Content uploaded by Charilaos Mertzanis
Author content
All content in this area was uploaded by Charilaos Mertzanis on Apr 30, 2020
Content may be subject to copyright.
Access to skilled labor,
institutions and firm performance
in developing countries
Charilaos Mertzanis
School of Business, Abu Dhabi University, Abu Dhabi, United Arab Emirates, and
Mona Said
Department of Economics, School of Business,
American University in Cairo, Cairo, Egypt
Abstract
Purpose –The purpose of this paper is to examine the role of access to skilled labor in explainingfirms’sales
growth subject to the controlling influence of a wide range of firm-specific characteristics and country-level
economic and non-economic factors.
Design/methodology/approach –The analysis uses a consistent and large firm-level data set from the
World Bank’s Enterprise Surveys that includes 138 developing countries. An instrumental variables model
with a GMM estimator is used for estimating the impact of access to skilled labor on firm performance.
In order to obtain more robust estimators, the analysis introduces country-level controls reflecting the
influence of economic and institutional factors, such as economic and financial development, institutional
governance, education and technological progress.
Findings –The results document a significant and positive association between access to skilled labor
and firm performance in the developing world. The explanatory power of access to skilled labor remains
broadly robust after controlling for a wide range of firm-specific characteristics: sectoral and geographical
influences matter. The results also show that the association between labor skill constraints and
firm performance is mitigated by country-level factors but in diverse ways. Development, institutions,
education and technological progress exert various mitigating effects on firm-level behavior regarding access
to skilled labor.
Originality/value –The paper’s novel contribution is threefold: first, it uses joint firm, sector and
country-level information to analyze the role of access to skilled labor on firm performance; second, it uses
consistently produced information at the firm level from 138 developing countries; and, third, it considers the
controlling impact of a wide range of country-level factors that reflect a country’s overall development,
institutions and evolution.
Keywords Institutions, Survey data, Firm performance, Labour skills
Paper type Research paper
1. Introduction
Access to skilled labor is a key determinant of firm performance. Nickell and Bell (1995),
Machin et al. (1996) and Leuven et al. (2004) argue that intensifying competitive pressures on
firm performance have increased the demand for skilled labor, which in turn has revealed a
problem of skill shortage. The situation is more aggravated in developing countries
where education–occupation mismatches as well as other labor market specificities result in
skill shortage fluctuations thereby affecting firm performance. For instance, Bhattacharya
and Wolde document that reducing the labor skill shortage from the MENA region
average to the world’s average could result in a 4 percent increase in per capita GDP
annually. Fakih and Ghazalian (2015) document for the same region the importance of
various firm-specific characteristics and institutional factors in explaining labor skill
shortages. Brixiova et al. (2009) show that labor skill shortages affect the pace of
convergence of transition economies in Central Europe toward European structures. Tan
et al. (2016) find that deficiencies in the education and training systems in Tanzania
compromise the quality of labor skills, giving rise to skill shortages, thereby constraining
International Journal of Manpower
Vol. 40 No. 2, 2019
pp. 328-355
© Emerald Publishing Limited
0143-7720
DOI 10.1108/IJM-11-2017-0301
Received 16 November 2017
Revised 21 August 2018
Accepted 28 October 2018
The current issue and full text archive of this journal is available on Emerald Insight at:
www.emeraldinsight.com/0143-7720.htm
328
IJM
40,2
the operations and growth of formal sector firms in the country. Fajnzylber and Fernandes
(2009) show that firms in Brazil and India that engage in international trade and foreign
direct investment activities experience more or less severe labor skill constraints depending
on their technology diffusion and extent of trade specialization, respectively.
This paper analyzes the relation between labor skill shortages of firms and their
performance in developing countries. The analysis of the relation is carried out under
different economic, educational and technological conditions. The novel contribution of the
paper is threefold: first, it uses firm, sector and country-level information to analyze the role
of firms’access to skilled labor on their performance; second, it uses consistently produced
information at the firm level from 138 developing countries; and, third, it considers the
controlling impact of a wide range of firm-specific characteristics and country-level factors
that reflect a country’s overall development, educational and technological conditions.
To our knowledge, this is the first paper that analyzes in detail this relation in a wide range
of developing countries. Our paper contributes to the literature on the impact of labor supply
conditions on firm performance in developing countries.
The results document a significant positive relation between access to skilled labor and
firm performance in the developing world that remains broadly robust to various firm-level
controls comprising a wide range of firm-specific characteristics (age, size, sector of
activity, location, export status, ownership, etc.). Importantly, the results also document that
the interaction between access to skilled labor and firm performance is mitigated by
country-level factors, such as the levels of development, income inequality, education and
technological progress. In order to develop a more solid basis for steering labor market
reform to improve firm performance in the developing world, policy makers need to
understand more adequately the influence of firm-specific factors and the economic and
institutional environment conditioning the interaction between labor skill constraints and
firm performance.
In what follows, Section 2 reviews the related literature; Section 3 describes the data and
the empirical methodology used for the analysis; Section 4 explores the power of labor skill
constraints and other firm-specific characteristics in predicting firm performance; Section 5
extends the analysis to include the controlling impact of country-level factors; and Section 6
concludes the paper.
2. Related literature
Labor skills are competencies needed to carry out the tasks and duties of a given job
(ILO, 2012, p. 11). These competencies include cognitive (i.e. literacy), non-cognitive (i.e. team
work, communication, language, IT use and other soft skills) and job-specific skills. Skill
formation is the result of a successive sequence of education, training and labor market
participation activities (Banerji et al., 2010). Few developing countries have comprehensive
information on these different competencies (World Bank, 2014).
The importance of labor skills for firm performance can be understood by reference to
the competitive pressure on firms to survive and grow in the modern globalized world. This
requires faster productivity growth and efficient strategy adaptation. Syverson (2011)
reviews the evidence on the determinants of firm productivity growth highlighting the
pervasive within-sector dispersion in productivity levels across firms. Hsieh and Klenow
(2009) show that this dispersion is larger in developing countries. The within-sector
dispersion in productivity levels among firms also appears to persist over time. The
evidence suggests that firms exhibit heterogeneous internal capabilities that persist over
time and respond with different efficiency to uncertain external demand.
Firms appear to offer two different responses to these challenges. The first focuses on the
competitive adjustment of their cost structure and output composition to align with the
broad shifts in overall demand brought about by the business cycle, trade liberalization
329
Firm
performance
in developing
countries
(Pavcnik, 2002) and technological change (Bloom et al., 2015). The second response focuses
on adjusting the internal organization and production strategy, by introducing new
management practices (Bloom and Van Reenen, 2010), in-service training programs
(Dearden et al., 2005) and investing in fixed capital improvements (Bartel et al., 2007), among
others. Both responses influence price and non-price factors of output and hence economic
performance. These responses are further elucidated below.
According to the first response, the availability and quality of appropriate labor skills
affect firm performance through their impact on unit labor cost (productivity effect), which
has confirmed evidence. Haskel and Martin (1996), Bennett and McGuinness (2009) and
Healy et al. (2015) argue that shortages of labor skills can cause a competitiveness loss as a
result of inadequate competences and excessive wage increases. Montt (2015) shows that in
the OECD countries higher and persisting mismatches in labor skills are associated with
lower labor productivity. Rehman (2015) documents that skill shortage is associated
with lower productivity growth in the Pakistani software industry. Mahy et al. (2015) find a
significant positive relationship between specialized training of employees and firm
productivity. This relation is stronger when firms belong to a technology-intensive industry,
deploy a relatively high share of tasks that require advanced skills, and evolve within an
overall discouraging business environment, as is the case in most developing countries.
The dispersion of labor skills too affects firm productivity. Bombardini et al. argue that
variations in labor skills help adjust mismatches between labor demand and supply, but the
adjusting effect varies among sectors of activity. Firms operating in industries with higher
complementarity respond more efficiently to labor skill dispersion. Iranzo et al. (2008) show
that the dispersion of labor skills in Italian firms has a positive effect on firm productivity
when it materializes between occupational groups within the same firm, i.e. between
productive and non-productive workers. In contrast, it has a negative effect when it
materializes across firms. Giesing and Laurentsyeva (2014) show that one standard
deviation increase in labor skill shortage results in a 6.1 percent decrease in total factor
productivity of the firm in the EU.
Moreover, the impact of labor skills on firm productivity is affected by the specific
characteristics of firms. Galindo-Rueda and Haskel (2005) argue that the share and
gender composition of full-time workers in the firm affects the impact of labor skill
dispersion on firm productivity. In a cross-country setting, Poschke (2014) shows that the
variation of firm size affects entrepreneurial activity and thereby the dispersion of
average demand for labor skills. Ouimet and Zarutskie (2014) find a positive association
between firm age, employee age with new skills and firm growth in new ventures.
Shleifer and Vishny (1994) and Prasnikar et al. (1994) argue that the firms’ownership
structure and especially state ownership have direct effects on worker incentives through
political constraints placed on the labor contracting process. Modified labor contracts can
in turn affect investment in human capital and hence labor skill acquisition and work
effort. Cai et al. (2008) argue that the formation and distribution of labor skills during the
reform process in China, especially in state-owned enterprises, has been driven rather by
government intervention than private sector initiatives. Kaiser (2002, Ch. 5) argues that
limited liability firms in the services sector in Germany tend to switch to less desirable
financial assets more often than unlimited liability firms, thereby affecting the structure
of skills required to manage these assets. Atkin et al. show that exporting firms in Egypt
realize learning-by-exporting economies that lead to specific skill acquisition that helps
improve their technical efficiency.
According to the second response, labor skills affect firm performance through their
impact on the range and composition of capabilities required to make efficient strategic
decisions (strategic management effect). These capabilities refer to a firm’s capacity to
deploy suitable resources and develop and disseminate task-specific information through
330
IJM
40,2
its human capital. Capabilities are intangible assets often developed in key functional
areas (e.g. production engineering, brand management in marketing) by combining
physical, human and technological skills. Teece (2017) argues that the capabilities view of
the firm looks beyond factors of production to recognize the importance of managerial
choices in enhancing labor productivity and better meeting consumer demand.
Technology adoption and expertise result from search, R&D and investment. Value
generation depends on the firms’response to competition, the extent to which they
appropriate and refine capabilities and the acquisition of suitable industrial knowledge
over time. Teece et al. (1997) further argue that improvements in internal coordination,
adjustment of labor skills to evolving production requirements, such as tacit knowledge
and asset complementarity, and the realization of increasing returns provide firms with a
competitive advantage thereby increasing productivity. The latter may be enhanced or
eroded depending on the stability of the business cycle, the ease of internal skill
adjustment and the extent of competitive replication of strategy. The strategic emphasis
shifts from merely having the right pool of human resources to internally generating new
human capital. Ahmad and Schroeder (2003) stress that success of some business
decisions (e.g. international mergers and acquisitions) necessitates recognition and
reconciliation of the differences among human resource management practices in different
countries and industries. Snow and Snell (2011) note that human capital improvements aid
strategy formulation by allowing novel approaches to the organization of production.
