Download as pdf or txt
Download as pdf or txt
You are on page 1of 23

Gender Wage Gap in the Indian Labour Market

Evidence from the NSS 66th Round Data

Shiney Chakraborty 1
Subrata Mukherjee 2

1. Introduction
During the last century theimpressive economic growth of Indian economy is unable to generate
sufficient ‘decent work’ 3 opportunities for the growing labour force and the latest National
Sample Survey Organization (hereafter NSSO) round (66th round:2009-10) data reveals the
decline in employment growth rate.The total decline in employment growth rate from 2.7 per
cent in 2004-05 to 0.8 per cent in 2009-10 of Indian economy is due to the slump in female
employment and the declining female work participation rates (hereafter FWPR) manifests
it(Chandrasekhar and Ghosh, 2011a, 2011b).Women workers generally face discrimination at the
time of entry or participation in the labour market and work force participation rates (hereafter
WPR)4of women are lower compared to men. There is also a sharp difference in work force
participation between rural and urban areas (See Figure 1). An investigation of theWPRfrom the
latest NSSO round (66th round) reveals a decline in FWPR from the last quinquennial round
2004-05,in both rural and urban areas. Figure 1 indicates that while the male WPR is persistently
above 50 per cent in both rural and urban areas throughout the years, female participation in
work force is substantially below their male counterparts. The female employment in the urban
areas has always been persistently lower than that in the rural areas.

A possible reason for different WPR between men and women could be rooted in gender
inequality in opportunity to work.This gender inequality hampers the development goals and
reduces economic growth via under-utilization of talent and hence decreases productivity of the

1
(corresponding author), MPhil student, Institute of Development Studies Kolkata, DD-27/D, Sector I, Salt Lake,
Kolkata 700064. Email address: shiney21chakraborty@gmail.com
2
Assistant Professor, Institute of Development Studies Kolkata, Email address: msubrta100@gmail.com
3
Decent work is defined by the International Labour Organization (ILO), as the employmentthat takes place under
conditions of freedom, equity, security and dignity, where rights are protected and adequate remuneration and social
coverage are also provided (ILO, 2000).
4
Work participation rates (WPR) or worker-population ratio (WPR) is defined as the number of persons/person-days
employed per thousand persons/person-days.

1
economy. Gender inequality is observed in different areas of human life one of which is in the
payment of wages and remunerations in the labour market. According to the International Trade
Union Confederation (hereafter ITUC) report (2008) world average gender wage gap is 16.5 per
cent. In Asian countries it is 21.2 per cent. Gender wage gap is the difference in wages between
men and women in paid employment and a review of various literatures (See for e.g. Kingdon
and Unni, 2001; Das, 2006; Chakraborty and Bhaumik 2008) depicts that gender wage gap
persists in the Indian economy which motivates us to unveil the aspect of disparity in male
female earnings.Here, earnings refer to the wage/salary income which is received for the
wage/salaried work done during the reference week by a wage/salaried employee and a casual
labourer. Table 1 indicates that gender disparity in wages is widespread and over the years
females are getting lower wages than males. Women casual labourers are getting less than men
and wage difference is large among casual workers than regular employees.

But from the ratio measures the wage gap can-not be measured exactly and the underpinning
factors related to wage difference can-not be determined.So in this study the determinants of
wages are identified by performing regression analysis and then the gender wage gap is divided
into the combined effect of difference in endowment and pure discrimination component by
applying the Blinder-Oaxaca (1973)decomposition technique.

There is a large body of literature on wage discrimination in the developed countries but
literature is limited in the context of developing countries. According to Oaxaca (1973)male-
female wage differential was mainly for the discrimination componentin the United States.
Peterson et al, (1997) suggested thatoccupational segregation was the main source of wage gapin
Norway.Paternostro and Sahn (1998) in Romania and Richard (2007) in Uganda found that much
of the wage gap was due to discrimination. In the Indian context Duraisamy and Duraisamy
(1996) and Das (2006) found a larger discrimination component.Bhaumik and Chakraborty
(2008)indicated a decline in gender earnings gap in India from 0.46 to 0.12by estimating the
earning function and by applying the Oaxaca decomposition technique from 1987 to 1999. A
review of these studies indicates that gender differential in wages still exists which needs further
examination and elaboration with the current data. In this study our objective is to examine the
role of various individual characteristics and macro-characteristics/contextual factors in the

2
determination of gender-based wage difference. This objective also motivates us to take up the
following question: Does male-female wage difference in a particular sector/employment
category get reduced with higher women participation in that sector?The reason behind this
argument is that with more number of women workers, bargaining power of women increases.
The increase in collective bargaining power of women will lead to an increase in female wage in
that particular sector/employment category. So gender wage gap will be reduced. This can be
perceived as a demand side argument.

The following sections of the paper are organized as follows: Section 2 describes the
Methodology adopted for addressing the questions and the briefly describe the data sources used
for this study. Section 3 presents empirical results and discusses them in the light of previous
findings. Section 4 concludes the paper and draws some policy lessons.

