. Modelling the Problem
.................................................................................14
4.3. Experimental
Design.....................................................................................15
4.4. Results
...........................................................................................................15
5. DISCUSSION AND CONCLUSIONS
.............................................................................18
6.
REFERENCES............................................................................................................20
7.
APPENDICES.............................................................................................................22
2
-
1. Introduction
For a company operating in fast-changing, particularly uncertain
high-tech market,
systematic management of project portfolio is an important
aspect of its strategic
decision-making process. Research and Development (R&D)
activities, for example,
represent a typical such environment.
A simple theoretical approach would model the portfolio
selection as a single
objective Capital Budgeting Problem that maximises portfolio
without violating
resource constraints (e.g. Luenberger 1998). The most used
Capital Budgeting
methods include Net Present Value (NPV), Internal Rate of Return
(IRR), Discounted
Cash Flow (DCF) and Payback Period (Investopedia, 2005).
In reality, however, this kind of methodology can rarely be
applied for several
reasons. Firstly, the value of projects often traces to their
performance against several
criteria rather than one (not all of which are necessarily
quantitative). Secondly, rather
than considering the portfolio selection as a single objective
problem, the company
may have multiple, conflicting and even incommensurate aims
(Liesiö et al, 2005)
likewise with their company-wide strategy. Thirdly, and perhaps
most importantly, a
standard Capital Budgeting Problem cannot accommodate incomplete
information
with regards to criteria weights and criterion-specific
performance levels. The
availability of incomplete, uncertain information about project
performance against
different criteria, however, is very often a realistic situation
– after all, most projects
are novel without reliable historical reference projects to give
indication of
performance levels.
Robust Portfolio Modelling (RPM), a recently developed extension
of Preference
Programming of the Helsinki University of Technology, takes into
account all of the
issues above and provides a flexible framework for mathematical
assessment of the
portfolio selection problem. RPM is a decision support framework
that makes little
demands on the data and, rather than suggesting a single optimal
portfolio, produces a
set of non-dominated (also Pareto-efficient) portfolios and
performance measures to
assess the robustness of the portfolio and the proposed
projects.
In this paper, RPM methodology is applied to support
Intellectual Property Right
(IPR) portfolio management. The study was conducted in
co-operation with
Asperation Oy, a joint Research & Development company of
Perlos Corporation and
3
-
Aspocomp Group Oyj1 (Asperation Oy, 2005). Asperation’s
innovations concentrate
around audio, optical, radio frequency and other technologies
connected to
mechanics, and printed circuit board technology (Perlos Oyj,
2005). Management of
the IPRs related to these innovations are of utmost strategic
importance for Asperation
to sustain competitive advantage in future. The purpose of the
study is to evaluate the
RPM framework in a real case and to present Asperation a
theoretical decision
support tool for patent portfolio optimisation and management.
These aims are in line
with Asperations’ larger scope of adapting a more systematic
methodology for their
project portfolio management practises.
The rest of this paper is structured as follows. Section 2
Discusses Intellectual
Property Rights background, while Section 3 introduces the
theoretical framework of
Robust Portfolio Modelling. In section 4, RPM is applied to the
case of IPR portfolio
management. Approach to modelling together with practical issues
and results are
discussed. Section 5 draws conclusions from the case study.
1 On September 1st, in the middle of the study, the functions of
Asperation were divided between
Aspocomp and Perlos (Perlos Oyj, 2005). The study was continued
with Perlos.
4
-
2. Patents as Intellectual Property
2.1. Background of Patenting
Patents, trade secrets, trademarks and copyrights are all a form
of intellectual property
that can be protected. This paper concentrates on patents.
Patents are intangible assets,
whose value lies within their right that prohibits others by law
to professionally
exploit the underlying invention. Producing, selling, importing
or using the patented
method is regarded as professional exploitation (National Board
of Patents and
Registration of Finland, 2005). Patents can be granted for
inventions that are new,
innovative and industrially practicable. In other words, just an
idea cannot be
patented.
Patents and other intellectual property are objects of
international co-operation and
corresponding regulation (Ernst & Young, 2000) administered
by World Intellectual
Property Organization (WIPO). Patent applications can be filed
nationally with the
country’s patent office, or internationally using systems such
as Patent Cooperation
Treaty (PCT) or European Patent Convention (EPC). The
prohibition law is
applicable within the geographical area it was granted, and it
is usually valid for a
maximum of 20 years, subject to yearly payments for keeping the
underlying patent
valid.
