In India, growth in the 1990s was accompanied by increasing
inequality across and within states, between rural and urban
areas, and within urban areas (Deaton & Dre`ze, 2002; Dhongde,
2007). Though growth led to considerable poverty reduction,
increasing inequality offset part of its effect. The slowdown
of poverty reduction is one reason to care about inequality,
but even in itself it is a key characteristic of the development
process and of actual concern to policy makers (Kanbur, 2000,
2007). Rising inequality puts stress on popular support for
growth strategies and threatens social and political stability.
As such, it may be detrimental to future growth and poverty
reduction (Nissanke & Thorbecke, 2006). Reducing
inequality and achieving more inclusive growth are in fact prime
objectives in India’s current Five Year Plan (Government of
India, 2008).
Inequality is related to the sectoral structure of growth because
different industries use different production factors and
different households differ in their supply of various production
factors. Ravallion and Datt (1996) show that during
1950–90 growth in the primary and tertiary sector reduced
poverty in India, while growth in the secondary sector did
not. They relate this to growth of the capital-intensive production
in manufacturing in this period, which was not beneficial
to the poor. Similar conclusions are drawn in Khan and Thorbecke
(1989) and James and Khan (1997). Their study of the
Indonesian Social Accounting Matrix confirms that traditional
labor-intensive technologies are more egalitarian than
modern capital-intensive technologies. The reason is that production
under traditional technology creates more employment,
directly and indirectly, and more income for rural
households. These studies focus on the distribution of value
added between capital and labor, but do not address inequality
among workers.
Inequality of earnings is an important source of total income
inequality (Gottschalk & Smeeding, 1997). Many studies
have shown that the wage rate of skilled relative to unskilled
workers, the skill premium, has risen in developing countries
(Anderson, 2005; Goldberg & Pavcnik, 2007). Kijima (2006)
finds that earnings inequality in India in the 1990s increased
due to a rising skill premium. Especially the returns to tertiary
education increased much, because relative demand outgrew
relative supply. Furthermore, and related to this, the service
sector has been the leading sector in terms of output and
employment growth and is the most skill-intensive sector.
During 1980–2000, labor moved out of agriculture into services,
while the employment share of manufacturing hardly
changed (Mazumdar & Sarkar, 2008, p. 225). Likewise,
Chamarbagwala (2006) finds that in the period 1983–2000
employment in India shifted from low-skilled into high-skilled
and medium-skilled occupations due to service sector expansion
and agricultural sector contraction. It seems clear that
inequality of earnings, and especially the skill premium, is
an important factor in the relationship between sectoral
growth and household income inequality. Therefore, the distribution
of skills across households should be taken into account
when analyzing this relationship.
The aim of this paper is to find out how the sectoral structure
of growth contributes to household income inequality in
India, and to show how one can—and should—account for
inequality among workers. The analysis is based on an extended
Social Accounting Matrix (SAM) for the years 2002–
03. The SAM is a suitable tool to analyze the distributive effects
of sectoral growth, as it captures the flow of income
and interdependence between industries, production factors,
and households, among others. The SAM has been widely
used for development planning, reflecting the view that aggregate
economic growth is an inadequate policy objective unless
attention is paid to distributional changes (see Defourny &
Thorbecke, 1984; Hayden & Round, 1982; Pyatt & Round,
1977). Due to its underlying assumptions of constant technology
and excess capacity, SAM-based multiplier analysis is
* I would like to thank Marcel Timmer and three anonymous referees for
helpful comments and suggestions. This paper also benefitted greatly from
participants’ comments at the IE&B Staff Seminar, University of
Groningen; at the IARIW 30th General Conference in Portoroz, Slovenia;
at the EUNIOS Workshop on Agricultural Policies and Social Accounting
Matrices in Developing Countries in Groningen, The Netherlands; and at
the 2nd Sino-Netherlands Workshop on Input–Output Techniques in
Beijing, China. Any remaining errors are my own. Final revision accepted:
July 22, 2009.
sometimes regarded as rather restrictive. However, it offers a
transparent framework of data with macroeconomic consistency.
Compared to regression-based studies of sectoral
growth and inequality or poverty (e.g., Loayza & Raddatz,
2009; Ravallion & Datt, 1996) the SAM analysis offers more
insight by taking into account sectoral interdependencies and
uncovering the channels through which income flows. Standard
SAMs for India are available and have been used, for
example, in Ten Raa and Sahoo (2007).
A methodological contribution is made by extending the
standard SAM through accounting for the skill-intensity and
the skill premium by sector and the education and sector of
employment of households. This is done using satellite accounts
for earnings and employment by sector and household
survey data for education and employment characteristics.
The extension of the SAM consists of dividing the single wage
account into 30 sub-accounts: three levels of educational
attainment and 10 sectors of employment. The main innovation
is the fact that each sector has its own wage account in
the extended SAM. It shows the distribution of wage income
between 30 different worker subgroups within the SAM’s representative
household groups.