2. Methodology and Data Sources


According to the payment nature of employment, the workers are classified into two main
categories: wage workers and non-wage workers. Wage workers consist of regular workers and
casual workers and non-wage workers can be termed as self-employed. Wage information for the
regular salaried workers and casual workers are available from surveys but as for the self-
employed category it is not easy to separate out the wage component, as there is no information
on wage for the workers who are also self-employed. As in this study we are mostly concerned
with gender based wage difference, we only consider, the regular wage / salaried employees and
the casual labourers of the age group 15-59 years for our analysis. 5 For the purpose of the gender
wage gap analysis first an ordinary least squares (hereafter OLS) estimation of a simple
unadjusted model of wage equation of the semi-log functional form is performed such that
(1)
Where ln Y i denotes the natural logarithm of daily wages of individual i, X i is a vector of
observed characteristics, β denotes the vector of regression coefficients, and ε i is a random error
2
term distributed N (0, e ). The natural logarithm of the daily wages (by adding Re 1 to each
wage rate in order to avoid non-feasibility of logarithmic transformation of ‘0’ wage rate) is

5
Only the age group 15-59 years is considered for the analysis as it is the most productive age group for wage based
employment.

3
usedas the dependent variable 6. There are nine independent variables considered for
determinants of the gender wage gap, of which two variables are quantitative and seven are
categorical variables. Two quantitative variables are age and age-squared. Unfortunately there is
no data on the experience level of the individual. Mincer (1974) proposed to estimate experience
as [Age-Schooling-6]. But National Sample Surveys (hereafter NSS) do not report years of
schooling, only the maximum level of schoolings is reported which is used to construct the
general education dummy variables. Since there is no uniform age of admission in schools
especially for rural sector females so age-squared is a good proxy for experience 7. Qualitative
variables include socio-religious group, marital status, general education, technical education,
principal status, industry8, country region (See Appendix 1 for details).

Most of the studies on gender wage gap use simple OLS regression technique to estimate the
wage equations, incorporating the wage earners only. But the presence of a large number of
unemployed people can lead to selectivity bias. In the presence of sample selection, OLS
estimation method will give biased and inconsistent estimators (Heckman, 1979). It is therefore
recommended to improve the wage equation by using Heckman correction procedure for
selectivity bias. In this paper we use the Heckman two-step estimation procedure. The wage
equation is, as previously given by

The basic idea of sample selection model is that the wages are only observed for those
individuals for whom Z* i >0, where Z* i (employment function) is given by:
(2)
Where is a latent variable associated with employment,H΄ i is a vector of determinants of
employment, α is the associated parameter vector and ε i is the error term.

Next the Oaxaca (1973) and Blinder (1973) decomposition analysis is performed to divide the
wage gap into explained component and unexplained/discrimination component. This
decomposition method explains the gap in the means of an outcome variable between two
6
Therefore, we consider log (wage rate) =log (wage rate+1) as a dependent variable.
7
The use of age instead of experience may lead to erroneous result for the individuals with interrupted labour force
participation.
8
Up to 2 digit NIC-04 classifications are taken for analysis.

4
groups, male and female. Decomposition method separates the wage gap into two factors. One
can be explained by different return to individual characteristics (endowment component),
another portion of the differential can-not be explained and is attributable to discrimination
(discrimination component). The gross wage differential can be written as

(3)

Where Y m and Y f are male and female wages respectively. Becker (1971) extended the model of
a competitive market discrimination coefficient for labour to include the influence of gender and
other personal characteristics. In the absence of discrimination wage difference between male
and female is the pure productivity difference (Q) which is defined as

(4)

Where Yo m / Yo f is the competitive wage ratio in the absence of discrimination. Becker defined
the discrimination coefficient as the difference or residual between observed male-female
average wage ratio and the wage ratio in the absence of discrimination. Blinder –Oaxaca stated it
as

(5)

Where Y m / Y f is the observed wage ratio. In logarithmic form this can be expressed as:
(6)
Therefore the difference in the log of observed wages is divided into two portions: the first
expression on the right hand side is for discrimination and the second component is for
difference in male-female productivity related characteristics. The decomposition technique can
be further expanded to incorporate the semi-logarithmic wage equations and estimated via OLS.
The B–O wage decomposition technique requires two separate regression functions for male and
female. So in order to investigate the sources of gender differentials in detail, men’s and
women’s wage functions are estimated separately in rural and urban sector such that:
(7)

(8)
Where Y denotes geometric mean earnings, X the vector of mean values of the regressors,
βvector of estimated coefficients,ε is the error term with zero mean and constant variance, m

5
represents men, f stands for women and r denotes rural sector whereas u refers urban sector. A
simple log mean wage difference between men and women can be written as:
(9)

Where
By adding and subtracting Xm βf in equation 9, the equation can be written as
(10)
Again the addition and subtraction of X β with equation 9 yields equation 11.
f m

(11)

