Seminar
Research seminar: Estimating the Dynamic Effects of a Job Training Program with Multiple Alternatives
- Speaker
- Kai Liu, University of Cambridge
- Date
- 28 August 2018
- When
- 11:45 - 12:45
- Where
- Auditoriet, SSB, Akersveien 26
Content
Kai Liu, University of Cambridge: https://sites.google.com/site/richardkailiu
“Estimating the Dynamic Effects of a Job Training Program with Multiple Alternatives “
Abstract:
This paper aims at estimating the causal effect of a job training program for disadvantaged youths on their subsequent earnings. Participants are randomized in treatment and control groups, but are allowed to choose when to leave the program and to participate (alternative) training when not enrolled in the program. We consider a multi-stage decision setting, where individuals sequentially select which training to obtain at every stage. The standard Wald estimator using initial random assignment as instrument identifies a weighted average of the effects of the treatment for the subgroups of individuals selecting different duration and different trainings outside the program. We develop a selection model which allows to separately estimate the effect of the treatment for the different subgroups. We use the estimated model to investigate dynamic complementarity/substitutability between different training programs. We show that considering the dynamic interaction between different trainings in this setting is relevant to study the heterogeneous effect of the program and to conduct a complete cost-benefit analysis.