Semi-Nonparametric Estimation of Binary Choice Models Using Panel Data : an Application to the Innovative Activity of German Firms

François Laisney, Michael Lechner and Winfried Pohlmeier This paper presents an extension of Chamberlain's approach to modelling correlated random effects in a dichotomous model estimated on panel data. The extension consists in relaxing the arbitrary normality assumption made on all error terms and on specifying instead a SNP distribution for the composite error term resulting in each wave from Chamberlain's treatment of correlated random effects. An application to a panel of 1325 German firms illustrates the feasibility of the approach.