The Determinants of Capital Intensity in Manufacturing:
The Role of Factor Endowments and Factor Market Imperfections
Rana Hasan
Asian Development Bank
Devashish Mitra†
Syracuse University
Asha Sundaram
University of Cape Town
August 16, 2010
ABSTRACT
In this study, we look at the determinants of industry-level capital intensities. Using cross-country
data, we find that the most important determinant is a country’s factor endowment. In addition, we
find that measures of labor regulation and financial development also matter: less restrictive labor
regulations and greater financial development are associated with lower and higher capital-intensity,
respectively. Furthermore, we establish that in developing countries less restrictive labor regulations
are associated with lower capital-intensity especially in sectors that require more frequent labor
adjustment. In addition, we find that hiring and firing regulations and centralized wage setting are
key elements of labor regulation affecting capital-intensity. In a case study, we show that India,
which has drawn attention for its stringent regulation of manufacturing labor markets, uses higher
capital-intensities in manufacturing than countries at its level of development and with similar factor
endowments. In a large proportion of industries, the gap between the actual capital-labor ratio used
and the capital-labor ratio predicted using factor endowments narrows when we control for labor
market rigidities and financial development. Our findings highlight the role played by factor market
imperfections in determining capital-intensity in manufacturing. In particular, labor regulations can
impose costs on labor use, thereby pushing firms towards greater capital intensity, in turn reducing
labor demand and curtailing gains from trade based on factor-abundance driven comparative
advantage.

The views presented here are those of the authors and not necessarily of the institutions they are affiliated
with.
Corresponding author: Department of Economics, The Maxwell School of Citizenship and Public Affairs,
Syracuse University, Eggers Hall, Syracuse, NY 13244, Email: [email protected]
†
1
1. Introduction
Factor markets that bring about an efficient allocation of resources play an important role in the
development process. In this study, we analyze the importance of factor market imperfections in
determining capital-intensities in manufacturing across countries. Though we look at both credit
and labor market imperfections, our focus for this paper is labor rigidities induced by labor market
regulation and their effects on the techniques of production used and the mix of varieties produced
within any industry. Labor market regulation has often been cited as one of the reasons for poor
performance of manufacturing in developing economies, especially those in South Asia and Latin
America.1 Though meant to protect labor, they can adversely affect it by reducing labor demand.
This may happen through various elements of labor regulation including restrictions on hiring and
firing, minimum wage laws, the rules governing collective bargaining etc. Thus, for example,
restrictions on layoffs have been blamed for hindering industrial expansion to economic scales of
production since firms may be reluctant to hire workers who they cannot fire or layoff easily (see
Panagariya, 2008 on the Indian case). In addition, restrictive labor laws can inhibit firms’ ability to
adjust their labor input to demand and technology shocks like those arising from trade liberalization.
Such shocks can therefore induce firms to hire informal workers who often operate in inferior
working conditions without basic labor protection (see Goldberg and Pavcnik, 2003 for Latin
America). Finally, firms may be more cautious about investment when workers are endowed with
more bargaining power.
Evidence that these laws depress employment and productivity or induce sluggishness in
employment adjustment, however, is tenuous. This is true both from the perspective of crosscountry analysis (see the survey of Freeman, 2009) as well as country case studies (see, for example,
Bhattacharjea, 2006, on a survey of the evidence for India). We take a different approach in this
study and focus on how the input mix gets affected by factor market imperfections.
We begin by carrying out a cross-country analysis that examines the relationship between capital
intensities, factor endowments, and measures of labor regulation and financial development for the
1
See, in particular, Besley and Burgess (2004) for India.
2
period 1994 through 2004.
We then use the results of this analysis to determine if Indian
manufacturing industries are more capital-intensive compared to manufacturing industries in
countries at India’s level of development and similar factor endowments.
We also test our
hypothesis that imperfections in labor and credit markets have been important determinants of
capital intensities in India, a country that has drawn attention in the literature for its labor market
rigidities. Though a labor abundant country, manufacturing in India is characterized by an absence
of large firms producing unskilled-labor intensive products, poor export performance in unskilledlabor intensive sectors (Panagariya, 2008), an overrepresentation in terms of value added and
employment in capital-intensive sectors (Kochhar et al, 2006) and by concentration of laborintensive manufacturing activity in the ‘unorganized’ or ‘unregistered’ manufacturing sector
comprising of small firms for whom labor regulations are less stringent and relatively less strictly
enforced.
We then present a case study that seeks to ascertain if actual capital-labor ratios prevailing in Indian
manufacturing in major industry groups from 1989 to 1996 were larger than predicted capital-labor
ratios for these industry groups based on relative factor demand functions estimated for the United
States (a country with relatively far less restrictive labor laws and a much more developed financial
system)evaluated at Indian wages.
Our study can be placed in the broad literature on the significance of smoothly functioning factor
markets on economic development. In particular, our study contributes to the literature on the
impact of labor regulation on economic performance (Besley and Burgess, 2004; Bhattacharjea,
2006: Hasan, Mitra and Ramaswamy, 2007; Mitra and Ural, 2008; Ahsan and Pages 2009; and
Gupta, Hasan, and Kumar, 2009 for India). We contribute to this literature by looking at an
alternative aspect of the impact of labor regulation on the economy by focusing on the technique of
production used. By comparing Indian capital-labor ratios to capital-labor ratios that would be used
in US manufacturing given Indian wages, we hope to determine if Indian labor regulations impose
costs on hiring labor and hence affect labor demand. We suggest that since the US has a relatively
free labor market, US capital-labor ratios, given factor prices, provide a useful comparison to
evaluate the choice of technique made by Indian firms and determine if these decisions are
suggestive of high labor costs in Indian manufacturing. We also use our cross-country analysis to
highlight the role played by labor regulation and credit market imperfections in determining the
3
capital-labor ratio used in production and show that stringent labor regulation can be particularly
distortionary in certain industries in developing economies. Additionally, we contribute to the
literature that points at higher capital-labor ratios used in manufacturing in India than in countries at
a similar level of development by showing that accounting for labor market regulation and capital
market imperfections can partly explain this discrepancy.2 Finally, we use a recently available dataset
to compare capital-labor ratios in Indian and Chinese manufacturing to investigate the behavior of
these two emerging Asian economies since 1980, when they started out with relatively similar socioeconomic conditions.
Our results indicate that besides factor endowments, labor market rigidity and credit market
imperfections are important determinants of techniques used in production across countries. We
find that a unit increase in the index measuring labor freedom (with 0 being the lowest possible
value and 10 being the highest possible value) is associated with a 20 percent decrease in the capitallabor ratio used in manufacturing industries. Also, lower labor freedom is associated with higher
manufacturing capital-intensity in industries that are prone to more frequent labor adjustment
especially in developing countries. Disaggregated analysis of the effects of labor market regulation
reveals that it is regulations governing hiring and firing and collective bargaining that are the
important aspects of labor regulation that affect capital-intensity. Interestingly, we find minimum
wage laws to have little impact on capital-intensity.
From our case study, we find that India uses a higher capital-labor ratio in manufacturing than
countries at its level of development with similar factor endowments.
India has also been
consistently using higher capital-labor ratios in a majority of manufacturing industries than China
since 1980. Besides, our results show that for each three digit manufacturing industry, India uses a
higher capital-labor ratio than predicted by its factor endowment. This gap between actual and
predicted capital-labor ratios narrows when we control for labor freedom and the level of financial
development for many industries, emphasizing that labor rigidity and capital market imperfections
are significant determinants of factor ratios.
Our comparative study of Indian and US
manufacturing indicates that in all broad manufacturing industry categories, India uses more capital
intensive techniques than those the US would use at Indian wages. Also, within broad industry
For instance, Kochhar et al (2006) use cross-country data to show that Indian industry generates more value
added and employment in capital-intensive, large-scale sectors compared to similar developing countries.
2
4
categories, India specializes in more capital intensive sub-categories in manufacturing compared to
the US, in contrast to what classical trade theory would predict3. We conclude that credit and labor
market imperfections are important determinants of capital-intensity in manufacturing. Specifically,
labor rigidities induced by stringent labor market regulation might push manufacturing firms to
move towards producing more capital intensive product varieties and/or using more capitalintensive techniques of production by indirectly raising the cost of labor to firms, thus reducing
labor demand. More flexible labor laws can reduce costs faced by firms in substituting labor for
other factors of production in response to shocks and can enable labor-abundant developing
countries to exploit gains from trade by specializing in more labor-intensive products that they have
a natural comparative advantage in.
From a broader policy perspective, our results serve to
emphasize that regulations to protect the welfare of workers need to be designed carefully.
Finally, it is important right at the outset to explain why we focus on India, both as an application of
our cross-country regressions as well as in our two comparative case studies. The main reason is that
India is a large, labor-abundant developing country, whose performance in labor-intensive
manufacturing has not been satisfactory. There is a wide range of opinions on the role played by
labor-market institutions in this regard. Opinions range from treating these institutions as being
totally irrelevant to blaming them completely for India’s lack-luster performance in manufacturing.
We thus use the case of India, with it’s stringent labor regulations but polarized debate on their
effects, to understand the role of labor-market institutions. We in fact show that labor regulations
have important effects. However, we also show that they only partially explain why India is not
specializing according to its factor-abundance based comparative advantage in its use of production
techniques and production of product varieties. Studying the case of India, therefore, helps us derive
useful policy lessons, not only for India itself but also for other developing countries with rigid labor
markets – such as many Latin American countries – on this important trade and development issue.
2. Theoretical Foundations
The theoretical framework for our cross-country analysis is based on Schott (2004). Schott (2004)
posits a Heckscher-Ohlin setting with the modification that product varieties within an industry vary
3
See Chapter 2 on The Heckscher-Ohlin Model by Feenstra (2004).
