Quantile Regression for Panel Data

Chapter 3 illustrates the use of the penalized quantile regression estimator. Angrist et al. (2002) points out that the primary incentive effect of the Colombia's voucher program should be on those who are near the margin for passing on to the next grade because vouchers were renewable as long as the students maintained good academic progress. Applying quantile regression, they report that increases in test scores are not observed in the lower quantiles of the conditional educational attainment distribution. They also estimate a classical Gaussian random effects model to account for individual heterogeneity, but this approach precludes estimating effects other than the mean. To get around the problem, we employ the quantile regression panel methods. The analysis shows that the program impact is largest in the lower tail of the conditional educational attainment distribution. This was conjectured by the original authors, but could not be confirmed empirically using conventional panel data methods that focused on the conditional mean.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Economics

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2015

Author dc:creator

• Lamarche, Carlos Eduardo

Contributors dc:contributor

• Koenker, Roger W.

• Economics, Theory

Language dc:language

eng

Identifier

(MiAaPQ)AAI3242908

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/85572