Yong Suk Lee Stanford University September, 2018 Working Paper No. 18-031 MODERN MANAGEMENT AND THE DEMAND FOR TECHNICAL SKILL
Yong Suk Lee Stanford University
September, 2018
Working Paper No. 18-031
MODERN MANAGEMENT AND THE DEMAND FOR TECHNICAL SKILL
! 1!
Modern Management and the Demand for Technical Skill
YONG SUK LEE*
Stanford University
September 18, 2018
Abstract This paper examines the relationship between modern management practices and the demand for different occupational skills utilizing a unique context in South Korea after the Asian financial crisis. Management practices in South Korea had traditionally emphasized the organizational harmony over individual performance, and firm growth over short-term profits. However, as South Korea opened up to foreign firms after the financial crisis, domestic firms started to adopt western or more “modern” management practices. Using the industry level variation in management practices generated by the average industry management index of five advanced economies (the US, Britain, France, Germany, and Italy), I find that modern management increases the demand for technical skill. Moreover, modern management practices help achieve various organizational changes that utilize information technology. I also find that performance measured as the return on asset increases with modern management practices, and document the complementarity between modern management practices and technical workers in increasing the return on assets. In short, this paper finds that modern management practices may increase the earnings difference between skilled - in particular, technically skilled - and unskilled workers. Keywords: modern management, management-technology complementarity, skill premium JEL Codes: M54, J31, J24
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!* Lee: Freeman Spogli Institute for International Studies, Stanford University, Encina Hall E309, 616 Serra St, Stanford, CA 94305, USA. Phone: 1-650-723-9741. Fax: 1-650-725-6530. Email: [email protected].
I thank Nick Bloom, Raffaella Sadun, Kathryn Shaw, Mitchell Hoffman, Shai Bernstein, Adriana Kugler, Chiaki Moriguchi, Mu-Jeong Yang, and participants at the Society of Labor Economist Meeting, Empirical Management Conference, Korea-America Economic Associations Workshop, and Stanford Economics Junior Faculty Workshop for comments.
! 2!
1. Introduction
The Asian financial crisis, which had originated in Southeast Asia in 1997, unexpectedly
and abruptly spread to South Korea, and by the end of that year a liquidity crisis was looming in
the then third largest economy in Asia. The International Monetary Fund (IMF) agreed to
provide an emergency loan package conditional on South Korea making progress on various
structural reforms surrounding financial openness and corporate governance. The domestic
market became more open to foreigners, and many US and European entities, taking advantage
of the deflated prices, invested in South Korean firms, equities, and assets (Kim 2006; Choi et al.
2007). Naturally, more foreigners were in managerial positions and western management
practices spread to many firms. Management practices in South Korea had traditionally
emphasized the organization over the individual worker. Individual salary contracts were rare
and bonuses were usually shared among the employees. Family owners had disproportionately
large control over shareholders and managers regarding firm operations. Firms tended to focus
on growth over short-term profits. However, many of these features changed around this period
as more firms adopted western or “modern” management practices that emphasize individual
performance and corporate profits.
Utilizing data from this unique period, this paper examines the relationship between
modern management practices and the demand for different occupational skills, and in particular
technical skill, in the manufacturing sector. The World Management Survey and research by
Bloom and Van Reenen (2007) have triggered an expanding volume of research that explores
how management practices affect the productivity and performance of firms and organizations.
However, there is relatively little research on how modern management practices affect the
demand for different occupational skills and, hence, inequality. Modern management practices
are a set of generally perceived best practices, such as, whether workers are incentivized and
compensated accordingly, whether targets are initially identified and later assessed, and whether
procedures to minimize production errors are in place. These practices could differentially affect
the demand for occupational skill groups, e.g., managers, technical workers, office workers,
production workers, etc., if modern management practices complement or substitute each
occupational skill to varying degrees.
However, estimating the impact of management practices on the demand for skill is
difficult because of endogeneity. Firms decide whether or not to adopt modern management
! 3!
practices based on the projected benefits and costs of adopting such practices, which are often
determined by unobservable firm specific characteristics. Without a randomized control trial or a
convincing quasi-experimental design it is difficult to estimate causal effects, but generating
exogenous variation in management practices is difficult. The goal of this paper is to holistically
show - through multiple correlational analyses, 2SLS regressions, and evidence of the
complementarity between modern management practices and technical skill – that modern
management practices play a role in increasing the demand for technical skill.
I adopt the questions developed by Bloom and Van Reenen (2007) and construct a
management index for South Korea using survey data from the early 2000s. In addition to firm
level variation, I use the industry level variation in management practices to examine the demand
for occupational skills. Management practices could vary across industries because of various
reasons, including the difference in technologies, multinational firm activity, labor unionization,
etc. As Figure 1 documents, there is substantial variation in management practices across
industries within countries. I construct an industry management frontier, i.e., the industry level
management practices of five advanced western economies, by averaging the industry level
management indexes of the US, Britain, France, Germany, and Italy using the World
Management Survey. Firms from this group of 5 (hereafter, G5) countries often invest and
operate in other countries. Hence, the industry management frontier is likely to be more
correlated with the management practices in countries with large western firm presence, which
was increasingly the case for post-financial crisis South Korea.
OLS regressions find that the management index is positively related to the employment
of skill - i.e., managers, technical workers, and office workers - relative to the employment of
production workers and simple manual laborers. The 2SLS regressions that use the variation
from the industry management frontier and further controls for industry level competition and
productivity find a positive impact of modern management practices on the employment of
skilled workers. The positive effect holds for relative wages as well. However, other industry
characteristics that affect the demand for occupational skill could still be correlated with the
industry management frontier. Technological change is considered a leading cause of the recent
rise in inequality and employment polarization (Acemoglu and Autor 2011; Autor and Dorn
2013), and researchers have found that technology and aspects of modern management, such as
decentralization or incentive-based human resource management, complement each other
! 4!
(Bresnahan, Brynjolfsson and Hitt 2002; Bloom, Sadun and Van Reenen 2012). Globalization,
corporate governance, and unionization could also be correlated with management practices and
influence the demand for different types of workers. When I control for technology using the
number of computers in the establishment, modern management increases the relative
employment and wage of technical workers only. Controlling for exports, corporate governance,
and union status, however, have virtually no effect on the estimates once technology is controlled
for. The empirical results suggest that modern management is likely one factor behind the
increase in the employment and wages of the high-skilled technical workforce. Finally, I show
that the complementarity between modern management practices and technical skill helps
achieve organizational change and improve firm performance. Such complementarity is likely
one of the underlying reason behind modern management’s demand for technical skill.
This paper is related to the literature that examines how organizational characteristics are
related to aspects of inequality. Caroli and Van Reenen (2001) find that delayering and multi-
tasking increases the demand for skill. Bandiera et al. (2007) and Lemieux et al. (2009) find that
performance pay increases wage inequality within firms. I believe this is the first paper that
examines how overall management practices, rather than a specific aspect of organizational
change, affect the demand for skill. The complementarity between modern management practice
and technical skill found in this paper is related to studies that show that the productivity gains
from technology differ based on the organizational characteristics of the firm. Bloom, Sadun and
Van Reenen (2012) find that information technology productivity is higher in firms that
implement incentive based human resource management, such as performance based pay and
promotion. Bresnahan, Brynjolfsson and Hitt (2002) show that productivity gains from
information technology are higher in more decentralized organizations.
Also related is the literature that examines labor market polarization by occupational skill,
where employment has been increasing in the high-skilled and low-skilled occupation categories
but decreasing in the middle-skilled occupation groups (Autor et al. 2003; Autor et al. 2006;
Autor and Dorn 2013; Michaels et al. 2014). In particular, Autor and Dorn (2013) show that the
reduced costs of routine and codifiable tasks from technological change disproportionately hurt
both blue- and white-collar workers in the middle-skill categories, relative to the low-skilled
service sector workers and high-skilled managerial, professional, and technical workers. I show
! 5!
that another factor - i.e., management practices - can explain employment polarization where
technical workers increase relative to unskilled workers.
Finally, this paper contributes to the expanding economics literature on management.
That literature has largely focused on the impact of management practices on the performance of
firms and organizations. For example, scholars have examined how management practices affect
firms (Bloom and Van Reenen 2011; Bloom et al. 2013), the health sector (McConnell et al.
2013; Bloom, Propper, Seiler, and Van Reenen 2015), schools and universities (McCormack et
al. 2014; Bloom, Lemos, Sadun, and Van Reenen 2015), and bureaucracies (Kahn et al. 2016;
Rasul and Rogger 2017). In this paper, I show that modern management not only improves firm
performance but also the earnings difference between the technically skilled and low skilled
workers.