5
in capital-intensity. Thus, some product varieties are more capital intensive than others. Under such
a scenario, a country’s factor endowment determines its product variety mix. More capital-abundant
countries specialize in the more capital-intensive product varieties. In the Lerner diagram in Figure
1 adapted from Schott (2004), the horizontal axis measures labor and the vertical axis measures
capital. X and Y are two product varieties manufactured using capital and labor. X=1/Px and
Y=1/Py represent the unit-value isoquants where Px and Py are world prices of X and Y under free
trade, respectively. Y is more labor intensive than X since for each wage-rental ratio, X warrants a
higher capital-labor ratio. The lines with vertical intercepts 1/r and 1/r’ and horizontal intercepts
1/w and 1/w’ are isocost lines that measure the amount of capital and labor that can be bought with
one dollar at rental rates r and r’ and wages w and w’ respectively.
The two unit-value isoquants result in three different cones, each of which represents the various
combinations of the two vectors (products) defining the cone. In the two good world depicted by
Figure 1, labor-intensive countries like India, with endowments like E’ that lie to the right of k2 and
above the horizontal axis, will specialize in the more labor-intensive product Y, use a capital-labor
ratio represented by the ray from the origin to E’ and will experience factor prices w’ and r’. Under
the assumptions of perfect competition in factor markets and constant returns to scale, profit
maximization occurs where the unit-value isocost line delineated by the intercepts 1/r’ and 1/w’ is
tangent to the unit-value isoquant Y.
More capital-abundant countries with endowment points like E, however, will produce both X and
Y with techniques κ1 and κ2 respectively.
For instance, the dotted line OME represents
combinations of capital and labor used for the production of product varieties X and Y. Note that
in the two good scenario, this cone in between Oκ1 and Oκ2 is the only true cone of diversification
as in the other two cones there is complete specialization. The wage-rental ratio in this economy will
then be w/r. In fact, under trade, all countries with endowment points between lines κ1 and κ2 will
have a wage-rental ratio of w/r in equilibrium. In other words, factor price equalization occurs
within the cone represented by product mixes in the region between Oκ1 and Oκ2. By a similar
argument, countries with factor endowment points to the left of κ1 and to the right of the vertical
axis will specialize in the production of the capital-intensive product variety X and will have a higher
wage-rental ratio than countries with endowments like E and E’.
6
This means that country factor endowments determine a country’s product and/or product variety
mix under trade. The analysis can be generalized for a situation with more than two products or
product varieties (different varieties of differing factor intensities of each broadly defined good).
Unit-value isoquants based on free trade world prices will determine cones of diversification and a
country’s endowment vector will determine its product and product variety mix and hence its factor
ratios in equilibrium. In fact, in a world with many (not just two) products and/or product varieties,
we can get several cones of diversification. Factor price equalization across countries results within
each cone of diversification except for extreme endowment points that result in countries
specializing completely in one product variety in which case different factor endowments could
result in different factor prices and factor ratios. This means that in a free trade equilibrium, capitalabundant countries will choose to specialize in the more capital-intensive product varieties within
each product category or industry and will use more capital-intensive techniques of production.
Hence, we can write the factor ratio employed by a country in an industry with different product
varieties as a function of its factor endowment:
k ic /lic = κi(Kc/Lc)
(1)
where ‘i’ refers to industry and ‘c’ to country, kic stands for capital used in an industry, lic for labor
used and Kc and Lc are country factor endowments.
For our comparative analysis of India and the US, we begin with a simple Cobb-Douglas production
function for a representative firm in industry ‘i’ in country ‘c’ with capital and labor as factors of
production.
qic = Fi(kic, τclic) = Aickicac(τclic)bc
(2)
qic stands for output of the single commodity produced by the firm, kic is capital used in production,
τclic is effective labor (where τIndia < τUS would capture the fact that labor productivity is lower in India
than in the US), Aic is a country specific technology parameter and ac and bc are country-specific
constants. Assuming perfect competition in factor and product markets, profit maximization would
yield:
7
k ic /lic = (w/r) ic (ac/bc)
(3)
3. Empirical Specification
We begin our cross-country comparison by estimating the equation:
Ln(Capital stock per worker)ict = α + βc + λi+ ζt + ε1,ict
(4)
where ‘i’ refers to a 3 digit ISIC manufacturing industry, ‘c’ refers to country and ‘t’ to time. ε1,ict is
the idiosyncratic error term. We then use the estimated βc, which are the average factor ratios used
in manufacturing in a country after controlling for industry and time specific factors affecting factor
ratios to estimate:
Estimated βc = α + η1Ln(Capital stock per worker)c + η2Private Creditc
+ η3Labor Freedomc + ε2,c
(5)
where ε2,ict is the idiosyncratic error term. Capital stock per worker at the country level is the average
capital stock per worker over the period 1994 through 2004. Equation (5) is derived from equation
(1) since it expresses factor use ratios in terms of country endowment ratios. Private Credit is claims
on the private sector as percentage of GDP and is an inverse measure of capital market
imperfections. Djankov, McLiesh and Shleifer (2007) have shown that the primary determinants of
this variable include the average time required to enforce contracts, the degree to which creditor
rights are protected and the existence of public and private credit registries that facilitate information
sharing through the distribution of data. High inflation leads to uncertainty in price levels and
adversely affects credit markets. The private credit to GDP ratio is found to be negatively related to
inflation. Thus, this variable captures the quality of credit market or the degree of financial
development. The variable, private credit to GDP ratio we use is an average across the period 1999
through 2003.
Labor freedom is an inverse measure of labor rigidities due to stringent labor regulations for the year
2004. Controlling for factor endowments, η2 and η3 capture the effect of factor market imperfections
on the factor intensity through the impact on technique of production employed or the types of
8
product varieties produced. Our hypothesis is that η3 < 0. Higher labor freedom is associated with
more labor-intensive techniques of production after accounting for country factor endowment and
capital market imperfections. Next, we exploit the richness of our labor freedom index to explore
the relative importance of different aspects of labor freedom in driving capital-intensity. We split
the labor freedom index into its components measuring the rigidity of minimum wage legislation,
hiring and firing regulation, conscription (use of draft), unemployment insurance legislation and
centralized collective bargaining over wage setting to determine the effect of each on the technique
of production.
We also estimate equation (5) by replacing Ln(Capital stock per worker) at the country level by
dummy variables based on grouping the countries into top, middle and bottom thirds on the basis of
their GDP per capita where GDP per capita is averaged across the period 1994-2004. We call these
variables Income Categories. We thus have three categories, Income Category 1: Low Income,
Income Category 2: Middle Income and Income Category 3: High Income. We do this to account
for the fact that within cones of diversification, factor prices are equalized. Assuming that high
income countries are also more capital-abundant, we argue that countries at similar levels of
development would tend to fall in the same cone of diversification and will experience factor price
equalization. We check for robustness of our results under this alternate specification.
Further, to account for the possibility that the effects of labor market and credit market
imperfections affect capital-intensity differently depending on industry characteristics we estimate:
Ln(Capital stock per worker)ic = αi + μiLn(Capital stock per worker)c + ε3,ic
(6)
and
Ln(Capital stock per worker)ic = αi + ν1iLn(Capital stock per worker)c
+ ν 2iPrivate Creditc+ ν 3iLabor Freedomc + ε4,ic
(7)
where Capital stock per worker both at the industry and at the country level is the average capital
stock per worker over the period 1994 through 2004 and the Private Credit and Labor Freedom
variables are as discussed previously. ε3,ic and ε4,ic are the idiosyncratic error terms. Equation (6) is
9
directly derived from equation (1).
Equation (7) includes our measures of factor market
imperfections and our hypothesis is that ν 3 < 0 for each ‘i’. We estimate equations (6) and (7) for
each of our twenty-eight 3 digit ISIC manufacturing industries separately since we expect the impact
of credit market imperfections and labor market rigidity to differ significantly across industries. We
also estimate equations (6) and (7) with Income Categories instead of Ln(Capital stock per worker)
at the country level as an explanatory variable to account for factor price equalization within cones
of diversification and check the robustness of our results.
Our next step is to investigate if and to what extent the predicted capital stocks per worker for each
manufacturing industry for India are lower than actual capital stocks per worker used. Equation (1)
indicates that country factor endowments determine industry level factor use. This means that in
the absence of factor market imperfections, equation (6) will closely predict factor ratios in
manufacturing for each country. For India, our hypothesis is that actual capital-labor ratios captured
by the capital stock per worker would be much higher than their predicted counterparts from
equation (6) given the existence of labor market rigidities due to stringent labor market regulation.
We therefore obtain predicted capital-labor ratios for each Indian manufacturing industry from
equation (7) to see if the gap, if any, between actual capital-labor ratios and those predicted with
country factor endowments narrows by accounting for credit market imperfections and labor market
rigidity. We argue that for India, the value taken by the labor freedom variable would likely account
for higher capital-labor ratios used in manufacturing industries after controlling for factor
endowment.
Next, we delve into the differential impact of labor market regulation across manufacturing
industries. We test the hypothesis that stringent labor market regulation is associated with greater
capital-intensity especially in industries that require more frequent labor reallocation. Haltiwanger,
Scarpetta and Schweiger (2008) show that employment protection legislation curbs job flows
particularly in those industries that require more frequent labor adjustment. For instance, they point
out that demand is more volatile or technological changes and upgradation more frequent in some
industries requiring labor reallocation. Alternatively, technological characteristics of some industries
may require firms to use highly specialized workers and thus make them less likely to frequently
adjust workforce to respond to idiosyncratic shocks. We argue that in addition to affecting job
flows, stringent labor market regulation may be associated with more capital-intensive techniques
10
and/or product varieties in these industries, especially in developing countries where labor markets
are more rigid. We hence estimate the following version of equation (7):
Ln(Capital stock per worker)ic = αi + δ1iIncome Categoryc + δ 2iPrivate Creditc+ δ 3iLabor Freedomc
δ 4iPrivate Creditc*US Job Flowi*Income Categoryc+ δ 5iLabor Freedomc* US Job Flowi*Income
Categoryc + ε4,ic
(8)
Here, to measure the ‘propensity’ of an industry for job-reallocation, we follow Haltiwanger et al
(2008) and use the gross job flows, measured as the sum of the job creation and destruction rates, in
an industry in the United States. The argument behind using U.S. industry-wise job reallocation rate
is that it can be argued to be a country with low policy-induced distortions. This means that the U.S.
job reallocation rate in an industry can capture factors other than labor policy, like technological and
market driven induced adjustment costs that make an industry more prone to frequent labor
adjustments. We assume, like in previous studies, that these technological and market driven
differences in the demand for job reallocation carry over to other countries in our sample. The
other variables are as defined in our previous specifications. Our hypothesis is that δ 5i < 0 (in
addition to δ3i being < 0) indicating that low labor freedom is associated with higher capital-intensity
and this negative effect is stronger in industries where job flows are high (in other words, in
industries that require frequent labor adjustment), especially in developing, or low and middle
income countries.