Barney and Wright (1998), Ployhart (2006) and Irvin and Michaels (1989) argue that, to the
extent that they are difficult to be imitated by competitors and to be substituted by other
resources, new capabilities enhance long-term competitive advantage. Amit and
Schoemaker argue that organizations must decide both on the selection of suitable
labor skills and capabilities and on the best way to make skilled labor an integral part of
organizational improvement. Ahmed et al. (2014) provide evidence that suitable
operational capabilities of workforce make the firms’performance more resilient to
shocks. Galunic and Riordan (1998) argue that heterogeneity in labor skills provides firms
with a higher ability of comprehensive problem solving and creative conflict resolution.
The cognitive diversity resulting from interaction among people with different
perspectives and capabilities facilitates the identification and formulation of better
organization strategies.
The studies surveyed above rely mostly on aggregate data and therefore do not
adequately consider the individual firms’perspective on the role of labor skill constraints for
their performance, especially in the developing countries. Further, they rarely consider the
role of a wide range of firm-specific characteristics. Moreover, the empirical association
between access to skilled labor and firm performance is mostly analyzed within individual
country settings and rarely on a comparative cross-country basis. Finally, the analysis of
economic and institutional factors across countries is rarely considered explicitly on a
cross-country setting. Based on a new and rich data set, this paper aims to fill the void to
some extent by examining in more detail the relation between access to skilled labor and
firm performance in developing countries. It tests for the impact of skilled labor constraints
experienced by firms on their performance in developing countries, subject to the influence
of a wide range of firm-specific characteristics and country-level institutional factors.
The large size of the data allows the analysis of the sectoral and geographical dimensions of
the association. The paper takes a behavioral perspective by considering firms’own
perceptions of labor constraints rather than relying on aggregated reporting information,
which may be either inadequate or not credible in many developing countries. The results
provide new evidence that contributes to the elucidation of certain aspects of the two types
of responses firms offer. The results have useful implications for labor market policy aiming
at enhancing labor supply conditions and firm performance in developing countries.
331
Firm
performance
in developing
countries
3. Data and research methodology
The analysis utilizes data from the World Bank’s Enterprise Surveys (ES hereafter).
The basic data include 131,908 non-financial firms from 138 developing countries, covered
by the ES between 2006 and 2016 over several surveys of firms from the manufacturing and
other non-financial sectors of economic activity. Survey rounds are essentially independent
cross-sections with only a limited number of firms appearing consistently throughout
survey rounds. However, the data set has the advantage of consistent identification of firms’
responses throughout many survey years and across a wide range of developing countries.
This allows the focusing on the variation in average firm-level responses within each
country and dynamically over time. The ES data reflect the individual firms’experience and
perception of the environment in which they operate. The ES facilitate the linking of firm
performance with other firm characteristics and the business environment. The ES use
updated and complete sampling frames for each country and strive to eliminate alien
elements from the frame prior to the sample selection. An important strength of the data is
that it focuses on small and medium enterprises, which are the bloodline of economic
activity in most developing countries. The ES stratifies firm size into small
(5–19 employees), medium (20–99) and large (100 and more) firms. A shortcoming of ES
data stems from the doubts often expressed regarding the representativeness of firms’true
constraints. While the responses of firms are private unaudited information, these concerns
could only be addressed by using firm census data, which are not available for most
developing countries. Instead, micro-survey data provide more valuable information, at
least in the developing countries.
The outcome variable for measuring firm performance is the real annual growth of sales,
GSALES. Based on the ES explanation, GSALES represents the real annual growth of sales
reported during the current fiscal year from a previous period. The variable captures a
dynamic aspect of firm performance. For most countries in the ES, the difference between
the two fiscal periods is two years. However, for some countries the interval is three years.
Hence, an annualized measure is used. All values for GSALES are converted to USD using
the exchange rate that corresponds to the fiscal year of the survey. GSALES is then deflated
to 2009 using the USD deflator.
The main independent variable is SKILLS. The latter is an ES categorical variable that
reflects the extent to which firms identify the current level of labor skills as a major
constraint for their performance. It is the answer of firms to the question: “How problematic
is the level of labor skills for the operation and growth of your business?”It takes the value
of 1 if the answer is “very problematic or moderately problematic”and 0 otherwise. It is
possible that the firms’answer does not capture all reality, as some firms may report labor
skill constraints while they are not actually constrained by them but rather facing
temporary skill shortages instead. Therefore, one must be cautious of this behavioral bias
and interpret the results carefully. Nevertheless, the measure is consistently applied across
firms and countries for the sample period. Table I reports the average value of SKILLS per
country; the average sample value is 21.3 percent and the standard deviation is 41.2 percent.
The data show a large divergence among countries. Most firms operating in Suriname
consider labor skills as the biggest obstacle to their operations (66.4 percent), while few
firms operating in Eritrea consider labor skills as an important obstacle (1.7 percent).
Interestingly, Table I shows that labor skills tend to be a serious obstacle to some of the
biggest developing economies, such as Brazil (57.3 percent), Argentina (49.5 percent) and
Syria (59.6 percent).
Control variables include a range of firm-specific characteristics and country-level
factors that account for the impact of national economic and non-economic conditions.
These latter control variables capture unobservable differences among countries and
account for any spurious relationships. They provide an improved measure of the impact of
332
IJM
40,2
firm-specific variables beyond the effects of others. Each individual firm is not large enough
to affect country-level measures of those development indicators. Thus, the regression
estimates represent within-country over time variation in the relationship between access to
skilled labor and firm performance. Table AI describes the variables used in the paper.
Country (num. of firms) Mean Country (num. of firms) Mean Country (num. of firms) Mean
Afghanistan (945) 32.3 Georgia (959) 17.2 Paraguay (974) 41.4
Albania (960) 17.5 Germany (1,196) 6.9 Peru (1,632) 33.4
Algeria (600) 36.8 Ghana (1,214) 9.4 Philippines (1,326) 6.0
Angola (785) 23.2 Greece (546) 8.6 Poland (1,960) 19.2
Antigua and Barbuda (151) 31.1 Grenada (153) 39.1 Portugal (505) 12.4
Argentina (2,117) 49.5 Guatemala (1,112) 31.3 Romania (1,721) 31.5
Armenia (1,128) 11.8 Guinea (223) 12.6 Russia (5,953) 27.7
Azerbaijan (1,196) 5.1 Guinea-Bissau (159) 13.2 Rwanda (453) 19.6
Bahamas (150) 33.3 Guyana (165) 48.8 Samoa (109) 27.8
Bangladesh (2,946) 16.8 Honduras (796) 23.6 Senegal (1,107) 9.1
Barbados (150) 28.9 Hungary (1,218) 10.5 Serbia (1,102) 13.2
Belarus (1,034) 23.6 India (13,515) 11.2 Sierra Leone (150) 17.4
Belize (150) 40.7 Indonesia (1,444) 6.1 Slovak Rep. (748) 20.4
Benin (150) 27.9 Iraq (756) 25.8 Slovenia (804) 8.7
Bhutan (503) 13.8 Ireland (501) 15.6 South Africa (937) 9.0
Bolivia (975) 32.9 Israel (483) 12.5 South Korea (598) 6.8
Bosnia and Herzegovina (999) 12.4 Jamaica (376) 21.5 South Sudan (738) 20.5
Botswana (610) 27.6 Jordan (1,076) 18.7 Spain (606) 13.8
Brazil (3,444) 57.3 Kazakhstan (1,768) 24.7 Sri Lanka (610) 19.3
Bulgaria (2,500) 25.6 Kenya (1,438) 16.3 St Kitts and Nevis (150) 45.0
Burkina Faso (533) 30.3 Kosovo (472) 17.8 St Lucia (150) 24.7
Burundi (427) 12.7 Kyrgyz Rep. (877) 22.6 St Vincent and Gren (154) 32.0
Cabo Verde (254) 39.7 Lao PDR (630) 21.5 Sudan (662) 15.5
Cambodia (974) 22.6 Latvia (836) 29.5 Suriname (152) 66.4
Cameroon (535) 26.8 Lebanon (943) 28.2 Swaziland (307) 13.4
Central African Rep. (150) 24.7 Lesotho (151) 17.3 Sweden (600) 23.2
Chad (150) 58.5 Liberia (150) 13.5 Syria (508) 59.6
Chile (2,050) 34.5 Lithuania (1,057) 29.9 Tajikistan (1,099) 18.9
China (2,700) 2.9 Madagascar (977) 12.7 Tanzania (1,232) 30.4
Colombia (1,942) 31.4 Malawi (673) 17.5 Thailand (1,043) 38.5
Congo (151) 48.8 Malaysia (1,115) 20.2 Timor-Leste (150) 18.0
Congo, Dem. Rep. (1,228) 33.7 Mali (850) 9.3 Togo (155) 18.4
Costa Rica (881) 24.7 Mauritania (387) 28.8 Tonga (150) 50.8
Côte d’Ivoire (526) 34.3 Mauritius (398) 46.9 Trinidad and Tobago (370) 40.7
Croatia (1,284) 14.9 Mexico (2,960) 24.7 Tunisia (592) 31.7
Czech rep. (846) 18.1 Micronesia (68) 44.1 Turkey (4,559) 21.4
Djibouti (266) 18.2 Moldova (1,238) 27.0 Uganda (1,325) 16.4
Dominica (150) 13.3 Mongolia (722) 18.7 Ukraine (2,556) 23.2
Dominican Rep. (360) 36.1 Montenegro (266) 6.0 Uruguay (1,228) 29.1
Ecuador (1,024) 33.1 Morocco (1,066) 30.5 Uzbekistan (1,251) 12.4
Egypt (5,766) 27.1 Mozambique (479) 18.0 Vanuatu (128) 29.4
El Salvador (1,053) 27.0 Myanmar (632) 16.0 Venezuela (820) 29.7
Eritrea (179) 1.7 Namibia (909) 13.7 Vietnam (1,053) 8.4
Estonia (804) 16.6 Nepal (850) 9.7 West Bank and Gaza (835) 15.4
Ethiopia (1,976) 9.6 Nicaragua (814) 18.6 Yemen (830) 30.9
Fiji (164) 15.3 Niger (275) 30.3 Zambia (1,204) 11.4
FYR Macedonia (982) 10.2 Nigeria (4,567) 8.2 Zimbabwe (599) 5.1
Gabon (179) 42.3 Pakistan (2,182) 15.7
Gambia (174) 11.5 Panama (969) 14.5
Source: Enterprise Surveys, World Bank. The figures report the percent of firms reporting labor skill as an
important constraint to do business
Table I.
Labor skills as a
constraint to firms’
business operations
(mean percent)
333
Firm
performance
in developing
countries
Table II shows summary statistics. The growth of sales varies considerably among countries,
whereby there is an average of 4.73 percent variation with a standard deviation of
27.8 percent. The firms in the sample are on average young medium-size firms, operating
mainly in themanufacturing sector in big cities. They are on average moderate exporters with
a minority foreign ownership stake. On average, about a third of their labor force consists of
unskilled workers, which are employed mainly as permanent production workers.