Equation 10 implies that in the absence of discrimination, female wage structure would prevail in
the market where as equation 11 indicates the prevalence of male wage structure in a non-
discriminatory market. Blinder and Oaxaca (1973) developed decomposition approaches to
partition the gender wage differential into components caused by two factors: The first term of
the right hand side of the equation (10 and 11) captures how the male-female wage differential
changes in response to changes in the men-women gap in characteristics. The first term is
sometimes called ‘observed X’s’ or ‘observed gender gap in characteristics’. The second term
measures the unexplained wage gap for differences in coefficients or returns. This term is
considered to measure the level of ‘gender discrimination’.The data used in this study comes
from the four quinquennial surveys (50th, 55th, 61st, 66th rounds) on employment and
unemployment (Schedule 10) of the NSSO.While we have used the published reports based on
the rounds, we have also used unit-record data from the 66thround for the detailed analysis. In the
analysis of the NSS unit record data we consider only those causal labourers who are engaged in
non-public works by excluding the casual workers involved in public works. 9

3. Results and Discussion


Before analyzing the estimated wage equations, we have estimated the mean and variation of
dependent and independent variables which are presented in Table 2. Table 2 shows that male-
female wage gaps can be computed and it is 32 per cent for the total samples – including workers

9
Casual workers employed in public works programme are excluded as their wages are administratively decided and
are not expected to vary by gender. Further, they are still not significant percentage of total casual work force.

6
in the rural and urban areas. 10 Out of the total 28, 8662 workers in the 15- 59 years age group
who are either regular salaried employee or casual wage labourersbased on their principal status,
51 percent are male and 49 percent are female.

The result of the selectivity corrected log wage equations for both males and females are
deciphered along with OLS regression results in rural and urban areas. Heckman regression
analysis is performed only on wage earners as the unobserved determinants of wages is expected
to vary between the wage earners and non-wage earners and this would tend to bias the
regression results if not controlled for. So sample selection is considered using Heckman two-
step selection model(Heckman, 1979). 11 In controlling for sample selection, three variables
[number of children (below five years) in the household, number of old members (above 60
years) in the household, people participated in the wage employment] are taken as selection
variables and the variables are expected to have a direct impact only on participation in the
labour market. Estimation results of wage rates for male and female using OLS and selectivity
correction are reported in Table 3.Econometric analysis highlights a concave relationship
between age and its quadratic with wages. In other words age and its square (age and age-
squared) have a significant positive and negative effect respectively on wages for both the sex
across sectors in both the OLS and selectivity corrected model.Coefficients of socio-religious
group show mixed (positive and negative) values and it can be inferred that urban sector labour
market is highly discriminated for the backward socio-religious group than the rural sector
labour market and this is consistent with the study on caste based discrimination in urban India
by Banerjee and Knight (1985). The result of caste and religious based discrimination in this
study is contrary to the previous study by Kingdon and Unni (2001) who found that neither the
people of the lower castes nor the Muslims faced wage discrimination in the labour market.

Marital status has a positive and significant effect on both males and females in the rural areas
and in the urban areas in all the eight equations.Kingdon (1998) have also found the same result
for male but she found an insignificant effect of marital status for female. However, according to

10
Differences in natural logarithm of daily wage rates can be converted by using the formula 100[exp (difference in
wage rate)-1] to percentage wage differences. The difference in the mean log daily wages for females and males in
the total sample in Table 4.2 is -0.39 both for rural and urban and which yields a 100[exp (-0.39)-1] =-32percent
wage difference and the negative sign indicates an existence of gender discrimination.
11
The models are estimated using STATA 10 (Hamilton, 2008).

7
her, marriage is endogenous to earnings i.e., those who earn more are more likely to get
married.The returns to education are all positive in the selectivity correction and OLS. The
coefficients of literate below primary and primary middle are not significantly different from
zero whereas the coefficients of secondary and higher secondary and specially the coefficients of
tertiary education are significantly different from zero, which implies that just being literate, or
literate with primary and middle schooling are not enough to earn better labour market rewards.
The result also indicates that returns to education increase with the level of education both for
males and females. This result is consonant with the previous studies in India by Kingdon
(1998). However, trend tends to favour rural male only up to secondary and higher secondary
education and favours females more than males with graduation and above qualification. In
urban area people having a university degree are getting more wages than their rural area
counterparts. Duraisamy and Duraisamy (2006) found that return to primary and middle
education is lower comparative to the secondary and higher secondary education for both the
genders and this is also corroborated in this study. For instance, the wages of rural females with
university degree rise more than 94 per cent in OLS and 75 per cent in Heckman selectivity
method relative to one with secondary and higher secondary education. While the wages of male
individuals rises by 49 per cent and 47 per cent respectively in OLS and selectivity correction
method to one with secondary and higher secondary education in rural areas. The difference is
more pronounced when it comes to urban sector. It is noted that earning of urban women with a
university degree increases by more than 129 per cent in OLS (125 per cent in other method)
compared to that of a men which rise by 78 per cent and 80 per cent in both the OLS and
selectivity corrected methods respectively.

