In this paper, we develop a panel data negative binomial count model that corrects for endogenous stratification and truncation. We also incorporate a latent class structure into our panel specification, which assumes that the observations are drawn from a finite number of segments, where the distributions differ in the intercept and the coefficients of the explanatory variables. The paper argues that count data panel models corrected for on-site sampling may still be inadequate and potentially misleading if the population of interest is heterogeneous with respect to the impact of the chosen explanatory variables.
|Number of pages||16|
|Publication status||Published - Feb 2013|