Religion in the labor market: evidence from India

ABSTRACT

In this paper, we analyse the education and wage gaps between Hindus and Muslims over the period 1983 to 2011–2012 in India. We find that Muslims are worse off than Hindus in terms of education and this disadvantage has increased over time. Our analysis shows that the wage gap is more pronounced at the higher end of the wage distribution which we interpret as a glass ceiling effect. We find that the wage gap has increased over time and the increase is more pronounced at the upper quantiles. Our decomposition analysis shows that difference in education has a significant contribution to the Hindu-Muslim wage gap. We find that the Hindu-Muslim gaps are more pronounced in urban areas relative to rural areas. Comparing Muslims with different Hindu castes, we find that while the disadvantaged castes have improved their condition relative to Muslims over time, the condition of Muslims relative to the advantaged castes has worsened. Dividing the sample into younger (aged 16–35) and older (36–65) cohorts, we find that the Hindu-Muslim gap in education is more acute in the younger cohort while the Hindu-Muslim wage gap is more pronounced for the older cohort.

KEYWORDS:

• Wage differential
• education
• Muslims
• inequality
• decomposition

JEL CLASSIFICATION:

• J15
• J31
• I24
• O15
• Z12

Acknowledgment

We are grateful to the editor, two anonymous referees, Siwan Anderson, Diane Coffey, Patrick Francois, Ashok Kotwal, Amartya Lahiri, Kevin Milligan, Marit Rehavi, Dean Spears and the seminar participants at the University of British Columbia and the Canadian Economics Association conference for their valuable comments and suggestions.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

¹ We focus on males in our analysis because of the low labour force participation rate for both Hindu and Muslim women, the significant differences between Hindu and Muslim female labor force participation rate throughout our sample period and the differential evolution of labor force participation rate of Hindu and Muslim women in the latter period. Appendix A compares the evolution of labor force participation rate across Hindus and Muslims.

² Scheduled Castes are a constitutionally recognized group of marginalized castes. Scheduled Tribes are constitutionally recognized tribal groups in India. Both these groups are historically economically disadvantaged and are beneficiaries of affirmative action (Deshpande, 2011).

³ See Blau et al. (2017), Lang et al. (2012), Fryer (2011), Altonji et al. (1999) for excellent reviews of the literature on gender and racial gaps in outcomes.

⁴ The evidence from the economics literature on discrimination against Muslims in the labour market is mixed. Thorat et al. (2007), using a correspondence study, in the context of private sector enterprises in different cities of India, find evidence of discrimination against Muslims. On the other hand, Banerjee et al. (2009), which undertakes a correspondence study in the context of the labor market in Delhi of two sectors, software and call-centres, find no evidence of discrimination against Muslims in callback rates.

⁵ The earliest available NSS employment-unemployment round micro-data is the 38th round (1983). The last available NSS employment-unemployment round micro-data is the 68th round (2011–12).

⁶ We do not use the data from the 43rd round (1987–88) for our wage analysis because of data issues. Firstly, there is a drastic fall in the number of people who report working for wages (Hnatkovska et al., 2012). Secondly, the wage estimates from the data are not comparable to even the official estimates published by NSS itself (Himanshu et al., 2013; Lanjouw et al., 2009).

⁷ The definitions of rural and urban areas in NSS are based on the most recently available Indian population census (NSSO, 2000). The criteria for being classified as an urban region have varied little over time (ORGI, 2018). All areas which are not defined as urban are considered rural areas..

⁸ The consumer price index (CPI) data, which is used to deflate wages, is available consistently for the 15 states used in our analysis. The base year for the calculation of CPI changed in 1995, and the pre-1995 series does not have data for the states of Haryana and Himachal Pradesh. The data is also not consistently available for the northeastern states (except Assam), New Delhi, Goa and the union territories. Three new states of Chhattisgarh, Jharkhand and Uttarakhand were created in India in 2000. We merge these states with their parent states since CPI data is not available separately for these states, either in the pre-1995 series or in the post-1995 series. Merging the newly created states with their parent states also ensures a consistent definition of states across our sample period, 1983 to 2011–12. See Appendix B for more details regarding CPI data.

⁹ In our sample, BJP has 94% of Hindu party seats.

¹⁰ This conversion is done by NSS.

¹¹ Tables A1a and A1b of the Appendix report the summary statistics of the female sample.

¹² We have included kernel densities of wages for Hindus and Muslims in Figure A2 and a corresponding discussion in Appendix C.

¹³ We discuss the details of the decomposition methodology, covariates used in the regressions underlying the decomposition and the quantile decomposition technique in Appendix D, Appendix E and Appendix F, respectively.

¹⁴ Appendix Table A2 shows the decomposition results for the four other intermediate NSS rounds for the overall male sample. These results are consistent with the results in Table 3.

¹⁵ For the all-male sample, we present the contribution of the rest of the covariates in Tables A3a and A3b and discuss the results briefly in Appendix G.

¹⁶ We report the estimated coefficients of the education dummies from the regressions used in decomposition in Tables A4-A10.

¹⁷ However, since the total gap has also increased, the contribution of education as a percentage of the total gap has fallen.