For our India-US comparison, we first estimate the capital used relative to labor by the average US
manufacturing firm in industry ‘i’ at time ‘t’ as a function of the relative wage rate using the
following specification obtained by taking logs of equation (3):
Ln(k/l)it = α + β1Ln(wage)it + δi+ γt + εit
(9)
k and l refer to capital and labor used respectively in industry ‘i’ at time ‘t’, δi and γt are industry and
year fixed effects respectively and εit is the error term. Ln(wage) is the logarithm of the wage rate.
The industry and year fixed effects control for industry and time specific shocks that affect factor
prices and factor ratios simultaneously.
We estimate equation (8) first for thirteen broad
manufacturing industry groups and then for twenty-one ISIC 3 digit manufacturing industries. Our
11
time period is from 1989 through 1996. We assume the user cost of capital (the rental rate) to be
constant across all industries and will hence be captured by the year fixed effects. Next, we obtain
predicted Ln(K/L)it for the US at Indian wages for industry ‘i’ at time‘t’. We then compare the
actual capital-labor ratios used in India for each industry at time‘t’ to capital-labor ratios predicted
for the US at Indian wages. We argue that if we find capital-labor ratios used in Indian
manufacturing to be larger than those predicted for the US at Indian wages, results are suggestive of
indirect labor costs brought about by labor regulation.
The existence of capital market
imperfections in India would only make the higher capital-labor ratios used in Indian manufacturing
more of an anomaly, consistent with the existence of indirect labor costs.
4. Data
Detailed definitions of the variables used in this analysis are available in the Data Appendix. For the
cross-country analysis of capital-intensity in the paper, we rely mainly on data on gross fixed capital
formation and the number of employees for twenty-eight three digit ISIC Revision 2 manufacturing
industries obtained from the Trade, Production and Protection database of Nicita and Olarreaga
(2006). Using the perpetual inventory method (as mentioned in the appendix,) data on gross fixed
capital formation have been converted to capital stock data. Capital stock per worker is then
measured as the ratio of capital stock to the number of workers. The production data are based on
domestic production data from the UNIDO International Yearbook of Industrial Statistics. Though
the data cover 100 countries, data availability varies across countries. We use data from 1994
through 2004 for our analysis.
We also carry out some of our cross-country analysis using economy-wide data on capital-intensity.
We obtain country level gross fixed capital formation and number of workers from the World
Bank’s World Development Indicators and convert gross fixed capital formation to capital stock
using the perpetual inventory method4. After accounting for missing investment data, we obtain a
Data on gross fixed capital formation are in thousands of current US dollars. We convert this data to
thousands of constant 2005 US dollars using the Wholesale Price Index for the United States from the World
Development Indicators. Similarly, we also deflate GDP per capita adjusted for PPP at current dollars to
2005 dollars.
4
12
set of 63 countries for our cross-country analysis. Data on GDP per capita, which we use in some
of our analysis, are also obtained from the World Bank’s World Development Indicators.
We capture the existence of capital market imperfections by a measure of financial development for
the years 1999 through 2003, defined as claims on the private sector by financial institutions as a
percentage of GDP, obtained from Andrei Shleifer’s website and described in Djankova, McLiesha
and Shleifer (2007). The data are computed by the authors from the IMF’s International Financial
Statistics. As noted earlier, this measure is inversely correlated with the existence of capital market
imperfections.
Labor freedom is measured by the Fraser Institute’s sub-index for Labor Freedom for the year 2004.
The Labor Freedom Indices measure labor freedom on a range of 0 to 10 in five different areas:
Minimum wage regulation, hiring and firing regulation, centralized wage setting, extension of union
contracts to nonparticipating parties and conscription (use of draft). We use both the composite
Labor Freedom sub-index and the indices for each of the five sub-components or sub-areas of labor
freedom in our analysis. We use data on U.S. industry-wise job flows from Haltiwanger, Scarpetta
and Schweiger, 2008.
The UNIDO production data do not cover China after the year 1997. In addition, investment data
for China are not available between 1994 and 2004. Given the emerging importance of China in the
global economy, we use recently available data from Wu, Lee and Rao (2007) for 18 manufacturing
industries to compare trends in capital stock per worker in India and China from 1980 through
2004. Wu, Lee and Rao (2007) construct comparable capital stock and employment figures for India
and China from the Annual Survey of Industries for India and the China Industrial Economic
Statistics Yearbook for China. They also compute PPP measures to express capital stock for both
countries in 1995 Chinese Yuan. A detailed account of how they construct capital stock and labor
estimates is available in their paper.
Finally, for our comparative analysis on US and Indian manufacturing, we use industry level data
from the Annual Survey of Industries for India and data from the NBER Manufacturing Industry
Productivity Database for the US. Data for both the US and India are for the years 1989 through
1996. The NBER data uses the US SIC 1987 industry classification while our Indian data uses the
13
Indian NIC 1987 classification. We use a concordance between US SIC 1987 and ISIC Rev 3 of the
United Nations used by Schaur, Xiang and Savikhin (2008) and concord the Indian NIC 1987 to
NIC 1998 which in turn follows ISIC Rev 3. Our main variables are capital stock, the number of
production workers and production worker wages for India and the US.
US capital stock data are
in 1987 US dollars. We use the PPI for the US from the United Nations Statistics Division to
deflate US wages to 1987 US dollars. For India, we use the Producer Price Index (PPI) for India
from the United Nations Statistics Division to deflate Indian capital stock and wages to 1987 Indian
rupees. We then apply the PPP exchange rate for 1987 of 3.79 Indian rupees to the US dollar
(obtained from the Penn World Tables with Base Year 2000) to express Indian capital stock and
wages in 1987 US dollars.
5. Results
5.1 Cross-country Analysis: Descriptive Analysis
Table 1 ranks the 63 countries in our cross-country analysis based on GDP per capita, Capital stock
per worker at the country level and estimated βc from equation (4). Columns (1) and (2) show that
India is placed very low in the per capita GDP and factor endowment (country ratio of capital to
labor endowment) rankings. In fact, only Cameroon, Kenya, Bangladesh, Nepal, Tanzania, Malawi
and Ethiopia have lower GDP per capita and capital stock per worker than India. However, from
Column (3), we see that India uses a higher overall capital-labor ratio (capital intensity) in the
twenty-eight manufacturing industries than 35 countries out of the 63 countries in our sample. The
discrepancy between India’s position in the GDP per capita and factor endowment rankings and its
position in the capital-labor use rankings point to a potential role for labor market imperfections and
possibly credit market imperfections in explaining the capital-intensive nature of India’s
manufacturing that we hope to explore through this paper.5
Table 2 provides the number of manufacturing industries in each of the 63 countries in our sample
that use lower and higher capital-labor ratios than the corresponding Indian manufacturing
industries. Capital stock per worker values are averages for the period 1994 through 2004. Even
It could be argued that part of this discrepancy is explained by the fact that manufacturing data for India
pertain to the formal sector. But then this can be expected to be true for most developing countries.
5
14
though we do not control for factor prices, endowments or the level of development, the table
shows that India uses a higher capital-labor ratio in 5 out of 21 industries than the UK (24 percent
industries for which data are available), 4 out of 19 (21 percent) industries than Germany, 6 out of
21 industries (almost 30 percent) than Spain and 4 out of 18 (18 percent) industries than the United
States, all OECD countries. India uses a higher capital-labor ratio in 11 out of 18 industries than
Kuwait, 8 out of 11 industries than the Czech Republic, 17 out of 22 industries than Hungary and 17
out of 18 industries than Latvia, all much higher-income countries compared to India. It uses a
higher capital-labor ratio in 20 out of 22 industries than its more affluent neighbor, Sri Lanka. The
observations from Tables (1) and (2) drive home one of the key points we try to make in this paper
– India uses higher capital-labor ratios in manufacturing than would be expected from a country of
its level of development and factor endowments.
5.2 Cross-country Analysis: Estimation Results
Tables 3(a) and (b) present results for estimation equation (5). Table 3(a) includes country factor
endowment along with labor freedom and private credit as percentage of GDP as explanatory
variables while Table 3(b) replaces country factor endowment with income category dummies as
explanatory variables. From column (1) of Table 3(a), the coefficient on country factor endowment
is positive and significant, indicating that a one percent change in capital stock per worker at the
country level is associated with a 0.4 percent increase in the average capital-labor ratio used in
manufacturing. The coefficient on private credit as percentage of GDP, a measure of financial
development and an inverse measure of the existence of capital market imperfections is positive and
significant indicating that a 1 percentage point change in private credit as a percentage of GDP is
associated with a 1 percent increase in the capital-labor ratio. We find evidence supporting our
hypothesis that η3 < 0. The coefficient on the index of labor freedom is negative and significant.
The negative coefficient indicates that a unit increase in the index measuring labor freedom (that
ranges from 0 to 10) is associated with a 20 percent decrease in the capital-labor ratio used in
manufacturing industries. Columns (2) through (7) use components of the labor freedom index to
measure the importance of different aspects of labor regulation that matter for capital-intensity in
manufacturing. Results indicate that minimum wage legislation, hiring and firing regulation and
collective bargaining are the most significant components affecting capital-intensity. Of these, the
largest effects on capital-intensity are on account of hiring and firing regulation.
15
In Table 3(b), we replace country factor endowment by income category dummies as explanatory
variables. These income dummies are intended to capture factor price equalization within cones of
diversification. The coefficients on the income dummies from column (1) have the expected signs
with the coefficient for the high income dummy larger in magnitude than the coefficient for the
middle income dummy. The coefficient on private credit as a percentage of GDP positive and
significant and close in magnitude to Table 3(a). Again, our hypothesis that η3 < 0 finds support.
