This paper estimates the employment effects of the Swedish trainee replacement schemes (an active labour market program that was in operation during the 1990s). The empirical analysis exploits a large and rich administrative data set, and we control for observed and unobserved selection bias by...