2. The Context
The Asian Financial Crisis, which started with the sudden drop in the value of the Thai
baht in 1997, quickly spread to neighboring Southeast Asian countries, as well as to South
Korea.1 Multiple South Korean banks and corporations that took out short-term loans in foreign
currencies were on the verge of going bankrupt and South Korea was unexpectedly in need of
cash to provide short-term liquidity into its financial sector. The IMF agreed to provide an
emergency loan package of $58.4 billion, but payments were to be made conditional on South
Korea making progress on a wide range of structural reforms. South Korea without much
alternative agreed to the conditional loans and embarked on a series of radical reforms. South
Korea's equity market became more open to foreign investors. Constraints on the purchase of
domestic assets by foreigners were relaxed. Corporate boards were instituted to enhance
accountability and transparency at banks and large firms. At the same time, US and European
firms, taking advantage of the deflated prices, purchased and invested in South Korean firms and
assets (Kim 2006; Choi et al. 2007). Naturally, more foreigners were in managerial positions
and western management practices quickly spread across the economy.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!1 A large literature examines the causes behind the Asian Financial Crisis and why it spread so quickly to many countries, including South Korea, which was geographically detached and generally perceived as a more stronger economy compared to Southeast Asia. Potential causes for the financial crisis include weak corporate governance, moral hazard in the financial sector, speculative attacks on currencies by hedge funds, as well as the grouping of countries under the so called emerging market funds. (Krugman 1999, Faccio et al. 2001).
! 6!
Some indication of this can be seen in how pay practices evolved in South Korea. Most
firms in Korea paid their employees based on a set wage table before the crisis. That is, worker
pay was primarily determined by the number of years one worked at the firm, along with a few
observable characteristics, such as, education, gender, military experience, etc. The notion that
individual pay among workers in the same entering cohort could differ substantially based on
performance was alien to many workers and firms at that time. However, with the spread of
western management practices, firms started to adopt salary systems where individual pay would
be negotiated based on performance. Figure 2 illustrates the year firms first adopted a salary
system based on a survey of South Korean establishments.2 Most establishments started to adopt
a salary system soon after the crisis and the number peaks in 2000 during the height of the
structural reforms.
The delayering of organizations and the introduction of teams are also aspects of
organizational change in modern firms (Caroli and Van Reenen 2001). The team structure
reduces the hierarchy within organizations and help expedite the decision making process.
Moreover, evaluating worker contribution is easier in smaller teams, since performance can more
easily be tracked within small groups. Figure 3 presents the year team systems were introduced.
It exhibits a similar pattern to Figure 2. The year teams were introduced also peaks at 2000. As
Figures 2 and 3 illustrate, two features often associated with modern management practices -
performance pay and delayering - were widely instituted in Korea soon after the financial crisis.
The share of occupation groups in the economy also changed soon after the financial
crisis. Figure 4 presents the employment patterns of four occupational skill groups - (a) managers,
(b) technical and professional workers, (c) office, service, and sales workers, and (d) production
and simple task workers - for all South Korean firms with 5 or more employees. The
employment dip in 1998 is the aftermath of the financial crisis, which negatively affected overall
employment. The dip is particularly sharp for the production and simple task workers and their
employment level does not reach its pre-crisis level until 2003. On the other hand, employment
of the technical and professional workers jumps after the crisis. The number of managers jumps
after the financial crisis as well, which is more clearly illustrated in Figure 6.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!2 Figures 3 and 4 are based on the workplace survey used in the empirical analysis of this paper. I provide more details on the survey in the following data section.
! 7!
Figure 5 presents the employment dynamics of each occupation group in terms of its
share relative to total employment. The share of production and simple task workers decrease by
about 10 percentage points from 50% to 40% after the crisis, reaching a new steady state. The
relatively lower skilled white-collar workers, i.e., office workers and service and sales workers,
also drop to a slightly lower steady state. On the other hand, the share of technical and
professional workers jumps to a higher steady state. Manager share also increases after the crisis,
which again is better illustrated in Figure 6. Figures 4 through 6 suggest that the demand for
managers, technical and professional workers may have increased after the financial crisis.
Though these figures are purely descriptive, they do hint that modern management practices may
be related to the composition of the occupational skill groups in the economy.
One thing to note is that there were shifts in the institutional arrangement related to South
Korea’s labor market after the Asian Financial Crisis. South Korea’s labor market was quite
inflexible. Firms generally hired workers at the entry level, and maintained them until retirement.
Labor laws made layoffs extremely difficult and hence the notion of life-time employment was
widely practiced. However, the massive bankruptcies and lay-offs following the Asian Financial
Crisis and the structural changes requested by the IMF resulted in labor market regulation
changes that eased some of the inflexibilities. In particular, the regulation on how long
businesses could use temporary workers was relaxed. Despite the effort to increase labor market
mobility, however, laying-off workers remained quite difficult and costly. Hence, the major
change in the labor market was the increased share of non-regular workers (temporary and part-
time employees). Such change was also accompanied by the increase in the gap of earnings,
especially benefits, between regular and non-regular workers. Such increase in the duality of the
labor market was often manifested through sub-contracting. For instance, when an auto
manufacturer needs to increase labor input to meet demand, it would increase temporary workers,
primarily by sub-contracting to firms that maintain a pool of temporary workers. Since firing
regular workers is costly and the cost of temporary workers is lower, companies maintain labor
flexibility and keep costs down by sub-contracting on a needs basis. The current paper focuses
on management practices and employees within the firm. The impact of modern management on
! 8!
temporary workers outside of the firm may be an important phenomenon, but in this paper those
effects are not captured and I focus on the within firm variation.3
3. Data
This paper’s empirical analysis uses a nationally representative workplace survey that
samples all South Korean establishments with 30 or more employees in 2002 and 2003. The
survey was conducted by the Korea Labor Institute as a pre-survey to the biannual Workplace
Panel Survey that starts in 2005. The 2002 and 2003 survey collected data on when firms first
adopted certain management practices after the financial crisis. It also provides detailed industry
classification, enabling it to be merged with the World Management Survey. These questions
were not asked in the main biannual panel survey that started several years later.4
The survey interviews human resource managers and employee representatives, and asks
about management practices, compensation policy, labor relations, and worker benefits. I
benchmark the World Management Survey to construct South Korea’s management index based
on questions that ask about employee evaluation, operations, and human resource management
strategies.5 These questions are listed in Table 1. Evaluation questions ask how important
individual performance is in promotion decisions and whether the establishment implements
Management by Objective (MBO) practices, where individuals set goals at the beginning of the
year and are evaluated based on performance on these goals at the end of the year. Operations
related question is whether the firm implements Six Sigma, which is a set of practices that aims
for process improvement.6 Human resource management questions ask about the organization’s
emphasis on the individual versus the team, performance versus harmony, and how flexible the
organization can hire and fire workers. I standardize each response and take the sum to construct
a standardized management index.
In addition, the survey provides the number of employees by occupational skill groups,
i.e., managers, technical workers, office workers, service and sales workers, production line
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!3 The increasing labor market duality, i.e., the divergence of regular workers and non-regular workers (temporary workers and part-time workers) is an important characteristic of the South Korean labor market that warrants separate examination. 4 More information on the survey and the data can be accessed at the Korea Labor Institute’s website 5 The World Management Survey does not yet cover South Korea. 6 One thing to note is that these management concepts were all devised or popularized in the west. MBO was popularized by the management guru Peter Drucker in 1954. Six Sigma was introduced by Motorola and Jack Welch used it as a central business strategy of General Electric.
! 9!
workers, and simple manual laborers. The manager category does not include executives, such as
the owner, CEO, or top executives. Technical workers are employees involved in technology,
professional, or research and development work. They include (1) science, technology, and
engineering professionals, (2) finance, accounting, and insurance professionals, and (3)
administrative and legal professionals. Office workers are other desk workers that are not
managers or technical workers. Simple manual laborers are low-skilled non-production line
workers, such as cleaning and janitorial personnel. I construct the relative demand of four
occupational skill groups, i.e., managers, technical workers, office workers, and service and sales
workers, relative to unskilled workers. Unskilled workers are comprised of the production line
workers and simple manual laborers.
I also construct the relative earnings. However, information on earnings is more limited
in the survey. The survey collects the average earnings of all six occupational skill groups, but
only for temporary workers in a subsample of the establishments. The survey does collect data
on the average earnings of managers and all non-managers combined, i.e., the rest of the workers,
for each establishment. I use these information to impute the average earnings of each
occupational group for each establishment. I first calculate the industry level relative earnings
based on the temporary worker information. I then use the number of employees in each
occupational skill group and the average earnings of managers and non-managers to impute the
average earnings of each occupational group for each establishment.7 Such imputation assumes
that the differences between temporary and permanent workers are constant across the non-
managerial occupations. In Appendix Table 1, I use data from the Korean Labor Income and
Panel Study (KLIPS) to examine the earnings differences between temporary and permanent
(regular contract) workers across non-managerial occupations. Earnings are the total
compensation that employees receive from the company and include wages/salaries, bonuses,
and benefits. In particular, I regress log earnings on the occupation group dummies, a dummy for
temporary work status, and the interactions terms between temporary work status and the
occupation group dummies. I also control for individual characteristics, region, industry, and
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!7 For example, I impute the average wage of technical workers as below
!!"#! =!!"!!!"#"$%&'×!!"!!!"#"$%&'
!!"#! + !!×!!!"#!!"#!!"#!!!"",!"#$,!"#$,!"#$
where ! denotes wage, !!"# the industry average wage from the temporary worker information, and ! employment, the subscripts indicate the occupational skill groups.
