Paper short abstract:

Do criminally accused Members of the Legislative Assembly harm local benefit delivery in India? Using a regression discontinuity design, I estimate the causal impact of electing a "criminal" politician on the distribution of NREGS projects. Accused politicians complete fewer NREGS welfare projects.

Paper long abstract:

Despite intense political competition, candidates facing criminal charges are routinely elected across India at higher rates than clean candidates. Once elected, how do "criminal" politicians perform in office? Using a mixed methods approach, I argue that criminal candidates' access to money and networks helps them win campaigns, even though they may underperform in office. To test if criminally accused politicians harm (or improve) benefit delivery at the local level, I construct a novel dataset detailing the criminal histories, wealth and electoral results of all state legislative candidates in India from 2003-2016. I combine the candidate dataset with original data on the geo-locations of over 20 million public works projects from India's largest anti-poverty scheme, the National Rural Employment Guarantee (NREGS). Using a regression discontinuity design, I estimate the causal impact of electing a criminally accused politician on the distribution of NREGS projects. Preliminary results indicate that constituencies that elect a criminally accused politician complete fewer NREGS projects.