The statistics show that the number of nation-wide patens
granted in Finland has been
decreasing slightly within the last 10 years from about 2400 to
about 2100 (National
Board of Patents and Registration of Finland, 2005). At the same
time, however, the
number of European patents granted in Finland has increased from
zero-level2 to
about 6000. This rapid increase of European and international
patents has prompted
companies to find methods to systematically evaluate and manage
patent portfolios as
global entities, part of their larger-scope strategies.
The creation, protection and utilisation of intellectual
property are recognised as
issues of all stakeholders, including government and industry
(Asian Productivity
Organization, 2004), because they directly affect country’s
competitiveness in global
markets. Therefore a number of governmental initiatives around
the world have been
2 Finland joined the EPC and the EPO in 1996 and is also a
member of WIPO.
5
-
started to not only support the risky process of intellectual
property protection, but
also to inform businesses about the importance of systematic
management of
intellectual property as part of their strategies. For example,
the government of Japan
has established an intellectual property strategic centre that
brings together the
government, academia and industry to tackle the problems.
Similarly, the Danish
government has addressed the issues of intellectual property in
their wider strategy for
improving business development.
2.2. Patent Portfolio Valuation Approaches
The management of tangible assets has a solid basis through many
years of research
and the exchange of experience, whereas the management and
evaluation of
intangible assets in a strategic way is still not much further
than in experimental stage.
Ernst & Young research (2000) revealed, for example, that
nine out of ten Danish
companies expected the importance of evaluating the value of
their patents and
management of patent portfolio to increase, but only a few could
identify actual
methods for doing this. Similarly, although it is estimated that
substantially more than
half of the total value of publicly listed US companies arises
from intangible assets,
three out of four companies do not assign any value to their
intellectual capital (Asian
Productivity Organization, 2004). The core of the problem
relates to the difficulties of
valuation, much of which is performed only in connection with
purchases, sales,
licensing agreements. The awareness and perceived benefits of
IPR are lacking
especially in many small and medium companies that do not see
the relevance of IPRs
to their business.
Considering theoretical perspectives, although decision analysis
has been widely
accounted for in the literature in connection to different areas
of tangible issues such
as healthcare, engineering and project management, very little
can be found about
decision analysis methods applied to intangible assets
management. Some individual
research contributions exist but there seems to be no clear
consensus. The few
mathematical valuation methods used in the field of IPR do come
from the field of
decision analysis and relate to evaluating preferences by
approximation (such as
PRIME) or through intervals (such as PAIRS) or indirectly by
considering
hypothetical alternatives’ value (such as SWING). More
quantitative models are some
extensions of Net Present Value or other basic financial
methods. Because of the
6
-
difficulties related to capture uncertainty in mathematical
models, subjective methods
of valuation such as Delphi technique, expert opinions and are
still widely used and
decisions are often made based on “feeling”.
Patenting institutions have begun to notice the need for
companies to develop patent
portfolio management methods. For example, as part of Danish
Government’s wider
strategy for improving business development, Danish Patent and
Trademark Office
recently addressed this issue by developing a software tool to
“identify untapped
business potential” (Nielsen, 2003). The tool, IPscore 2.0,
presents a systematic
evaluation – both qualitative and quantitative – in the form of
financial forecast
estimating the internal net present value (i.e. not market
value) of the evaluated
technology. Input is gathered with questionnaires to ensure
comprehensive evaluation
(such as evaluations at various stages during the project
lifetime), and output is
provided both numerically and graphically.
To take a new, academically orientated approach to patent
portfolio valuation and
management, Robust Portfolio Optimisation is presented and
applied in this paper.
7
-
3. Robust Portfolio Modelling – a Theoretical Framework
Multiple Criteria Optimisation (also referred to as Multiple
Objective Optimisation)
has been and continues to be a classical field of study in
Operational Research (Liesiö
et al, 2004). Over the years this area has evolved, and recently
the focus has been to
provide easily applicable, user-friendly methods that
accommodate the use of
incomplete information with regards to required data (Liesiö,
2004). In other words,
the methods seem to be developing more and more useful to real
problems rather than
offering a theoretical consideration that may be of little value
to decision makers.
