Estimation of Panel Data Models with Two-Sided Censoring

Abstract

It is straightforward to construct moment conditions for two-sided censored panel data regression models with strictly exogenous explanatory variables. The contribution of this note is to show that one set of these moment conditions uniquely identify the parameters of the model under a natural full-rank condition. The identification result suggests an estimator that is then applied to evaluate the portfolio allocation effect of a Danish tax reform that increased the after-tax capital income of bonds relative to stocks.

Keywords: Censoring, Panel Data, Tax Reform, Portfolio Allocation

JEL Classification: C33, C34, G11, H24