Modelling Companies’ Dividend Policy Using Account Panel Data

Abstract

Econometric testing in finance is mainly carried out using time series or cross section data. The availability of long time series of disagreggated price data partially explains this phenomenon. Very little empirical work for testing capital asset pricing models, for instance, have been done using panel data methods. Early studies, which have used panel data in finance, have mainly concentrated on dividends. In finance, empirical testing relies mostly on firm specific samples limited to stock market quoted firms. As Bond and Meghir [1987] suggest: “A sample of exclusively quoted firms is non-random, since firms may only receive a stock market listing if they satisfy specified criteria and even they may choose to remain unquoted”. As a matter of fact, this decision, to be officially quoted or not, can be controlled by allowing for fixed effects.