Preference Programming is one such an area of Multiple Criteria
Optimisation that
accommodates incomplete information about criterion-specific
performance levels
and criteria weights in the selection of a single alternative
(Liesiö et al, 2005).
Incomplete information is captured by set inclusion meaning that
the ‘true’ value of a
particular parameter is contained within a feasible set derived
from the decision
maker’s preferences.
Robust Portfolio Modelling (RPM) is an extension of Preference
Programming to the
problem of project portfolio selection. In RPM, the total value
of each project and the
project portfolio are modelled by an additive value tree
analysis as a weighted average
of the criterion-specific scores. Linear inequalities are used
to capture incomplete
information about criteria weights and criterion-specific
performance levels are
modelled by using score intervals (Liesiö et al, 2005). The use
of incomplete
information leads to a recommendation of multiple non-dominated
portfolios rather
than one. Non-dominated portfolios have an overall value not
less than any other
portfolio for any combination of feasible weights and scores. To
offer concrete
decision recommendations, measures for individual project and
project portfolio
robustness are employed.
A graphical representation of RPM (figure 1) gathers together
the main ideas and the
multi-step process of the model application. The first phase
includes a choice of the
criteria against which the potential projects are evaluated.
Incomplete information is
first modelled by reasonably wide score intervals and loose
weight statements in order
to take clearly into account the uncertainties. Computation of
non-dominated
portfolios follows, together with gradual analysis of projects
and project portfolios by
robustness measures. Once the so-called borderline projects have
been identified, the
8
-
focus moves on to give additional information, i.e. narrow score
intervals and give
more precise weight information with respect to these borderline
projects. The final
project selection method is left for the decision maker, but by
this phase he or she can
concentrate on analysing the set of remaining borderline
projects (i.e. the uncertain
zone), which usually is much smaller in number than the original
set of potential
projects. The projects identified as core and exterior project
require less attention.
Figure 1. RPM framework. Source: http://www.rpm.tkk.fi/
3.1. Representation of Additive Value
Let { }mxxX ,,1 K= denote a set of m available projects that are
evaluated against n criteria. A performance score of project j with
regards to I:th attribute (criterion) is
represented by . jiv
In addition, each project j is associated with k:th resource
consumption , and the
total amount of resource k is limited by a budget B
jkc
k. The overall value of a project is
the weighted average of its criterion-specific scores, i.e.
9
http://www.rpm.tkk.fi/
-
( ) ∑=
=n
i
jii
j vwxV1
This represents an additive value model in which the weight wi
represents the relative
importance of criterion i. The weights of all n criteria form a
weight vector
that can be normalised so that it belongs to a set ],,[ 1 nwww
K=
⎭⎬⎫
⎩⎨⎧
≥=∈=∈ ∑=
0,1|1
0i
n
ii
nw wwRwSw
that is, weight information is modelled by linear
constraints.
3.2. Incomplete Information
One of the most difficult issues of applying mathematical models
to support decision-
making is that many of them require complete information in the
sense of fixed values
of weights and scores. However, this kind of data is not often
available, nor is it
reasonable to always assume that such point estimates are
accurate. For example, it is
quite impossible to give a zero-variance point estimate for a
completely new project
that has not been started yet. A key feature of RPM methodology
is to allow for
incomplete information with regards to weights of the criteria
and the criterion-
specific scores of each project.
Incomplete weight information is modelled as a feasible (and
convex) weight set
described by linear constraints derived from decision makers’
preference
statements. Incomplete criterion-specific performance
information is modelled
through score intervals
0ww SS ⊆
],[ji
ji vv that are assumed to contain the ‘true scores . The
interval for the value of a portfolio p, therefore, is given
by
jiv
)],(max),,(min[),,( wpVwpVvwpVww SwSw ∈∈
∈
where
( ) ( )
( ) ( )∑ ∑∑
∑ ∑∑
∈ ∈ =
∈ ∈ =
==
==
px px
n
iiji
j
px px
n
iiji
j
j j
j j
vwwxVwpV
vwwxVwpV
1
1
,,
,,
10
-
3.3. Additional Constraints
The RPM framework allows taking into account additional
information such as:
o minimum budget usage
o project synergies
o balanced allocation of resources
and others that can be modelled by adding relevant logical
constraints to the set of
constraints.
3.4. Non-dominated Portfolios
Because of using incomplete information, the model does not
suggest a single optimal
portfolio but offers a number of non-dominated portfolios that
lie within a feasible
information set. Non-dominated portfolios are feasible (i.e.
constraint-satisfying)
portfolios whose overall values are higher than any other
portfolios’ values for any
combination of feasible weights and scores.
