An empirical re-examination of the dividend-investment relation

Modigliani and Miller show that, in perfect capital markets, the optimal investment decisions of a firm are not affected by how these investments are financed. Miller and Modigliani further imply that, under the assumption of perfect capital markets, a firm's investment decisions are not affected by its dividend decisions, although dividend decisions may or may not be influenced by investment decisions. Fama and Miller label this result the separation principle. Most recent studies of the separation principle that take into account the existence of market imperfections report sharply contradictory results. This paper tests for linear and nonlinear causality between dividends and investments using both firm-specific and aggregate data for a sample of 417 firms over the 1962 to 2004 period. In general, linear causality tests support the separation principle, whereas nonlinear causality test results contradict the separation principle by revealing strong bi-directional linkages between dividends and investments.