! 10!
business characteristics. Since KLIPS is an individual level panel, I also present results that use
individual fixed effects in lieu of the individual characteristics. The omitted category is the sales
and service workers, so the coefficient estimates indicate the earnings differential between each
occupation group relative to this omitted category. The coefficient estimate on each interaction
term indicates the difference between the earnings differential between temporary and permanent
workers for each occupation group and the earnings differential between temporary and
permanent workers for service and sales workers. If the differences between temporary and
permanent workers are constant across the non-managerial occupations, the estimates on the
interaction terms would not be statistically different from zero. Indeed none of the interactions
terms are significantly different from zero in Appendix Table 1.
Despite the paper’s focus on manufacturing establishments, there are establishments with
unusually high ratios of skilled to unskilled workers, often due to the very low number of low-
skilled workers. For example, quite a few establishments have zero or only one production
worker. I drop establishments with zero or one production worker. I also drop establishments if
the ratio of any skilled occupation group to unskilled workers is greater than 10.8 The final
sample size is 1,430 establishments and Table 2 presents the descriptive statistics.
Establishments on average have 390 employees. Production workers comprise 52% of the
employment, followed by office workers at 13%, and simple manual laborers at 12%. Managers
comprise 10 % of the employment and technical workers 7%. Service and sales workers
comprise the lowest share at 6%. The earnings of managers relative to unskilled workers in log
differences, i.e., ln(manager earnings) – ln(unskilled earnings), is 0.5. The relative premium is
0.46 for technical workers, 0.01 for office workers, and 0.06 for sales and service workers. In
other words, manager pay is over 50% higher than that of unskilled workers and is closely
followed by technical workers. The average pay of office workers is very close to that of
unskilled workers.
4. The Empirical Framework
I consider the following equation to examine the relationship between modern
management practices and the demand for different occupational skill groups:
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!8 There were 280 establishments with zero production worker, and 10 establishments with exactly one production worker. This process in total drops 308 observations. !
! 11!
!!"# = !!!!" + !!"Π+ !!"#. (1)
! indexes the occupational skill group, ! the establishment, and ! the industry. !!"# is
establishment i's share of employees in occupation group ! relative to the unskilled occupation
group. !!" is the set of control variables that include establishment age and log employee size, as
well as region dummies, and a year dummy for 2003 (since the survey spans two years). The
dummy variables capture unobserved region and time characteristics that affects the demand of
each occupational skill group. !!" is the establishment's management index. In addition to
examining labor share !!"# , I examine earnings share !!"# , i.e., the relative earnings of
occupational skill group ! relative to that of the unskilled occupation group. By examining both
relative employment and earnings, one can check whether the responses are driven by the
demand for or supply of the different occupational skill groups. If modern management increases
the relative demand for a certain occupational skill, then both the relative employment and
earnings of that occupational skill group would increase.
The main challenge in examining the impact of modern management practices on the
demand for occupational skill groups is the fact that management practices are endogenous.
Firms weigh the benefit against the cost of introducing modern management practices and
choose whether or not to adopt new practices. There would be firm specific characteristics that
factor into that decision, many of which the econometrician may not be able to control for. In
addition to firm level OLS regressions of equation (1), this paper examines the industry level
variation in management practices and the demand for occupational skill. Modern management
practices would vary across industries due to various reasons that render modern management
practices more beneficial to some industries relative to others.9 Industries that often use
complicated machineries and production processes may reap the benefits of modern management
more than others. Industries that use a multitude of intermediate goods sourced from many
regions may benefit more from modern management than industries that do not. Multinational
firm activity or labor unionization could also generate industry level variation in management
practices. As Figure 1 illustrates there indeed is substantial variation in management practices
across industries within each of the five advanced economies. I construct the industry !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!9 The idea of utilizing industry level variation for identification has been used to examine the impacts of financial dependence. Rajan and Zingales (1998) identify the industry’s need for external finance to examine the impact of external financing on growth. Manova (2004) examines how financial market imperfections distort trade across industries and countries.
! 12!
management frontier by averaging this group of 5 - the US, Britain, France, Germany, and Italy-
countries' industry level management indexes from the World Management Survey. These five
economies are generally at the frontier of production technologies and business practices.
Modern management practices were widely established among firms in these countries and many
multinational firms originate from these advanced economies. Moreover, firms and funds from
these western economies invested in South Korea’s firms and assets after the financial crisis.
Hence, the industry management frontier is likely to be more correlated with the management
practices in countries with large western firm presence, which was increasingly the case for post-
financial crisis South Korea.
Figure 7 presents the correlation between the industry management frontier (the G5
industry management index) and South Korea’s management index. The x-axis indicates that
there is substantial industry level variation in management practices in the Group of 5 countries.
Also, there is a positive correlation between the industry management frontier and South Korea’s
management index. Table 3 presents the regression results of this relationship. In panel A of
column (1), South Korea’s management index is regressed on the G5 industry management index
and the control variables. Standard errors are clustered at the industry level. A standard deviation
increase in the G5 index increases South Korea’s management index by about 0.13 standard
deviation, and the effect is statistically significant at the 1 percent level. The other panels in
column (1) examine the correlation using each country’s industry management index. The
estimates are similar across all 5 countries. I next split Korea’s management index into two
components – an industry average and the establishment specific residual– and examine how
each component is related to the foreign management indexes. The foreign industry level
management indexes are strongly correlated with South Korea’s industry level index but not with
the establishment specific residual. Finally, I examine the scatterplots between the employment
share of each occupational skill group and the G5 industry management index in Figure 8. The
positive relation between the share of technical workers and the G5 index, and the negative
relation between the share of production workers and the G5 index particularly stands out. This
potentially suggests that modern management practices may be complementary to technical skill
while substituting production workers.
These evidence all illustrate that the G5 index can potentially serves as an instrument for
the Korean management index. The other condition for the G5 index to be a valid instrumental
! 13!
variable is that there are no additional omitted variables, i.e., the G5 index is not correlated with
the error term in equation (1). It is worth emphasizing that the G5 index must be uncorrelated
with the error term conditional on the control variables in the regression. Hence, in the
sensitivity analysis I include potential omitted variables - such as, industry level characteristics,
variables that proxy for technology, trade, corporate governance and labor unions - to the base
regression and examine the consistency of the estimates.
5. The Empirical Results
5.1 Modern management and the level of employment and wages by skill groups
I first examine how modern management is related to the level of employment and
earnings of each occupational skill group in Table 4. The base OLS results in Panel A indicate
that modern management is positively related to the number of managers, technical workers, and
office workers, but negatively related to the number of production workers. On the other hand,
modern management is positively and significantly related to the earnings of all six occupational
skill groups.
In Panel B, I separate the management index into two components - the industry average
and the establishment specific residual component. In general, employment is more strongly and
significantly related to the industry average than the residual component. Establishments in
industries with a higher average management index have significantly more managers, technical
workers, and office workers, but less production workers. On the other hand, earnings tends to be
more significantly related to the establishment specific residual. An interesting feature from the
earnings result is that the coefficient estimates on the establishment specific residual index for all
six occupation groups are quite similar, whereas the estimates on the industry average index are
substantially more varied. One explanation may be that the establishment specific residual index
is closely related to the establishment's total productivity that affects all workers similarly,
whereas the industry average index better captures the differential demand for the occupational
skill groups.
Panel C examines how the industry management frontier, i.e., the G5 industry
management index, is related to the level of employment and earnings. The industry
management frontier is most strongly and positively related to the number of technical workers,
as well as managers, but negatively related to the number of production workers. The industry
! 14!
management frontier is also positively and significantly related to the earnings of skilled workers,
especially, technical workers and office workers.
Finally, Panel D presents the regression results that use the variation in South Korea’s
management index generated by the industry management frontier, i.e., the 2SLS results. As
Table 3 illustrated the first stage is strong with an F-statistic of 13.47. Overall, the 2SLS results
are qualitatively similar to the reduced form results of Panel C. The evidence is consistent with a
positive impact of modern management practices on the employment and earnings of skilled
workers, notably, the technical workers, managers, and office workers.
5.2 Modern management and the relative demand for skill
Table 5 examines the relative demand for the different occupational skill groups. The
dependent variables in columns (1) through (4) are the natural logarithms of the employment of
managers, technical workers, office workers, and sales and service workers relative to the
employment of unskilled workers, i.e., production workers and simple manual laborers. Columns
(5) to (8) present the relative earnings results for the corresponding occupation groups. The OLS
results in Panel A indicate that modern management is most strongly associated with the relative
demand for technical workers and office workers. For managers the estimate is significant only
for relative employment, and for sales and service workers only the relative earnings result is
significant.
