Abstract: Selective school admissions give rise to a Regression Discontinuity (RD) design that nonparametrically identifies causal effects for marginal applicants. Without stronger assumptions nothing can be said about causal effects for inframarginal applicants. Estimates of causal effects for inframarginal applicants are valuable for many policy questions, such as affirmative action, that substantially alter admissions cutoffs. This paper develops a latent factor-based approach to RD extrapolation that is then used to estimate effects of Boston exam schools away from admissions cutoffs. Achievement gains from Boston exam schools are larger for applicants with lower English and Math abilities. I also use the model to predict the effects of introducing either minority or socioeconomic preferences in exam school admissions. Affirmative action has modest average effects on achievement, while increasing the achievement of the applicants who gain access to exam schools as a result.
Wanna Get Away? RD Identification Away from the Cutoff (with Joshua Angrist, IZA Discussion Paper 7429)
Abstract: In the canonical regression discontinuity (RD) design for applicants who face an award or admissions cutoff, causal effects are nonparametrically identified for those near the cutoff. The effect of treatment on inframarginal applicants is also of interest, but identification of such effects requires stronger assumptions than those required for identification at the cutoff. This paper discusses RD identification away from the cutoff. Our identification strategy exploits the availability of dependent variable predictors other than the running variable. Conditional on these predictors, the running variable is assumed to be ignorable. This identification strategy is illustrated with data on applicants to Boston exam schools. Functional-form-based extrapolation generates unsatisfying results in this context, either noisy or not very robust. By contrast, identification based on RD-specific conditional independence assumptions produces reasonably precise and surprisingly robust estimates of the effects of exam school attendance on inframarginal applicants. These estimates suggest that the causal effects of exam school attendance for 9th grade applicants with running variable values well away from admissions cutoffs differ little from those for applicants with values that put them on the margin of acceptance. An extension to fuzzy designs is shown to identify causal effects for compliers away from the cutoff.
Changes in Job Stability: Evidence from Lifetime Job Histories (with Roope Uusitalo, IZA Discussion Paper 4721)
Abstract: We use lifetime job histories from the pension records to evaluate changes in job stability in Finland between 1963 and 2004. We specify a duration model and estimate the effects of elapsed duration, age, and calendar time on the hazard of job ending using individual-level panel data spanning over four decades. We find that this hazard increased during the recession years in the early 1990s but has now returned to the level that prevailed in the 1970s. We also demonstrate that the fluctuations in the hazard rate together with the changes in labor market entry rates have complicated dynamic effects on the tenure distribution, and that analysing the changes in job stability based on the elapsed duration of ongoing jobs may be quite misleading.
Research in Progress
(Un)Ordered Treatments in the LATE Framework: Getting Beyond Average Causal Responses
Abstract: In this paper I generalize the covariate-based approach to external validity of instrumental variable estimates by Angrist and Fernandez-Val (2010) to settings with ordered and unordered treatments. I assume that treatment effects are mean independent of potential treatment status conditional on covariates. Given sufficient variation in treatment status of individuals, this allows one to identify the full profile of incremental average treatment effects. These effects can then be used to extrapolate the effects for different populations characterized by their covariate distribution.
Adaptive Bandwidth Choice Estimation in the Regression Discontinuity Design
Abstract: In this paper I propose an adaptive bandwidth choice algorithm for the regression discontinuity design when the estimation is based on local polynomial regression. The algorithm can be easily adapted to settings with various choices regarding the order of polynomial. In addition, the algorithm automatically takes into account different choices for the kernel function, the inclusion of additional covariates, and alternative assumptions on the variance-covariance structure of the error terms. I show that the algorithm produces a consistent estimator of the asymptotically optimal bandwidth and that the resulting regression discontinuity estimator satisfies the asymptotic optimality criterion of Li (1987). Finally, I provide Monte Carlo evidence suggesting that the proposed algorithm also performs well in finite samples.
Towards Equality of Educational Opportunity: Distributional Consequences of the Finnish Comprehensive School Reform
Abstract: During the 1970s Finland moved from a highly selective school system characterized by early assignment of students into academic and non-academic tracks to a uniform comprehensive school system. In this paper I utilize regional variation in the timing of reform implementation to identify causal effects of comprehensive schooling on the distribution of cognitive skills and earnings. I find that comprehensive schooling had no impact on cognitive skills as measured by Army test scores. However, the reform decreased earnings in the top quartile of the distribution while keeping the bottom half of the distribution unchanged. I also find substantial heterogeneity in the earnings effects by parental background: most of the estimated effects come from individuals with educated or high-earning parents. I find no evidence of heterogeneity in cognitive skills effects along this dimension.