Estimating Endogenous Ordered Response Panel Data Models with an Application to Income Gradient in Child Health

In this paper, a control function approach is developed to deal with endogeneity issue in estimating panel data models with ordered response. This methodology is characterized by a two-step procedure, requiring only standard regression to be implemented in each step. To demonstrate the usefulness of our method, we investigate income gradient in child health based on U.S. data from the Panel Study of Income Dynamics. We find empirical evidence that income gradient rises during early childhood, followed by a decline after age 12. Ignoring endogeneity of family income and unobserved individual heterogeneity would underestimate the true gradient considerably