As before, I separate out the industry average and establishment specific residual
management index in Panel B. Again the industry average, compared to the establishment
specific residual, has a substantially larger relation with the relative employment of each skill
group, and the estimate on the industry average is largest for the technical workers. The relative
earnings results are not as strong statistically, except for one estimate on the industry average
index for sales and service workers. However, similar to before the magnitude of the estimate for
technical workers is larger than that for managers or office workers. There is no establishment
specific residual effect on the relative earnings of any skill group.
Finally, the reduced form results of Panel C and the 2SLS results of Panel D all return
positive and significant estimates for the relative employment shares of skilled workers. Similar
to previous results, the effect is strongest for the technical workers. Moreover, the positive
estimates for both relative employment and relative earnings suggest that the effects are driven
! 15!
by modern management’s demand for skilled workers, rather than the relative supply of skilled
workers. The coefficient estimates for the technical worker group is consistently the largest in
Table 5, regardless of whether one examines relative employment or relative earnings, or
whether one examines the OLS result, reduced form result, or 2SLS result.
The dataset used in this paper does not have information on hours worked. However,
there could be substantial variation in hours worked across the different occupation groups
across industries. To better account for hours worked in the analysis, I use the KLIPS data to
calculate the average work hours of each occupation group by industry and weight employment
and normalize earnings. However, individuals in KLIPS do not populate all manufacturing
industries and occupation groups, and hence, there are many missing weights. In such cases, I
use the average across industries as the weights for each occupation group. I rerun the
regressions in Table 5 and present the results in Appendix 2. The results are overall quite similar
to that of Table 5.
5.3 Sensitivity analysis and controlling for technology
There are likely unobserved factors that correlate with the industry level management
index and the demand for skill in the Table 5 Panel D results. Especially, there could be other
industry level omitted variables, such as industry level competition and productivity. I construct
the industry level revenue based Herfindahl-Hirschman index to control for the degree of
competition and concentration. I construct industry labor productivity to capture productivity
differences across industries. Table 6 Panel A presents the results when I include these industry
level variables. Appendix Table 3 presents the first stage regression. Industries that are more
competitive and more productive have significantly better management practices. In fact, the
first-stage statistic of the 2SLS regression actually increases to 15.54. However, including these
industry variables barely changes the 2SLS estimates in the second stage. The estimates on Table
5 Panel D and Table 6 Panel A are quite similar. The inclusion of these variables may not fully
alleviate omitted variable concerns, but the consistency of the 2SLS estimates is reassuring.
Technology adoption varies across industries and influences the demand for skill.
Moreover, there is evidence that organizational characteristics, such as delayering and
multitasking, are complementary to technology. The rest of Table 6 presents the 2SLS results
with additional controls for technology as well as other potential channels that could influence
! 16!
the demand for technical skill. In Panel B, I additionally control for the log number of computers
at the establishment to proxy for technology usage. The relative employment share results in
columns (1) through (4) indeed indicate that technology increases the demand for skill,
especially managers and office workers. The coefficient estimates on the management index
become smaller in magnitude, but only the estimate for the technical workers remains significant.
On the other hand, the relative earnings results in columns (5) through (8) indicate that, the
number of computers is not significantly related to the relative earnings of skill groups, and if
any tends to reduce the earnings gap. However, the estimates on the management index, as well
as their standard errors become larger. Overall, Table 6 Panel B indicates that modern
management increases the demand for technical skill even when controlling for a proxy for
technology.
I further explore how controlling for exports, corporate governance, and labor unions
change the Table 5 Panel D results. I include the export share of revenue as an additional control
in Panel C. The literature has found exports to be associated with better quality and the demand
for skilled workers (Verhoogen 2008). However, the export share variable has no significant
relationship with the relative employment or wage share of all occupational skill groups when
management is controlled for. Moreover, the coefficient estimates on the management index is
very similar in terms of both magnitude and significance to the 2SLS results in Table 5.
I next examine whether controlling for an aspect of corporate governance, i.e., whether
the firm had a Chief Executive Officer (CEO) independent from the owner, changes the impact
of modern management on the demand for skill in Panel D. Many firms in Korea were directly
governed by their owners rather than independent CEOs. Holding management practices
constant, having a CEO tends to be negatively associated with the relative demand of skill.
However, the coefficient estimates on the management index barely changes. Panel E examines
whether having a labor union influences the relative demand for skill. I find no impact of union
status on relative employment and earnings. Again, the coefficient estimates on the management
index are virtually unchanged. Finally, I control for all of the above potential channels in Panel F.
All coefficient estimates are very similar to that from Panel B. This suggests that technology is
the critical factor that needs to be accounted for in the 2SLS regressions that uses the industry
management frontier as an instrumental variable. However, once technology is controlled for,
other factors identified in the literature are not major concerns, and the 2SLS regression that
! 17!
controls for technology suggests that modern management increase the demand for skill. The
magnitude of the effect is quite large. A 0.2 standard deviation increase in the management index
increases the earnings difference between technical workers and low-skilled workers by about a
full standard deviation.
While controlling for the number of computers at the establishment, I examine whether
modern management’s relative demand for technical skill is primarily due to the increase in the
demand for technical workers and/or the decrease in the demand for unskilled workers. Table 7
Panels A and B present the results on employment and wages. Modern management has a
significant and positive effect on the demand for technical workers. However, modern
management's effects on the employment and earnings of unskilled workers are not significant.
This indicates that modern management's relative demand for skill is primarily driven by the
increased demand for technical workers.
If establishments had temporary employees, those numbers were included in the
employment numbers. In Table 7 Panels C and D, I separate out the temporary employees from
the regular employees. First of all, column (7) shows no evidence that modern management
practices substitute regular employees for temporary employees. Columns (1) through (6)
indicate that modern management’s demand for the technical workers is for the regular
employees and not the temporary employees.
6. Organizational change and firm performance
6.1 Modern management and organizational change
Why might modern management increase the demand for technical skill? Caroli and Van Reenen
(2001) show that organizational delayering and multi-tasking increases the demand for skill. One
mechanism by which modern management could increase the demand for technical skill is
through the complementarity between modern management and technical skill to achieve various
organizational changes, e.g., digitization of administrative, financial, and production tasks,
decentralization and delayering of the organization, and information sharing.
As previously noted, technical workers perform a wide range of high-skill activities and
include not only the science, technology, and engineering professionals, but also finance,
accounting, and insurance professionals, and administrative and legal professionals. Modern
management practices aim to improve firm performance through better measurement of workers
! 18!
and tasks. Firms are adopting information technology into its organization to improve
performance and reduce costs. This is where modern management practices and technical skill
can complement each other, i.e., in achieving IT based organizational changes. For example,
information sharing within the firm can reduce redundant tasks and errors, and ultimately reduce
costs. To set up a system of information sharing, the firm needs IT technicians to set up and
manage the computer system, administrative professionals to implement the system, and
management practices that can measure and assess the workers and their tasks. In other words,
new organizational objectives like information sharing, decentralization, and digitization often
mobilize various technology, financial, and administrative professionals and accompany
management practices that emphasize measurement.
In Table 8, I examine whether modern management practices are positively related with
how well organizational change objectives that utilize information technology are achieved.
Several questions in the 2003 survey examine information technology use. I compile a set of
questions that asks how effective information technology is in helping accomplish various
organizational objectives. I regress each of these variables on the modern management index
including the same set of control variables used in the previous tables.
Establishments with better management practices utilize information technology to
digitize modular tasks, e.g., accounting, inventory management and sales management (Panel A),
employ Enterprise Resource Planning (ERP) or systems that comprehensively support workplace
tasks (Panel B), and Electronic Data Interchange (EDI) or systems that help connect and support
clients and suppliers (Panel C). Establishments with better management practices were also more
likely to say that information technology helped in setting up task force teams (Panel D),
delayering and decision making (Panel E), increasing multi-tasking (Panel F), encouraging
employee involvement in company operations (Panel G), and sharing information (Panel H).
And as more direct evidence, establishments with better management practices were more likely
to achieve what the company had intended to achieve through its investment in information
technology personnel (Panel I). The results in Table 8 suggest that modern management practices
complement technical skill to help achieve organizational changes that utilize information
technology. The ultimate goal of such organizational changes would be to improve firm
performance. In Table 9, I directly examine the complementarity between modern management
and technical workers in terms of firm performance.
! 19!
6.2 Modern management and firm performance
I first examine the relationship between modern management and firm performance
measured as the return on assets (ROA). ROA is constructed as net income, i.e., profits post-tax,
divided by total assets. ROA is one measure of firm performance and indicates how well
management is utilizing all of the company’s resources to generate profit. Table 9 columns (1)
and (2) present the OLS results where I include the same set of control variables as before. The
estimates on modern management are positive but not statistically significant. When I instrument
using the industry management frontier in columns (3) and (4), modern management positively
and significantly increases the ROA. In column (3), A standard deviation increase in the
management index increases the ROA by 0.304, and the impact is statistically significant at the
10% level. Given that the average ROA is about 0.12 with a standard deviation of 0.57, this
amounts to more than half a standard deviation increase. Once I control for the number of
computers in column (4) a standard deviation increase in the management index increases the
ROA by 0.64, which is greater than a full standard deviation, and the impact is statistically
significant at the 5 percent level.