The coefficient value of technical education is positive in both the methods indicating that rate of
return to technical education positively influences the wages and persons having this are better
paid.The rate of return to casual labour market is negative in all the eight equations, and it
implies that as workers deviate from regular salaried market to casual wage labour market, their
wage rate decreases and its negative impact is more in urban areas. So it can be argued that the
wage difference is pronounced for casual wage labourers. The wage difference between regular
and casual workers may be due to the educational and skill differences between them.Industry
wise analysis suggests that both in rural and in urban areas male wages are higher in ‘other
services’ industry followed by construction industry in both the two methods and industry wise
8
return is higher in rural areas. While in OLS method female get higher wages in ‘construction’
sector followed by transport, storage and communication industry in both rural and urban areas,
but rate of return is higher in urban areas. After controlling the selection bias rate of return for
female is higher in transport, storage and communication industry followed by construction
industry and again in urban areas return is higher. Across industries as workers are deviating
from agriculture to trade, hotels and restaurant industries their wages decrease, for both the sexes
in rural sector as also in urban sector. In this study regional analysis is also done by taking into
account of regional dummies. The associated F and t statistics in the OLS models suggest a good
fit of the model.

An important finding from Table 3 is the insignificant value of the selectivity variables for male
and female in both the sectors. So the problem of selectivity bias does not exist. Alternatively
there is no effect of selection (i.e. those who get select into the wage sample have no higher
wages relative to those with average characteristics drawn at random from the population).Given
the wide spread persistence of the gender wage gap, the final step is to decompose the wage gap
between men and women into two parts i.e. the part of the gap that is due to characteristics and
that can be explained due to differences in relative endowments of women and men and the other
part is unexplained and it may be due to discrimination in employment and wages.The
coefficient estimates from the OLS wage regressions 12 are used to decompose the wage gap
between male and female. In Table 4,a positive number indicates the percentage by which the
gender wage gap would be reduced if male and female are equal in respect to the characteristic
assuming that the characteristic is rewarded according to the estimated wage function for
female/male. Negative number implies that if women are more like men in this respect but the
wage functions remains the same will lead to an increase in gender wage gap. The average
gender wage gap in rural area is 38.8 per cent which is divided into two parts-the portion that is
explained due to difference in relative endowments (0.8 per cent) and the other portion is due to
different returns to productive factors (38.0 per cent).

Table 4 indicates that the discrimination component is larger than the endowment component in
rural areas. An alternative explanation is using the pure discrimination approach (male-wage

12
Decomposition analysis can also be performed after taking into account for selection bias. Following Neuman and
Oaxaca (2003) deduct the selection effects from the overall differential and then perform the standard decomposition
method to this adjusted differential.

9
structure), the proportion of the wage differential due to discrimination is 98 per cent and the
proportion of the wage gap for endowment factor is two per cent. Standardizing by female wage
structure (pure nepotism), gives a lower estimate of discrimination, namely 91 per cent of wage
gap arises due to discrimination and nine per cent is due to different return to characteristics.

On the other hand, in urban areas the average gender wage gap is 39.5 per cent. This is
segregated into the portion that is explained for differences in characteristics (-12.6 per cent) and
the other portion is due to unexplained factors (discrimination) (52.1 per cent). This result
implies that in urban sector, endowment component would tend to favour women and if returns
to endowments are same for males and females, the expected female wage would be greater than
the male wage. But differences in returns to relative endowments, along with the factors that the
model can-not capture would result in a net reduction of average female wages by 52.1 per cent.
The result of B-O decomposition technique in urban area implies that by using male wage
structure discrimination component takes into account 132 per cent of the wage gap whereas
explained component comprises of 32 per cent of the wage gap.Using the pure nepotism (female
wage structure) 90 per cent of the difference in male-female mean of log wage rates is due to
unexplained factors, while 10 per cent is due to explained factors. In both rural and urban areas
using the two different wage structures (male wage structure and female wage structure), there is
a difference between the explained components and the unexplained components which lead to
the conclusion that B-O decomposition method suffers from index number problem.

After analyzing the determinants of wage rate and the underpinning factors of gender wage gap,
in this section the objective is to find out whether this wage gap is expected to reduce with
increased women’s participation in the job market.In a study Mahajan (2011) found that for
greater supply of female labourers to agriculture in southern states gender differential in wages is
also much more compared to northern states where women’s participation in the job market is
lower. She suggested that higher supply of female labourers to agriculture would reduce female
wages more and affect female to male wage ratio negatively.Here an OLS regression analysis is
performed to underpin the determinants of the wage gap. Here the dependent variable is
difference between male and female wage rate and the independent variables are percentage of
women workers, percentage of rural workers, male and female average years of education and
male and female average age. The data is taken across industries (Fifteen industries are for the
10
regular workers and sixteen industries are for the casual workers) (See Appendix 2 for list of the
industries). Summary statistics of the variables used in the regression analysis are presented in
Table 5. Table 6 reveals that the coefficient value of percentage of women workers is positive
and highly significant implying that with increased number of women workers men women wage
gap increases. Alternatively, the hypothesis, that ‘as the number of women workers increases in
a sub-sector or industry, their collective bargaining skill will increase which will eventually
lower the wage gap’ -is rejected. In the Indian context increase in number of women workers in a
sector or industry does not lead to reduction in the gender-wage, rather, women get higher
wages in sub-sectors/industries where their relative share is low.