Columns (2) through (7) echo results from Table 3(a) except that the coefficient on minimum wage
legislation is no longer significant.
On the other hand, the coefficient on hiring and firing
regulations is the largest sized of the various components of labor regulation. Overall, results in
Tables 3 (a) and (b) show that labor market rigidities and credit market imperfections explain factor
ratios used in manufacturing after controlling for country factor endowment. In addition, hiring and
firing regulation and collective bargaining resulting in centralized wage setting are the important
components of labor regulation affecting capital-intensity in manufacturing.
Next, we estimate equations (6) and (7) for each of the twenty eight ISIC 3 digit manufacturing
industries in our sample. Tables 4(a) and (b) present the results for equation (7). Table 4(a) includes
country level capital stock per worker averaged over the period 1994 thorough 2004 as an
explanatory variable while Table 4(b) replaces this with the income category dummies. From Table
4(a), country level capital per worker is positively associated with capital stock per worker used in all
except one manufacturing industry. Also, coefficients are significant in fifteen out of twenty-eight
industries. The private credit variable is positive for all except 4 out of twenty-eight industries,
however, it is precisely estimated in only 6 industries. We argue that this can be explained by the
fact that capital stock per worker at the country level is most likely determined by financial
development. To ensure that our results are not contaminated by this endogeneity between financial
development and country factor endowment, we perform our estimation without including the
financial development variable and find that our results for labor freedom are qualitatively
unchanged. The labor freedom variable is of the expected negative sign in all but 2 of the twentyeight industries and is significant in 9, providing stronger evidence for the impact of labor market
rigidities over credit market imperfections on factor ratios in manufacturing. For both the private
credit variable and the labor freedom variable, where significant, the coefficients always have the
16
right sign. Results suggest that a 0.01 increase in the labor freedom index is associated with between
0.1 and 0.6 percent decreases in the capital-labor ratio used in manufacturing.
From Table 4(b), focusing on the key variables of interest, the labor freedom index is negative for all
but 3 industries and is significant for 10 industries. The private credit variable is now positive for all
but 1 industry and is significant for thirteen industries. Overall, we find compelling evidence that ν 3
< 0 for most industries, though the effects are not precisely estimated for all industries. The overall
flavor of results in this section demonstrates that controlling for country factor endowments or
income categories, rigid labor market conditions and greater capital market development are
associated with higher capital-labor ratios in manufacturing.
We next move on to determine if stringent labor regulation is associated with higher capital-intensity
in industries that require more frequent labor adjustment. Results for equation (8) are presented in
Table 5. Note that Income Category 3 or high-income countries are the omitted category in this
table. Column (1) shows that greater labor freedom is associated with lower capital-intensity and
that greater levels of financial development are associated with higher capital-intensity in
manufacturing. The magnitudes of the effects are consistent with Tables 3 (a) and (b). From
columns (2) and (3), we find that the triple interaction between the labor freedom index and U.S.
industry level job flow is negative and is significant for developing countries (column (2)) and
especially for middle-income countries (column (3)). (Note that the coefficient estimate of the
double interaction between labor freedom and US job flow is also negative.) This provides evidence
for our hypothesis that lower labor freedom is associated with greater capital-intensity in
manufacturing in industries whose demand fundamentals and technology require more frequent
labor adjustment especially for developing countries. It also provides some reassurance that our
labor freedom index is picking up the impact of stringent labor market regulation on capitalintensity.
To conclude our cross-country analysis, we obtain predicted values of capital stock per worker used
for each industry for India first from equation (6) and next, by including only private credit and then
both private credit and labor freedom as explanatory variables (equation (7)). We then compare
these predicted values to actual capital-labor ratios used by Indian manufacturing industries. Results
are presented in Tables 6 (a) and (b). As before, the difference between Tables 6 (a) and (b) is that
17
Table 6 (b) substitutes income category dummies for the country factor endowment to capture the
idea of factor price equalization within cones of diversification. Results are striking. In the 23
industries for which data for India are available, actual capital-labor ratios used are higher than
predicted capital-labor ratios from just including country factor endowment as an explanatory
variable in 21 industries. Hence, in 21 out of 23 industries, country factor endowment under
predicts the capital-labor ratio actually used by the Indian industry.
Next, from Table 6 (a) the gap between the actual and predicted values narrows in 11 out of these
21 industries by just adding private credit as an explanatory variable. This could be explained by the
fact that India’s capital markets, while not well developed by developed country standards, are fairly
advanced when compared to other countries at India’s stage of development (with similar relative
factor endowments). Adding both private credit and labor freedom as explanatory variables narrows
the gap in 12 out of 21 industries. Additionally, in 17 industries, adding the labor freedom variable
over and above the private credit variable reduces the gap between actual and predicted capital-labor
ratios further. This means that labor market rigidities and capital market imperfections are able to
explain the difference between actual capital-labor ratios used and capital-labor ratios predicted
using country factor endowments for a majority of the industries. A similar story is apparent from
Table 6 (b). Actual capital-labor ratios are higher than predicted capital-labor ratios in 21 out of 23
industries for which we have data for India. In 13 of these industries, the gap between actual and
predicted capital-labor ratios narrows if we control for labor freedom and financial development.
It is important to note that though labor market rigidity and capital market imperfections largely
seem to explain the gap between actual and predicted capital-labor ratios used in Indian
manufacturing industries, they do not explain it completely. This is not hard to believe for two
reasons. First, our labor freedom index is an imperfect measure of actual labor freedom. For
instance, our variable does not capture factors like the ease of writing part-time contracts. However,
we argue that given existing data, we do our best in establishing the importance of labor regulation
induced rigidities in the factor ratios used in production. Second, other government policies besides
labor policy can affect capital-intensity in manufacturing. India long followed a development
strategy that focused on self-sufficiency in heavy industry, resulting in concentration of
manufacturing in capital-intensive sectors. Even after liberalization in 1991, the Indian government
encouraged the use of imported capital inputs in manufacturing at low custom duty rates for export18
oriented production and credit was subsidized for technology upgradation, especially for small and
medium sized firms (Palit, 2008). In addition to stringent labor regulation, these government
schemes could have incentivized the substitution of capital for labor by Indian industry.
5.3 Comparison of Indian and Chinese Manufacturing
We then compare trends in capital stock per worker in Chinese and Indian manufacturing for the
period 1980 through 2004. Figure 2 presents capital stock per labor in thousands of 1995 Chinese
Yuan per labor adjusted for PPP for India and China from 1989 through 2004 in 19 manufacturing
industries. Capital stock per labor in Indian manufacturing is consistently above capital stock per
labor in China. Even though both countries were comparable in many ways in 1980 with similar
income and development indicators, also reflected in the similar levels of capital stock per labor in
1980, from 1980 through 2000, India’s growth in capital stock per labor was higher than that of
China with significant divergence from 1990 through 2000. This observation is consistent with our
story of higher costs to employing labor in Indian manufacturing due to labor market rigidities.
Figures 3 (a), (b), (c) and (d) present capital stock per labor in thousands of 1995 Chinese Yuan per
labor by industry for 1980, 1990, 2000 and 2004 respectively. In 11 out of 19 industries, India has
consistently had higher capital stock per labor than China throughout the period. However, in most
of these 11 industries including Paper and Printing, Leather, Rubber and Plastics, Chemicals, Non
metallic Minerals, Basic Metals, Metal Products, Electrical equipment and Instruments, though India
exhibits consistently larger capital stock per labor than China, the gap between Indian and Chinese
capital stock per labor narrows after 2000, except for Machinery, where the gap starts to narrow in
the mid nineties. In Petroleum and Coke, where India is more capital intensive, the gap keeps
growing very rapidly. As far as the remaining 8 industries are concerned, in Tobacco, China
consistently exhibits higher capital stock per labor than India. In Food and Beverages and Apparel,
though Indian industry starts out more capital-intensive in 1980, by 2004, it is less capital-intensive
than its Chinese counterpart.
This is reversed for Textiles, Wood Products and Transport
Equipment, where India comes out more capital-intensive in 2004 after having started out with
lower capital per labor in 1980. In summary, these figures are in the spirit of our basic finding that
Indian manufacturing is more capital-intensive than Chinese manufacturing in most industries even
though we perceive the gap to be narrowing in the last decade.
19
5.4 Comparative Study of US and Indian Manufacturing
Last, we turn to our comparative study of Indian and US manufacturing. We compare actual capital
labor ratios used in Indian manufacturing to capital-labor ratios for the US predicted at Indian
wages. Table 7 presents a comparison of the actual capital-labor ratios used in India for each
industry to capital-labor ratios predicted for the US at Indian wages for 13 broad industry groups.
Columns 2 and 3 provide the mean (over time) actual capital-labor ratio for India and predicted
capital-labor ratio for the US at Indian wages for each industry group respectively. Results show
that actual capita-labor ratios in Indian manufacturing are larger than those predicted for the US at
Indian wages. What is interesting about our results is that the actual capital-labor ratio is larger for
every single industry group we consider, suggesting potential indirect labor costs in Indian
manufacturing6. Again, even though we do not explicitly consider capital market imperfections,
their existence would induce firms to adopt more labor-intensive techniques, which would mean
that our results of higher actual capital-labor ratios in Indian manufacturing compared to the US are
strongly suggestive of the existence of hidden labor costs due to stringent labor regulation.
To probe our results further, we focus on more disaggregated industry groups and perform the same
analysis. We report results in Table 8, with columns 1 and 2 giving the actual and predicted capitallabor ratios respectively. At this more disaggregated level of industrial classification, we find that the
actual capital-labor ratios for Indian manufacturing are higher than the predicted ratios for the US at
Indian wages for the majority of the industries. We next posit that within broad industry groups in
manufacturing, India specializes in the more capital intensive sub-groups. To see this, we look at
output shares of each of the manufacturing industry sub-groups within each broad group for both
India and the US. Our output variable for the US is the total value of shipments for each industry
and the corresponding variable for India is total output for each industry7. We present the output
shares along with industry capital-labor ratios in Table 8. Table 8 confirms that within broad
industry categories, India specializes more in capital-intensive sub-categories in comparison to the
6
Results hold even if we consider capital-labor ratios for the year 1996 instead of the average over time.
