Economic liberalization and urban–rural inequality in India: a quantile regression analysis

Abstract

We examine India’s urban–rural inequality in welfare in 1993–1994 and 2004, a period which coincides with the country’s economic liberalization reforms and rapid economic growth. Using real monthly per capita household consumption expenditure as our measure of welfare, we estimate quantile regressions to analyze the urban–rural welfare gap across the entire welfare distribution. While the urban–rural welfare gap was fairly convex across the welfare distribution in 1993–1994, it became more concave in 2004, with the gap narrowing for the lowest and highest quintiles and widening for the middle three quintiles. The urban–rural gap in returns to all levels of education widened substantially for the bottom four quintiles but became increasingly negative for the top quintile. Applying the Machado and Mata (J Appl Econom 20:445–465, 2005) decomposition technique to decompose the urban–rural welfare gap at each percentile, we find that for the bottom 40% of the distribution, differences in the distribution of covariates became less important while differences in the distribution of returns to covariates became more important in explaining the gap. The opposite was true for the top 40% of the distribution. Our analysis suggests that while the rural poor appear to be catching up with their urban counterparts in terms of labor market characteristics, ten years of economic reforms have intensified the urban–rural gap in returns to these characteristics. On the other hand, the rural rich lag even further behind the urban rich with respect to their labor market characteristics even though the urban–rural gap in the returns to these characteristics has diminished during the reform period. Future efforts to generate urban–rural equality may require policies that seek to equalize returns to labor market characteristics between the two sectors at the lower half of the distribution and improve rural labor market characteristics at the top half of the distribution.