Work

Bayesian treatment effects models with variable selection for panel outcomes with an application to earnings effects of maternity leave