Impact Evaluation of National Rural Employment Guarantee Scheme in India

This study aims to thoroughly evaluate the impacts of India’s National Rural Employment Guarantee Scheme (NREGS), the largest public works project in the world. NREGS was started in 2006, following the 2005 Act that provides each rural household with a legal right to be employed up to 100 days per year at state-level minimum wage rate. NREGS has the following characteristics. First, it is demand-driven. Second, all rural households are eligible for NREGS and self-selection is the major mechanism to target the poor. Third, NREGS offers equal wage rates to women for the same work as men and makes payment directly to the individual workers. Fourth, NREGS projects are selected at the local level.

View more We aim to explore the following research questions:

1. Targeting and Implementation

How well does NREGS target its intended group? Are the poor (non-poor) self-selecting into (out of) the program, as intended?
2. Partial equilibrium effects
How does NREGS affect the labor allocation, migration, and welfare of its participants? Does participation in NREGS work “crowd out” participants’ labor supply for other purpose such as working on their own or neighbors’ fields?

3. General equilibrium effects and income distributional effects
What are the general equilibrium effects? And what are the distributional effects of NREGS over different types of households such as land owners, laborers, etc.?

4. Gender-specific effects
What are the gender-specific effects? We aim to provide evidence on the magnitude of economic and non-economic benefits derived by females in different income groups from NREGS and the mechanisms that may underlie such effects.

We use two existing data sets: a three-round (2004, 2006, and 2008) panel data set from 4,800 households in Andhra Pradesh and the National Sample Survey data which are collected annually and cover all districts in India.

To examine partial equilibrium effects, we compare NREGS participants with non-participants. Methodologically, we use propensity score matching, double (or triple) differences, and instrumental variable methods. To examine general equilibrium effects (or spillover effects), we explore the phase-in structure of NREGS implementation to compare non-participants in the treatment areas with non-participants in the control areas, using propensity score matching and double (or triple) differences.

The panel data will allow us to control for observed confounding factors, and to examine heterogeneous impacts due to initial conditions. Propensity score matching and instrumental variable methods are used to deal with the selection bias due to non-random program placement and self-selection of program participation.

Research Findings

From the top to bottom of India's MGNREGA in one week

Megan Sheahan's piece in Economics that Really Matters 



नरेगा से लोगों को, महिलाओं को ताकत मिली है : The fight against bonded labour

In this video, Gyarsi bai from Baran district in Rajasthan (with Nikhil Dey of Mazdoor Kisan Shakti Sangthan) talks about the role played by the NREGA in the Sahariya community's struggle against bonded labour and poverty. 





Politics and MNREGA: A limited Link in Andhra Pradesh

This post first appeared as a column in Ideas for India on November 19, 2014:

The Ministry of Rural Development claims that MNREGA needs to be changed in order to reduce politics and corruption in the scheme. One of the studies cited by the Ministry is an analysis of the extent of political manipulation of MNREGA funds in Andhra Pradesh. In this column, the authors of the study assert that while politics may influence programme expenditure in some places and to a small extent, this is not universally true and does not undermine the effective targeting and good work of the scheme at large.

In India, politicians are debating if and how the Mahatma Gandhi National Rural Employment Guarantee Act (MNREGA) — the “right to work” legislation which grants work on small-scale and government-funded public works projects — will evolve under the direction of the new Prime Minster. Some of the currently proposed changes include limiting the scheme to only the 200 poorest districts and increasing spending on materials at the expense of wages. These proposals have been criticised by a group of 28 economists who wrote an open letter to the Prime Minister detailing their concerns about the potential impact of such reforms. The central government responded in turn that the scheme needed to be changed in order “to reduce politics and corruption”, setting off a major public debate on the topic. 



Investigating extent of political manipulation of MNREGA funds in Andhra Pradesh

Our new working paper and shorter policy note investigate the presence of political manipulation of funds under MNREGA in Andhra Pradesh (AP), a state in southern India. We were motivated to investigate the extent of preferential resource allocation — even before the government’s claims — for two major reasons: MNREGA is designed to be a self-targeting and demand-driven programme where labour is aggregated and public works are selected at the local level before final approval at higher levels of government is sought; and MNREGA also put in place a suite of accountability and transparency mechanisms, including but not limited to publicly available data and social audits, with the objective of restricting any possible political manipulation of funds. Our goal was to evaluate the extent to which these unique features of MNREGA have eliminated avenues for taking advantage of the programme for political gain.


Using a range of publicly available data sources detailed in the paper, we established the relationship between project spending at the mandal (sub-district) level and two major components of clientelism — ‘vote buying’ prior to elections and political patronage following a vote — versus the stated target of the programme, addressing human needs. The timely occurrence of a national and state-level election in 2009, several years into the implementation of MNREGA, allowed us the opportunity to test for the incidence of vote buying by the national and state-level incumbent coalition, the United Progressive Alliance (UPA), leading up to the election. Then, because we observed several years of MNREGA implementation following the 2009 election won by the incumbent UPA coalition, we also tested for the existence of patronage effects.



MNREGA expenditures largely well-aligned with needs of sub-districts

Our findings suggest that vote buying was not apparent before the 2009 election. However, in the post-2009 election years, we do find evidence (although the effect is small in magnitude) that the distribution of funds was partially politically motivated, either as a patronage effect following the 2009 election or as a vote buying effect leading up to the by-election in 2012 or full election in 2014. At the same time, we also observe that expenditures were well-aligned with the needs of the mandal in terms of not only characteristics of the population, land, and infrastructure before the start of the programme, but also the changing labour market dynamics across years and agricultural seasons. Indeed, even in the post-election period where clientelism seems to exist, we find that the objective needs of the mandals explain 63% of the variation in MNREGA expenditures, while all of the political factors taken together explain only 2.5%. The fact that clientelism does not appear to have a major influence on spending levels means that the self-targeting, transparency, and accountability mechanisms, including widespread information disclosure and social audits integrated into MNREGA, appear to be working and reducing the potential for large-scale clientelism to take hold in AP.


In the context of claims made by the new Indian government about the political nature of MNREGA implementation — claims which included a selective and misleading portrayal of our findings — we reiterate that our study shows that while politics may influence programme expenditure in some places and to a small extent, this is not universally true and does not undermine the effective targeting and good work of the scheme at large. Extensive research, including an article by two members of our research team, documents the variation in implementation records across states; pointing to some states (like AP) where MNREGA works well to help the rural poor, and others where it fails. Overhauling a programme — indeed, a right — for all, based on the experiences of a few, may harm the rural livelihoods that have been fundamentally and meaningfully enhanced by the employment guarantee. A more thoughtful policy response may be to bolster the transparency mechanisms already in place and target reforms to those areas where MNREGA has been excessively politically manipulated as demonstrated by the abundant data available on MNREGA participation.


Further Reading