The estimation analysis must deal with a number of potential econometric problems.
First, there is a possibility that the skilled labor constraints of firms may be endogenously
influenced by the growth of sales, giving rise to a problem of reverse causation. This, in
turn, might cause regressors to be correlated with the error term. However, the large
cross-section of firms and countries limits the possibility of reverse causation. Second,
time-invariant country characteristics, such as institutions, may exert considerable
unobservable cross-country effects on individual firm behavior and therefore be correlated
with the regressors. In order to deal with this potential problem more effectively, we add
firm-level characteristics as first-level controls and country-level factors as second-level
controls, corrected for collinearity and capturing nonlinearities (log values) and we
apply an instrumental variables analysis. The OLS estimator may result in upwards biased
estimates, whilst the use of a within-group estimator may result in downwards biased
estimates (Nickell, 1981).
Endogeneity concerns are dealt with the use of instrumental variables estimation.
However, finding proper external instruments across countries and over time is a
formidable task and thus the ordinary instrumental variables estimator does not adequately
Variable Obs. Mean SD Min 0.25% Median 0.75% Max
GSALES 85,191 4.73 27.75 −100 −7.87 3.89 16.19 100
SKILLS 147,468 0.21 0.41 0 0 0 0 1
AGE 143,689 2.54 0.82 0 2.08 2.56 3.09 5.37
SIZE 151,274 0.81 0.4 0 1 1 1 1
SECTR 151,707 0.59 0.61 0 0 1 1 3
LEGAL 145,892 0.48 0.5 0 0 0 1 1
LOCAT 115,165 0.53 0.5 0 0 1 1 1
EXPR 149,208 15.5 36.19 0 0 0 0 100
OWNF 148,620 10.61 30.8 0 0 0 0 100
OWNG 151,707 0.03 0.16 0 0 0 0 1
OWND 151,707 0.87 0.33 0 1 1 1 1
RSKILL 73,262 33.36 32.22 0 0 25 58.33 100
RPROD 63,444 2.14 17.03 0 0.15 0.42 1.33 31
RPERM 120,954 0.23 1.06 0 0 0 0.15 80
GDPCP 151,707 3.48 0.49 2.2 3.11 3.5 3.85 4.77
GINI 139,923 39.49 8.36 16.64 33.6 38.47 45.3 65
HDI 149,741 0.66 0.13 0.3 0.59 0.68 0.75 0.91
CREDIT 151,063 1.51 0.32 0.31 1.29 1.54 1.71 2.2
VOICC 150,859 −0.21 0.79 −2.16 −0.88 −0.18 0.41 1.67
POLSTB 150,859 −0.57 0.88 −2.69 −1.19 −0.62 0.07 1.33
GVEFF 150,859 −0.24 0.66 −1.73 −0.68 −0.27 0.12 1.89
RGQLT 150,859 −0.18 0.69 −2.26 −0.64 −0.28 0.27 1.89
RLLW 150,859 −0.35 0.68 −1.95 −0.82 −0.46 −0.03 1.95
CNTCR 150,859 −0.4 0.67 −1.64 −0.9 −0.56 −0.12 2.29
FRCTHN 149,892 0.45 0.23 0 0.25 0.42 0.63 0.93
FRCLNG 150,819 0.41 0.31 0 0.13 0.37 0.72 0.92
FRCRLG 150,969 0.41 0.22 0 0.21 0.38 0.61 0.86
Sources: World Development Indicators; World Bank Enterprise Surveys
Table II.
Descriptive statistics
of variables
334
IJM
40,2
address endogeneity bias. Arellano and Bover (1995) and Blundell and Bond (1998) argue
that lagged values of the endogenous variable may by themselves be poor instruments in
large data sets and they propose the system GMM estimator. The latter can account for
endogeneity bias by producing valid internal instruments under the assumption that
current-period shocks in the error term do not affect past values of the regressors and
the past values of the regressors do not directly affect current values of the outcome
variable. In other words, instruments include both lagged levels and lagged differences of
the variables. However, this causes a potential problem of instrument proliferation.
Roodman (2009) discusses the problem of too many instruments. He argues that the system
GMM estimator assumes no contemporaneous correlation among cross-sectional error
terms, so period dummies can be used to control for it. It also requires that there is first-order
serial correlation but no second-order serial correlation in the error term. But, as the time
dimension increases, the number of instruments can be too large compared to the sample
size making some asymptotic results and specification tests invalid. Too many instruments
can overfit endogenous variables and fail to expunge their endogenous components,
resulting in biased estimates. He moreover shows that in large-T data samples, a shock to
the country’s fixed effect would decline with time and hence the correlation of the lagged
outcome variable with the error term would be insignificant. The system GMM estimator
produces efficiency gains when the number of time series observations is small relatively to
cross-section observations. It also reduces the adverse effect of information gaps in the
unbalance panel data. The validity of the instruments collectively can be tested by using the
Hansen’sJstatistic of over identifying restrictions. However, the potential instrument
proliferation problem may also lead to the overfitting of the endogenous variables and
thereby weaken the validity of the Jtest. A return to the Sargan test might be contemplated
as it is less vulnerable to instrument proliferation, but this test requires homoscedastic
errors for consistency, which is a difficult assumption to make. As a rule of thumb, the
number of instruments should be less than the number of countries. On the other hand,
Bond et al. (2001) argue that in small samples the asymptotic standard errors may be
downwards biased. Windmeijer (2005) provides a finite-sample correction procedure for
the variance of the linear efficient two-step system GMM estimator. In our analysis, the
existence of a very large N makes the potential instrument proliferation problem less likely
to occur. Thus, given the small T per country and the large N of countries in our weakly
balanced sample, we use an instrumental variables model with a two-step GMM estimator
for cross-section analysis and correct for finite-sample bias. Only observations with
non-missing values are considered. Standard errors are corrected for heteroskedasticity and
clustered at the country level. In general, the estimated model assumes that the following
equation describes the firm’s underlying response:
Yij ¼ajþSKILLSijb1þX2
0b2þX3
0b3þmij;(1)
where Y
i,j
(GSALES) is the annual growth rate of sales reported by firm iin country j;
SKILLS is the binary variable that reflects the perception of labor skills as significant or
insignificant constraints for performance of firm iin country j;X
2
0is the vector of
firm-specific characteristics; X
3
0is the vector of country-level control variables. Τhe term μ
ij
is the two-way error term component that comprises the sum of η
i
,λ
t
and e
i,t
, where η
i
accounts for unobservable country-specific effects, λ
t,
accounts for time-specific effects and
the ε
i,t
is a disturbance parameter that is assumed to vary across countries and years.
An additional variable is deployed as external instrument in the endogeneity analysis. This
variable is the extent to which labor regulations affect firms’business operations
(LABREG), obtained from the ES. It exhibits a significant correlation with SKILLS
(31.8 percent) and no correlation with GSALES (0.083 percent). Hansen’sJstatistic will be
335
Firm
performance
in developing
countries
used for checking the validity of instruments and the model’s specification. Given that our
analysis contains several regressors, we are also mindful of Roodman’s (2009) concerns that
too many instruments tend to weaken the power of the Hansen Jtest. Finally, we are well
aware of the difficulty in interpreting the observed correlations as causal effects. Our results
are therefore interpreted as strength of association rather than causation, and the use of the
words “predict”or “impact”or “effect”is only made to simplify exposition. Table III
presents the pairwise correlations between the independent variable and the firm-specific
control variables. The results show small pairwise correlations (considerably lower than 0.5)
and therefore do not document severe collinearity problems between the firm-specific
variables. Therefore, all these variables can be included in regression analysis.
4. Analysis of results
The regression results report the estimated impact of access to skilled labor on firm
performance depending on the firms’own characteristics. In order to obtain a better
understanding of the relevant impact, the total sample of developing countries is divided
between high-income and low-income ones based on the sample mean level of GDP per
capita. Table IV reports the results of the baseline analysis. Several regression models are
sequentially estimated and the results are shown in columns A1–A11. In all models access
to skilled labor is positively and significantly associated with the growth of firms’sales in
developing countries. Better access to skilled labor increases the growth of sales.
The magnitude of the effect is on average small (1.9 percent) but strongly significant.
The Hansen Jstatistic is low and therefore the null hypothesis, that the models’
over-identification restrictions are valid, is not rejected. This implies that the instruments
are valid instruments, i.e., uncorrelated with the error term, and that the excluded
instruments are correctly excluded from the estimated equation. The predictive power of
access to skilled labor remains robust after the successive addition of firm-specific
characteristics in the analysis. The latter are significant determinants of sales growth.
The younger the firm, the higher its sales growth. This confirms the product life-cycle
hypothesis. The larger the firm, the more buoyant its sales growth. Large firms can realize
better organization, economies of scale and scope and establish more expanded sales
networks and marketing policies. Sales growth is faster as we move from manufacturing to
services and commerce. The latter find it easier to expand sales as they face lower costs of
production and distribution of products and services. Export-oriented firms realize higher
sales growth as they have access to wider markets and benefit from the challenges of open
competition. Sales growth is also stronger in public joint-stock companies and in those
companies located in the capital city or other major cities in a country. Further, high
ownership stakes by private domestic and foreign owners as well as by the state tend to
retard sales growth. On the other hand, the higher the stakes of dominant shareholders, the
stronger the growth of sales in the firm. It appears that dominant shareholders take a more
proactive stance in boosting sales growth performance of their firms. The inclusion of
firm-specific characteristics affects the magnitude but not the direction of the labor skill
effect. However, more pronounced differences emerge when the sample is split between
high- and low-income developing countries. While access to skilled labor remains robust, the
magnitude of its effect is twice as large in low-income countries as compared to that in
high-income countries. It appears that labor skill improvement has a stronger effect in
less-developed countries and therefore presents a more powerful case for public
intervention. Further, the role of the sector of activity and export status of firms is
stronger in high-income developing countries, whilst that of the legal status, location of
operation and ownership structure is stronger in low-income countries. This new
cross-country evidence highlights the important role of labor skills for sales growth as well
as of the role of firm-specific characteristics in explaining firm performance across different
336
IJM
40,2
Pairwise correlation 1234567891011121314
1 GSALES 1
2 SKILLS 0.041* 1
3 AGE −0.082* 0.041* 1
4 SIZE −0.028* −0.038* −0.205* 1
5 SECTR −0.033* 0.045* 0.123* −0.131* 1
6 LEGAL 0.078* 0.112* 0.113* −0.192* 0.059* 1
7 LOCAT 0.045* 0.024* 0.026* −0.042* 0.030* −0.022* 1
8 EXPR 0.018* 0.028* 0.111* −0.296* 0.189* 0.147* −0.020* 1
9 OWNF 0.035* 0.013* −0.022* −0.181* 0.007* 0.118* 0.042* 0.195* 1
10 OWNG 0.021* 0.027* 0.038* −0.058* −0.012* 0.036* 0.024* −0.002 −0.005 1
11 OWND 0.007 −0.019* −0.101* 0.121* −0.034* −0.143* −0.013* −0.081* −0.025* 0.038* 1
12 RSKILL 0.029* 0.051* 0.041* −0.087* 0.046* 0.104* 0.068* 0.048* 0.026* −0.010* −0.046* 1
13 RPROD 0.014* 0.011* 0.039* −0.154* −0.001 0.046* 0.007 0.054* 0.055* 0.046* −0.011* 0.015* 1
14 RPERM 0.0041 0.006* −0.031* 0.018* −0.009* −0.027* −0.002 0.002 −0.002 −0.008* −0.005 −0.018* −0.013* 1
Notes: Pairwise correlation coefficients of firms-specific characteristics with GSALES. Correlation rates are in all cases considerably lower than 0.50. *Significant at the
5 percent level
Table III.
