Specification and simulated likelihood estimation of a non‐normal treatment‐outcome model with selection: Application to health care utilization

Abstract

We develop a specification and estimation framework for a class of nonlinear, non‐normal microeconometric models of treatment and outcome with selection. A latent factor structure is used to accommodate selection into treatment and a simulated likelihood method is used for estimation. The methodology is applied to examine the causal effect of managed care, a multinomial discrete choice process, on the utilization of health care services, measured as binary indicators and counts. The results indicate that there are significant unobserved self‐selection effects and these effects substantially change the estimated effects of insurance on utilization.