One of the important findings of our econometric analysis is significant positive coefficient value
of male average years of education and male average age. The other is the negative coefficient
value of female average years of education and female average age. This implies a one year
increase in male average years of education will lead to an increase of 9 rupees wage gap
whereas one year increase in female worker’s education will lead to 12 rupees reduction in wage
gap. This has an important policy implication, namely, that an increase in female average years
of education will help to reduce the wage gap. On the other hand, the negative coefficient of the
average age of female workers implies that an increase in average age of female workers will
help to reduce the wage gap. A close look at the coeffcient value of the variables in above
regression suggests that it would be more effective to increase the educational level of the
women in order to reduce the wage gap than other variables such as average age etc. The value
of the F and t statistics and R2 indicate that the model fits the data well.

4. Conclusion
This study is one of the first few comprehensive studies on gender wage gap in India which has
attempted to correct the sample selection bias – a methodological improvement in the model
which is not normally found in such studies. The labour economics literature is replete with
numerous studies on wages and earnings in India by legion of eminent scholars by specially
focusing on agricultural wages or on wages in the manufacturing sector but there are very few
studies on wages and earnings across different sections of labour markets.The present study has
tried to bridge the evidence gap in this regard. The study reiterates the earlier finding of gender

11
difference in wages in India andgenerates further evidence and explanation on sector and
industry specific gender wage gap. The results of gender wage determination show that male-
female wage difference is calculated to be 32 per cent when rural and urban areas are considered
together.The relevance of ‘human capital’ in the determination of wage is clearly indicated by
our results of increasing returns to education and of significant returns to experience for male
and female workers in the urban and rural areas. Our results from the determinants of wages
imply that education has a highly significant relationship with wages for both the sexes i.e.
wages increase with increasing educational level. As for females education is more important for
increasing their wages, it has implications for reducing gender based wage difference. On an
average, women with university degree show significantly higher returns to education than men
with similar educational qualification. Our analysis suggests that policies to enhance women’s
education will in turn help to increase their wage rate. Therefore policies to reduce gender
inequalities in access to education are paramount in this respect.Despite vast investment in
elementary educationto ensure the access to primary education, this continues to be of poor
quality with regard to labour market return. However, this is not to implicate that in India
emphasis on elementary education is not needed for the low returns of education up to primary
and middle as the elementary education is a necessary input for higher studies.Rather, we would
argue for policies which are essential for improving quality of elementary education. The
negative significant result of the final status of the worker implies that some stringent regulation
is needed in the labour market so that the casual labourers will get wages according to their
ability. There is a need for deliberate government policies and efforts to reduce the wage gap
among regular and casual workers. The significant values of the explanatory variable country-
region on wage gap indicated a regional segmentation of Indian labour market.

Our gender based wage decomposition results highlight the higher incidence of discriminatory
practices in the urban areas.In the rural areas discrimination component is larger than the
endowment component.In the urban areas the negative contribution of the endowment
component implies that endowment component would tend to favour women and if returns to
endowments are same for both the sexes, the expected wage for the female workers would be
greater than that of the male workers. Differences in returns to relative endowments, along with
other unobserved factors explain a net reduction of average wage for the female workers by 52.1

12
per cent in the urban areas.The presence of high discrimination component in both rural and
urban areas highlights that gender based wage difference is pervasive and unless the stereotype
behaviour of society changes or women’s position in the labour market undergoes radical
changes, the wage structure will continue to be imbalanced and unequal in spite of the presence
of the ‘equal remuneration act’. So any effort to reduce wage gap, must address gender
inequalities from a multi-dimensional perspective which accounts for changing perceptions and
notions regarding women’s role and contribution among different agents of the labour markets,
in addition to the enhancement of women’s employment. A deliberate government policy and
efforts are needed to reduce the wage difference which is strong indication of discrimination and
it should be aimed at empowering of the women who suffer from discrimination. Only an
inclusive growth strategy shall lead to lowering of wage differentials and removal of disparities
in living standards of people.The analysis of this studysuggests that it is highly recommended to
increase the female average years of education than to focusing on increase their average age in
labour market particiopation.

The findings of this study has some implications of policy relevance: First, investment in
education: since education has a highly significant relationship with wages, so one policy
implication is to focus on higher and technical education for women. Quality of education and
lack of technical education hampers women’s access to job market. With the new and rapidly
changing technology, one is required to link the education with labour market demands.
Investment in education is necessary to increase the competitiveness of the countries. Second,
reducing discrimination: the analysis suggests gender difference in wages exists in India andthe
magnitude of gender based wage gap is different in the rural and urban areas. A further
segregation of the wage gap highlights a larger amount of unexplained portion. In Indian context
detailed information on the type of work women are engaged in are often missing and there is
limited micro-level information on the discriminatory practices by the employers of various
types and sizes. So there must be effective implementation of the equal pay for equal work and to
revise the minimum wages. Some protection and social security is also need to be given to the
casual wage labourers. Third, increase the productivity of the agriculture: India is a
predominantly agriculture and primary sector based economy as far as labour employment is
concerned. Most of the workers live in the rural areas and gender based wage differences are

13
very common in the agricultural sector where more women are actively engaged (FAO, 2010 and
2011). Any policy with the objective of improving productivity in the agricultural sector may
lead to shift of the workforce to non-farm sectors with lesser gender-based wage difference.