Correlation among
firm-specific variables
337
Firm
performance
in developing
countries
GSALES A.1 A.2 A.3 A.4 A.5 A.6 A.7 A.8 A.9 A.10 A.11
All countries High-income Low-income
SKILLS 0.019*** 0.021*** 0.017*** 0.018*** 0.019*** 0.018*** 0.019*** 0.020*** 0.018*** 0.011*** 0.021***
(8.41) (9.24) (7.69) (8.13) (8.32) (7.98) (7.14) (8.53) (6.68) (3.04) (5.42)
AGE −2.238*** −3.230*** −3.323*** −3.263***
(−12.85) (−15.61) (−9.81) (−12.57)
SIZE 1.467*** 1.964*** 2.115*** 1.967***
(12.82) (12.98) (9.62) (9.70)
SECTR 0.339*** 0.290*** 0.651*** 0.003
(4.86) (2.66) (3.64) (0.02)
EXPRT 0.027*** 0.009* 0.020** 0.002
(6.54) (1.81) (2.51) (0.29)
LEGAL −0.499*** −0.599*** −0.172 −0.654***
(−6.19) (−6.13) (−1.18) (−5.04)
LOCAT −0.398*** −0.361*** 0.362*** −0.895***
(−5.03) (−4.46) (3.21) (−7.95)
OWNP −0.057*** −0.058*** −0.003 −0.074***
(−4.47) (−4.39) (−0.18) (−4.52)
OWNF −0.037*** −0.050*** 0.013 −0.069***
(−2.79) (−3.56) (0.67) (−3.96)
OWNG −0.044** −0.034 0.009 −0.040
(−2.32) (−1.62) (0.38) (−1.25)
OWND 0.001 0.007* −0.017*** 0.022***
(0.32) (1.81) (−2.80) (4.03)
adj. R
2
0.0017 0.0090 0.0023 0.0028 0.0067 0.0036 0.0020 0.0028 0.021 0.017 0.022
Hansen’sJ0.011 0.438 0.438 0.002 0.009 0.002 0.274 0.051 0.016 3.992 3.824
p( J) 0.918 0.508 0.508 0.994 0.925 0.964 0.601 0.821 0.900 0.045 0.051
n93,849 93,043 93,849 93,588 93,273 93,848 76,757 89,807 72,075 24,704 47,371
Notes: Outcome variable is GSALES. The estimation uses a GMM model with instrumental variables. The extent to which labor regulations affect an individual firm’s
business operations is used as an external instrument. tvalues in parentheses. *,**,***Significant at 10, 5 and 1 percent level, respectively
Table IV.
Labor skills, firm-
specific characteristics
and firm performance
338
IJM
40,2
microeconomic conditions in developing countries. The results tend to confirm those found
in the productivity-enhancement approach (Haskel and Martin, 1996; Bennett and
McGuinness, 2009; Healy et al., 2015), even though the latter are drawn from individual
countries and different data.
An important question is whether the impact of skilled labor constraints depends on
labor structures internal to the firm. In order to study this question, Equation (1) is re-
estimated to include separately the impact of labor structure information at the firm level
that is available in the ES. These include the ratio of unskilled over skilled permanent
production workers in the firm (RSKILL), the ratio of non-production over production
permanent workers in the firm (RPROD) and the ratio of temporary over permanent
full-time workers in the firm (RPERM). Table V reports the results of the estimation (models
B.1–B.6). Access to skilled labor remains a robust positive predictor of sales growth.
The magnitude of the effect is small but significant. It also varies considerably between
high- and low-income developing countries. However, the predictive power of the internal
labor structure indicators is weak. The ratios of non-production over production permanent
workers in the firm and of temporary over permanent full-time workers in the firm do not
appear to affect sales growth. The ratio of temporary over permanent workers in the firm is
significant only in high-income countries. It appears that temporary employment rises
relative to permanent one along with the level of a country’s economic development. On the
other hand, the ratio of unskilled over skilled permanent production workers in the firm is a
significant positive predictor of sales growth. Surprisingly, the higher the percentage of
unskilled labor in the firms’labor force, the higher the growth of firms’sales. This is so in
both high- and low-income countries. This finding may be explained by the informal labor
conditions and depressed labor costs due to abundant and cheap labor in most developing
countries. It may also be explained by lax labor regulations. Further, the low-tech,
low-manufacturing structure of production in developing countries may also justify the
importance of unskilled labor. An ILO (2010) study highlights how different incentive-based
labor structures within the firm affect its productivity and long-term performance. This is
new evidence on developing countries, which deserves further analysis based on proper
accounting of the determinants of internal labor structures.
In order to see the sectoral distribution of the impact of labor skill constraints, Equation
(1) is re-estimated separately for six sectors of activity: manufacturing, services, retail and
wholesale commerce, construction and the rest of the economy. Table VI presents the results
of the baseline sectoral model separately for the firm-specific characteristics (Panel A) and
the internal labor structures (Panel B). The results in Panel A show a significant positive
association between access to skilled labor and sales growth for the manufacturing and
GSALES B.1 B.2 B.3 B.4 B.5 B.6
All countries High-income Low-income
SKILLS 0.020*** (8.50) 0.019*** (6.29) 0.019*** (6.02) 0.018*** (5.69) 0.011** (2.38) 0.021*** (4.72)
RPERM 0.090 (0.91) 0.064 (0.41) −0.924* (−1.82) 0.186 (1.16)
RPROD 0.137 (1.48) 0.106 (1.17) 0.176 (1.25) 0.066 (0.66)
RSKILL 0.114*** (3.14) 0.112*** (2.99) 0.135*** (2.69) 0.095* (1.79)
adj. R
2
0.0025 0.0018 0.0018 0.0025 0.0054 0.0013
Hansen’sJ0.002 0.209 0.157 0.293 4.460 4.809
p( J) 0.965 0.648 0.692 0.588 0.035 0.028
n89,664 49,658 46,315 44,437 14,668 29,769
Notes: Outcome variable is GSALES. The estimation uses a GMM model with instrumental variables. The
extent to which labor regulations affect an individual firm’s business operations is used as an external
instrument. tvalues in parentheses. *,**,***Significant at 10, 5 and 1 percent level, respectively
Table V.
Labor skills, internal
labor structure and
firm performance
339
Firm
performance
in developing
countries
GSALES Manufacturing Services Wholesale Retail Construction Other
Panel A. firm-specific characteristics
SKILLS 0.022*** (6.17) 0.013** (2.38) 0.010 (1.39) −0.012 (−0.43) 0.056 (1.48) 0.013 (0.77)
AGE −3.505*** (−13.85) −2.959*** (−6.46) −1.355** (−2.17) −9.368*** (−4.45) −5.212** (−2.06) −4.338*** (−3.05)
SIZE 1.622*** (8.45) 2.450*** (7.77) 2.560*** (5.56) 3.426** (2.22) 3.190* (1.92) 0.002 (0.00)
EXPRT 0.004 (0.67) 0.034*** (2.64) 0.092** (2.16) −0.013 (−0.25) −0.196 (−0.39) −0.013 (−0.25)
LEGAL −0.309** (−2.52) −0.973*** (−4.76) −1.239*** (−4.15) −2.935*** (−3.61) 0.698 (0.78) 0.261 (0.33)
LOCAT −0.203* (−1.90) −0.690*** (−4.30) −0.761*** (−3.72) 2.200*** (2.70) 2.693* (1.87) 0.530 (0.82)
OWNP −0.044** (−2.34) −0.078*** (−3.17) −0.089*** (−2.79) −0.042 (−0.59) 0.348 (1.08) 0.069* (1.95)
OWNF −0.031 (−1.58) −0.079*** (−3.05) −0.086** (−2.44) 0.090 (0.88) 0.485 (1.34) 0.073* (1.85)
OWNG −0.021 (−0.75) −0.099** (−2.33) 0.080 (1.43) −0.353 (−1.51) 0.303 (0.92) −0.090 (−0.47)
OWND −0.004 (−0.85) 0.020** (2.32) 0.039*** (3.42) −0.131*** (−3.48) 0.114*** (2.91) 0.018 (0.73)
adj. R
2
0.0134 0.0115 0.0161 0.0112 0.0625 0.0197
Hansen’sJ1.363 0.017 2.485 5.483 0.827 3.171
p( J) 0.243 0.897 0.115 0.019 0.363 0.075
n40,935 18,576 9,830 1,254 814 666
Panel B. internal labor structures
SKILLS 0.017*** (5.17) −0.026 (−0.26) −0.026 (−0.26) −0.026 (−0.26) −0.026 (−0.26) 0.009 (0.69)
RPERM 0.052 (0.33) 24.951 (1.31) 24.951 (1.31) 24.951 (1.31) 24.951 (1.31) 0.168 (0.12)
RPROD 0.097 (1.09) 1.930 (1.52) 1.930 (1.52) 1.930 (1.52) 1.930 (1.52) 1.613* (1.91)
RSKILL 0.107*** (2.76) −0.768 (−0.59) −0.768 (−0.59) −0.768 (−0.59) −0.768 (−0.59) 0.164* (1.66)
adj. R
2
0.0101 0.0103 0.0132 0.0107 0.0581 0.0173
Hansen’sJ0.902 0.608 0.608 0.608 0.608 6.147
p( J) 0.342 0.436 0.436 0.436 0.436 0.013
n42,786 82 82 82 82 1,324
Notes: Outcome variable is GSALES. The estimation uses a GMM model with instrumental variables. The extent to which labor regulations affect an individual firm’s
business operations is used as an external instrument. tvalues in parentheses. *,**,***Significant at the 10, 5 and 1 percent level, respectively
Table VI.
Labor skills and firm
performance by
sector of activity
340
IJM
40,2
services sectors only. On the other hand, access to labor skills does not predict sales
growth in the commercial, construction and other sectors of the economy. These latter
sectors appear to be adequately served by low-skill local workforce that is widely
abundant in developing countries. This is reasonable to expect, as sales growth in most
non-manufacturing sectors depends mainly on distribution and other non-production
channels, which in turn do not greatly depend on high-level skills. For example, a recent
report on the commercial sector in Europe documents considerable differences in the
dynamics of the manufacturing, services and commerce sectors with significant
implications for labor requirements. Moreover, the role of firm-specific characteristics
differs among sectors of activity as regards both their significance and direction of
influence. Firm age, size and location of operation seem to be robust predictors of sales
growth across sectors. The results in Panel B show that, when the internal labor structures
are accounted for, access to skilled labor predicts sales growth only in the manufacturing
sector. Moreover, the unskilled vs skilled workers ratio is the only significant factor and this
is true only in manufacturing. Overall, the results document that the relation between access
to skilled labor and firms’sales growth differs across sectors of activity in the developing
countries and it is relatively more robust in the manufacturing sector. If the association is to
be better understood, additional sectoral factors, such as regulation, competition and
concentration of firms across sectors, should be accounted for.
