Time-Varying Coefficients for Discrete Panel Data / Zeitvariierende Koeffizienten für diskrete Panel Daten

Summary For panels which are observed over a long time it has to be assumed that the effects of covariates vary across time. An estimation method based on local likelihood principles is proposed which allows to estimate the variation of coefficients in the case of discrete responses. Approximate confidence intervals are derived for the estimates. The method is applied to business tendency surveys and the results are compared to the extended Kaiman filtering approach.