This course reviews fixed effects approaches to estimation of nonlinear panel data models.
In the course, we consider both static and dynamic panel data versions of standard nonlinear models such as qualitative response models, multiplicative models, models with selection, and duration models. The importance of the panel aspect is that it allows us to investigate models with error terms that have an unrestricted individual-specific effect. The main question that will be addressed is whether it is possible to identify and estimate the “structural” parameters of these models when the number of time-periods is much smaller than the number of cross sectional units. We will consider both parametric and semiparametric versions of the models; if time permits, we will also discuss the corresponding nonparametric models.
We will emphasize practical aspects of the available methods, and the course will be taught at a level of rigor appropriate for well-trained graduate students interested in empirical microeconomics.