An interesting question is the geographical manifestation of the relation between access
to labor skills and sales growth. The determinants of labor skills and firm performance vary
across countries throughout the world for various factors governing the availability of
skills, the structure of production and sales practices. In order to investigate the
geographical impact of access to skilled labor on sales growth, Equation (1) is re-estimated
for each of the main UN-classified regions: Africa, East Asia and the Pacific, Central and
East Asia, Europe, Middle-East and North Africa and South-East Asia. Table VII presents
the results for each region. Considerable differences emerge across geographical regions.
Access to skilled labor appears to be a significant predictor of sales growth in firms
operating only in Africa, East and Central Asia and Europe. Where significant, access to
labor skills improves sales growth. However, the magnitude of the effect differs
considerably from 3.4 percent in Africa to 1.8 percent in Europe. Further, the role of
firm-specific characteristics differs too across regions. Firm age, size and location of
operation seem to be robust predictors across regions. Similarly, when the firms’internal
labor structures are accounted for, access to skilled labor is a significant predictor of sales
growth in firms operating only in Africa and East Asia and the Pacific. Further, only the
ratio of unskilled over skilled labor is significant. These regional differences may also be
explained by the considerable labor skill mismatch around the world observed with respect
to not only job seekers vs employment opportunities, but also to the different methods of
measuring and assessing labor skill constraints (ILO, 2014). These differences may also be
the result of choice of invalid instruments for some models, which are associated with high
values of the Jstatistic. Nevertheless, this new evidence shows a strong geographical
dimension in the analysis of access to skilled labor and sales growth that deserves
further analysis, taking into consideration trade and investment flows, the technological
characteristics of infrastructure and the social and environmental conditions on a
global scale.
5. Country-level factors and robustness checks
There may be concerns that the empirical results obtained above depend on the nature of
labor skill constraints within the firm. Macroeconomic and institutional conditions have a
role to play too. They affect domestic patterns of consumption, the organization of
production and the effectiveness of infrastructure and technological progress among
341
Firm
performance
in developing
countries
GSALES Africa East-Asia and Pacific East and Central Asia Europe Latin America Middle-East and North Africa South-East Asia
Panel A. firm-specific characteristics
SKILLS 0.034*** (3.87) −0.003 (−0.35) 0.023*** (3.83) 0.018* (1.94) −0.006 (−1.44) −0.004 (−0.47) 0.004 (0.57)
AGE −1.105* (−1.82) −3.583*** (−6.78) −6.966*** (−10.65) −7.231*** (−6.88) −4.681*** (−11.15) −0.160 (−0.28) −3.132*** (−8.95)
SIZE 1.397*** (2.67) 1.418*** (3.95) 2.284*** (5.93) 2.324*** (4.30) 2.123*** (6.81) 3.822*** (8.66) 1.444*** (5.30)
SECTR 0.353* ( (1.65) −0.160 (−0.56) 0.239 (0.80) −0.368 (−0.73) 0.073 (0.24) 0.689** (2.12) 0.174 (0.64)
EXPRT 0.011 (0.51) −0.002 (−0.14) 0.030** (2.38) 0.059*** (3.66) −0.038*** (−3.24) 0.039*** (3.24) 0.011 (1.13)
LEGAL −1.227*** (−4.01) 0.722*** (3.11) 0.547* (1.76) 0.365 (0.94) 0.010 (0.05) 0.485* (1.77) −0.339* (−1.65)
LOCAT −4.069*** (−11.87) 0.282 (1.05) −0.647*** ( (−3.32) −0.258 (−0.98) −0.372*** (−2.79) 0.734*** (3.77) −0.003 (−0.02)
OWNP −0.093*** (−4.60) 0.029 (1.12) −0.083** (−2.55) 0.007 (0.24) 0.014 (0.47) −0.019 (−0.47) −0.102* (−1.86)
OWNF −0.070*** (−3.20) 0.003 (0.11) −0.075** (−2.14) −0.004 (−0.11) 0.004 (0.14) 0.022 (0.51) −0.012 (−0.20)
OWNG −0.010 (−0.14) 0.026 (0.83) −0.054 (−1.30) −0.055 (−0.91) 0.145 (1.46) −0.018 (−0.31) −0.068 (−0.68)
OWND 0.037** (2.50) −0.002 (−0.24) 0.008 (0.77) −0.016 (−1.12) 0.006 (0.77) −0.073*** (−6.73) 0.018** (2.27)
adj. R
2
0.0154 0.0137 0.0271 0.0179 0.0692 0.0127 0.0154
Hansen’s
J
0.713 14.758 0.537 0.003 8.654 3.446 1.331
p( J) 0.398 0.001 0.464 0.955 0.003 0.063 0.249
n14,822 12,155 11,158 4,044 11,650 6,201 12,045
Panel B. internal labor structures
SKILLS 0.032*** (2.97) 0.034*** (2.67) 0.014 (1.58) 0.014 (1.02) −0.002 (−0.33) 0.016 (1.39) 0.007 (1.02)
RPERM −0.268 (−0.68) 0.382 (1.27) 1.246 (1.03) 7.141*** (3.67) 0.339 (1.05) −1.442 (−1.57) 0.019 (0.04)
RPROD −0.198 (−1.11) 1.029** (2.33) 0.039 (0.13) −0.784 (−1.09) 0.219* (1.76) 0.630 (1.02) 0.357 (0.95)
RSKILL 0.297*** (2.82) 0.184*** (2.64) 0.056 (0.35) 0.151 (0.94) −0.018 (−0.36) 0.217* (1.87) −0.148** (−2.19)
adj. R
2
0.0017 0.0045 0.0053 0.0024 0.0228 0.0016 0.0017
Hansen’s
J
0.020 9.430 1.018 1.140 6.034 5.362 1.123
p( J) 0.887 0.002 0.313 0.286 0.014 0.021 0.289
n8,012 7,816 4,763 1,486 10,187 3,208 8,965
Notes: Outcome variable is GSALES. The estimation uses a GMM model with instrumental variables. The extent to which labor regulations affect an individual firm’s
business operations is used as an external instrument. tvalues in parentheses. *,**,***Significant at 10, 5 and 1 percent level, respectively
Table VII.
Labor skills and firm
performance by
geographical region
342
IJM
40,2
countries, all of which affect labor market outcomes. Therefore, it is possible that, by
accounting for the impact of economic and institutional country-level factors, the association
between access to skilled labor and firm performance is affected. In order to test for these
conjectures, a series of additional robustness tests in groups are applied. Based on the
literature suggestions, Equation (1) is expanded to include separately the impact of a
country’s: level of economic, financial and human development as well as its openness;
access to and quality of education and aspects of technological progress. The variables
corresponding to each of the above groups of factors are described in Table AI. Correlation
checks preceded the choice of variables in each group to minimize collinearity problems.
Interaction effects are included to capture the dynamic influences.
5.1 Impact of the development, finance and openness conditions
Economic development and growth dynamics in open economies affect the size and
composition of labor skills demanded by firms. Development is associated with the rise and
decline of industrial sectors, the introduction of new technology and structural
transformation in the production and distribution of goods and services. The ILO (2010)
argues that economic growth in the future will depend more heavily than today on the
productivity of the workforce, complemented by longer labor force participation rates,
especially among women and older workers. Shierholz (2014) notes that, by determining the
overall level of aggregate demand, economic growth rather than structural changes affect
the demand for labor skills. Murphy et al. (1991) show that important recent financial
innovations led to the development of new markets and high incomes, which changed the
demand for skills in the financial sector, which has become an attractive destination of
highly talented individuals. Finally, Tybout notes that a country’s economic openness tends
to enhance labor productivity. In order to test for some of these conjectures, Equation (1) is
expanded to include regressors capturing the impact of economic, financial and human
development and an open environment of the sample countries. Table VIII presents the
results of the baseline model after controlling sequentially for economic growth (GDPG),
economic development (GDPCAP), financial development (CREDIT), income inequality
(GINI), human development (HDI), the extent of globalization (GLOBAL) and the openness
of capital account (FINOPEN) (models C.1–C.10). The results show that, when the individual
effects are taken into consideration, access to labor skills is a significant predictor of sales
growth in few models. The rate of economic growth and economic and human development
in a country do not appear to affect the association between labor skills and sales growth in
developing countries. In contrast, financial development turns out to be significant with a
strong positive interaction effect that points to accelerating influence. It seems that access to
finance and financial development mitigate strongly the effect of access to skilled labor on
sales growth and it may be an important policy factor in the labor market of developing
countries. Further, the openness of the capital account does not appear to mitigate the
impact of access to skilled labor, but higher globalization does so. However, the latter is
associated with a negative coefficient for both access to skilled labor and itself. The luck of
significance may also be attributed to instrumental variables misspecification as the large
value of the Jstatistic shows for some models. It is important to note that while the Jstat
may reject the null hypothesis, it does not provide guidance on the likely sources of the
model’s misspecification. This leads to the inclusion of all firm-specific characteristics and
controls as regressors. In this case, access to labor skills turns significant and positive for
both high- and low-income developing countries. The Jstatistic does not reject the null
hypothesis and therefore the models do not suffer serious misspecification bias. The impact
of GDP growth is significant but with a positive sign for low-income countries and a
negative one for high-income countries. Economic development is also significant but with
inverse signs for low- and high-income countries. Both the Gini and human development
343
Firm
performance
in developing
countries
GSALES C.1 C.2 C.3 C.4 C.5 C.6 C.7 C.8 C.9 C10
All countries High-income Low-income
SKILLS 0.015 −6.352 0.103 0.784 −1.259*** −2.470 −4.608*** 0.017*** 0.021*** 0.004**
(0.26) (−1.32) (0.48) (0.53) (−4.37) (−1.57) (−3.09) (3.84) (3.21) (2.13)
GDPG −0.080 −0.376 −2.259*** 4.855***
(−0.04) (−1.52) (−8.88) (10.81)
SKILLS ×GDPG −0.549
(−0.13)
GDPCAP −14.103 −8.608*** 0.675 −8.234***
(−1.27) (−14.91) (0.47) (−6.67)
SKILLS ×GDPCAP 70.020
(1.32)
GINI 0.171 0.051** 0.004 0.028
(1.50) (2.13) (0.12) (0.45)
SKILLS ×GINI −0.236
(−0.45)
HDI 34.409 0.662 −2.282 26.447***
(0.74) (0.19) (−0.58) (3.93)
SKILLS ×HDI −14.79
(−0.52)
CREDIT 1.805** 4.563*** 7.712*** 3.515***
(−2.18) (14.37) (13.99) (4.57)
SKILLS ×CREDIT 34.81***
(4.42)
FINOPEN 19.188 −0.005 −1.370*** −0.852
(1.16) (−0.02) (−3.08) (−1.61)
SKILLS ×FINOPEN −75.68
(−1.13)
GLOBAL −1.481*** 0.626*** 0.569*** 0.492***
(−2.62) (18.34) (11.59) (6.48)
SKILLS ×GLOBAL 7.805***
(3.10)
RPERM 0.182 0.470 0.162 0.161 0.362 1.475 0.385 0.320 1.397** 0.406*
(0.80) (1.55) (0.72) (0.53) (1.54) (1.30) (1.49) (1.59) (2.36) (1.86)
(continued )
Table VIII.
