Caste, corruption and political competition in India

Voters in India are often perceived as being biased in favor of parties that claim to represent their caste. We incorporate this caste bias into voter preferences and examine its influence on the distributive policies and corruption practices of the two major political parties in the North Indian st...

Full description

Saved in:
Bibliographic Details
Published inResearch in economics Vol. 69; no. 3; pp. 336 - 352
Main Authors Acharya, Avidit, Roemer, John E., Somanathan, Rohini
Format Journal Article
LanguageEnglish
Published Elsevier Ltd 01.09.2015
Subjects
Online AccessGet full text

Cover

Loading…
More Information
Summary:Voters in India are often perceived as being biased in favor of parties that claim to represent their caste. We incorporate this caste bias into voter preferences and examine its influence on the distributive policies and corruption practices of the two major political parties in the North Indian state of Uttar Pradesh (U.P.). We begin with a simple two-party, two-caste model to show that caste bias causes political parties to diverge in their policy platforms and has ambiguous effects on corruption. We then develop the model to make it correspond more closely to political reality by incorporating class-based redistributive policies. We use survey data from U.P. that we collected in 2008–2009 to calibrate voter preferences and other model parameters. We then numerically solve for the model׳s equilibria, and conduct a counterfactual analysis to estimate policies in the absence of caste bias. Our model predicts that the Bahujan Samaj Party (BSP), which was in power at the time of our survey, would be significantly less corrupt in a world without caste-based preferences. •We model an election in Uttar Pradesh, India.•The competition is between four parties over a four dimensional policy platform.•We investigate the effect of “caste bias” that voters have for particular parties.•If voters support a party because of caste bias, that party will be more corrupt.
ISSN:1090-9443
1090-9451
DOI:10.1016/j.rie.2015.02.001