政治代表与司法结果：来自印度的证据

Political Representation and Judicial Outcomes: Evidencefrom India 27th May, 2025 Abstract The economic impact of electing members of minority groups into positions of political poweris well-established. However, the impact of political representation on broader civil rights andliberties, and particularly access to justice, remains unexplored. This paper employs a close-election regression discontinuity design to explore whether female political representation canexplain judicial outcomes in the Indian context, focusing on crimes against women.Despitepoliticians having no formal influence over the judiciary, I find that the election of a femalepolitician generates a large and statistically significant increase in the likelihood of convictionfor crimes against women, relative to the election of a male politician.I do not find similardifferences in the likelihood of conviction for gender-neutral crimes, suggesting that femalepoliticians shape judicial outcomes within issue areas that align with gendered spheres ofinfluence and interest.Additional analysis—on whether female politicians cater to genderedpreferences in public goods and whether the effect of female representation on the likelihoodof conviction varies with local gender bias—points to two potentially important mechanisms.These include a policy channel, whereby female politicians actively attempt to act in women’sinterests, and an exposure channel, whereby observing female representatives positively informscitizens’ views on women’s competencies. This study emphasises the importance of politicalrepresentation in expanding vulnerable groups’ access to justice. 1Introduction A well-established literature has highlighted the economic impact of electing members ofsocially disadvantaged groups into positions of political power, especially for public goodsprovision (Pande, 2003; Chattopadhyay and Duflo, 2004; Beaman et al., 2007, 2012). However,the impact of political representation on access to justice remains unexplored.While it isoften argued that vulnerable groups, particularly women, face major barriers to being heardwithin the criminal justice system (Jassal, 2024; Sharma et al., 2025; Dutt, 2018; Kadyan &Unnithan, 2019), no study has sought to test whether political representation can alleviate theseconstraints. This study seeks to address a gap in the literature by estimating the impact of femalepolitical representation on judicial case outcomes, focusing on crimes against women in theIndian context. I employ a close-election regression discontinuity (RD) design to isolate plausibly exogenousvariation in Indian electoral constituencies’ exposure to male or female politicians. The quasi-experimental RD design circumvents several of the empirical challenges associated with causalestimation of the effect of political representation. As numerous studies have previously argued,narrowly determined election outcomes may plausibly be attributed to idiosyncratic factorsrather than systematic constituency characteristics correlated with judicial outcomes, such thatthe sample conceivably simulates random assignment to male or female political representation(Lee, 2008; Lee and Lemieux, 2010; Dell, 2015; Baskaran et al., 2018; Asher and Novosad,2017).Balance tests conducted on this study’s dataset verify that demographic, economic,and geographic characteristics vary smoothly across the win-loss threshold, indicating that pre-determined confounding variables are not driving results. Furthermore, sorting tests confirmthat neither male nor female candidates are able to precisely manipulate close election outcomes.Data derived from detailed case records allows me to address the temporal misalignment problemarising from judicial delays and political turnover by linking each case to the politician in officeduring the end of the trial period. Baseline RD estimates suggest that constituencies with exposure to female political repre-sentatives at the state constituency level exhibit a 5% higher conviction likelihood for casespertaining to crimes against women, relative to constituencies with exposure to male represen-tatives.I record considerable heterogeneity by crime type, with treatment effects of female (relative to male) politicians estimated at 18% for rape cases and 4% for non-rape sexual harass-ment cases. I observe these effects despite the fact that the Indian constitution emphasises theseparation of state powers, and elected politicians therefore have no formal influence over anyaspect of judicial proceedings. Results are robust to a range of specifications, including alter-native polynomial forms and bandwidths. For cases related to gender-neutral crimes, includingcrimes against property and and crimes against religion, I do not identify significantly higherconviction likelihoods in constituencies with female relative to male representation; indeed, forsome crimes I estimate a significantly negative treatment effect. This suggests that female politi