Impact of economic
development, finance
and globalization
344
IJM
40,2
GSALES C.1 C.2 C.3 C.4 C.5 C.6 C.7 C.8 C.9 C10
All countries High-income Low-income
RPROD 0.335* 0.665** 0.259 0.211 0.439** 0.887 0.532** 0.282 0.278 0.284
(1.84) (2.11) (1.47) (1.00) (2.24) (1.61) (2.08) (1.45) (1.30) (0.85)
RSKILL 0.021 −0.040 0.011 0.020 −0.054 −0.293 −0.097 −0.073* −0.013 −0.135***
(0.49) (−0.57) (0.23) (0.36) (−1.09) (−1.26) (−1.27) (−1.77) (−0.21) (−2.59)
Firm-specific characteristics Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
adj. R
2
0.032 0.080 0.024 0.040 0.116 0.038 0.065 0.039 0.032 0.080
Hansen’sJ6.251 0.199 10.643 21.053 1.437 7.025 3.411 3.504 2.758 1.043
p( J) 0.012 0.656 0.001 0.000 0.231 0.003 0.066 0.055 0.034 0.594
n22,184 21,648 22,184 22,184 21,674 21,611 22,142 21,075 7,479 13,596
Notes: Outcome variable is GSALES. The estimation uses a GMM model with instrumental variables. The extent to which labor regulations affect an individual firm’s
business operations is used as an external instrument. The coefficients of firm-specific characteristics are not reported. tvalues in parentheses. *,**,***Significant at 10, 5
and 1 percent level, respectively
Table VIII.
345
Firm
performance
in developing
countries
measures are significant and positive for both low- and high-income developing countries.
Financial openness is significant only for high-income countries, whilst the extent of
globalization seems to matter for both low- and high-income countries. Internal labor
structures appear to be somewhat important in predicting the impact of access to skilled
labor on sales growth, except for the ratio of non-production to production workers. Thus,
the relative proportion of non-permanent workers appears to affect positively sales growth,
whilst the relative proportion of unskilled workers appears to have a negative effect in
low-income countries. Overall, the levels of economic and financial development as well as
globalization of countries appear to mitigate the association between access to skilled labor
and sales growth of firms in developing counties. This impact does however differ between
low- and high-income countries. Thus, the role of access to skilled labor in determining firm
performance in developing countries can be better understood by taking into consideration
the mitigating impact of national economic, financial and human development levels. The
results do not contradict the skill-biased technological progress hypothesis and document a
significant role of a more globalized environment.
5.2 Impact of the educational environment
The extent of access to education as well as the quantity and quality of provided
education affect the demand and supply of labor skills. The ILO (2010) argues that in
many parts of the developing world, the growing size of the youth population challenges
both existing education and training capacities as more young people enter the world of
work. The importance of schooling for raising labor skills has long been documented
(see Psacharopoulos, 1994, for a survey). In order to check for some of these conjectures,
Equation (1) is expanded to include regressors capturing the impact of a country’s
various aspects of quantity and quality of education. Table IX presents the results of the
baseline regression model after controlling for a country’s years of schooling of adult
population (SCHLNG), level of adult literacy (LTRCY), quality of education (QLEDUC),
extent of having access to advanced education (ACCEDUC), extent of inequality in
attaining education (EDINEQ) and the extent of reliance on professional management
(PROFMNG) (models D1.–D.9). The results show a mixed impact. When control factors
are individually considered, access to skilled labor remains a significant predictor of sales
growth only after controlling for years of schooling, reliance on professional
management, access to advanced education and inequality in attaining education.
Their interaction effect is also high and significant indicating strong dynamic influence.
Access to skilled labor appears to be insignificant when the quality of education and the
literacy rate are considered. Further, all education control variables appear to be
significant, except for the quality of education and the literacy rate. However, even when
access to skilled labor is significant, the respective Jstatisticisforsomemodelstoohigh
casting doubt on the validity of the estimates. After including all control variables,
surprisingly access to skilled labor ceases to be a significant predictor of firms’sales
growth, except for low-income countries. Again, given that the Jstatisticistoohighin
these models where access to skilled labor is insignificant, the latter may be due to
instrument misspecification. However, the results document an important role for the
educational environment in mitigating the association between access to skilled labor
and sales growth in low-income countries. Implicitly this tends to confirm the findings of
Easterly and Pritchett, who raised caution that education may not have much impact on
skills in less-developed countries that lack other supporting factors, such as functioning
institutions for markets and legal systems. These results tend to contradict the findings
of Hanushek and Woessmann, who argue that initially low-skill levels retard the
effectiveness of subsequently higher skill training and education. The evidence on
developing countries seems different from that in the developed ones.
346
IJM
40,2
GSALES D.1 D.2 D.3 D.4 D.5 D.6 D.7 D.8 D.9
All countries High-income Low-income
SKILLS −0.451** (−2.45) −1.356 (−1.63) 1.059** (2.44) −2.396*** (−2.59) −0.402** (−2.23) −1.839 (−1.45) 0.002 (0.41) 0.009 (1.40) 0.001* (1.87)
SCHLNG −1.599** (−2.50) −0.060 (−0.53) 0.358** (1.98) −0.330* (−1.70)
SKILLS ×SCHLNG 7.647** (2.50)
QLEDUC −6.801 (−1.54) 0.485 (1.01) 4.777*** (3.25) 0.064 (0.07)
SKILLS ×QLEDUC 35.580* (1.65)
PROFMNG 6.104*** (2.66) 1.296*** (3.49) 0.260 (0.44) 2.434** (2.40)
SKILLS ×PROFMNG −23.983** (−2.41)
ACCEDUC−−0.590** (−2.34) 0.240*** (12.41) 0.124*** (2.84) 0.334*** (10.97)
SKILLS ×ACCEDUC 3.283*** (2.60)
EDINEQ −44.621*** (−2.76) −9.299*** (−3.56) −33.658** (−2.28) −12.27*** (−2.82)
SKILLS ×EDINEQ 178.875** (2.27)
LTRCY −0.347 (−1.32) −0.270*** (−8.08) −1.921*** (−4.24) −0.330*** (−7.62)
SKILLS ×LTRCY 2.158 (1.46)
RPERM 0.256 (1.28) 0.366* (1.83) 0.359* (1.70) 0.818** (2.44) −0.173 (−0.53) 0.165 (0.80) 0.580*** (2.91) 1.137 (1.40) 0.510** (2.54)
RPROD 0.344* (1.87) 0.314* (1.66) 0.256 (1.45) 0.451* (1.75) 0.045 (0.24) −0.003 (−0.03) 0.065 (0.38) 0.044 (0.27) 0.208 (0.62)
RSKILL −0.020 (−0.44) 0.052 (0.91) 0.024 (0.50) −0.158* (−1.76) 0.031 (0.47) −0.039 (−0.67) −0.057 (−1.29) −0.053 (−0.77) −0.033 (−0.58)
Firm-specific
characteristics
Yes Yes Yes Yes Yes Yes Yes Yes Yes
adj. R
2
0.023 0.012 0.020 0.011 0.031 0.056 0.026 0.016 0.068
Hansen’sJ0.105 4.621 1.453 2.133 9.366 0.001 7.822 8.820 0.040
p( J) 0.746 0.032 0.228 0.304 0.001 0.523 0.005 0.003 0.841
n22,184 21,200 21,200 22,076 20,057 34,333 19,181 6,242 12,939
Notes: Outcome variable is GSALES. The estimation uses a GMM model with instrumental variables. The extent to which labor regulations affect an individual firm’s business operations is
used as an external instrument. The coefficients of firm-specific characteristics are not reported. tvalues in parentheses. *,**,***Significant at 10, 5 and 1 percent level, respectively
Table IX.
Impact of education
conditions
347
Firm
performance
in developing
countries
5.3 Impact of technological progress
The pace and diffusion of technological progress affects the quality and composition of
labor skills demanded by firms. Acemoglu argues that labor skills and inequality are linked
through technology and its impact on the demand for high-skill labor. Although other
factors, such as changes in the minimum wage, the extent of unionization and globalization,
play a role in affecting skilled labor, technology is the major driving force. This is due to the
existence of technology-skill complementarities. Technical change favors more skilled
(educated) workers, replaces tasks previously performed by the unskilled ones and
increases the demand for skills. Machin and van Reenen (1998) suggest that technological
factors alone can account for a third or less of labor skill changes in the USA and UK, but far
more elsewhere. They obtain results that broadly support the skill bias hypothesis across
countries. In order to check for some of these conjectures, Equation (1) is expanded to
include regressors capturing the impact of a country’s various aspects of technological
progress. Table X presents the results of the baseline regression model after controlling for a
country’s efficient use of talent (USETLNT), the extent of firm-level technology adoption
(TECHABS), the pace of technological innovation (TECHINNV ), the extent of technological
achievement (TECHACHV ) and the level of knowledge economy (KEI) (models E.1–E.8).
The results show a significant positive effect of access to skilled labor in almost all models.
The magnitude of the effect is quite strong too. The individual effects of the technological
progress variables are significant, except for the knowledge economy index. The interaction
effects are significant and strong indicating strong dynamic influences. When all
technological control variables are included in the regression, access to skilled labor remains
positive and significant with a lower quantitative effect. The direct impact of the technology
variables changes too. The talent index and the technology absorption index remain
significant only in low-income developing countries. This implies that countries at lower
stages of development make better use of talent enhancement in boosting competitiveness
and performance - an unexpected result. On the other hand, technology absorption appears
to act as a constraint on sales growth. Technological achievement is more pronounced in
high-income developing countries but not conducive to higher sales growth. Apart from
confirming the importance of access to skilled labor, these results show a quite diverse
influence of the technological progress variables. As some of the models are associated with
high values of the Jstatistic, this diverse picture may be the result of instrument
misspecification. Nevertheless, even in models with very low values of the Jstatistic, the
sign and magnitude of coefficients change among models. It appears that the role of
technological progress in affecting firm performance is complex and could be better
understood if analyzed along a multidimensional framework of influences and interactions.
The data for developing countries appear to suggest that some technological progress
indices are complementary with skilled labor changes, whilst others are not. Based on this
evidence, it is hard to say whether skill-biased technological change is present in developing
countries. The results tend to align with those of Goux and Maurin who do not find robust
evidence for France and we share the concerns of Doms et al. and Dunne et al. (1997)
who stress the importance of unobservable influences in driving the interaction of skills and
technological change.
