Date of Award

May 2019

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Economics

Committee Member

Curtis Simon

Committee Member

Chungsang Lam

Committee Member

Devon Gorry

Committee Member

Aspen Gorry

Abstract

This dissertation is comprised of two essays, examining trends in wages, education, and labor supply of married women in India. The first chapter examines the key question\textemdash why Indian married women's labor force participation(LFP) declined from 35% in 1983 to 24% in 2011, during a period accompanied by high economic growth (between 6% and 7% per year), increasing educational attainment, and falling fertility. Importantly, real wages of both men and women grew over the same period, which economic theory suggests should have opposite effects on the incentive of married women to participate in market work. As straightforward as this proposition may appear, direct evidence is surprisingly scarce. Using data from India's National Sample Survey (NSS) Organization's Employment and Unemployment surveys, I construct a unique sample of 681,204 matched husband-wife pairs. I employ a household model of labor supply and examine whether, with rising wages, income effects depressed female LFP, and to what extent did an own-wage positive substitution effect offset this decline. As a simpler explanation and to showcase my data, I conduct a shift-share dual-decomposition of trends in wives' LFP with respect to trends in the education of husbands and wives. I find that changes in the educational composition of neither husbands nor wives explain the observed decline in wives LFP. Hence, an explanation based on the evolution of wages and elasticities is more appropriate. Two-stage least squares estimates of the model of household labor supply reveal positive wives' own-wage elasticities, rising from 0.25 in 1983 to 0.67 in 2011, and negative cross-wage elasticities, "rising" from -0.54 in 1983 to -0.75 in 2011. Estimating the model for the period as a whole, trends in wives' wages imply an increase in women's LFP of 8.6 percentage points, and trends in husbands' wages imply a decrease of 20.8 percentage points, for an overall decline of 12.2 percentage points. In other words, though wives' rising own-wage elasticities and wages should have implied an increase in their LFP, the wives' rising cross-wage elasticities coupled with their husbands' rising wages account for the same magnitude as the decline in the wives' LFP.

The second essay within this dissertation examines how the change in observed wages and gender wage gap reflects the changing composition of the female workforce in India. Between 1983 and 2011, the average real wage increased by 149% and 5\% for Indian married women and men, respectively. At the same time, there was rapid economic growth (between 6% and 7% per year) as well as rising educational attainment in India. I use my unique sample of matched husband-wife pairs to study the change in standardized selection bias, which, by definition changes, because wives' behavior has changed in terms of the relationship between standardized wages and employment status. I employ the Gronou-Heckman-Roy model to show how growing inequality within gender can increase measured wages via a changing selection bias\textemdash even if the aggregate female employment rate is held constant- by changing the relative importance of market and nonmarket factors for explaining which women are employed. The empirical analysis shows how the Heckman two-step estimator, when applied to Indian data, suggests that selection into the wives' wage employment is consistently negative, however, is becoming less negative from -0.379 log points in 1983 to -0.181 log points in 2011. The selection bias being negative implies that the less-skilled wives are a part of the wage-earning female workforce, and that the average market wage of the wage-earning wife is less than the average (potential) market wage of the non-employed wives. In addition, I find that the gender wage gap (using two-step estimator) shows a moderate increase (in absolute value) from a -0.455 to a -0.503 log wage point differential over the study period.

Share

COinS