Finally, I examine the complementarity between modern management and technical
workers in columns (5) and (6). I run an OLS regression that additionally includes the log
number of technical workers, and the management index interacted with the log number of
technical workers. The coefficient estimates on the interaction term are positive and statistically
significant and are similar regardless of whether I control for log computers in column (6).
Overall, Table 9 indicates that modern management and firm performance are positively related
and that there is complementarity between modern management and technical workers.
7. Conclusion
This paper examined the relationship between modern management practices and the
demand for different occupational skills. Focusing on the industry level variation in management
practices generated by the industry management frontier, i.e., the average industry management
index of 5 advanced economies (the US, Britain, France, Germany, and Italy), I examine how
modern management practices affect to the demand for managers, technical workers (R&D
workers and technicians), office workers, sales and service workers relative to the unskilled
! 20!
production workers and simple manual laborers. As widely found in the literature, controlling for
technology is important for the analysis. I find that modern management has a positive effect on
the demand for workers with technical skill, but has relatively little effects on the other
occupational skill groups. I show that the complementarity between modern management
practices and technical skill helps achieve organizational changes that utilize information
technology. I also find that performance measured as the return on assets increases with modern
management practices, and that modern management practices complement technical workers in
increasing the return on assets. A relatively large literature finds that modern management
practices increase the efficiency of firms and organization. However, what this paper finds is that
such practices may increase inequality between skilled - in particular, technically skilled - and
unskilled workers.
The complementarity between modern management practices and technical workers
suggests why modern management might increase the demand for technical skill - the marginal
productivity of technical workers might be higher than that of other occupational skill groups,
and modern management practices are better at identifying that. Firms adopt modern
management practices to better monitor and track the performance of workers. These firms are
making the additional effort to identify the types of workers that generate relatively more value
to the firm, and eventually hire and pay more for such workers. If workers with technical skill
create more value to the firm, by contributing to organizational change objectives and ultimately
firm performance, firms that implement modern management practices would more likely
increase the demand for technical workers relative to the other occupational skill groups. It
seems that more firms around the world are adopting modern management practices, either
because of guidance by foreign investors, multinational firms, management consultancy, MBA
education, and even international organizations like the World Bank. Future research that further
examines how modern management practices increase the demand for technical skill using
detailed data in other settings could help shed light on how labor market inequality and
polarization may evolve in the future.
! 21!
References
Acemogul, Daron and David Autor. 2011. “Skills, Tasks, and Technologies: Implications for Employment and Earnings.” In Handbook of Labor Economics, vol. 4, ed. David Card and Orley Ashenfelter, 1043-1166. Amsterdam: Elsevier.
Autor, David and David Dorn. (2013). "The Growth of Low-Skill Service Jobs and the Polarization of the U.S. Labor Market," American Economic Review, 103(5): 1553-97.
Autor, David, Frank Levy, and Richard Murnane. 2003. "The Skill Content of Recent Technological Change: An Empirical Exploration." Quarterly Journal of Economics, 118(4): 1279-1333.
Autor, David H., Lawrence F. Katz and Melissa S. Kearney. 2006. "The Polarization Of The U.S. Labor Market," American Economic Review 96(2): 189-194.
Bandiera, Oriana, Iwan Barankay, and Imran Rasul. 2007. Incentives for Managers and Inequality among Workers: Evidence from a Firm-Level Experiment. Quarterly Journal of Economics, 122 (2): 729-773 Bloom, Nicholas and John Van Reenen (2007). “Measuring and Explaining Management Practices Across Firms and Countries.” Quarterly Journal of Economics, 122, 1351–1408.
Bloom, Nicholas and John Van Reenen (2011). “Human Resource Management and Productivity.” In Handbook of Labor Economics, Vol. 4B, edited by Orley Ashenfelter and David Card. North-Holland, Chapter 19, pp. 1697–1769.
Bloom, N., Sadun, R. and van Reenen, J. (2012). ‘Americans do IT better: US multinationals and the productivity miracle’, American Economic Review, 102(1): 167–201. Bloom, N., Eifert, B., Mahajan, A., Mckenzie, D. and Roberts, J. (2013). ‘Does management matter: evidence from India’, Quarterly Journal of Economics, vol. 128(1), pp. 1–51. Bloom, N., Propper, C., Seiler, S. and van Reenen, J. (2015). ‘The impact of competition on management practices in public hospitals’, Review of Economic Studies, vol. 82(2), pp. 457–89. Bloom, N., Renata Lemos, Raffaella Sadun and John Van Reenen. 2015. “Does management matter in schools?” Economic Journal, 125: 647-674. Bresnahan, Timothy, Erik Brynjolfsson, and Lorin Hitt (2002). “Information Technology, Workplace Organization and the Demand for Skilled Labor: Firm-Level Evidence.” Quarterly Journal of Economics, 117(1): 339–376.
Caroli, Eve, and John Van Reenen. 2001. “Skill-Biased Organizational Change? Evidence from a Panel of British and French Establishments.” Quarterly Journal of Economics, 116(4): 1449-1492.
! 22!
Choi, Jongmoo Jay, Sae Woon Park, and Sean Sehyun Yoo. 2007. "The value of outside directors: Evidence from corporate governance reform in Korea." Journal of Financial and Quantitative Analysis 42(4): 941-962. Faccio, Mara, Larry H.P. Lang, and Leslie Young. 2001. “Dividends and Expropriation.” American Economic Review 91(1): 54-78. Khan, Adnan, Asim Khwaja, and Ben Olken. 2016. “Making Moves Matter: Experimental Evidence on Incentivizing Bureaucrats through Performance-Based Postings.” Unpublished. Kim, Kihwan. 2006. "The 1997-98 Korean financial crisis: Causes, policy response, and lessons." In IMF Seminar on Crisis Prevention in Emerging Markets. Krugman, Paul. 1999. The Return of Depression Economics. New York: W. W. Norton & Company. Lazear, Edward (2000). “Performance Pay and Productivity.” American Economic Review, 90, 1346–1361. Lemieux, Thomas, W. Bentley MacLeod, Daniel Parent. 2009. “Performance Pay and Wage Inequality.” Quarterly Journal of Economics, 124(1): 1-49. Manova, Kalina. 2013. “Credit Constraints, Heterogeneous Firms, and International Trade.” Review of Economic Studies 80: 711-744. McConnell, John, Rich Lindrooth, Doug Wholey, Tom Maddox, and Nicholas Bloom. 2013 “Management practices and the quality of care in Cardiac units.” Journal of the American Medical Association: Internal Medicine, 173(8):684-692. McCormack, John, Carol Popper and Sarah Smith. 2014. “Herding Cats? Management and University Performance.” Economic Journal, 124: 534-564. Michaels, Guy, Ashwini Natraj, and John Van Reenen. 2014. “Has ICT Polarized Skill Demand? Evidence from Eleven Countries over 25 Years.” Review of Economics and Statistics 96(1): 60-77. Rajan, Raghuram G. and Luigi Zingales. 1998. “Financial Dependence and Growth.” American Economic Review, 88(3): 559-586. Rasul, Imran and Daniel Rogger. 2017. “Management of Bureaucrats and Public Service Delivery: Evidence from the Nigerian Civil Service.” Economic Journal, doi:10.1111/ecoj.12418 Verhoogen, Eric. 2008. "Trade, Quality Upgrading, and Wage Inequality in the Mexican Manufacturing Sector." Quarterly Journal of Economics 123(2): 489-530.
! 23!
Figure 1. The distribution of industry level management practices across countries
Source: World Management Survey.
0.2
.4.6
.81
Density
-3 -2 -1 0 1 2US
0.2
.4.6
.81
Density
-2 -1 0 1 2Britain
0.5
1Density
-4 -2 0 2France
0.2
.4.6
.8Density
-2 -1 0 1 2Germany
0.2
.4.6
.81
Density
-3 -2 -1 0 1 2Italy
! 24!
Figure 2. Number of establishments that introduce performance pay by year
Figure 3. Number of establishments that introduce team systems by year
0!
50!
100!
150!
200!
250!
1977!
1979!
1982!
1983!
1986!
1987!
1988!
1989!
1990!
1991!
1992!
1993!
1994!
1995!
1996!
1997!
1998!
1999!
2000!
2001!
2002!
2003!
salary!system!start!year!
0!
50!
100!
150!
200!
250!
1977!1982! 1986!1988!1990!1992!1994! 1996!1998!2000!2002!
team!system!start!year!
! 25!
Figure 4. Employment by skill groups
Figure 5. Employment share by skill groups
0!
500000!
1000000!
1500000!
2000000!
2500000!
3000000!
1994! 1995! 1996! 1997! 1998! 1999! 2000! 2001! 2002! 2003!
Managers!
Technical!and!professional!workers!
Office,!service,!and!sales!workers!
ProducGon!and!simple!task!workers!
0!
0.1!
0.2!
0.3!
0.4!
0.5!
1994! 1995! 1996! 1997! 1998! 1999! 2000! 2001! 2002! 2003!
