Estimation of Fractional Dependent Variables in Dynamic Panel Data Models with an Application to Firm Dividend Policy

Abstract

Fractional dependent variables and models with state dependence arise in many economic applications. However, estimating models with fractional dependent variables is complicated by the presence of two corner solution outcomes. When coupled with a dynamic panel data setting, estimating quantities of interest can be quite complex or computationally difficult. This paper demonstrates a method for estimating fractional response variables, which is easy to implement, and presents an application of the technique to the determination of firm dividend policy. The estimation demonstrates that neglecting dynamics, unobserved heterogeneity, or the doubly-censored nature of the dependent variable can generate misleading conclusions.

Keywords: Tobit, nonlinear model, dual corner solution, doubly-censored

JEL Classification: C1, C2, C5