The e�ects of a training program on employment and wages are evaluated, using data
from a randomized study, the National Job Corps Study, and using the principal stratification
approach to address simultaneously the complications of noncompliance, truncation of wages
by nonemployment, and unintended missing outcomes, where the first two complications are
of substantive interest, whereas the third is a nuisance. We conduct a likelihood-based analysis
using finite mixture models estimated by the EM algorithm. Monotonicity of compliance
holds by design; we maintain an exclusion restriction assumption for the e�ect of assignment
on employment and wages. We provide estimates under the Missing at Random assumption,
and assess the robustness of our results to deviations from it. The plausibility of meaningful
restrictions is investigated by means of scaled log-likelihood ratio statistics. The substantive
conclusions are as follows. For compliers, the e�ect on employment is negative in the short
term, but becomes positive in the long term, with these e�ects being negligible and not strongly
supported by the data. There is, however, a nonnegligible subgroup of complying participants
for whom the program is detrimental on employment. For the subgroup of those compliers
who would be employed whether trained or not, the e�ect of training on wages is small but
positive at all weeks. Previous analyses on this data set that did not address all complications
in a principled manner led to di�erent conclusions about Job Corps.
Presented by:
Fabrizia Mealli (University of Florence)
Date & time:
September 27, 2010 3:00 am
External seminars home