Definition. Portfolio p dominates p’ with regards to information
set S if and only if
V(p,w,v) ≥ V(p’,w,v) for all (w,v) ∈ S, (p,p’) ∈ P and
V(p,w,v) > V(p’,w,v) for some (w,v) ∈ S, (p,p’) ∈ P.
Non-dominated portfolios are therefore optimal solutions in some
parts of the
information space. A thorough discussion of theory behind
dominance structures and
the computation of non-dominated portfolios can be found in
Liesiö et al, 2005.
To illustrate the effect of incomplete weight information on
non-dominated portfolios,
consider figure 2. It depicts a situation where three portfolios
are evaluated against
two criteria of which criterion 1 is more important than
criterion 2 (i.e. w1 > w2). The
feasible weight area is therefore delimited to the left-hand
side of vertical dotted line,
which means that portfolio 3 can be excluded from the set of
non-dominated
portfolios with regards to given information set.
11
-
Figure 2. Additive value of portfolios with preference
information w1 > w2
3.5. Additional Information
Should the number of non-dominated portfolios be too large,
decision makers can be
asked to reduce the information set by proving narrower score
intervals and/or further
weight constraints. It should be noted that while these actions
are likely to reduce the
number of non-dominated portfolios, they do not add any new
portfolios to it.
3.6 Robustness
The set of non-dominated portfolios itself does not give the
decision maker any
straight recommendation about which projects should be included
in the chosen
portfolio – after all, there may be a large number of different
non-dominated
portfolios. However, one can calculate a core index for each
individual project. The
core index simply measures in how many of the non-dominated
portfolios each
project was contained. A core index of 100% implies that no
matter what the ‘true’
weights and scores are, as long as they lie within the given
limits, this project would
be included in all of the non-dominated portfolios and should
therefore be selected by
the decision maker in any case. These projects are called core
projects. Using the
same logic, a project with 0% core index is not included in any
of the non-dominated
portfolios and should therefore be discarded. These projects are
referred to as exterior
projects. The projects whose core index lies between these two
extremes are so-called
borderline projects.
12
-
The identification of core, exterior and borderline projects
facilitates the work of
decision maker greatly because s/he can concentrate on
evaluating further the
borderline projects that belong to optimal portfolios for some
but not all weight and
score regions. This narrowing of the scope of analysis and
further investigation is the
real value of RPM methodology. Even though the computation of
non-dominated
portfolios requires a complicated algorithm, the end results are
clearly understandable
even without mathematical background and can therefore be easily
communicated.
The robustness of a non-dominated portfolio can be measured by
maximin and
minimax-regret rules that are employed in Preference Programming
(Salo and
Hämäläinen, 2001). Maximin rule by definition recommends the
portfolio that
performs the best in the worst-case scenario (i.e. the lower
bound of its value defined
by feasible weights and scores). The recommendation is
therefore
( )wpVPwn SwPp
,minmaxargmin ∈∈=
Minimax-regret on the other hand recommends a portfolio whose
greatest possible
loss of value with regards to some other portfolio over the
information set is smallest.
The recommendation is therefore
( ) ( )[ ]wppVwppVPwNn SwPpPp
regret ,'/,/'maxminarg,
maxmin '−=
∈∈∈−
It should be noted that maximin and minimax-regret rules may
recommend different
portfolios.
13
-
4. Case Study on Patent Portfolio
4.1. Problem Statement
For a company whose success is closely tight to the innovative
Research &
Development activities, systematic managing of Intellectual
Property Rights (IPR)
related to these new technologies is vital to ensure positive
future perspectives. The
problem is twofold. On micro level, the company needs a
methodology to value IPR
in order to assess whether it is profitable to patent a
particular IPR in some possible
technological and/or geographical extent. Secondly, because of
limited (financial)
resources, not all IPRs can be patented – at least in global
scale. Therefore the
company needs a methodology to recommend a selection of
individual projects to
form an efficient patent portfolio.
A recent academic study on IPR valuation concentrated on
building a methodology
for valuing individual projects (Antila et al, 2005). In this
section, RPM methodology
is applied to a problem of supporting patent portfolio
management, whereby patents
are considered as projects. Previously Asperation Oy had chosen
individual patents to
project portfolio so that they fit within the budget. One of the
objectives of this study
is provide comparison for the previous heuristic.