As noted in the Data section, we deflate output variables to 1987 values using the PPI and use the PPP
exchange rate to express Indian output in 1987 US dollars.
7
20
US in three of the six aggregate industries for which we have data on more disaggregate sub
industries (in three out of five if we leave out the miscellaneous group). For instance, in the broad
industry of Paper, Printing and Publishing, India specializes in the capital-intensive industry of paper
production unlike the US, which specializes in Printing/Publishing. This is very surprising since the
US is much more capital abundant than India.
In our discussion of the theoretical framework, we show that a labor abundant country such as India
would specialize in labor-intensive product varieties within an industry and the US, which is capital
abundant in comparison to India, would specialize in capital intensive product varieties. By a similar
argument, we would expect India to specialize in the more labor intensive sub-categories within
broad industry groups than the US, which would be expected to specialize in the more capitalintensive sub-categories.
Our results are not consistent with these predictions. We argue that
restrictive labor regulations raise the cost of hiring workers, thereby explaining this anomaly.
6. Conclusion
In this study, we show that labor rigidities due to stringent labor regulation can lead countries to
specialize in more capital intensive product varieties and/or use more capital intensive techniques in
production by imposing costs on the employment of labor. Our results indicate that labor market
rigidities, induced specially by restrictions on hiring and firing and collective bargaining over wages,
as well as credit market imperfections are important in explaining a country’s production technique
in manufacturing.
Stringent labor regulation can affect capital-intensity particularly for those
industries that require frequent labor adjustment in developing countries. For India, we find that it
uses higher capital-labor ratios in manufacturing than would be expected of a country at its level of
development with its factor endowment and that labor freedom explains this discrepancy in most
industries.
In addition, we show that for broad manufacturing industry categories, actual Indian capital-labor
ratios are higher than capital-labor ratios used in US manufacturing at Indian wages and that within
these broad industry categories, India specializes in the more capital-intensive sub-categories
compared to the US. We conclude that this anomaly suggests indirect labor costs faced by Indian
21
manufacturing firms due to restrictions on hiring and firing workers. Our results imply that though
more capital per worker is associated with higher wages, excessive regulation can hurt workers by
lowering labor demand.
Besides, rigid labor markets may prevent labor abundant developing
countries from fully exploiting gains from trade by specializing in and exporting labor intensive
commodities and product varieties in which they have a comparative advantage and reap the
benefits of globalization. In this way, regulation aimed at protecting the welfare of labor may end
up hurting it.
References
Ahsan, A. and C. Pagés (2009), “Helping or Hurting Workers?: Assessing The Effects Of De Jure
and De Facto Labor Regulation in India”, World Bank.
Bartelsman E. J. and W. Gray (1996), “The NBER Manufacturing Productivity Database”, NBER
Technical Working paper # 205, Cambridge, USA.
Besley T. and R. Burgess (2004), “Can Regulation Hinder Economic Performance? Evidence from
India”, Quarterly Journal of Economics, 119(1), p 91-134.
Bhattacharjea A. (2006), “Labor Market Regulation and Industrial Performance in India – A Critical
Review of the Empirical Evidence”, The Indian Journal of Labor Economics, 49(2), p. 211-32.
Djankov S., McLiesh C. and A. Shleifer (2007), ‘Private Credit in 129 Countries’, Journal of Financial
Economics 84, 299–329.
Feenstra R. C. (2004), “Advanced International Trade: Theory and Evidence”, Princeton University
Press, Princeton, USA.
Freeman R. B. (2009), “Labor Regulations, Unions and Social Protection in Developing Countries:
Market Distortions or Efficient Institutions’, NBER Working Paper 14789.
Goldberg P. K. and N. Pavcnik (2003), ‘The response of the informal sector to trade liberalization’,
Journal of Development Economics, 72(2003), 463-496.
Gupta, P., R. Hasan, and U. Kumar (2009), “Big Reforms but Small Payoffs: Explaining the Weak
Record of Growth in Indian Manufacturing”, India Policy Forum, 5, 59-108.
Haltiwanger J, Scarpetta S and H Schweiger (2008), ‘Assessing Job Flows Across Countries: The
Role of Industry, Firm Size and Regulations’, NBER Working Paper # 13920.
Hasan R., Mitra D. and K. V. Ramaswamy (2007), “Reforms, Labor Regulations and Labor Demand
Elasticities: Empirical Evidence from India”, Review of Economics and Statistics, 89(3), p. 466-481.
22
Key Global Indicators, United Nations Statistics Division, United Nations.
Kochhar, K, U. Kumar, R. Rajan, A. Subramanian, and I. Tokatlidis. 2006. “India’s pattern of
development: What happened, what follows?” Journal of Monetary Economics, 53: 981-1019.
Mitra D. and B. P. Ural (2008), “Indian Manufacturing: A Slow Sector In A Rapidly Growing
Economy,” Journal of International Trade and Economic Development, 17(4), p. 525-559.
Nicita A. and M. Olarreaga (2006), ‘Trade, Production and Protection 1976-2004’, World Bank,
Washington D.C.
Palit A. (2008), ‘Policies Responsible for Excessive Use of Capital Relative to Labor in India’s
Manufacturing Industries’, mimeo, International Labor Organization, Geneva.
Panagariya A. (2008), ‘India, The Emerging Giant’, Oxford University Press, New York
Penn World Tables 6.2, Center for International Comparisons of Production, Income and Prices,
University of Pennsylvania, USA.
Schaur G., Xiang C. and A. Savikhin (2008), “Factor Uses and Pattern of Specialization”, Review of
International Economics, 16(2), p. 368-382.
Wu H. X., Lee B. L. and D. S. Prasada Rao (2007), ‘Comparative Performance of Indian and
Chinese Manufacturing Industries, 1980 – 2004’, mimeo.
23
Table 1: Country Rankings
(1)
Country Rank: GDP Ln(GDP
per Capita
per
capita)
United States
3.674
Norway
3.670
Austria
3.492
Japan
3.429
United Kingdom
3.423
Australia
3.398
France
3.396
Hong Kong, China
3.390
Germany
3.383
Italy
3.373
Finland
3.362
Israel
3.268
Singapore
3.257
Spain
3.209
Macao, China
3.152
Cyprus
3.090
Portugal
3.016
Greece
3.011
Kuwait
2.978
Malta
2.974
Slovenia
2.958
South Korea
2.929
Czech Republic
2.907
Hungary
2.715
Oman
2.702
(2)
Country Rank:
Factor Endowment
Japan
Singapore
Hong Kong, China
Germany
Austria
Norway
Macao, China
Israel
France
United States
Italy
Australia
South Korea
Finland
Malta
United Kingdom
Spain
Malaysia
Cyprus
Portugal
Greece
Slovenia
Chile
Uruguay
Kuwait
Ln(Capital
stock per
worker)
5.377
5.177
5.096
4.837
4.789
4.686
4.577
4.462
4.393
4.354
4.297
4.250
4.144
4.144
4.081
3.995
3.933
3.730
3.727
3.683
3.649
3.528
3.352
3.331
3.312
24
(3)
Country Rank:
Country Dummy
South Korea
Japan
Oman
Singapore
Germany
Norway
Malawi
Austria
France
Israel
Finland
Italy
United States
Hong Kong, China
United Kingdom
Portugal
Turkey
Mexico
Spain
Malaysia
Chile
Greece
Cyprus
Ecuador
Panama
Country
Dummy
Coefficient
0.87
0.38
0.231
0.0574
0.00721
-0.275
-0.286
-0.288
-0.415
-0.529
-0.595
-0.612
-0.752
-0.869
-0.919
-1.153
-1.175
-1.184
-1.247
-1.512
-1.529
-1.551
-1.574
-1.579
-1.589
Poland
Chile
Uruguay
Trinidad, Tobago
Mexico
Malaysia
Latvia
Botswana
Romania
Colombia
Turkey
Gabon
Bulgaria
Tunisia
Panama
Iran
Venezuela
Peru
El Salvador
Ukraine
Jordan
Philippines
Morocco
Egypt
Ecuador
Sri Lanka
Indonesia
Azerbaijan
Bolivia
2.464
2.374
2.322
2.319
2.317
2.306
2.227
2.141
2.044
2.026
2.019
2.011
1.983
1.968
1.966
1.938
1.924
1.730
1.696
1.665
1.560
1.538
1.445
1.398
1.398
1.361
1.303
1.101
1.039
Czech Republic
Gabon
Hungary
Oman
Panama
Trinidad, Tobago
Mexico
Colombia
Turkey
Poland
Venezuela
Botswana
Jordan
Tunisia
Peru
Iran
Ecuador
Latvia
El Salvador
Philippines
Morocco
Egypt
Indonesia
Romania
Ukraine
Sri Lanka
Bolivia
Bulgaria
Cote d'Ivoire
3.178
2.986
2.945
2.930
2.891
2.805
2.772
2.659
2.631
2.533
2.515
2.488
2.485
2.434
2.354
2.219
1.943
1.922
1.797
1.736
1.677
1.677
1.656
1.634
1.629
1.297
1.206
1.018
0.762
25
Uruguay
India
Tunisia
Morocco
Slovenia
Czech Republic
Venezuela
Iran
Malta
Cameroon
Cote d'Ivoire
Kuwait
Poland
Hungary
Peru
Indonesia
Gabon
Colombia
Romania
Macao, China
El Salvador
Philippines
Tanzania
Bolivia
Botswana
Trinidad, Tobago
Jordan
Egypt
Bulgaria
-1.784
-1.865
-1.887
-1.922
-2.004
-2.02
-2.046
-2.09
-2.167
-2.265
-2.276
-2.295
-2.3
-2.438
-2.502
-2.508
-2.592
-2.607
-2.769
-2.77
-2.801
-2.845
-2.876
-2.93
-2.974
-3.038
-3.151
-3.233
-3.436
India
Cameroon
Cote d'Ivoire
Bangladesh
Nepal
Kenya
Ethiopia
Malawi
Tanzania
1.031
0.780
0.614
0.576
0.423
0.199
-0.229
-0.382
-0.469
Azerbaijan
India
Cameroon
Kenya
Bangladesh
Nepal
Tanzania
Malawi
Ethiopia
0.602
0.524
0.469
0.020
-0.066
-0.139
-0.670
-0.819
-0.979
Latvia
Sri Lanka
Ethiopia
Kenya
Nepal
Bangladesh
Ukraine
Azerbaijan
Australia
-3.866
-3.973
-4.014
-4.139
-4.335
-4.641
-5.203
-5.654
Notes: 1) Column (1) presents rankings by Ln(GDP per capita) in thousands of constant 2005 PPP adjusted International Dollars with GDP per capita
averaged over 1994 through 2004 from the WPI. 2) Column (2) presents rankings by Ln(Capital stock per worker) at the country level in thousands of
constant 2005 US Dollars with Capital stock per worker averaged over 1994 through 2004 and generated from WDI data. 3) Column (3) presents
rankings by coefficients on the country dummy from a regression of Ln(Capital stock per worker) in thousands of constant 2005 US Dollars generated
from UNIDO data for each country in twenty-eight 3 digit ISIC manufacturing industries from 1994 to 2004 on country, 3 digit industry and time
dummies (with Australia as the omitted country).