%!Manager!
%!Technical!and!professional!
%!Office,!service,!and!sales!
%!ProducGon!and!simple!task!
! 26!
Figure 6. Employment and employment share of the managers
Figure 7. Korea management index and the Group of 5 countries industry management index
170000!
220000!
270000!
320000!
370000!
420000!
0.036!
0.041!
0.046!
0.051!
0.056!
0.061!
0.066!
1994! 1995! 1996! 1997! 1998! 1999! 2000! 2001! 2002! 2003!
% Manager Managers
-20
24
Kore
a m
anag
emen
t ind
ex
-1 -.5 0 .5Group of 5 industry management index
! 27!
Figure 8. Employment share by skill group and the G5 industry management index
! 28!
Table 1. Questions used in constructing the management index for Korea Category Survey question
Evaluation
How important are individual performance evaluation scores in promotion decisions? (0 to 100 scale)
Do you implement Management by Objectives (MBO)? (Yes/No)
MBO is a practice where individuals set goals at the beginning of the year and are evaluated based on performance on these goals at the end of the year.
Operations Do you implement Sigma 6 practices? (Yes/No)
Human resource management
HRM’s main objective is to reduce labor costs, as opposed to promoting loyalty to the firm (1 to 7 scale) Hire and fire qualified personnel based on firm needs, as opposed to develop personnel by hiring new recruits and maintaining long-term employment. (1 to 7 scale) Utilize temporary workers as much as possible, as opposed to use permanent workers as much as possible. (1 to 7 scale)
HRM is based on individual performance, as opposed to teamwork. (1 to 7 scale)
HRM focuses on maximizing employee’s short-term performance, as opposed to long-term development and nurturing of employees
Notes: Management by Objectives (MBO) is a practice where employees set goals at the beginning of the year and are evaluated based on performance towards the set goals at the end of the year.
! 29!
Table 2. Descriptive statistics
Variable Mean Std. Dev. Min Max Obs
Mangement index 0.18 1.07 -2.10 3.53 1430
G5 industry management index 0.02 0.39 -1.02 0.61 1430
Total employment 390.14 1726.89 6 42150 1430
Manager share 0.10 0.08 0 0.5 1430
Technical (R&D and tech) worker share 0.07 0.09 0 0.63 1430
Office worker share 0.13 0.09 0 0.72 1430
Service and sales worker share 0.06 0.10 0 0.74 1430
Production worker share 0.52 0.26 0.008 0.98 1430
Simple task worker share 0.12 0.21 0 0.91 1430
Log(manager earnings relative to unskilled earnings) 0.50 0.23 -0.12 1.85 1430
Log(technical worker earnings relative to unskilled earnings) 0.46 0.29 -0.08 1.17 1430
Log(office worker earnings relative to unskilled earnings) 0.01 0.12 -0.41 0.46 1430
Log(service and sales worker earnings relative to unskilled earnings) 0.06 0.19 -0.34 0.56 1430
Manager earnings (1,000 KRW) 33308.46 9640.78 2000 97000 1430
R&D and technical worker earnings (1,000 KRW) 33633.58 15545.59 1915 135452 1430
Office worker earnings (1,000 KRW) 20868.58 7509.21 1488 53899 1430
Service and sales worker earnings (1,000 KRW) 22359.40 9331.58 1422 70032 1430
Return on capital 0.04 5.28 -172.78 78.93 1326
Log(revenue) 10.60 1.98 4.28 17.52 1359
Number of computers 138.34 400.84 0 8000 1353
Use team system 0.58 0.49 0 1 1430
Use salary system 0.45 0.50 0 1 1424
Hired consultants 0.38 0.49 0 1 1430
Has a CEO separate from the owner 0.49 0.50 0 1 1430
Notes: Salary and revenue numbers are adjusted to 2002 values. Primary source of data is the 2002 and 2003 establishment level pilot survey for the Korea Workplace Panel. The G5 industry management index is constructed based on the US, Britain, France, Germany, and Italy data in the World Management Surveys.
! 30!
Table 3. The correlation of management indexes (1) (2) (3)
Management index Industry average management index
Firm specific management index
Foreign management indexes
G5: US, Bri, Fra, Ger, Ita 0.368*** 0.406*** -0.0376 (0.100) (0.127) (0.0621)
Observations 1,430 1,430 1,430 R-squared 0.229 0.268 0.153
US 0.162** 0.148* 0.0135 (0.0653) (0.0753) (0.0345)
Observations 1,338 1,338 1,338 R-squared 0.227 0.189 0.156
Britain 0.145* 0.175* -0.0297 (0.0765) (0.0986) (0.0439)
Observations 1,363 1,363 1,363 R-squared 0.218 0.181 0.152
France 0.170** 0.196*** -0.0260 (0.0656) (0.0674) (0.0597)
Observations 1,204 1,204 1,204 R-squared 0.255 0.229 0.178
Germany 0.169** 0.225** -0.0566 (0.0830) (0.0992) (0.0452)
Observations 1,085 1,085 1,085 R-squared 0.239 0.269 0.156
Italy 0.147 0.238** -0.0916** (0.0896) (0.0926) (0.0427)
Observations 1,152 1,152 1,152 R-squared 0.247 0.259 0.178
Notes: Each regression includes as control variables the age and size (log employment) of the establishment, dummy variable for survey year 2003, and province fixed effects. Standard errors clustered at the industry level are in parentheses. *, **, *** denote significance at the 10, 5, and 1 percent level.
! 31!
Table 4. Employment and salary effects from modern management - by skill group (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
log(# of managers)
log(# of technical workers)
log(# of office workers)
log(# of service and
sales workers)
log(# of production workers)
log(# of simple task workers)
log(manager earnings)
log(technical worker
earnings)
log(office worker
earnings)
log(service and sales worker
earnings)
log(production worker
earnings)
log(simple task worker earnings)
Panel A: Establishment level management index
Management index 0.0999*** 0.215*** 0.101*** -0.0755 -0.0613*** 0.00701 0.0385*** 0.0582*** 0.0475*** 0.0632*** 0.0282*** 0.0394***
(0.0233) (0.0539) (0.0203) (0.0540) (0.0227) (0.0460) (0.00837) (0.0188) (0.0129) (0.0169) (0.00892) (0.00924)
R-squared 0.645 0.394 0.648 0.217 0.718 0.078 0.289 0.219 0.257 0.264 0.266 0.279 Panel B: Industry average index and establishment specific residual
Industry average management index
0.227*** 0.952*** 0.251*** 0.0127 -0.138* -0.223 0.0661** 0.177 0.125* 0.288*** -0.00383 0.0769
(0.0670) (0.123) (0.0815) (0.162) (0.0734) (0.138) (0.0266) (0.113) (0.0698) (0.0947) (0.0470) (0.0495)
Firm specific management index
0.0828*** 0.117** 0.0812*** -0.0873* -0.0509** 0.0378 0.0348*** 0.0423*** 0.0371*** 0.0330*** 0.0326*** 0.0344***
(0.0237) (0.0497) (0.0191) (0.0490) (0.0203) (0.0477) (0.00862) (0.0120) (0.0105) (0.0115) (0.00919) (0.00930)
R-squared 0.647 0.427 0.651 0.218 0.719 0.080 0.290 0.230 0.264 0.307 0.267 0.280 Panel C: G5 industry index (Reduced form results)
G5 management index 0.164** 0.845*** 0.139 0.297* -0.146** 0.0222 0.0427* 0.261** 0.172*** 0.169 -0.0245 0.00716
(0.0667) (0.131) (0.0861) (0.154) (0.0704) (0.191) (0.0249) (0.101) (0.0580) (0.126) (0.0393) (0.0611)
R-squared 0.641 0.416 0.644 0.220 0.718 0.078 0.277 0.253 0.272 0.266 0.261 0.268 Panel D: Predicted index (2SLS results using G5 industry index as IV)
Management index 0.447*** 2.294*** 0.378* 0.807 -0.397* 0.0602 0.116* 0.708*** 0.468*** 0.458 -0.0666 0.0195
(0.144) (0.547) (0.208) (0.494) (0.222) (0.520) (0.0608) (0.263) (0.150) (0.280) (0.114) (0.161)
First stage F-statistic 13.47
Notes: Each regression includes age, size (log employment) of establishment, a 2003 year dummy, and province fixed effects as control variables. The number of observations is 1,430, except for Panel E which is 1,353. Kleibergen-Paap F-statistics are reported for the first stage regressions in Panel D. Standard errors clustered by industry are in parentheses. *, **, *** denote significance at the 10, 5, and 1 percent level.
! 32!
Table 5. Modern management and the demand for skill (1) (2) (3) (4) (5) (6) (7) (8)
ln!( !"#$%&!!"!!
!"#$%&!!"!!"#$%&&'(!!"#$%#&)
where X=
ln!( !"#$%&#!!"#$%$&'!!"!!
