The Dividend Initiation Decision : Theory and Evidence

This paper develops and tests a dynamic, sequential equilibrium model of corporate cash payout policy that endogenizes a firm's dividend initiation decision, and its extreme reluctance to subsequently cut dividends in a sequential equilibrium. After payment of dividends, all excess cash is disgorged via stock repurchases that elicit no price reactions. The theoretical model generates results consistent with many stylized facts related to dividend initiations, including: a positive announcement effect associated with a dividend initiation; a larger (in absolute value) negative announcement effect for a dividend cut/omission than for an initiation; and a probability of dividend initiation that is increasing in the firm's profitability and assets in place, and decreasing in the personal tax rate on dividends relative to capital gains. The model also generates additional novel predictions, one of which is that the probability of dividend initiation is decreasing in managerial ownership of the firm, and this effect is stronger the weaker is (external) corporate governance. A second novel prediction is that the dividend initiation probability is decreasing in the potential loss in value from the "two-audience-signaling" information disclosure costs associated with secondary equity issues. These new predictions are tested empirically using a predictive logit model of dividend initiations. Moreover, the paper tests and finds additional empirical support for the information-disclosure result using a regression discontinuity design. Thesis Supervisor: Nittai Bergman Title: Nanyang Technological University Professor, Associate Professor of Finance