[Acknowledgement: We are grateful to Neetha N,AchinChakraborty, SubhonilChowdhury,


IndraniChakraborty and ZakariaSiddique for their valuable comments, suggestions and guidance
at various stages of this work. We are responsible for all errors if any.]

References:
Banerjee, B. and Knight, J.B. (1985), ‘Caste Discrimination in the Indian Labour Market’,
Journal of Development Economics, Vol. 17, 277-307.
Becker, G.S. (1971)” ‘The Economics of Discrimination’, Chicago: University of Chicago Press.
Bhaumik, S. K. and Chakraborty, M. (2008), ‘Does move to market have an impact on earnings
gap across gender? Some evidence from India’, Applied Economics Letters, No-15:8,
601-605.
Blinder, A.S (1973), ‘Wage Discrimination: Reduced Form and Structural Estimates’, Journal of
Human Resources, VIII, 436-55.
Chandrasehar, C.P and Ghosh, J. (2011b), ‘Womens’s Work in India” Has Anything Changed?’.
Available at www.macroscan.com, (accessed on 23 November 2011).
Chandrasehar, C.P. and Ghosh, J. (2011a), ‘The Latest Employment Trends from the NSSO’.
Available at www.macroscan.com, (accessed on 23 November 2011).
Das, M.B. (2006), ‘Do Traditional Axes of Exclusion Affect Labour Market Outcomes in India’,
Social Development Paper, No. 97, South Asia Series.
Duraisamy, M. and Duraisamy, P. (1996), ‘Sex Discrimination in Indian Labour Markets’,
Feminist Economics, Vol. 2, No. 2, 41-61.
Duraisamy, M. and Duraisamy, P. (2006), ‘Regional Difference in Wage Premia and Returns to
Education by Gender in India’, Indian Journal of Labour Economics, Vol. 48, No. 2,
335-347.
Heckman, J. (1979), ‘Sample Selection Bias as a Specification Error’, Econometrica, 47, 153-63.

14
ILO (2000), ‘Modular Package on Gender, Poverty and Employment. Investing in Human
Capital: Focus on Training’, ILO, Geneva. Available at http:// www.ilo.org / public /
English / employment / skills / informal / gpe / informa / pack/ english.htm.
International Trade Union Confederation report (2008), The Global Gender Pay Gap, Belgium.
Kingdon, G. G. (1997), ‘Does the Labour Market Explain Lower Female Schooling in India’.
Available at http://www.econ.ox.ac.uk/Members/geeta.kingdon/Published
Papers/JDevstudLabourlowerfemaleschooling.pdf.
Kingdon, G.G. and Unni, J. (2001): “Education and Women’s Labour Market Outcomes in
India”. Education Economics, Vol. 9 (2), 173-195.
Mahajan, K. (2011), ‘The gender Gap in Agricultural wages in India: Spatial Variation, Caste
and Non-Farm Employment’, Paper Presented at 7th Annual Conference on Economic
Growth and Development, December, Indian Statistical Institute, New Delhi.
Mincer, J.A. (1974), ‘Schooling, Experience, and Earnings’, NBER Books, National Bureau of
Economic Research. Available at http://www.nber.org/books/minc74-1.
Oaxaca, R. (1973) ‘Male-Female Wage Differentials in Urban Labour Markets’, International
Economic Review, 14, 693-709.
Paternostro, S. and Sahn, D. E. (1998), ‘Wage Determination and Gender Discrimination in a
Transition Economy: The Case of Romania’, Cornell University, Ithaca, New York.
Peterson, T. et al, (1997), ‘Within Job Wage Discrimination and the Gender wage Gap: The case
of Norway’, European Sociological Review, Vol-13, No-2, 199-213.
Richard S. (2007), ‘Wage Determination and Gender Discrimination in Uganda’, No. 50,
Economic Policy and Research Centre.

15
Tables, Figure and Appendix

Table 1: Average daily regular and casual wage in (Rs.) and Ratio of female to male wages
(per cent) at current prices
Year Regular salaried employees Ratio Casual wage labourer Ratio

Rural Urban Rura Urba Rural Urban Rura Urba


l n l n
Male Femal Male Femal Male Femal Male Femal
e e e e
1993 58.48 34.89 78.12 62.31 0.60 0.8 23.18 15.33 32.38 18.49 0.66 0.57
-94

1999 127.3 78.61 169.7 140.3 0.62 0.83 45.48 29.39 63.25 38.22 0.65 0.60
-
2000
2004 144.9 85.53 203.3 153.2 0.59 0.75 55.03 34.94 75.1 43.88 0.63 0.58
-05

2009 249.1 155.87 377.1 308.79 0.63 0.82 101.5 68.94 131.9 76.73 0.68 0.58
-10 5 6 3 2

Source: Calculated by using various rounds of NSSO report.