4.2. Modelling the Problem
In the studied case, there are 21 available projects, each of
which is evaluated against
5 criteria that are most important to the company. With regards
to ordinal preference
order, these criteria are Primary Business Potential, Necessity,
Technical Coverage,
Strategic Fit and Secondary Business Potential, respectively.
The resource usage cj
represents the cost of securing the patent and the total cost is
constrained by a budget
B that is not to be exceeded. The patent-specific costs are
constructed so that they
represent the “useful” geographical patenting area for each
patent (project). In other
words, for the given cost (i.e. geographical area) of each
patent it is profitable to
secure the patent if it is decided to be included in the
portfolio. This kind of
formulation helped to keep the number of projects reasonable
with regards to
computational time (if patents in different geographical extent
were modelled as
mutually exclusive projects, with 5 extents the number of
projects would have been
already 105). The criteria weights were obtained both as
ordinary preference
14
-
statements (most important, second most important etc) and
precise point estimates.
This allowed testing the model with different set-ups to see if
and how the results are
affected.
The criterion-specific score estimate intervals were obtained
after careful
consideration by the decision makers. The intervals represent an
area of assumable
but tolerable variation about uncertain future performances. In
other words, from
strategic perspective, it is unrealistic to imagine that a
project’s performance with
regards to given criteria would not lie within the suggested
interval.
4.3. Experimental Design
A computational study of case was performed with RPM-Solver – a
software
developed at the Systems Analysis Laboratory in Helsinki
University that uses RPM
framework to calculate non-dominated portfolios and different
robustness measures
for individual portfolios. A short description of the software
is presented in
www.rpm.tkk.fi.
The first test run was conducted with the data in the exact form
it was given (see
appendices, table 1). However, given that the score intervals
were rather
homogeneous in width and that levels of scores did not vary
much, the computational
time grew unrealistic. Therefore several slightly simplifying
experimental designs
were developed to analyse the case in numerical terms. Three
different score intervals
representing criterion-specific variation (that is, uncertainty)
of 5% (low), 10%
(medium) and 15% (high) were constructed. All cases were run
with three different
budgets (low, medium, high). Criteria weights were modelled as
ordinary preference
statements (RICH-method), with the restriction that all of the
weights were at least
one fifth of the average of the weights.
4.4. Results
The results are analysed mainly by comparing core indices.
Although some test runs
produce promising results, in general the recommendations are
not so obvious. In fact,
in most cases almost all projects fall into borderline category
meaning that the model
is not able to produce many core or exterior projects that would
reduce the amount of
further consideration. For example, the results of a “middle
case” test run (medium
15
http://www.rpm.tkk.fi/
-
score variation and budget) are presented in figure 2. Only one
project appears as so-
called core project and another as exterior project.
Figure 2. Core indeces with a score interval of 10% and a budget
constraint of 107.5
In general the projects 23, 26 and 37 perform very well in most
of the test runs.
Similarly, projects 13 and 33 fall into exterior project class
in most cases. The model
therefore helps to reduce the amount of further investigation by
about 25% in general.
Effect of budget constraint
Changing the budget constraint does not seem to change the
number of core and
exterior projects meaningfully (see appendices, figures 2 and
3). On the other hand,
borderline projects’ core indices get bigger when increasing the
budget.
16
-
Effect of score intervals
Increasing the score intervals seems to have a dramatic effect
in not only decreasing
the number of exterior projects but also decreasing the number
of core projects. In
fact, regardless of the budget constraint, a score variation of
15% yields neither core
nor exterior projects, (see appendices, figure 3) whereas a
score variation of 5%
produces both core and exterior projects (see appendices, figure
4). The reason
translates back to the data being quite homogeneous. For more
variable data a 15%
score interval would produce clearly better results.
17
-
5. Discussion and Conclusions
Although the numerical results appear somewhat inconclusive, it
should be kept in
mind that the model in this study was somewhat simplified,
partly because of lack of
available data and resources needed with regards to the scope of
the course. In
addition, the data was perhaps not gathered in the way that
would benefit the model
best (that is, the data was not enough variable). Therefore more
important than
concentrating on the results of the test runs is to consider how
to improve and extend
the model. Some ideas of these ideas are discussed below.