26
Table 2: Number of 3 digit ISIC industries with lower and higher Ln(Capital stock per
worker) than India
Lower Ln(K/L) Higher Ln(K/L)
Country
than India
than India
Australia
1
11
Austria
3
16
Azerbaijan
10
1
Bangladesh
9
1
Bulgaria
14
3
Bolivia
14
2
Botswana
2
1
Chile
9
12
Cote d'Ivoire
8
6
Cameroon
7
7
Colombia
4
2
Cyprus
6
11
Czech Republic
8
3
Germany
4
15
Ecuador
8
12
Egypt
5
1
Spain
6
15
Ethiopia
12
1
Finland
3
17
France
3
12
Gabon
2
1
United Kingdom
5
16
Greece
7
10
Hong Kong, China
2
7
Hungary
17
5
Indonesia
12
9
Iran
14
9
Israel
3
9
Italy
3
18
Jordan
16
5
Japan
2
14
Kenya
3
0
South Korea
0
16
Kuwait
11
7
Sri Lanka
20
2
Latvia
17
1
Macao, China
5
6
Morocco
12
8
Mexico
8
15
27
Malta
Malawi
Malaysia
Norway
Nepal
Oman
Panama
Peru
Philippines
Poland
Portugal
Romania
Singapore
El Salvador
Slovenia
Trinidad, Tobago
Tunisia
Turkey
Tanzania
Ukraine
Uruguay
United States
Venezuela
7
2
6
2
16
3
5
6
8
6
5
5
2
8
1
8
6
5
14
16
10
4
15
9
5
17
14
2
18
6
1
4
2
15
0
15
6
2
2
7
16
8
1
7
18
6
Notes: (1) Ln(Capital stock per worker) are in thousands of constant 2005 US dollars and are generated from
UNIDO data. Capital stock per worker is the average capital stock per worker from 1994 through 2004
28
Table 3 (a): Capital Stock per Worker and Labor Freedom: With Country Capital per Worker
Ln(Capital per worker)
country level
Composite Labor Freedom
Index
Private credit (% of GDP)
Labor Freedom Index –
Minimum Wage
Labor Freedom Index –
Hiring and Firing
Labor Freedom Index –
Unemployment Insurance
Labor Freedom Index –
Collective Bargaining
Labor Freedom Index Conscription
Constant
Observations
R-squared
(1)
Country
Dummy
0.437**
(0.203)
-0.235***
(0.063)
1.035**
(0.439)
-2.405***
(0.632)
51
0.558
(2)
Country
Dummy
0.777***
(0.166)
(3)
Country
Dummy
0.711***
(0.172)
(4)
Country
Dummy
0.475**
(0.202)
(5)
Country
Dummy
0.756***
(0.166)
(6)
Country
Dummy
0.472**
(0.205)
(7)
Country
Dummy
0.585***
(0.182)
0.421
(0.387)
-0.042
(0.086)
-0.189
(0.136)
0.010
(0.089)
0.044
(0.091)
-0.038
(0.036)
-3.429***
(0.706)
38
0.726
0.371
(0.397)
-0.157**
(0.072)
0.842*
(0.467)
0.109
(0.349)
0.728*
(0.422)
0.323
(0.438)
-0.204**
(0.092)
-0.081
(0.057)
-0.151***
(0.050)
-3.096***
(0.508)
55
0.550
-2.818***
(0.627)
51
0.563
-3.692***
(0.535)
38
0.690
-2.654***
(0.684)
51
0.548
0.012
(0.037)
-3.738***
(0.475)
55
0.516
Notes: (1) Robust standard errors in parentheses. (2) *** p<0.01, ** p<0.05, * p<0.1, +p<0.12. (3) Country dummies are from a first stage regression
of Ln(Capital stock per worker) in thousands of constant 2005 US dollars generated from UNIDO data for each country in twenty-eight 3 digit ISIC
manufacturing industries from 1994 to 2004 on country, 3 digit industry and time dummies.
29
Table 3 (b): Capital Stock per Worker and Labor Freedom: With Income Categories
Income Category 2
Income Category 3
Composite Labor Freedom
Index
Private credit (% of GDP)
Labor Freedom Index –
Minimum Wage
Labor Freedom Index –
Hiring and Firing
Labor Freedom Index –
Unemployment Insurance
Labor Freedom Index –
Collective Bargaining
Labor Freedom Index Conscription
Constant
Observations
R-squared
(1)
Country
Dummy
0.794**
(0.364)
1.307***
(0.381)
-0.267***
(0.086)
1.462***
(0.459)
-1.991***
(0.497)
51
0.551
(2)
Country
Dummy
0.895*
(0.500)
1.360**
(0.598)
(3)
Country
Dummy
1.195***
(0.385)
1.775***
(0.390)
(4)
Country
Dummy
0.828**
(0.353)
1.426***
(0.414)
1.433*
(0.711)
-0.041
(0.115)
-0.123
(0.147)
-0.017
(0.130)
-0.019
(0.101)
-0.050
(0.040)
-2.158**
(0.837)
38
0.623
1.123**
(0.448)
-0.087
(0.083)
1.271***
(0.470)
(5)
Country
Dummy
0.960**
(0.407)
1.445***
(0.486)
1.049*
(0.549)
(6)
Country
Dummy
0.882**
(0.371)
1.301***
(0.427)
(7)
Country
Dummy
1.005**
(0.395)
1.575***
(0.434)
1.252***
(0.448)
1.043**
(0.472)
-0.214**
(0.097)
-0.099
(0.111)
-0.151**
(0.062)
-3.027***
(0.484)
55
0.497
-2.486***
(0.478)
51
0.548
-2.730***
(0.575)
38
0.586
-2.397***
(0.511)
51
0.528
-0.014
(0.041)
-3.293***
(0.423)
55
0.486
Notes: (1) Robust standard errors in parentheses. (2) *** p<0.01, ** p<0.05, * p<0.1, +p<0.12. (3) Country dummies are from a first stage regression
of Ln(Capital stock per worker) in thousands of constant 2005 US dollars generated from UNIDO data for each country in twenty-eight 3 digit ISIC
manufacturing industries from 1994 to 2004 on country, 3 digit industry and time dummies. (4) Income categories 2 and 3 are two dummies each for
the respective income categories based on 3 parts of Per Capita GDP in 2005 international dollars averaged across the period 1994 through 2004.
30
Table 4 (a): Industry wise Capital Stock per Worker and Labor Freedom with country level
Capital Stock per Worker
Dependent variable: Ln(Capital stock per worker) for each 3 digit ISIC manufacturing
industry
ISIC
Ln(Capital per
Labor
Private credit Observations
worker)
freedom
(% of GDP)
country level
index
0.3062
-0.1566
0.9893*
44
311 Food
(0.219)
(0.095)
(0.566)
0.3969**
-0.0566
1.0401*
36
313 Beverage
(0.164)
(0.122)
(0.539)
0.7553**
-0.1749
0.4142
29
314 Tobacco
(0.308)
(0.233)
(0.972)
0.3647
-0.5635***
2.2393***
39
321 Textiles
(0.255)
(0.161)
(0.806)
Apparel
-0.1897
-0.6002***
3.0033***
38
322
(0.256)
(0.143)
(0.822)
0.5079
0.1637
-0.2878
30
323 Leather
(0.449)
(0.182)
(0.945)
0.7214**
-0.4709***
0.8570
28
324 Footwear
(0.261)
(0.131)
(0.647)
0.5697**
-0.2715*
0.5710
39
331 Wood
(0.221)
(0.157)
(0.566)
0.7328***
-0.0774
0.3783
38
332 Furniture
(0.155)
(0.102)
(0.496)
Paper
0.7306**
-0.1770
0.1109
41
341
(0.271)
(0.122)
(0.697)
0.4396*
-0.0869
0.3812
36
342 Publishing
(0.247)
(0.121)
(0.622)
0.4231*
-0.1859*
0.9778
36
351 Industrial Chemicals
(0.243)
(0.106)
(0.612)
0.1176
-0.1127
1.2785*
38
352 Chemicals
(0.289)
(0.092)
(0.639)
Petroleum
0.8360***
-0.2563
0.6004
25
353
(0.247)
(0.251)
(0.806)
0.7144
-0.0684
-0.4569
10
354 Petroleum products
(0.516)
(0.501)
(1.920)
0.3052
-0.4241**
1.2709
36
355 Rubber
(0.294)
(0.182)
(1.014)
0.4099
0.0772
0.1103
42
356 Plastic
(0.309)
(0.172)
(0.797)
31
361 Pottery
362 Glass
369 Non-metallic minerals
371 Iron, Steel
372 Other metals
381 Metal products
382 Machinery
383 Electrical
384 Transport equipment
385 Scientific equipment
390 Other
0.5604
(0.337)
0.3510
(0.282)
0.1622
(0.276)
0.8772**
(0.377)
0.7149***
(0.256)
0.2455
(0.295)
0.6272**
(0.263)
0.3425
(0.215)
0.9450***
(0.194)
0.5792***
(0.195)
0.6403***
(0.213)
-0.5255**
(0.226)
-0.1390
(0.151)
-0.2165*
(0.111)
-0.2710
(0.212)
-0.1603
(0.225)
-0.1351
(0.146)
-0.0810
(0.095)
-0.1835
(0.130)
-0.1056
(0.094)
-0.0751
(0.113)
-0.3565**
(0.148)
1.0911
(1.136)
0.9559
(0.894)
1.4139*
(0.726)
-0.4039
(1.074)
0.1295
(0.950)
0.7912
(0.711)
0.2912
(0.648)
1.1365
(0.680)
-0.0242
(0.531)
0.3647
(0.622)
0.7041
(0.637)
25
35
35
30
31
40
38
42
41
32
36
Notes: (1) Robust standard errors in parentheses. (2) *** p<0.01, ** p<0.05, * p<0.1. (3) The dependent
variable is Ln(Capital stock per worker) in thousands of constant 2005 US dollars where Capital stock per
worker is averaged across the years 1994 through 2004. (3) The independent variable Ln(Capital stock per
worker) at the country level is in thousands of constant 2005 US dollars with Capital stock per worker
averaged across the years 1994 through 2004.
