Association for Public Policy Analysis & Management

Public-works programs are increasingly becoming popular in developing countries as a means of combating long-term poverty. These programs are being designed as a means of creating employment and income trajectories that will pull people permanently out of poverty. The Mahatma Gandhi National Rural Employment Guarantee Act (MGNREGA) in India is the world’s largest public works program. The MGNREGA came into effect in April 2006 and was implemented in districts across the country in three phases, with nationwide coverage attained in the third phase in April 2008. The Act mandates local government bodies to provide 100 days of employment to each rural household per year in the nature of manual unskilled labor.

Existing literature on the MGNREGA has shown that the scheme has significantly increased employment and wages for rural households. This impact has been larger for women’s employment and wages. The scheme has also been shown to function more as a safety net than as a general employment-generating program by providing an employment alternative in districts that face rainfall shocks. Given the findings that the MGNREGA works as an employment and income safety-net, the employment scheme has the potential to positively impact dimensions of well-being that are related to poverty and income levels. Higher income or less variability in income across the year can enable families to make better investments in infant and maternal health. Increases in women’s employment can also impact infant and maternal health through channels such as shifts in the intra-family bargaining power distribution and women’s time away from home. Moreover, if the program improves children’s health outcomes, the program has the potential to weaken the cycle of poverty that persists across generations.

I analyze the impact of the MGNREGA on infant and maternal health using a regression discontinuity framework based on the algorithm that was used by the Government to choose districts for treatment in each phase. According to the algorithm each state was given a fixed number of districts to treat, which was determined by the share of the country’s poor living in the state. Given this number, each state ranked its districts based on a development index (a sum of three sub-indices measuring agricultural output, agricultural wages and the proportion of the population that belongs to the Scheduled Tribe/Scheduled Caste group) and treated the lowest ranking districts till it exhausted the number that could be treated. This creates a discontinuity in treatment at the rank value that equals the number of districts that are to be treated in a given state. I include a second difference in time to control for any differences in districts across the discontinuity points.

I find that the MGNREGA reduced under-5 child and maternal mortality by 0.9 and 0.2 percentage points on a sample base mean of 9.9% and 0.82% respectively.  Breaking child mortality further into neonatal and infant mortality I find that the MGNREGA reduced infant mortality by 0.7 percentage points and neonatal mortality by 0.4 percentage points on a sample base mean of 5.2% and 3.7% respectively.