Firstly, the applied model in this paper simplifies the problem
from a geographical
perspective. The patent-specific costs (annual payments) were
formed to represent a
“useful” patenting area. In other words, the cost of a
particular patent, i.e. the sum for
the useful patenting area, is such it that guarantees profitable
protection of the patent
should it be patented. In reality, each patent has different
cost and criterion-specific
performance level depending on geographically how widely the
patent is protected. A
full examination of the geographical issue would be to treat
each project in each
geographical area independently and model them as mutually
exclusive projects. For
example, if projects 1,2,3,4 and 5 represent the same patent in
different geographical
extent, an additional logical constraint tells that only one of
these 5 projects can be
selected for any portfolio. However, because this approach would
require the
collection of many times the original data it was decided to
leave it outside this study.
Another curiosity to consider is the assumption of the applied
model that there are no
interdepencies between the individual projects. However, in
reality such
interdepencies exist. For example, it is quite likely that two
related projects, if both
started, produce together greater overall value than the sum of
their individual values.
In other words, these projects have positive synergies. In RPM
framework, synergies
can be modelled by adding aditional project for which there are
no costs but a value
equal to the supplemental synergy value of the two projects if
both started. This
additional “synergy project” is thus started if and only if both
(or all) of the
underlying projects are started.
Even further modelling challenge arises from the issue that
R&D company may have
rights to patents that they do not own. In other words, one
company owns the patents
and another uses them for some cost, and vice versa.
18
-
With regards to numerical results, the study shows that although
there is no need to
provide accurate data, too homogenous data leads to results that
do not say much.
Emphasising the fact that preference statements about weights
and score intervals for
criterion-specific performance levels are enough may lead to a
situation where the
given information is not diverse enough. For example, if the
score intervals are close
to the same width for many project-criterion combinations, even
though the levels of
the intervals vary, the model is not able to find a clear set of
core or exterior projects.
This finding is important when communicating how the decision
maker should gather
the information needed.
In conclusion, RPM framework provides a flexible, easily
understandable but process-
wise challenging framework for portfolio analysis. The real
merits of RPM seem to be
its low demand on data and the ability to simply communicate all
phases of the
modelling process to all decision makers. The implementation of
RPM is a multi-step
challenge that requires careful planning. The study proved a
useful introduction to
these challenges and opportunities in applying RPM to real
cases.
19
-
6. References
1 Antila, M., Beletski, A., Isola, T., Janhonen, H., Leino, L.,
(2005). Seminar on
Case Studies in Operational Research, Final Report, Systems
Analysis Laboratory,
Helsinki University of Technology,
http://www.sal.tkk.fi/Opinnot/Mat-
2.177/projektit2005/Loppuraportti_Asperation.pdf
2 Asian Productivity Organization, (2004). Intellectual Property
Rights, Report of
the APO Symposium Intellectual Property Rights 11-14 November
2003,
Bangkok, Thailand, ISBN: 92-833-7020-1.
3 Asperation Oy website, (2005), http://www.asperation.com/.
4 Ernst & Young and Ementor Management Consulting, (2000).
Management and
evaluation of patents and trademarks, Consultant’s Analysis
Report for the Danish
Patent and Trademark Office.
5 Investopedia website, (2005),
http://www.investopedia.com/terms/c/capitalbudgeting.asp.
6 Liesiö, J., (2004). Robust Multicriteria Optimization of
Project Portfolios,
Master’s Thesis, Systems Analysis Laboratory, Helsinki
University of
Technology.
7 Liesiö, J., Mild, P., Salminen, L., Ikonen, L., (2004). Final
Report, Seminar on
Case Studies in Operational Research, Systems Analysis
Laboratory, Helsinki
University of Technology, http://www.sal.tkk.fi/Opinnot/Mat-
2.177/projektit2004/Loppuraportti_Inframan.pdf.
8 Liesiö, J., Mild, P., Salo, A., (2006). Preference Programming
for Robust
Portfolio Modeling and Project Selection, manuscript, European
Journal of
Operational Research (forthcoming).
9 Luenberger, D., (1998). Investment Science, Oxford University
Press.
10 Multiple Criteria Portfolio Analysis research project’s
website (2005),
http://www.rpm.tkk.fi/.
11 National Board of Patents and Registration of Finland,
(2005), http://www.prh.fi/.
