The estimation of the causal effect of an endogenous treatment based on an instrumental variable (IV) is often complicated by the non-observability of the outcome of interest due to attrition, sample selection, or survey non-response. To tackle the latter problem, the latent ignorability (LI)...