Did the nationwide implementation of electronic fund management in the Indian employment guarantee scheme result in reduced expenditures? A re‐examination of the evidence

• Published date: 01 November 2024
• Author: Deepti Goel,J. V. Meenakshi,Zaeen Souza
• Date: 01 November 2024
• DOI: http://doi.org/10.1111/rode.13105

SPECIAL ISSUE ARTICLE

Did the nationwide implementation of

electronic fund management in the Indian

employment guarantee scheme result in

reduced expenditures? A re-examination of the

evidence

Deepti Goel

1

| J. V. Meenakshi

2

| Zaeen de Souza

3

1

Department of Economics, Pitzer College, Claremont, California, USA

2

Department of Social Sciences and Humanities, Indraprastha Institute for Information Technology, Delhi,

New Delhi, India

3

Fellow, ODI, Banjul, The Gambia

Correspondence

J. V. Meenakshi, Department of Social

Sciences and Humanities, Indraprastha

Institute for Information Technology,

Delhi, New Delhi, India.

Email: meena@econdse.org;

meena@iiitd.ac.in

Abstract

Digital tools are increasingly being used in welfare

programmes to reduce corruption and increase transpar-

ency. Banerjee et al. (2020, E-governance, accountability,

and leakage in public programmes: experimental evi-

dence from a financial management reform in India.

American Economic Journal: Applied Economics.

https://doi.org/10.1257/app.20180302) evaluate the effec-

tiveness of one such intervention. In a later section of

their paper, they use a Two-Way Fixed Effects (TWFE)

specification to examine the consequences of the nation-

wide scale of an electronic funds management system in

India's workfare programme, and report that it reduced

expenditures by 19%. The present paper extends its anal-

ysis by (a) exploiting the recent literature that disaggre-

gates the TWFE coefficient in the presence of staggered

treatment timing to pinpoint sources ofidentifying varia-

tion and (b) attempting to uncover heterogeneity in

treatment effects. We find that certain problematic com-

parisons have a large weight in the TWFE coefficient.

Further, an event study analysis of the six constituent

Received: 1 August 2022 Revised: 27 March 2024 Accepted: 28 March 2024

DOI: 10.1111/rode.13105

1994 © 2024 John Wiley & Sons Ltd. Rev Dev Econ. 2024;28:1994–2013.wileyonlinelibrary.com/journal/rode

and valid comparisons shows that there is no support for

parallel trends, so lag coefficients cannot be vested with

causal interpretation. Our results imply that large-scale

evaluations, because of the very diversity they encom-

pass, need to explicitly account for the factors that are

responsible for programme effectiveness.

KEYWORDS

e-governance, MGNREGS, public funds management

JEL CLASSIFICATION

D73, D78, H53, H75, I38

1|INTRODUCTION

There is an emerging literature that attempts to assess the impacts of technology and policy

interventions implemented at scale in order to address concerns of external validity and to cap-

ture any general equilibrium effects that the intervention may have induced. One strand of this

literature uses experimental methods. For example, Muralidharan and Niehaus (2017) make a

detailed case for using large-scale randomised control trials (RCTs) because these typically

involve randomising over villages or larger geographical units, rather than households, and can

therefore account for general equilibrium effects more effectively. The other strand uses non-

experimental methods, relying on multiple sources of data, including administrative and sec-

ondary sources. This literature often considers national or global level impacts; examples

include Bharadwaj et al. (2020) and Gollin et al. (2021).

In India, the last couple of decades have seen an increasing use of digital tools in

implementing welfare programmes to minimise leakages and corruption, and enhance trans-

parency. By their very design, these interventions are large-scale in nature. The experimental

literature in the Indian context has attempted to assess the effectiveness of these interventions

mainly in the form of ‘embedded experiments’wherein the research team works in close col-

laboration with the relevant government agency in designing a randomised rollout.

1

For exam-

ple, Muralidharan et al. (2023b) conducted an RCT in state of Jharkhand to determine if

biometric identification helped reduce leakages in the public (food) distribution system. Their

results indicate that although corruption fell, it came at the cost of loss of access to benefits by

legitimate beneficiaries, at least temporarily. There is also work on the use of smartcards to

effect payments in the Mahatma Gandhi National Rural Employment Guarantee Scheme

(MGNREGS), India's flagship workfare scheme that guarantees 100 days of work to any rural

household that demands it. Muralidharan et al. (2016) and Muralidharan et al. (2023a) collabo-

rated with the state government of Andhra Pradesh to randomise the rollout of the reform to

conclude that smartcards resulted in lower leakages and faster payments, and that it improved

incomes, not just from the programme itself, but from spillover effects into private

employment.

Another such embedded experiment is by Banerjee et al. (2020), who employ an RCT in the

state of Bihar to examine the impacts of the ‘Bihar Experimental Transfer System’(BETS). The

GOEL ET AL.1995