!"#. !"#$%$&'!!"!!"#$%&&'(!!"#$%#&)
where X=
Managers Technical workers
Office workers
Sales and
service workers
Managers Technical workers
Office workers
Sales and service workers
Panel A: Establishment level management index
Management index 0.144*** 0.260*** 0.146*** -0.0310 0.00533 0.0251* 0.0143** 0.0301**
(0.0335) (0.0609) (0.0297) (0.0616) (0.00775) (0.0149) (0.00714) (0.0117)
R-squared 0.266 0.163 0.181 0.320 0.059 0.074 0.086 0.086
Panel B: Industry average index and establishment specific residual
Industry average management index
0.394*** 1.119*** 0.419*** 0.180 0.0334 0.144 0.0922* 0.255***
(0.0992) (0.150) (0.106) (0.187) (0.0364) (0.102) (0.0489) (0.0713) Establishment specific management index
0.111*** 0.144** 0.109*** -0.0594 0.00156 0.00914 0.00390 -
0.000157
(0.0331) (0.0560) (0.0266) (0.0562) (0.00791) (0.00657) (0.00324) (0.00462)
R-squared 0.275 0.209 0.192 0.322 0.061 0.102 0.126 0.256 Panel C: G5 industry index (Reduced form results)
G5 management index
0.279*** 0.959*** 0.254** 0.412** 0.0507 0.268*** 0.180*** 0.177*
(0.102) (0.162) (0.121) (0.187) (0.0369) (0.0973) (0.0389) (0.0966)
R-squared 0.259 0.189 0.173 0.326 0.065 0.190 0.266 0.161
Panel D: Predicted management index (2SLS results using G5 industry index as IV)
Management index 0.758*** 2.606*** 0.689** 1.119* 0.138 0.729** 0.489*** 0.480**
(0.258) (0.652) (0.312) (0.609) (0.116) (0.297) (0.161) (0.214) First stage F-statistic 13.47
Notes: Each regression includes age, size (log employment) of establishment, a 2003 year dummy, and province fixed effects as control variables. Unskilled workers include production workers and simple manual laborers. The number of observations is 1,430. Kleibergen-Paap F-statistics are reported for the first stage regressions in Panel D. Standard errors clustered by industry are in parentheses. *, **, *** denote significance at the 10, 5, and 1 percent level.
! 33!
Table 6. Sensitivity of the 2SLS results (1) (2) (3) (4) (5) (6) (7) (8)
ln!( !"#$%&!!"!!
!"#$%&!!"!!"#$%&&'(!!"#$%#&)
where X=
ln!( !"#$%&#!!"#$%$&'!!"!!
!"#. !"#$%$&'!!"!!"#$%&&'(!!"#$%#&)
where X=
Managers Technical workers
Office workers
Sales and
service workers
Managers Technical workers
Office workers
Sales and service workers
A. Control for industry level competition and labor productivity
Management index 0.782*** 2.903*** 0.723** 1.254** 0.166 0.740** 0.543*** 0.530** (0.280) (0.646) (0.338) (0.618) (0.123) (0.314) (0.148) (0.227)
HHI (revenue) 0.114 1.600** 0.169 0.764 0.143 0.0610 0.310* 0.265 (0.290) (0.672) (0.313) (0.501) (0.0903) (0.230) (0.167) (0.236)
Labor productivity 0.00758 -0.327 -0.00219 -0.228 -0.0128 -0.0215 -0.107** -0.0455 (0.0781) (0.207) (0.0826) (0.168) (0.0302) (0.0802) (0.0446) (0.0475)
First stage F-statistic 15.54 B. Additionally control for technology
Management index 0.327 2.765*** 0.127 0.958 0.211 1.124* 0.811** 0.783** (0.345) (1.017) (0.475) (0.911) (0.199) (0.603) (0.326) (0.392)
ln(computer) 0.452*** 0.0400 0.596*** 0.227 -0.0319 -0.320 -0.232** -0.219 (0.120) (0.360) (0.173) (0.313) (0.0658) (0.204) (0.113) (0.140)
First stage F-statistic 7.12 C. Additionally control for export
Management index 0.751*** 2.822*** 0.705** 1.270** 0.164 0.745** 0.539*** 0.542** (0.291) (0.631) (0.348) (0.642) (0.121) (0.325) (0.149) (0.212)
Export share 0.165 0.285 0.0732 -0.0914 0.00700 -0.0536 -0.00327 -0.0853 (0.123) (0.300) (0.131) (0.285) (0.0426) (0.105) (0.0578) (0.0573)
First stage F-statistic 15.24 D. Additionally control for governance
Management index 0.790*** 2.955*** 0.731** 1.289** 0.171 0.760** 0.555*** 0.541** (0.287) (0.663) (0.344) (0.638) (0.127) (0.325) (0.153) (0.233)
Has a CEO -0.0770 -0.537** -0.0739 -0.361* -0.0533 -0.201** -
0.128*** -0.109* (0.0998) (0.246) (0.107) (0.205) (0.0377) (0.0991) (0.0459) (0.0624) First stage F-statistic 15.15 E. Additionally control for labor unions
Management index 0.756*** 2.841*** 0.695** 1.248** 0.166 0.723** 0.528*** 0.514** (0.277) (0.616) (0.335) (0.595) (0.118) (0.305) (0.142) (0.225)
Has a labor union 0.0335 0.0825 0.0122 -0.0489 -6.50e-05 0.0280 0.00910 0.0154 (0.0517) (0.129) (0.0420) (0.0828) (0.0119) (0.0313) (0.0224) (0.0230) First stage F-statistic 16.28 F. Control all of the above
Management index 0.394 1.731** -0.00108 0.437 0.114 1.579* 0.841* 1.063** (0.436) (0.815) (0.576) (1.005) (0.217) (0.908) (0.468) (0.508)
First stage F-statistic 9.3 Notes: Each regression includes age, size (log employment) of establishment, a 2003 year dummy, and province fixed effects as base control variables. Unskilled workers include production workers and simple manual laborers. Kleibergen-Paap F-statistics are reported for the first stage regressions. Standard errors clustered by industry are in parentheses. *, **, *** denote significance at the 10, 5, and 1 percent level.
! 34!
Table 7. Additional results (1) (2) (3) (4) (5) (6) (7)
Panel A. Employment
log(# of managers)
log(# technical workers)
log(# of office workers)
log(# of service and sales workers)
log(# of production workers)
log(# of simple task workers)
Management index 0.236 2.340*** -0.00269 0.740 -0.438 0.800 (0.174) (0.898) (0.322) (0.763) (0.384) (0.801)
ln(computer) 0.228*** -0.120 0.379*** 0.0142 -0.0165 -0.541* (0.0647) (0.330) (0.116) (0.268) (0.150) (0.287)
Panel B. Earnings
log(manager earnings)
log(technical worker earnings)
log(office worker earnings)
log(service and sales worker
earnings)
log(production worker earnings)
log(simple task worker earnings)
Management index 0.0428 0.980* 0.620** 0.590 -0.179 -0.0742 (0.0866) (0.502) (0.270) (0.437) (0.190) (0.246)
ln(computer) 0.0737** -0.220 -0.129 -0.108 0.108 0.0854 (0.0308) (0.175) (0.0980) (0.157) (0.0674) (0.0861)
Panel C. Employment - regular employees
log(# of
managers) log(# technical
workers) log(# of office
workers)
log(# of service and sales workers)
log(# of production workers)
log(# of simple task workers)
log(# of employees
Management index 0.214 2.366*** 0.118 0.750 -0.427 0.727 0.0721 (0.173) (0.917) (0.343) (0.717) (0.382) (0.721) (0.0692)
ln(computer) 0.238*** -0.124 0.328*** 0.0101 0.0162 -0.572** 0.00455 (0.0638) (0.340) (0.123) (0.250) (0.145) (0.270) (0.0287)
Panel D. Employment - temporary employees
log(# of
managers) log(# technical
workers) log(# of office
workers)
log(# of service and sales workers)
log(# of production workers)
log(# of simple task workers)
log(# of employees
Management index 0.139 -0.0333 -0.118 0.0936 -0.570 0.392 -0.427 (0.127) (0.144) (0.251) (0.209) (0.384) (0.464) (0.500)
ln(computer) -0.0549 0.0340 0.161* -0.0374 0.111 -0.153 0.150
(0.0529) (0.0663) (0.0922) (0.0832) (0.172) (0.163) (0.196) Notes: The number of observation is 1,345. The regressions in this table use the full specification in Table 6 Panel F and controls for the age and size (log employment) of the establishment, the industry competition index (HHI) and labor productivity, measures of technology, export, governance, and unions, a dummy variable for the survey year 2003, and province fixed effects. Results are from 2SLS regressions that use the industry management frontier as the instrumental variable. The Kleibergen-Paap F-statistics is 9.3. In Panel D, when there are zero part-time employees I add one before taking logs. Standard errors are clustered at the industry level. *, **, *** denote significance at the 10, 5, and 1 percent level.
! 35!