6. Conclusions
Access to skilled labor affects the structure and cost of production of firms and provides the
capabilities required for productivity growth and efficient strategic decisions thereby
affecting their performance. Based on a large and consistent data set from World Bank’s
Enterprise Surveys, the paper explores the empirical relation between access to skilled labor
and firm performance, measured by the annual growth of sales, in 138 developing countries.
The results document a significant positive relation between access to skilled labor and
348
IJM
40,2
GSALES E.1 E.2 E.3 E.4 E.5 E.6 E.7 E.8
SKILLS All countries High-income Low-income
SKILLS 2.120*** (2.80) 0.852** (2.34) 2.014** (2.55) −2.352*** (−2.86) −0.435 (−1.24) 0.017*** (4.35) 0.013** (2.32) 0.019*** (3.57)
USETLNT 8.387*** (2.71) 0.889*** (3.39) 0.369 (0.60) 1.320*** (3.52)
SKILLS ×USETLNT −53.61*** (−2.78)
TECHABS 6.230*** (3.53) 0.149 (0.42) −0.638 (−0.77) −0.854* (−1.76)
SKILLS ×TECHABS −18.16** (−2.31)
TECHINNOV 11.215** (2.54) −1.404*** (−3.61) −1.272 (−1.37) −0.169 (−0.35)
SKILLS ×TECHINNV −60.952** (−2.53)
TECHACHV −7.610*** (−2.59) −8.476*** (−3.64) −31.79*** (−5.27) 4.378 (1.31)
SKILLS ×TECHACHV 32.216*** (2.87)
KEI −0.639 (−0.37) 1.359*** (6.34) 2.305*** (6.47) −0.216 (−0.51)
SKILLS ×KEI 9.800 (1.25)
RPERM 0.377* (1.92) 0.317 (1.52) 0.357* (1.69) 0.881** (2.41) 0.552*** (2.83) 0.087 (0.49) 1.337** (2.39) 0.006 (0.03)
RPROD 0.256 (1.32) 0.224 (1.30) 0.157 (0.86) 0.570* (1.94) 0.269 (1.40) 0.014 (0.17) 0.187 (1.07) −0.022 (−0.26)
RSKILL −0.010 (−0.20) 0.023 (0.51) −0.043 (−0.89) −0.101 (−1.35) −0.060 (−1.23) −0.019 (−0.50) 0.037 (0.66) −0.063 (−1.20)
Firm-specific
characteristics
Yes Yes Yes Yes Yes Yes Yes Yes
adj. R
2
0.034 0.027 0.056 0.032 0.043 0.062 0.057 0.045
Hansen’sJ1.497 0.104 0.608 4.305 6.828 1.233 2.448 3.221
p( J) 0.221 0.747 0.436 0.104 0.009 0.267 0.163 0.113
n21,200 21,200 21,200 22,015 21,651 32,041 10,978 21,063
Notes: Outcome variable is GSALES. The estimation uses a GMM model with instrumental variables. The extent to which labor regulations affect an individual firm’s
business operations is used as an external instrument. The coefficients of firm-specific characteristics are not reported. tvalues in parentheses. *,**,***Significant at 10, 5
and 1 percent level, respectively
Table X.
Impact of
technological progress
conditions
349
Firm
performance
in developing
countries
sales growth that remains broadly robust after various controls at both the firm and country
levels. The firm-specific characteristics appear to be significant predictors of firms’sales
growth. However, their inclusion does not substantially affect the relation between access to
skilled labor and sales growth. Sectoral and geographical influences are documented too.
Internal labor structures seem to be less important in predicting sales growth, except for the
ratio of unskilled over skilled workers, which appears to be relatively robust in predicting
sales growth. The impact of access to skilled labor on sales growth remains robust in most
models after controlling for the impact of various country-level economic and non-economic
factors; however, the magnitude of the effect varies considerably among the different
controls as well as between low- and high-income countries. Economic, financial and human
development, financial openness and globalization, the extent of access to education and its
quality as well as different technological factors appear to exert a varying mitigating effect
on the relation between access to skilled labor and sales growth. The evidence shows that
institutions and technology matter. They affect the supply of and demand for skilled labor
and thereby firm performance across sectors and regions. Once their role is taken into
consideration, a more solid basis for understanding labor market phenomena, such as the
mismatch between job seekers and employment opportunities and skill-biased technical
change, may be obtained for the developing countries. Finally, the analysis would need an
extension to include the role of regulation, global trade and investment, as well as the impact
of national culture and political conditions that influence the interaction of labor market
institutions and firm performance.
References
Ahmad, S. and Schroeder, R.G. (2003), “The impact of human resource management practices on
operational performance: recognizing country and industry differences”,Journal of Operations
Management, Vol. 21 No. 1, pp. 19-43.
Ahmed, M.U., Kristal, M.M. and Pagell, M. (2014), “Impact of operational and marketing capabilities on
firm performance: evidence from economic growth and downturns”,International Journal of
Production Economics, Vol. 154, pp. 59-71.
Arellano, M. and Bover, O. (1995), “Another look at the instrumental variable estimation of
error-components models”,Journal of Econometrics, Vol. 68 No. 1, pp. 29-51.
Banerji, A., Cunningham, W., Fiszbein, A., King, E., Patrinos, H., Robalino, D. and Tan, J.-P. (2010),
Stepping up Skills for More Jobs and Higher Productivity, World Bank, Washington DC,
available at: http://documents.worldbank.org/curated/en/2010/06/12515032/stepping-up-
skillsmore-jobs-higher-productivity
Barney, J.B. and Wright, P.W. (1998), “On becoming a strategic partner: the role of human resources in
gaining competitive advantage”,Human Resource Management, Vol. 37 No. 1, pp. 31-46.
Bartel, A., Ichniowski, C. and Shaw, K. (2007), “How does information technology affect productivity?
Plant-level comparisons of product innovation, process improvement and worker skills”,
Quarterly Journal of Economics, Vol. 122 No. 4, pp. 1721-1758.
Bennett, J. and McGuinness, S. (2009), “Assessing the impact of skill shortages on the productivity
performance of high-tech firms in Northern Ireland”,Applied Economics, Vol. 41 No. 6,
pp. 727-737.
Bloom, N. and Van Reenen, J. (2010), “Why do management practices differ across firms and
countries?”,Journal of Economic Perspectives, Vol. 24 No. 1, pp. 203-224.
Bloom, N., Darca, M. and van Reenen, J. (2015), “Trade induced technical change? The impact of
Chinese imports on innovation, IT and productivity”,Review of Economic Studies, Vol. 83 No. 1,
pp. 87-117.
Blundell, R. and Bond, S. (1998), “Initial conditions and moment restrictions in dynamic panel data
models”,Journal of Econometrics, Vol. 87 No. 1, pp. 115-143.
350
IJM
40,2
Bond, S.R., Bowsher, C. and Windmeijer, F. (2001), “Criterion-based inference for GMM in
autoregressive panel data models”,Economic Letters, Vol. 73 No. 3, pp. 379-388.
Brixiova, Z., Li, W. and Yousef, T. (2009), “Skill shortages and labor market outcomes in Central
Europe”,Economic Systems, Vol. 33, pp. 45-59.
Cai, F., Park, A. and Zhao, Y. (2008), The Chinese Labor Market in the Reform Era, Cambridge
University Press, Cambridge, pp. 167-214.
Chinn, M.D. and Ito, H. (2006), “What matters for financial development? Capital controls, institutions
and interactions”,Journal of Development Economics, Vol. 81 No. 1, pp. 163-192.
Dearden, L., Reed, H. and van Reenen, J. (2005), “The impact of training on productivity and wages:
evidence from British panel data”,Oxford Bulletin of Economics and Statistics, Vol. 68 No. 4,
pp. 397-421.
Fajnzylber, P. and Fernandes, A.M. (2009), “International economic activities and the demand
for skilled labor: evidence from Brazil and China”,Applied Economics, Vol. 41 No. 5,
pp. 563-577.
Fakih, A. and Ghazalian, P. (2015), “What factors influence firm perceptions of labor market
constraints to growth in the MENA region?”,International Journal of Manpower, Vol. 36 No. 8,
pp. 1181-1206.
Galindo-Rueda, F. and Haskel, J. (2005), “Skills, workforce characteristics and firm-level productivity:
evidence from the matched ABI/employer skills survey”, IZA Discussion Paper No. 1542,
March, Bonn.
Galunic, D.C. and Riordan, S. (1998), “Resource recombinations in the firm: knowledge structures and
the potential for Schumpeterian innovation”,Strategic Management Journal, Vol. 19 No. 12,
pp. 1193-1201.
Giesing, Y. and Laurentsyeva, N. (2014), Brain Drain and Firm Productivity: Evidence from the
Sequential Opening of EU Labour Markets, Springer.
Haskel, J. and Martin, C. (1996), “Skill shortages, productivity growth and wage inflation”, in Booth, A.
and Snower, D. (Eds), Acquiring Skills: Market Failures, their Symptoms and Policy Responses,
Cambridge University Press, Cambridge, pp. 147-174.
Healy, J., Mavromaras, K. and Sloane, P.J. (2015), “Adjusting to skill shortages in Australian SMEs”,
Applied Economics, Vol. 47 No. 24, pp. 2470-2487.
Hsieh, C.-T. and Klenow, P.J. (2009), “Misallocation and manufacturing TFP in China and India”,The
Quarterly Journal of Economics, Vol. 124 No. 4, pp. 1403-1448.
ILO (2010), A Skilled Workforce for Strong, Sustainable and Balanced Growth: A G20 Training Strategy,
The International Labor Organization, Geneva.
ILO (2012), International Standard Classification of Occupations ISCO-08, Vol. 1, The International
Labor Organization, Geneva.
Iranzo, S., Schivardi, F. and Tosetti, E. (2008), “Skill dispersion and firm productivity: an analysis
with employer-employee matched data”,Journal of Labor Economics, Vol. 26 No. 2, pp. 247-285.
Irvin, R.A. and Michaels, E.G. (1989), “Core skills: doing the right things right”,The McKinsey
Quarterly, Summer, pp. 4-19.
Kaiser, U. (2002), Innovation, Employment, and Firm Performance in the German Service Sector,
Springer-Verlang, Berlin.
Leuven, E., Oosterbeek, H. and van Ophem, H. (2004), “Explaining international differences in male skill
wage differentials by differences in demand and supply of skill”,Economic Journal, Vol. 114
No. 495, pp. 466-495.
Machin, S. and van Reenen, J. (1998), “Technology and changes in skill structure: evidence from seven
OECD countries”, Berkeley Centre for Labour Economics Working Paper No. 3, Berkeley.
Machin, S., Ryan, A. and van Reenen, J. (1996), “Technology and changes in skill structure: evidence
from and International panel of sectors”, Discussion Paper No. 4, The Labor Market
Consequences of Technical and Structural Change.
351
Firm
performance
in developing
countries
Mahy, B., Rycx, F. and Vermeylen, G. (2015), “Educational mismatch and firm productivity: do skills,
technology and uncertainty matter?”,De Economist, Vol. 163 No. 2, pp. 233-262.