16
Table 2: Descriptive statistics
Variable Rural Urban
Male(n=87083) Female(n=86395 Male(n=59830) Female(n=55354)
Wage rate 4.60(.84) 4.21(.61) 5.32(.93) 4.93(1.07)
Age 31.71 (12.02) 31.95(11.87) 32.94 (12.05) 34.04(12.30)
Age-squared 1150.06(828.65) 1161.88(846.38) 1230.26(848.16) 1309.70(884.05)
Socio-religious group (%)
Others 56.44 56.50 68.31 68.33
ST 10.78 10.84 2.76 2.82
SC 21.66 21.55 14.32 14.09
Muslim 11.12 11.10 14.61 14.75
Marital status (%)
Unmarried 26.99 22.94 33.02 27.93
Married 69.99 72.02 64.44 67.54
Others 3.03 5.05 2.54 4.53
General education (%)
Illiterate 32.56 34.78 13.80 14.69
Literate below primary 9.78 10.00 5.90 5.58
Primary & middle 33.60 32.09 27.92 27.19
Second & high second 19.81 18.92 32.42 32.26
Graduation & above 4.24 4.22 19.96 20.28
Technical education (%)
No 99.03 99.04 94.85 95.05
Yes 0.97 0.96 5.15 4.95
Principal Status (%)
Regular wage/salaried 19.37 11.78 72.09 70.11
Casual wage labour 80.63 88.22 27.91 29.89
Industry
Agriculture and allied activities 46.51 32.00 5.42 5.64
Manufacturing 5.12 1.68 16.30 3.56
Construction 9.86 1.46 7.94 0.95
Trade, hotels and restaurants 6.49 1.50 21.39 2.27
Transport, storage and 4.06 0.07 7.22 0.29
Communications
Other services 27.96 63.30 41.74 87.30
Country Region (%) Yes
Source: Estimated from NSSO 66th round unit record data.
Note: The figures within parentheses are SDs.

17
Table 3: Wage rate estimation: OLS and selectivity corrected results
Explanatory variables Rural(OLS) Rural(Selectivity Urban(OLS) Urban(Selectivity
corrected) corrected)
Male Female Male Female Male Female Male Female
Coeff. Coeff. Coeff. Coeff. Coeff. Coeff. Coeff. Coeff.
Age 0.0089*** 0.0029*** 0.01273** 0.0068** 0.0203*** 0.0155*** 0.0205*** 0.0175***
Age-squared - -
0.0002*** -0.0001 -.00002 -0.0001 0.0003*** -0.0002* -0.0001 -0.0001
Socio-religious group
Others
ST - - -
0.0668*** -0.0117 -0.02347 0.0505 0.0952*** 0.1310*** 0.0649 0.0216
SC - - -
0.0270*** -0.0166 -0.04332 -0.0107 0.1466*** 0.1915*** -0.0814*** -0.1743***
Muslim - -
0.0601*** 0.0546* 0.01474 0.0051 0.0663*** 0.1851*** -0.0195 -0.1263***
Marital status
Unmarried
Married 0.0684*** 0.0667*** 0.04501 0.0337 0.1005*** 0.1038*** 0.1061*** 0.1063***
Others 0.1180*** 0.1115*** 0.09010 0.0328 0.0904*** 0.0950*** 0.1405* 0.0503
General education
Illiterate
Literate below primary
0.0175 0.0322 0.04444 0.0381 0.0474** 0.0008 0.0291 0.1017*
Primary & middle 0.0752*** 0.0416*** 0.12264 0.0354 0.1009*** 0.1788*** 0.1636*** 0.2086***
Second & high second
0.1560*** 0.1458*** 0.23983 0.2676*** 0.3533*** 0.7565*** 0.4079*** 0.6984***
Graduation & above 0.4908*** 0.9418*** 0.47018* 0.7523*** 0.7855*** 1.2972*** 0.8019*** 1.2591***
Technical education
No
Yes 0.0962** 0.0188 0.06030 0.1397 0.1484*** 0.0932** 0.1357** 0.1032*
Principal Status
Regular wage/salaried