32
Table 4 (b): Industry wise Capital Stock per Worker and Labor Freedom with Income
Categories
Dependent variable: Ln(Capital stock per worker) for each
ISIC 3 digit manufacturing industry
ISIC
Income
Income
Labor
Private
Observations
Category 2: Category 3:
freedom
credit
Middle
High
index
% of GDP
Income
Income
0.5370
1.5067***
-0.1552
0.7885
44
311 Food
(0.389)
(0.434)
(0.094)
(0.541)
0.2499
0.8107
-0.1105
1.6054***
36
313 Beverage
(0.461)
(0.555)
(0.127)
(0.489)
1.7208**
2.0442
-0.1627
1.1459
29
314 Tobacco
(0.669)
(1.229)
(0.259)
(1.098)
0.6226
0.4961
-0.6448***
3.0905***
39
321 Textiles
(0.546)
(0.620)
(0.165)
(0.768)
Apparel
-0.3057
0.1685
-0.5350***
2.2462***
38
322
(0.629)
(0.849)
(0.144)
(0.736)
0.8753
1.3697
0.0976
0.3715
30
323 Leather
(0.895)
(0.944)
(0.153)
(0.890)
0.6559
1.6222**
-0.4965***
1.5941**
28
324 Footwear
(0.460)
(0.707)
(0.138)
(0.628)
0.7833
1.5610*
-0.3133*
1.0288
39
331 Wood
(0.511)
(0.869)
(0.184)
(0.807)
Furniture
0.9684**
1.6456**
-0.1637
1.3703*
38
332
(0.469)
(0.676)
(0.163)
(0.764)
0.9468
1.4533**
-0.2845**
1.2201*
41
341 Paper
(0.603)
(0.690)
(0.137)
(0.718)
0.5968
0.8901
-0.1452
1.0425*
36
342 Publishing
(0.529)
(0.633)
(0.118)
(0.520)
0.8388
1.1822
-0.2150*
1.4046*
36
351 Industrial Chemicals
(0.599)
(0.766)
(0.117)
(0.772)
Chemicals
0.4827
1.3195***
-0.0769
0.6197
38
352
(0.425)
(0.479)
(0.111)
(0.582)
1.5487**
2.3941**
-0.3214
1.2978
25
353 Petroleum
(0.690)
(0.866)
(0.270)
(0.845)
2.1272
0.2623
0.1470
0.5105
10
354 Petroleum products
(1.305)
(1.641)
(0.413)
(1.140)
1.1144*
1.0474
-0.4105*
1.4791
36
355 Rubber
(0.627)
(1.043)
(0.210)
(1.198)
Plastic
0.5746
1.7407**
0.0794
-0.0182
42
356
(0.692)
(0.651)
(0.164)
(0.623)
33
361 Pottery
362 Glass
369 Non-metallic minerals
371 Iron, Steel
372 Other metals
381 Metal products
382 Machinery
383 Electrical
384 Transport equipment
385 Scientific equipment
390 Other
1.2939
(0.916)
1.0497
(0.754)
0.5150
(0.586)
2.1720**
(0.790)
1.1431
(0.861)
0.4318
(0.536)
0.7807
(0.531)
-0.1041
(0.697)
1.4592**
(0.542)
1.1678**
(0.470)
0.8142
(0.533)
1.4284
(1.016)
0.9032
(0.827)
1.0464*
(0.566)
1.2527
(1.301)
1.4988
(1.021)
0.3347
(0.660)
2.0228***
(0.585)
0.3958
(0.722)
1.5817**
(0.653)
1.8633***
(0.576)
1.2766*
(0.722)
-0.5731**
(0.260)
-0.1850
(0.157)
-0.1935
(0.120)
-0.3908
(0.298)
-0.2965
(0.228)
-0.1869
(0.153)
-0.0770
(0.106)
-0.2546
(0.153)
-0.2546**
(0.120)
-0.0888
(0.152)
-0.4303**
(0.160)
1.7276
(1.252)
1.4449*
(0.822)
1.1025
(0.675)
1.2645
(1.305)
1.1955
(1.011)
1.3409*
(0.671)
0.4631
(0.695)
1.8486**
(0.693)
2.0749***
(0.722)
0.8103
(0.710)
1.7488***
(0.625)
25
35
35
30
31
40
38
42
41
32
36
Notes: (1) Robust standard errors in parentheses. (2) *** p<0.01, ** p<0.05, * p<0.1. (3) The dependent
variable is Ln(Capital stock per worker) in thousands of constant 2005 US dollars where Capital stock per
worker is averaged across the years 1994 through 2004. (3) The independent variables Income Categories 2
and 3 are two dummies each for the respective income categories based on Ln(Per Capita GDP) in 2005
international dollars averaged across the period 1994 through 2004.
34
Table 5: Labor Freedom and Industry Labor Reallocation
(Composite) Labor Freedom Index
Private credit (% of GDP)
Labor Freedom *US Job flow
Private credit*US Job flow
Labor Freedom *US Job flow* Income Category 1
Private credit*US Job flow*Income Category 1
Labor Freedom *US Job flow*Income Category 2
Private credit*US Job flow*Income Category 2
Ln(Capital stock per worker) for
each 3 digit ISIC manufacturing
industry
-0.246***
-0.117
-0.067
(0.033)
(0.134)
(0.129)
1.300*** 1.338***
0.955*
(0.152)
(0.494)
(0.491)
-0.387
-0.696
(0.880)
(0.852)
-3.063
-1.107
(3.126)
(3.132)
-0.112
(0.511)
-3.167
(2.951)
-1.144***
(0.405)
7.587***
(1.870)
Labor Freedom *US Job flow* Developing (Income Categories
1 & 2)
-0.999**
(0.409)
Private credit*US Job flow* Developing (Income Categories 1
& 2)
7.020***
(1.865)
-1.256***
(0.145)
-0.433***
(0.118)
Income Category 1
Income Category 2
Developing (Income Categories 1 & 2)
4.369***
(0.206)
Yes
970
0.476
Constant
3 digit industry fixed effects
Observations
R-squared
-1.386***
(0.425)
-0.241
(0.281)
-0.638**
(0.289)
4.249***
(0.280)
Yes
970
0.451
4.311***
(0.276)
Yes
970
0.492
Notes: (1) Robust standard errors in parentheses. (2) *** p<0.01, ** p<0.05, * p<0.1. (3) The dependent
variable is Ln(Capital stock per worker) in thousands of constant 2005 US dollars where Capital stock per
worker is averaged across the years 1994 through 2004. (4) The omitted category for Income categories is
Income category 3. (5) The variable ‘Job flows’ refers to the gross job creation and destruction rate for each
industry in the US.
35
Table 6 (a): Actual and predicted Ln(Capital stock per worker)
ISIC
Actual
Predicted Ln(Capital stock per worker)
Ln(Capital
Factor
Factor
Factor
stock per
Endowment
Endowment,
Endowment,
worker)
Financial
Financial
Development
Development,
Labor Freedom
2.285
2.093
2.194
2.226
311 Food
3.224
2.636
2.635
2.647
313 Beverage
Tobacco
-0.451
1.868
1.675
1.706
314
1.161
1.067
1.173
321 Textiles
1.850
0.882
1.126
1.267
322 Apparel
3.544
1.346
1.258
1.238
323 Leather
1.581
0.417
-0.396
-0.189
324 Footwear
2.408
1.051
0.900
0.987
331 Wood
-0.141
0.391
0.152
0.185
332 Furniture
5.725
2.182
1.887
1.957
341 Paper
Publishing
3.086
1.785
1.769
1.800
342
Industrial
Chemicals
2.713
3.034
3.041
351
2.644
2.587
2.824
2.849
352 Chemicals
4.449
2.766
2.738
2.761
353 Petroleum
3.805
2.240
2.404
2.376
354 Petroleum products
3.814
1.952
1.880
1.912
355 Rubber
6.349
2.450
2.524
2.508
356 Plastic
1.137
0.721
0.814
361 Pottery
Glass
4.005
2.165
2.229
2.241
362
Non-metallic
minerals
3.116
2.312
2.567
2.590
369
2.245
1.897
1.915
371 Iron, Steel
3.134
2.039
1.986
2.033
372 Other metals
2.099
2.123
2.146
381 Metal products
2.816
1.061
1.137
1.133
382 Machinery
3.116
1.798
1.799
1.856
383 Electrical
3.034
1.228
0.779
0.805
384 Transport equipment
3.397
1.175
1.062
1.063
385 Scientific equipment
Other
1.777
0.840
0.536
0.684
390
Notes: (1) The variable Capital stock per worker is in thousands of constant 2005 US dollars and is generated
from UNIDO data. Values are averages from 1994 to 2004. (2) Predicted Ln(Capital stock per worker) are
from cross-country regressions of Ln(Capital stock per worker) in thousands of constant 2005 US dollars for
each of the twenty-eight 3 digit ISIC manufacturing industries on three sets of explanatory variables. The
first column of predictions includes Ln(Capital stock per worker) at the country level as the only explanatory
variable. The second column adds on the percentage of private sector credit as a percentage of GDP as an
additional explanatory variable and the third column includes labor freedom as a third explanatory variable.