Table 8. Modern management and the efficacy of information technology in achieving organizational objectives
Dependent variable Coefficient estimate
on management index
Observations R-squared
A. Use information technology to digitize modular tasks, e.g., accounting, inventory management, sales management.
0.0559**
(0.0248) 613 0.091
B. Use Enterprise Resource Planning (ERP) or systems that comprehensively support your workplace's tasks
0.101***
(0.0169) 613 0.096 C. Use Electronic Data Interchange (EDI) or information technology systems that help connect and support clients and suppliers
0.0488**
(0.0225) 613 0.062 D. Did information technology at your workplace help with the organization and operation of project based task force teams
3.334**
(1.611) 611 0.150
E. Did information technology at your workplace facilitate organizational delayering and decision making?
0.0640** (0.0271) 613 0.089
F. Did information technology at your workplace help increases workers that can multitask?
0.0407** (0.0200) 613 0.074
G. Did information technology at your workplace encourage worker involvement in company operations?
0.0615** (0.0249) 613 0.083
H. Degree to which on the job experiences are collected and shared among employees through information technology
0.0978***
(0.0321) 557 0.088 I. Degree to which information technology capital and personnel investment achieved what the company had intended to achieve
0.114***
(0.0359) 558 0.128 Notes: Each row represents an OLS regression where the first column describes the dependent variable, and the second column indicates the coefficient estimate on the management index. All regressions controls for the age and size (log employment) of the establishment, the industry competition index (HHI) and labor productivity, and province fixed effects. Standard errors clustered by industry are in parentheses. *, **, *** denote significance at the 10, 5, and 1 percent level.
! 36!
Table 9. Complementarity between modern management and technical skill
(1) (2) (3) (4) (5) (6) OLS OLS 2SLS 2SLS OLS OLS
Return on assets
Management index 0.00346 0.0272 0.304* 0.639** -0.0436* -0.0145 (0.0182) (0.0171) (0.172) (0.322) (0.0227) (0.0228)
Management index*ln(tech workers) 0.0252*** 0.0203** (0.00729) (0.00791)
ln(tech workers) -0.0294 -0.00209 (0.0212) (0.0150)
ln(computer) -0.0972** -0.297*** -0.0944** (0.0455) (0.110) (0.0416)
First stage F-statistic 16.02 16.02 Observations 1,324 1,260 1,324 1,260 1,324 1,260 R-squared 0.013 0.022 0.016 0.024
Notes: Each regression controls for the age and size (log employment) of the establishment, the industry competition index (HHI) and labor productivity, measures of technology, export, governance, and unions, a dummy variable for the survey year 2003, and province fixed effects. Kleibergen-Paap F-statistics are reported for the first stage regressions in columns (3) and (4). Standard errors clustered by industry are in parentheses. *, **, *** denote significance at the 10, 5, and 1 percent level.
! 37!
Appendix Table 1. Earnings differences between regular and temporary workers across occupation groups
(1) (2) (3) (4) (5) (6)
Log (earnings)
Technical worker 0.132*** 0.129*** 0.0530* 0.0626 0.0608 0.0449 (0.0203) (0.0206) (0.0290) (0.0499) (0.0497) (0.119)
Office worker 0.121*** 0.115*** 0.104*** 0.0764** 0.0738** -0.00458 (0.0136) (0.0135) (0.0218) (0.0357) (0.0357) (0.0744)
Production worker -0.0351** -0.0392*** -0.0513** -0.0224 -0.0247 -0.0712 (0.0145) (0.0143) (0.0239) (0.0351) (0.0351) (0.0933)
Simple task worker -0.225*** -0.229*** -0.277*** -0.105*** -0.103*** -0.169 (0.0188) (0.0187) (0.0362) (0.0354) (0.0354) (0.106)
Temporary work -0.287*** -0.284*** -0.277*** -0.0604** -0.0598** -0.0592 (0.0227) (0.0227) (0.0355) (0.0247) (0.0247) (0.0456)
Temporary * Technical worker -0.0677 -0.0937 0.0599 -0.0300 -0.0304 0.0168 (0.108) (0.112) (0.111) (0.102) (0.101) (0.161)
Temporary * Office worker -0.0506 -0.0472 -0.0664 0.00279 0.00385 -0.0366 (0.0320) (0.0320) (0.0429) (0.0397) (0.0396) (0.0615)
Temporary * Production worker -0.00426 0.00592 -0.0164 -0.00532 -0.000657 -0.00872 (0.0261) (0.0266) (0.0460) (0.0309) (0.0309) (0.0588)
Temporary * Simple task worker
0.00695 0.0153 0.00913 0.000125 0.00128 0.0719 (0.0305) (0.0307) (0.0510) (0.0348) (0.0348) (0.0668)
Gender, education level, and age fixed effects Yes Yes Yes No No No Individual fixed effects No No No Yes Yes Yes Region fixed effects Yes Yes Yes Yes Yes Yes Industry (2 digit) fixed effects Yes Yes Yes Yes Yes Yes Business type fixed effects No Yes Yes No Yes Yes Business size fixed effects No No Yes No No Yes Observations 15,591 15,584 6,512 14,268 14,261 5,140 R-squared 0.470 0.474 0.510 0.835 0.835 0.868
Notes: The above regressions use data from the Korean Labor and Income Panel Study (KLIPS) covering the years 2000 to 2006. The sample is comprised of workers holding non-managerial position. The omitted category, i.e., the reference occupation group is the service and sales workers. Business types are personal business, incorporated company, foreign company, government enterprise, self-employed, and others. Business size is categorized into eleven bins. Robust standard errors are in parentheses. *, **, *** denote significance at the 10, 5, and 1 percent level.
! 38!
Appendix Table 2. Results that account for average hours worked (1) (2) (3) (4) (5) (6) (7) (8)
ln!( !"#$%&!!"!!
!"#$%&!!"!!"#$%&&'(!!"#$%#&)
where X=
ln!( !"#$%&#!!"#$%$&'!!"!!
!"#. !"#$%$&'!!"!!"#$%&&'(!!"#$%#&)
where X=
Managers Technical workers
Office workers
Sales and service workers Managers Technical
workers Office
workers Sales and service workers
Panel A: Establishment level management index
Management index 0.173*** 0.507*** 0.159*** -0.100 -0.00663 0.0187 0.0172* 0.0374**
(0.0375) (0.119) (0.0337) (0.121) (0.0135) (0.0126) (0.0100) (0.0144)
R-squared 0.233 0.114 0.129 0.159 0.050 0.061 0.049 0.088
Panel B: Industry average index and residual establishment management index
Industry average management index
0.499*** 2.070*** 0.415*** 0.292 -0.0207 0.117 0.136** 0.352***
(0.126) (0.295) (0.126) (0.399) (0.0796) (0.0768) (0.0620) (0.0875)
Firm specific management index
0.130*** 0.299*** 0.125*** -0.152 -0.00476 0.00564 0.00143 -0.00437
(0.0360) (0.109) (0.0324) (0.108) (0.00926) (0.00729) (0.00557) (0.00634)
R-squared 0.246 0.154 0.136 0.160 0.051 0.079 0.094 0.280 Panel C: G5 industry index (Reduced form results)
G5 management index
0.321** 1.856*** 0.188 0.797** 0.0531 0.254*** 0.253*** 0.229**
(0.124) (0.312) (0.135) (0.371) (0.0622) (0.0713) (0.0533) (0.103)
R-squared 0.223 0.140 0.118 0.165 0.055 0.163 0.226 0.163
Panel D: Predicted management index (2SLS results using G5 industry index as IV)
Management index 0.876*** 5.055*** 0.511 2.172* 0.145 0.693*** 0.689*** 0.625***
(0.308) (1.172) (0.344) (1.204) (0.183) (0.244) (0.216) (0.217) First stage F-statistic 13.34
Panel E: Panel D + additional control variables
Management index 0.377 4.942*** -0.144 1.999 0.217 1.035** 1.087*** 0.941**
(0.428) (1.565) (0.511) (1.714) (0.315) (0.450) (0.374) (0.371) First stage F-statistic 9.29
Notes: In this table employment is weighted by the industry average hours worked and earnings is normalized by the industry average hours worked for each occupation group. Each regression includes age, size (log employment) of establishment, a 2003 year dummy, and province fixed effects as control variables. Unskilled workers include production workers and simple manual laborers. The number of observations is 1,426. Kleibergen-Paap F-statistics are reported for the first stage regressions in Panel D and E. Panel E additional controls for all the variables used in Table 6 Panel F. Standard errors clustered by industry are in parentheses. *, **, *** denote significance at the 10, 5, and 1 percent level.
! 39!
Appendix Table 3. First-stage regression that includes industry level competition and productivity variables
Management index
G5 management index 0.328*** (0.0833)
HHI (revenue) -0.594** (0.236)
Labor productivity 0.140*** (0.0515)
Establishment size Yes Establishment age Yes Year dummy Yes Region fixed effects Yes Observations 1,430 R-squared 0.241
Notes: The regression reports the first stage of the 2SLS regression that corresponds to Table 6 Panel A. Standard errors clustered by industry are in parentheses. *, **, *** denote significance at the 10, 5, and 1 percent level.