We provide an econometric framework to identify causal treatment effects in situations where multiple outcomes are available and all the outcomes depend on the same endogenous regressor. The finite sample performance of alternative causal estimators – with and without instrumental variable – in terms of the percentage bias, efficiency, and coverage probability are compared using Monte Carlo simulations. The simulations provide suggestive evidence on the complementarity of instrumental variable (IV) and latent factor methods and how this complementarity depends on the number of outcome variables and the degree of contamination in the IV. We apply the causal inference methods to assess the impact of mental illness on work absenteeism and disability, using the National Comorbidity Survey Replication data from the US.