Job Market Paper
Judicial quality and recidivism for high-profile criminals: structural evidence from US federal courts

Policy concern regarding recidivism is considered to be one of the recent debates in the literature. This paper aims to look at the recidivism of high-profile criminals as a strategic action related to the expected composition of the federal bench and, therefore, punishment. Empirical findings suggest that a 1 standard deviation increase in the Republican-associated share of justices on the bench leads to an increase in recidivism for fraud by 3.8 percentage points. The proposed mechanism is the strategic anticipation of lower punishment for white-collar crimes, estimated in a single-agent dynamic discrete choice (DDC) framework. Counterfactual experiments suggest that regulating equal composition of the benches by party nominations in the federal courts leads to a decrease in recidivism by 4.04 percentage points.

Working papers:

Career concerns and judicial biases in labor litigation: evidence from US federal courts

Party polarization in the U.S. judicial system has become a salient institutional concern. Using a novel dataset of published federal district court opinions and text--mined case outcomes, I document that conservative judges exhibit a 3.5% lower probability of ruling pro--Employee in labor disputes against corporate parties, and that this ideological gap is dynamic over electoral cycles. I estimate a structural life--cycle model in which judges choose rulings and retirement decision while facing promotion incentives. The model matches the observed election--year rise in type--violating rulings and the cross--party patterns in harshness. Structurally, semi--retirement carries utility roughly eight times that of full retirement, and Lenient types bear the largest punishment cost for deviating toward harsher rulings, consistent with the cross--type gradients in the data. Counterfactuals isolating the mechanism show that removing election--year violation incentives reduces election--year mismatch by about 8--12 percentage points across types—effectively collapsing the election/non--election wedge—whereas shifting semi--retirement eligibility (Rule--of--70/90) has only second--order effects. The results highlight career concerns tied to the electoral calendar as a key driver of preference--and suggest that separating promotions from elections would meaningfully cut bias, even if retirement policy stays the same.

Work in progress:

Estimating the effect of import sanctions on Russian metals using LASSO