Dynamics of Inequality of Opportunity in Wage Earnings in India: A Machine Learning Approach.

This paper examines how inequality of opportunity (IoP) in wage earnings changed in India between 2018-19 and 2022-23. This is especially important in a country like India where more than 80 % of workforce is in the informal sector with little-to-no security of work and wages. We use machine learning models on large nationally representative datasets to capture the effects of parental background, caste, location and gender in ways the standard regression models cannot. A critical problem with this data is its reliance on co-resident households which introduces a strong selection bias. Our novelty lies in developing a new reweighting procedure to correct for this sample selection bias, making our estimates representative of the entire population. Our findings show that parental education is the primary driver of wage outcomes, although its dominance has weakened. Gender has become a more important factor, particularly in rural labour markets. While our findings suggest a drop in overall IoP measure from 19 % to 15 %, this apparent improvement, however, masks some heterogeneity. We argue that inequality is not simply declining but reconfiguring. Inherited disadvantages such as caste now operate through subtler channels, where social identity and family capital interact in new ways. This evolution of inequality demands a more nuanced policy response, one that looks beyond simple fixes to address the changing architecture of disadvantage in India.