Dividend payments and share repurchases of US firms:An econometric approach

Abstract

The analyses of dividends paid by firms and decisions to repurchase their own shares require an econometric approach because of the complex dynamic interrelationships. This chapter begins by, first, highlighting the importance of developing comprehensive econometric models for these interrelationships. It is common in finance research to spell out “specific hypotheses” and conduct empirical research to investigate validity of the hypotheses. However, such an approach can be misleading in situations where variables are simultaneously determined as is often the case in financial applications. Second, financial and accounting databases such as Compustat are complex and contain useful longitudinal information on variables that display considerable heterogeneity across the firms. Empirical analyses of financial databases demand the use of econometric and computational methods in order to draw robust inferences. For example, using longitudinal data on the same US firms, it was found that dividends were neither “disappearing” nor “reappearing” but were relatively stable in the period 1992–2007 Bhargava (Journal of the Royal Statistical Society A, 173, 631–656, 2010). Third, the econometric methodology tackled the dynamics ofrelationships and investigated endogeneity of certain explanatory variables. Identification of the model parameters is achieved in such models by exploiting the cross-equations restrictions on the coefficients in different time periods.Moreover, the estimation entails using nonlinear optimization methods to compute the maximum likelihood estimates of the dynamic random effects models and for testing statistical hypotheses using likelihood ratio tests. For example, share repurchases were treated as endogenous explanatory variable in the models for dividend payments, and dividends were treated as endogenous variables in the models for share repurchases.Theempirical results showed that dividends are decided quarterly at the first stage, and higher dividends payments lowered share repurchases by firms that are made at longer intervals. These findings cast some doubt on evidence for the simple “substitution” hypothesis between dividends and share repurchases. The appendix outlines some of the econometric estimation techniques and tests that are useful for research in finance.