Distribution analysis based on the standard SAM shows
that growth of any sector will slightly reduce inequality, suggesting
that the sectoral structure of growth does not matter
for inequality. After extending the SAM, however, we find
that growth in several sectors increases inequality between
and within household groups. The effects are largest for community,
social, and personal services; followed by heavy manufacturing
and the other services sectors. Growth in these
sectors increases inequality because they pay relatively high
wages (the sector premium), they are skill-intensive, and pay
a high skill premium. Only agricultural growth reduces
inequality. The results confirm the importance of our extension
for an analysis of income distribution, and emphasize that
employment creation is not sufficient to secure equitable
growth. They strengthen the call for the development of unskilled-
labor-intensive manufacturing, as India’s current pattern
of growth offers too little opportunities for low-skilled
workers.
The rest of the paper is organized as follows. In Section 2 the
structure of the SAM, multiplier analysis, and the method for
extension of the SAM are discussed. In Section 3 the data are
presented, and in Section 4 the results are discussed. In Section
5 the results are related to India’s policies and pattern of
growth. Finally, Section 6 concludes.
Rapid growth in India in the past decades has led to the relative
decline of agriculture and growing importance of services.
This has fueled inequality, which is likely to increase
further in the future. Based on an extended SAM, we find that
growth of community, social, and personal services raises
inequality between and within household groups most, followed
by heavy manufacturing and other services sectors.
Only agricultural growth reduces inequality, but the growth
potential of agriculture is limited compared to manufacturing
and services.
Much of our findings are explained by the high skill-intensity
and skill premium of some sectors, combined with a high
sector premium. This stresses the importance of taking into account
the educational level and sector of employment of
households, as well as sector- and education-specific wages,
when analyzing the effects of sectoral growth on inequality.
We find that the sector premium is higher in sectors with a
higher public share of GDP, which is related to the observations
by Deaton and Dre`ze (2002) that public salaries grew
twice as fast as the agricultural real wage in the 1990s. It is
not at all clear that public wages are driving wage differentials
between sectors or to what extent, but this may be an interesting
question for further research.
We do not conclude that India’s development strategy
should focus solely on agricultural growth to reduce
inequality. Of course, investing in agriculture to increase
the sector’s labor productivity could benefit the agricultural
workers. But demand growth for agricultural products will
be lower than for other sectors, so there is limited scope
for expansion. Therefore, our results confirm the importance
of expansion in unskilled-labor-intensive manufacturing, as
argued by Kochhar et al. (2006) and Krueger (2007). Inclusive
growth requires employment opportunities for lowskilled
workers outside agriculture, in both rural and urban
India.
As a more general conclusion, our results confirm that the
sectoral composition of growth and the production technology
of sectors matter for inequality, in line with Ravallion and
Datt (1996), Datt and Ravallion (2002), and James and Khan
(1997). However, we find that the skill-intensity of a sector is
more important for inequality than the capital-labor ratio.
This is in line with the increasing wage inequality that has
accompanied globalization in many developing countries
(Anderson, 2005; Goldberg & Pavcnik, 2007). Since countries
can differ greatly in terms of the sectoral structure of growth
and production technologies, one should question the relevance
of cross-country studies that assume a common relationship
between GDP growth and inequality (e.g., Barro,
2000).
We also draw an important methodological lesson from
this study, namely, that a social accounting matrix (SAM)
with a single wage account will produce misleading results:
the level of aggregation in the wage account matters for the
distributive effects of growth that can be measured. The standard
SAM for India has a single wage account and shows
that between-group inequality is neutral to the sectoral structure
of growth. Only once labor is divided according to educational
level and sector of employment, it becomes clear that
the sectoral structure of growth matters for inequality.
Though SAMs for other countries usually have separate labor
accounts for different educational levels and sometimes
for agricultural versus non-agricultural labor (e.g., Khan &
Thorbecke, 1989, for Indonesia; Jensen & Tarp, 2005, for
Vietnam), the subdivision according to sector of employment
is never made. Especially when households are grouped on
the basis of geography or ethnicity, the link with their sector
of employment is too indirect. Besides improving this link,
another advantage of our extension is that it adds information
on within-group inequality of earnings. For India, we
find within-group inequality constitutes more than one-fourth
of total earnings inequality.
This study is a simple exercise, and the results are best considered
as an indication of the relationships we are interested
in. Despite the limitations of the SAM model, the results are
intuitive and emphasize the importance of inequality among
workers due to differences in educational attainment and sector
of employment in India’s economic structure. The increasing capital-intensity of India’s industrial sector is a concern to
the government (Government of India, 2008), but it should
be clear that employment creation will not be enough to secure
equitable growth. Rather, a focus on employment
growth for low-skilled labor is essential for reducing inequality.
