This paper proposes a nonparametric estimator for treatment effects on censored outcomes when the treatment may be endogenous and have arbitrarily heterogeneous effects within a local average treatment effects framework. The proposed quantile treatment effects estimator is based on Kaplan-Meier estimators of the latent outcome cdf, and, relative to existing estimators, is efficient and robust to departures from the identifying assumptions. The paper derives the estimator's asymptotic distribution, illustrates its performance using Monte Carlo simulations, and applies it to a real world example.
Adjacent gas stations would be expected post similar gas prices under standard models of either Bertrand competition or collusive behavior. The posted price of an adjacent pair would not be expected depend on the distance to the next nearest gas station under Bertrand competition, however, but would under collusive behavior. We use a unique dataset derived from crowd-sourced information on gas station prices and locations to test this distinguishing implication and find evidence for significant collusion among gas stations.
Are classroom peer effects driven by low-ability students benefitting from exposure to high-ability students by "osmosis" or by cooperation and collaboration in well-functioning groups? Using data on randomly assigned student groups we find little evidence that students' performance responds to the ability of other students in their groups, but strong evidence that it does respond to the degree to which the group worked together.
I estimate the long term effect of military service on quantiles of earnings and education using the Vietnam draft lottery eligibility status as an instrument. I compare the local quantile treatment effect estimator studied by Abadie, Angrist, and Imbens (2002) to the instrumental variables quantile regression technique developed by Chernozhukov and Hansen (2008). Ordinary quantile regression shows a large negative association between service in Vietnam and earnings of white men, with the effect increasing in magnitude for the upper quantiles. Quantile treatment effects estimates show the opposite pattern, although much smaller in magnitude, with a small negative effect at the lower end of the distribution, and a small positive effect at the upper end. This suggests the ordinary quantile result is due to heterogeneous selection effects. The two methods of quantile treatment effects estimation give similar results.
We estimate the quantile treatment effects of widespread remedial education interventions—summer school and grade retention—using two alternative sets of identifying assumptions and estimators. The first requires that the intervention not systematically alter the rank ordering of individuals. The second requires that exceeding the discontinuity threshold have a monotonic effect on treatment status. While the two estimators have different interpretations and require different assumptions, they both suggest that summer school and retention have a greater impact on the higher end of the distribution, conditional on being near the threshold.
This research takes the union's primary objective to be growth of the union as an institution and the political survival of the union officers. The union's primary instrument is the wage schedule, while the constraints are the employer's investment and employment decisions. The optimal union wage schedule is derived as a function of the underlying distribution of workers' human capital and preferences. Skewness in the distribution of ability as well as employer discretion over merit pay lead to an optimal union wage schedule which depresses the return to productivity with respect to the competitive sector, suggesting a wage-compressing effect of unionization.