36
Table 6(b) : Actual and predicted Ln(Capital stock per worker)
ISIC
Actual
Predicted Ln(Capital stock per worker)
Ln(Capital
Income
Income
Income Category
stock per
Category
Category
Dummies, Financial
worker)
Dummies
Dummies,
Development,
Financial
Labor Freedom
Development
2.285
2.187
2.270
2.259
311 Food
3.224
2.944
2.966
2.964
313 Beverage
-0.451
2.274
1.961
1.971
314 Tobacco
1.445
1.503
1.354
321 Textiles
1.850
1.112
1.185
1.121
322 Apparel
3.544
1.358
1.359
1.394
323 Leather
Footwear
1.581
0.749
0.476
0.439
324
Wood
2.408
1.304
1.381
1.352
331
-0.141
0.677
0.702
0.679
332 Furniture
5.725
2.525
2.486
2.491
341 Paper
3.086
1.953
2.082
2.091
342 Publishing
2.946
3.214
3.184
351 Industrial Chemicals
2.644
2.617
2.729
2.725
352 Chemicals
4.449
3.041
3.013
2.977
353 Petroleum
Petroleum
products
3.805
2.633
2.690
2.744
354
Rubber
3.814
1.953
1.870
1.832
355
6.349
2.544
2.629
2.635
356 Plastic
1.719
1.224
1.064
361 Pottery
4.005
2.381
2.290
2.231
362 Glass
3.116
2.390
2.537
2.517
369 Non-metallic minerals
2.207
2.285
2.225
371 Iron, Steel
3.134
2.329
2.468
2.415
372 Other metals
2.264
2.322
2.281
381 Metal products
Machinery
2.816
1.415
1.623
1.590
382
3.116
2.316
2.365
2.343
383 Electrical
3.034
1.582
1.376
1.322
384 Transport equipment
3.397
1.316
1.269
1.228
385 Scientific equipment
1.777
1.192
1.102
1.113
390 Other
Notes: (1) The variable Capital stock per worker is in thousands of constant 2005 US dollars and is generated
from UNIDO data. Values are averages from 1994 to 2004. (2) Predicted Ln(Capital stock per worker) are
from cross-country regressions of Ln(Capital stock per worker) in thousands of constant 2005 US dollars for
each of the twenty-eight 3 digit ISIC manufacturing industries on three sets of explanatory variables. The
first column of predictions includes three income category dummies based on Ln(GDP per capita) as the only
explanatory variable. The second column adds on the percentage of private sector credit as a percentage of
GDP as an additional explanatory variable and the third column includes labor freedom as a third explanatory
variable.
37
Table 7: Capital-labor ratios for India and predicted capital-labor ratios for the US at Indian
wages
Industry Group
Actual capitalPredicted
labor ratio:
capital-labor
India
ratio: US
Food and Beverages/Tobacco
9.40
7.39
Textiles/Wearing Apparel
9.79
8.09
Leather
9.48
7.24
Wood
9.28
6.49
Paper/Printing/Publishing
10.51
8.67
Coke/Petroleum/Nuclear Fuel
12.06
10.75
Chemicals
11.40
9.34
Rubber
10.75
8.10
Non-metallic minerals
10.62
7.59
Basic metals
11.60
9.01
Metal
10.22
8.71
products/Machinery/Communication
equipment
Transport Equipment
10.28
8.61
Medical
10.13
8.04
instruments/Watches/Furniture/Other
manufacturing
Notes: (1) Ratios are averaged over the period 1989 through 1996
38
Table 8: Capital-labor ratios and Output shares for India and the US by industry
India
Actual
Ln(K/L)
Aggregate
Industry
Industry
Output
share
Food/Beverage
and Tobacco
Food/Beverage
Tobacco
9.75 0.93
7.19 0.07
Textiles and
apparel
Textiles
Apparel
Leather
Wood
9.87
8.88
9.48
9.28
Paper, printing
and publishing
Paper
Publishing/Printing
US
Ln(K/L)
at Indian
wages
9.75
9.45
Actual
Ln(K/L)
Output
share
11.65
12.56
0.93
0.07
10.89
9.43
10.25
10.75
0.61
0.39
1.00
1.00
10.93 0.65
9.64 0.35
10.40 12.10
10.02 11.29
0.45
0.55
Coke/Petroleum
12.06 1.00
12.40 13.95
1.00
Chemicals
11.40 1.00
10.93 12.44
1.00
Rubber/Plastics
10.75 1.00
9.67 11.16
1.00
Non-metallic minerals
10.62 1.00
9.64 11.57
1.00
Basic metals
11.60 1.00
10.53 11.96
1.00
Metal products,
machinery,
equipment
Metal products/Machinery
Office/accounting machinery
Electrical machinery
Communication equipment
10.04
10.83
10.38
10.73
9.88
11.10
10.09
10.66
11.30
12.58
11.17
12.06
0.55
0.11
0.18
0.16
Transport
equipment
Other transport equipment
Motor vehicles/trailers
9.86 0.43
10.71 0.57
10.00 11.53
10.38 11.72
0.31
0.69
Medical
10.46 0.27
instruments/watches/clocks
Furniture and other manufacturing
9.91 0.73
Notes: (1) Ratios are averaged over years 1989 through 1996
9.95 11.49
0.58
9.16 10.49
0.42
Miscellaneous
manufacturing
39
0.88
0.12
1.00
1.00
0.57
0.04
0.26
0.13
9.67
7.96
8.79
8.68
K
κ1
1/r
●E
κ2
X=1/Px
M
1/r’
● E’
Y=1/Py
O
1/w
1/w’
L
Figure 1: Cones of diversification and specialization in a Heckscher-Ohlin setting.
40
0
50
100
150
Capital stock per labor, 1980-2004
1980
1985
1990
1995
2000
2005
year
India
China
Figure 2: Capital stock per labor for the period 1980 through 2004 in thousands of 1995
Chinese Yuan per labor.
41
Figure 3 (a): Capital stock per labor by industry for 1980 in thousands of 1995 Chinese Yuan
per labor.
Figure 3 (b): Capital stock per labor by industry for 1990 in thousands of 1995 Chinese Yuan
per labor.
42
Figure 3 (c): Capital stock per labor by industry for 2000 in thousands of 1995 Chinese Yuan
per labor.
Figure 3 (d): Capital stock per labor by industry for 2004 in thousands of 1995 Chinese Yuan
per labor.
43
Data Appendix
Trade, Production and Protection Database, Nicita and Olarreaga (2006), data at the 3 digit
ISIC industry level for each country over time:
Number of Employees for each 3 digit ISIC industry: Total number of persons who worked in
or for the establishment on average during the reference year. The number of employees is
including all persons engaged other than working proprietors, active business partners and unpaid
family workers.
Capital Stock: Gross fixed capital formation reported in thousands of US dollars, defined as the
value of purchases and own-account construction of fixed assets during the reference year less the
value of corresponding sales, is converted to capital stock in thousands of 2005 US dollars assuming
a 15 percent depreciation rate for capital and a steady growth rate of investment (gross fixed capital
formation). The fixed assets covered are those (whether new or used) with a productive life of one
year or more.
World Development Indicators, World Bank, data at the country level over time:
GDP Per Capita: GDP per capita is the PPP adjusted GDP per capita in thousands of current
dollars.
Number of Workers: Total labor force comprising of people ages 15 and older who meet the
International Labor Organization definition of the economically active population: all people who
supply labor for the production of goods and services during a specified period. It includes both the
employed and the unemployed.
Capital Stock: Gross Fixed Capital Formation, defined as outlays on additions to the fixed assets
of the economy plus net changes in the level of inventories reported in thousands of current US
dollars, is used to generate capital stock in thousands of 2005 US dollars assuming a 15 percent
depreciation rate for capital and a steady growth rate of investment (gross fixed capital formation).
44
The Fraser Institute, Economic Freedom of the World:
Labor Freedom: Labor Freedom sub-index that is a part of the Economic Freedom Index (see
Economic Freedom of the World 2009 Annual Report, http://www.freetheworld.com/release.html
and Freeman (2009)). We use both the composite Labor Freedom sub-index and the indices for
each of the sub-components for labor freedom: Minimum wage regulation, hiring and firing
regulation, centralized wage setting, extension of union contracts to nonparticipating parties and
conscription (use of draft).
Djankov, McLiesh and Shleifer (2007), data at the country level from IMF, International
Financial Statistics, averaged for the period 1999 through 2003:
Private Credit (% of GDP): claims on the private sector by commercial banks and other financial
institutions. The variable is expressed as a percentage of GDP.
Haltiwanger, Scarpetta and Schweiger (2008), data on U.S. industry level job flows:
Industry-wise Gross Job Flow for the United States: Sum of the job creation rate (defined as the
increase in employment from the previous period divided by the average employment in the two
time periods) averaged over the 1989-2001 and the job destruction rate (defined as the negative of
the decrease in employment from the previous period divided by the average employment in the two
time periods) averaged over 1989-2001.
NBER Manufacturing Industry Productivity Database, data for the US at the 3 digit US SIC
level and Annual Survey of Industries, data for India at the 3 digit NIC level, 1989 through
1996:
Real Net Capital Stock (US): Obtained using a perpetual inventory model and data on real
industry investment8 in 1987 US dollars.
The NBER Technical Working Paper by Bartelsman and Gray (1996) provides details on the calculation of
the real net capital stock from investment data for the US.
8
45
Real Capital Stock (India): Book value of fixed capital deflated to 1981 Indian rupees by a whole
sale price index for machinery, transport equipment and construction as in Hasan, Mitra and
Ramaswamy (2007).
Workers (US and India): Total number of production workers.
Wages (US): Production worker wages in current US dollars.
Wages (India): Total wages and salaries in current Indian Rupees.
46