Montt, G. (2015), “The causes and consequences of field-of-study mismatch: an analysis using PIAAC”,
OECD Social, Employment, and Migration Working Paper No. 167, Paris.
Murphy, K.M., Shleifer, A. and Vishny, R. (1991), “The allocation of talent: implications for growth”,
Quarterly Journal of Economics, Vol. 106 No. 2, pp. 503-530.
Nickell, S. (1981), “Biases in dynamic models with fixed effects”,Econometrica, Vol. 49 No. 6, pp. 1417-1426.
Nickell, S. and Bell, B. (1995), “The collapse in demand for the unskilled and unemployment across the
OECD”,Oxford Review of Economic Policy, Vol. 11 No. 1, pp. 40-62.
Ouimet, P. and Zarutskie, R. (2014), “Who works for startups? The relation between firm age, employee
age, and growth”,Journal of Financial Economics, Vol. 112 No. 3, pp. 386-407.
Pavcnik, N. (2002), “Trade liberalization, exit, and productivity improvements: evidence from Chilean
plants”,Review of Economic Studies, Vol. 69 No. 1, pp. 245-276.
Ployhart, R.E. (2006), “Staffing in the 21st century: new challenges and strategic opportunities”,Journal
of Management, Vol. 32 No. 6, pp. 868-897.
Poschke, M. (2014), “The firm size distribution across countries and skill-biased change in
entrepreneurial technology”, IZA Discussion Paper No. 7991, Institute of Labor Economics, Bonn.
Prasnikar, J., Svejnar, J., Dubravko, M. and Prasnikar, V. (1994), “Behavior of participatory firms in
Yugoslavia: lessons for transforming economies”,Review of Economics and Statistics, Vol. 76
No. 4, pp. 728-741.
Psacharopoulos, G. (1994), “Returns to investment in education: a global update”,World Development,
Vol. 22 No. 9, pp. 1325-1343.
Rehman, N.U. (2015), “‘Drivers of firms’growth: a case study of software firms in Islamabad/
Rawalpindi regions”,Journal of Management Development, Vol. 34 No. 8, pp. 901-921.
Roodman, D. (2009), “A note on the theme of too many instruments”,Oxford Bulletin of Economics and
Statistics, Vol. 71 No. 1, pp. 135-158.
Shierholz, H. (2014), “Is there really a shortage of skilled workers?”, in Palley, T.I. and Horn, G.A. (Eds),
Restoring Shared Prosperity: A Policy Agenda from Leading Keynesian Economists, CreateSpace
Independent Publishing, pp. 143-150.
Shleifer, A. and Vishny, R.W. (1994), “Politicians and firms”,Quarterly Journal of Economics, Vol. 109
No. 4, pp. 995-1025.
Snow, C.C. and Snell, S.A. (2011), “Strategic human resource management”, in Kozlowski, S.W.J. (Ed.), Oxford
Handbook of Industrial-Organizational Psychology, Oxford University Press, New York, NY, pp. 30-38.
Syverson, C. (2011), “What determines productivity”,Journal of Economic Literature, Vol. 49 No. 2,
pp. 326-365.
Tan, H.W., Sajitha, B. and Tanaka, N. (2016), “Skill use, skill deficits, and firm performance in formal
sector enterprises: evidence from the Tanzania enterprise skills survey”, Policy Research
Working Paper No. 7672, World Bank Group, Washington DC.
Teece, D.J. (2017), “A capability theory of the firm: an economics and (strategic) management
perspective”, Working Paper Series No. 20, Tusher Center on Intellectual Capital, University of
California, Berkeley, CA, April 24.
Teece, D.J., Pisano, G. and Shuen, A. (1997), “Dynamic capabilities and strategic management”,
Strategic Management Journal, Vol. 18 No. 7, pp. 509-533.
The International Labor Organization (2014), “Report III: Report of the Conference”,19th International
Conference of Labor Statisticians,Geneva,October 2-11, 2013.
The World Bank (2014), “STEP skills measurement surveys: innovative tools for assessing skills”,
Social Protection and Labor Discussion Paper No. 1421, Washington DC.
Windmeijer, F. (2005), “A finite sample correction for the variance of linear efficient twostep GMM
estimators”,Journal of Econometrics, Vol. 126 No. 1, pp. 25-51.
352
IJM
40,2
Further reading
Acemoglu, D. (1998), “Why do new technologies complement skills? Directed technical change and
wage inequality”,Quarterly Journal of Economics, Vol. 113 No. 4, pp. 1055-1089.
Besley, T. and Burguess, R. (2004), “Can labor regulation hinder economic performance? Evidence from
India”,Quarterly Journal of Economics, Vol. 119 No. 1, pp. 91-134.
Bombardini, M., Gallipoli, G. and Pupato, G. (2012), “Skill dispersion and trade flows”,The American
Economic Review, Vol. 102 No. 5, pp. 2327-2348.
The World Bank (2015), “Labor market impacts and effectiveness of skills development programs in
5 States in India: Assam, Andhra Pradesh, Madhya Pradesh, Odisha and Rajasthan”, Education
Global Practice Report No. 94682-IN, The World Bank, April.
Wolde, H. and Bhattacharya, M.R. (2010), “Constraints on trade in the MENA region”, International
Monetary Fund Working Paper No. 10-31.
(The Appendix follows overleaf.)
353
Firm
performance
in developing
countries
Appendix
Name Description and source
AGE Logarithm of the number of years since the year of a firm’s establishment, from the World
Bank Enterprise Surveys
SIZE Ordinal variable that takes the value of 1 if the firm is small (5–19 employees), 2 if it is
medium (20–99 employees) and 3 if it is large (over 99 employees), from the World Bank
Enterprise Surveys
SECTR Ordinal variable that takes the value of 1 if the firm is in manufacturing, 2 in services, 3 in
retail, 4 in wholesale, 5 in construction and 6 in other sectors, from the World Bank
Enterprise Surveys
EXPRT Percent of a firm’s total sales that are directly exported, from the World Bank
Enterprise Surveys
LEGAL Ordinal variable that takes the value of 1 if the firm is a traded joint-stock co., 2 if it is a
non-traded joint-stock co., 3 if it is a sole proprietorship, 4 if it is partnership, 5 if it is a limited
partnership and 6 otherwise, from the World Bank Enterprise Surveys
LOCAT Ordinal variable that takes the value of 1 if the firm is located in the capital city, 2 in a city of
more than 1m, 3 in a city between 1/4 and 1m, 4 in a city between 1/20 and 1/4 of 1m and 5 in
a city less than 1/20m, from the World Bank Enterprise Surveys
OWNP Percent of a firm’s equity capital owned by domestic private owners, from the World Bank
Enterprise Surveys
OWNF Percent of a firm’s equity capital owned by foreign owners, from the World Bank
Enterprise Surveys
OWNG Percent of a firm’s equity capital owned directly by the state, from the World Bank
Enterprise Surveys
OWND Percent of a firm’s equity capital owned by the largest owners, from the World Bank
Enterprise Surveys
RSKILL Ratio of unskilled over skilled permanent production workers in the firm, calculated from
data in from the World Bank Enterprise Surveys
RPROD Ratio of non-production over production permanent workers in the firm, calculated from data
in from the World Bank Enterprise Surveys
RPERM Ratio of temporary over permanent full-time workers in the firm, calculated from data in the
World Bank Enterprise Surveys
GDPG Logarithm of the annual GDP growth rate (%), from the World Bank Development Indicators
GDPCAP Logarithm of GDP per capita (current USD), from the World Bank Development Indicators.
It is generally considered a measure of a country’s level of economic development
GINI Gini coefficient, from the World Bank Development Indicators. It is a measure of a country’s
income distribution and it is generally considered an indicator of income inequality
HDI Index ranging from 0 to 1, with higher values corresponding to higher human
development, form the UNDP. It is a summary measure of average achievement in key
dimensions of human development: a long and healthy life, knowledge and decent
standard of living
CREDIT Logarithm of domestic credit to the private SECTR (% GDP), from the World Bank
Development Indicators. It is generally considered a measure of a country’s level of
financial development
GLOBAL The KOF Index of globalization, ranging from 0 to 100, comprises three dimensions of
globalization: economic, social and political. Higher values indicate higher globalization.
Published by the ETH Zurich
FINOPEN The Chinn-Ito index measures a country’s degree of capital account openness, from Chinn
and Ito (2006) (annually updated). This index takes on negative and positive values;
higher values imply that countries are more open to (fewer restrictions on) cross-border
capital transactions
SCHLNG Years of schooling of a country’s adult population, from the Global Competitiveness Report
(continued )
Table AI.
Definitions of
variables
354
IJM
40,2
Corresponding author
Charilaos Mertzanis can be contacted at: charilaos.mertzanis@aucegypt.edu
Name Description and source
LTRCY Percent of adult literacy, from the World Development Indicators. It reflects the percentage of
people ages 15 and above who can both read and write with understanding a short simple
statement about their everyday life
QLEDUC Index of the quality of education, ranging from 0 to 7, from the Global Competitiveness
Report. It reflects a country’s education achievements based on education results (e.g. PISA
tests, etc.)
ACCEDUC Index of access to advanced education, ranging from 0 to 7, from the Global Competitiveness
Report. It reflects a country’s extent of advanced education based on the percent of people
receiving advanced education
EDINEQ Index of inequality in attaining education, ranging from 0 to 7, from the Global
Competitiveness Report. It reflects a country’s extent of inequality in attaining education
based on test scores, poverty levels and other issues
PROFMNG Index of reliance on professional management, ranging from 0 to 7, from the Global
Competitiveness Report. It reflects the extent of a country’choice of professional business
management based on merit and qualifications
USETLNT Index of the extent of effective use of talent, ranging from 0 to 100, from the Global
Competitiveness Report. It measures the extent to which countries grow, attract and retain
talent for boosting their talent competitiveness
TECHABS Index of the extent of technological absorption at the firm level, ranging from 0 to 7, from the
WEF Global Competitiveness Report. It reflects the extent to which business adopt new
technology in their business activity
TECHINNV Index of technological innovation, ranging from 0 to 100, from the WEF Global
Competitiveness Report. It reflects the extent of technological innovation based on both the
innovation input and output indices, each comprising several factors
TECHACHV Index of technological achievement, ranging from 0 to 1, from the United Nations
Development Program. It reflects the extent of a country’s creation and diffusion of
technology and building of human skills in order to facilitate technological innovation in the
network age
KEI Index of the level of knowledge economy, ranging from 0 to 10, from the World Development
Indicators. It reflects a country’s ability to generate, adopt and diffuse knowledge for
facilitating economic development
LABREG Percent of firms identifying labor regulations as a constraint to business operations, from the
World Bank Enterprise Surveys Table AI.
For instructions on how to order reprints of this article, please visit our website:
www.emeraldgrouppublishing.com/licensing/reprints.htm
Or contact us for further details: permissions@emeraldinsight.com
355
Firm
performance
in developing
countries
