Nonlinear Panel Data Models with Dynamic Heterogeneity.

We introduce nonlinear panel data models for individual risl management and control. Our approach extends traditional models by allowing for nonlinearities in input variables and unobserved individual heterogeneity with possible temporal dependence. This permits us to develop nonlineat models for the assessment of individual risks in the presence of morla hazard and adverse selection. In practice, these phenomena are confronted by banks and insurance companies in their basic activities of fund supply amd loss coverage.