12 Nielsen, P.-E., (2004). Evaluating patent portfolios – a
Danish initiative, World
Patent Information, Vol. 26, Issue 2.
20
http://www.sal.tkk.fi/Opinnot/Mat-2.177/projektit2005/Loppuraportti_Asperation.pdfhttp://www.sal.tkk.fi/Opinnot/Mat-2.177/projektit2005/Loppuraportti_Asperation.pdfhttp://www.asperation.com/http://www.investopedia.com/terms/c/capitalbudgeting.asphttp://www.sal.tkk.fi/Opinnot/Mat-2.177/projektit2004/Loppuraportti_Inframan.pdfhttp://www.sal.tkk.fi/Opinnot/Mat-2.177/projektit2004/Loppuraportti_Inframan.pdfhttp://www.rpm.tkk.fi/http://www.prh.fi/
-
13 Perlos Oyj website (2005), http://www.perlos.com/.
14 Salo, A., Hämäläinen, R.P, (2001). Preference Ratios in
Multiattribute Evaluation
(PRIME) – Elicitation and Decision Procedures under Incomplete
Information,
IEEE Transactions on Systems, Man, and Cybernetics, Vol. 31, pp.
553 – 545.
21
http://www.perlos.com/
-
7. Appendices
Budget limits Importance 80 65 100 50 70 90 125 2 4 1 5 3
Project Annual payment Necessity Strategic fit
Primary business potential
Secondary business potential
Technical coverage
002ab 10 4 4,5 3,5 4,5 3 4,5 3 4 2,5 3,5 007 7,5 4 5 2,5 3,5 2 3
3 4 2 3 011 7,5 3,5 4,5 3 4 2 3 1,5 2 3 4 012 7,5 3,5 4,5 3 4 2 3
1,5 2 3 4 013 10 4 4,5 2,5 3,5 2,5 3,5 2,5 3,5 1,5 2,5 014 10 4 4,5
2,5 3,5 3,5 4,5 3 4 1,5 2,5 020 7,5 4 4,5 3,5 4,5 2 2,5 2 3 1,5 2,5
021 7,5 2 3 3 4 2 3 1,5 2 1,5 2,5 022 5 2,5 3,5 2,5 3,5 2 2,5 1,5
2,5 3,5 4,5 023 7,5 3 4 3 4 3,5 4,5 3 4 2,5 3,5 026 7,5 3 4 3,5 4,5
3 4 1,5 2,5 3 4 027 7,5 4 4,5 3,5 4,5 1,5 2,5 1,5 2,5 3,5 4,5 028
10 4 5 4 4,5 3,5 4,5 3 4 2 3 029 10 4 5 4 4,5 3 4 2,5 3,5 2 3 030
10 4 5 3,5 4,5 2,5 5 2 4,5 3 3,5 033 10 4 4,5 4 4,5 1,5 4,5 2 3 1,5
3 034 5 1,5 2,5 2,5 3,5 1 3,5 1 3,5 1,5 3,5 035 10 4 4,5 2,5 3,5 3
4 3,5 5 3,5 4,5 036 10 4 4,5 3,5 4,5 3,5 4,5 3 4 3,5 4,5 037 5 3 4
2,5 3,5 2 3 2 3 3,5 4,5 038 10 3,5 4,5 3,5 4,5 3 4 2,5 3,5 3,5
4,5
Table 1. Original data. Note that the data was modified for test
runs
22
-
Figure 1. Core indeces with a score interval of 5% and a budget
constraint of 90
23
-
Figure 2. Core indeces with a score interval of 5% and a budget
constraint of 125
24
-
Figure 3. Core indeces with a score interval of 5% and a budget
constraint of 107.5
25
-
Figure 4. Core indeces with a score interval of 15% and a budget
of 107.5
26
Contents1. Introduction2. Patents as Intellectual Property2.1.
Background of Patenting2.2. Patent Portfolio Valuation
Approaches
3. Robust Portfolio Modelling – a Theoretical Framework3.1.
Representation of Additive Value3.2. Incomplete Information3.3.
Additional Constraints3.4. Non-dominated PortfoliosV\(p,w,v\) \(
V\(p’,w,v\) for all \
3.5. Additional Information3.6 Robustness
4. Case Study on Patent Portfolio4.1. Problem Statement4.2.
Modelling the Problem4.3. Experimental Design4.4. ResultsEffect of
score intervals
5. Discussion and Conclusions6. References7. Appendices