Casual wage labour - -0.0319 -0.40834** -0.2284** - - -0.4014*** -0.4796***


0.2870*** 0.3666*** 0.4871***
Industry
Agricultutre and allied
Manufacturing -
0.1498*** 0.0233 0.11899 0.0062 0.0944*** 0.1284*** 0.1094 -0.1471***
Construction 0.3881*** 0.3332*** 0.35400** 0.3494*** 0.3341*** 0.4320*** 0.3391*** 0.4500***
Trade, Hotels and - - -
Restaurants 0.0956*** -0.1764** -0.15409 -0.1944 0.0812*** 0.2545*** -0.1181* -0.1901***
Transport, Storage and
Communications 0.2580*** 0.2830*** 0.21863 0.3771 0.2411*** 0.3214*** 0.2556*** 0.4559***
Other Services 0.5854*** 0.2099*** 0.59165*** 0.2308** 0.4602*** -0.0996** 0.4446*** -0.0261
Country Region Yes Yes Yes Yes Yes Yes Yes Yes
Constant 4.5765*** 4.0585*** 4.3709*** 4.3734*** 4.8468*** 4.3435*** 4.9549*** 4.2858***
^ (lambda) 0.3830 -0.0561 -0.2969 -0.0052
R2 0.21 0.16 0.35 0.41
Adjusted R2 0.21 0.15 0.35 0.41
Source: Estimated from NSSO 66th round unit record data.
Note: All coefficients are significant at greater than 10per cent level except, ***signifies at 1per cent level of
significance, ** signifies at 5per cent level of significance, * signifies at 10per cent level of significance.

18
Multicollinearity corrected results & no heteroskedasticity.Here the coefficient values of country regions are not
reported. They are avoided to reduce the size of the table.

Table 4: Blinder and Oaxaca decomposition method-


Components of Decomposition Male–weighted Female-weighted
Value Value
Rural Urban Rural Urban

Amount attributable: 1.9 16.0 1.9 16.0

-due to endowments (E): 0.8 -12.6 3.4 3.9

-due to coefficients (C): 1.0 28.6 -1.5 12.0


Shift Coefficient (U): 36.9 23.5 36.9 23.5
Raw differential (R):{E+C+U} 38.8 39.5 38.8 39.5
Adjusted differential (D):{C+U} 38.0 52.1 35.5 35.6
Endowments as per cent total (E/R): 2.2 -32.0 8.7 9.9
Discrimination as per cent total (D/R): 97.8 132.0 91.3 90.1
Source: Estimated from NSSO 66th round unit record data.
Note: a) Positive number indicates advantage to male group; negative number indicates advantage to female group.
b) The results from decomposition are presented using Blinder’s (1973) original formulation of E, C, U, and D. The
endowments (E) component of the decomposition is the sum of (the coefficient vector of the regressors of the male-
wage group) times (the difference in group means between the male-wage and female-wage groups for the vector of
regressors). The coefficients (C) component of the decomposition is the sum of the (group means of the female-wage
group for the vector of regressors) times (the difference between the regression coefficients of the male-wage group
and the female-wage group). The unexplained portion of the differential (U) is the difference in constants between
the male-wage group and the female-wage group. The portion of the differential due to discrimination is C+U. The
raw (or total) differential is E+C+U.

Table 5: Summary statistics

Variable Mean and SD


Percentage of women workers
23.11 (20.44)
Percentage of rural workers
52.72 (22.44)
Male average years of education
7.31 (2.89)
Female average years of education
7.25 (3.54)
Male average age
34.63 (3.15)

19
Female average age
35.11 (3.51)
Source: Estimated from NSSO 66th round unit record data.
Note: The figures within parentheses are SDs.

20
Table 6: Wage difference estimation
Variables All workers
Coeff.
Percentage of women workers
1.5018***
Percentage of rural workers
-0.1318
Male average years of education
9.4667*
Female average years of education
-12.5089**
Male average age
11.5677**
Female average age
-10.8080***
Constant
10.1180
F-value 2.39
R2 0.37
Adjusted R2 0.22
N 31
Source: Estimated from NSSO 66th round unit record data.
Note: All coefficients are significant at greater than 10 per cent level except, ***signifies at 1 per cent level of
significance, ** signifies at 5 per cent level of significance, * signifies at 10 per cent level of significance.

Source: Employment and Unemployment various reports, NSSO (Report No. 515 and NSS KI)

21
Appendix 1: Independent Variables and Coding
Variable Coding
Age In years

Socio-religious Group Others, ST, SC, Muslim


Marital Status Unmarried, Married, Others
General Education Illiterate, Literate below primary
Primary & middle,
Second & high second, Graduation &
above
Technical Education No, Yes

Principal Status Regular wage/salaried


Casual wage labour
Industry Agriculture and allied activities,
Manufacturing, Construction
Trade, hotels and restaurants
Transport, storage and Communications
Other services

Country Region Central, North, South, East


West, North-east
Source: Estimated from NSSO 66th round unit record data.

Appendix 2:National Industrial Classification (2004)-up to 2 digits.


1) Agriculture.
2) Fishing.
3) Mining and Quarrying.
4) Manufacturing.
5) Electricity gas and water.
6) Construction.
7) Trade, hotels and restaurant.
8) Transport storage and
communication.
9) Financial Intermediation.
10) Real estate, renting,
business.
11) Public administration.
12) Education.
13) Health and social work.
14) Community social and
personal service.
15) Private households.
16) Extra territorial.
Note: Regular workers are considered for the first 15 industries (as no regular workers are involved in extra
territorial industry) and casual workers are considered for all the 16 industries.

22
Total Words: 7087

23

You might also like