Abstract
1- Hypothesis development
2- Methodology and variable construction
3- Data
4- Empirical results
5- Robustness tests
6- Conclusions
References
Abstract
This study documents the fact that large dividend increases are followed by a significant increase in leverage, consistent with management increasing the dividend to use up excess debt capacity. However, the leverage increase is not captured by a standard partial adjustment model of leverage. Nor does it reflect variables known to be related to dividend increases, such as firm maturity, investment, and risk. Instead, the dividend increase signals a complex change in the way firms adjust to their leverage target, but it does not signal a change in the target
Hypothesis development
Firms that make large dividend increases do so voluntarily. Models of dividend signaling suggest that this should be a signal of quality, implying increased future profits (Bhattacharya 1979; Miller and Rock 1985). However, the work of GMS fails to find the hypothesized increase in profits. Instead, they find a change in firm maturity and related variables. But a large part of firm maturity is predictable and so need not be signaled. The behavior they find which cannot be predicted before the dividend change includes declining reinvestment, so the dividend increase could be a signal of an unanticipated change in investment policy. Since investment policy and financing policy are closely linked through the pecking order view of financing, we hypothesize that the unanticipated change being signaled by a large dividend increase could include a change in financing policy. Specifically, we hypothesize that large dividend increases or dividend initiations signal a change in leverage policy of a type that could not be forecasted based on standard models of leverage. To be consistent with the move towards a more mature type of firm that is related to dividend increases, we hypothesize that the change is an increase in leverage. We require that the change is not related to predictable firm maturity variables, so that the dividend increase is a genuine signal: Hypothesis 1A: Firms making large dividend increases subsequently increase leverage in a way not predicted by standard models and not related to predictable firm maturity variables. The null hypothesis against which we test is that any change in leverage policy subsequent to a large dividend increase either is explained by standard models of leverage (trade-off, partial adjustment, or standard pecking order) or reflects the predictable maturing of the firm.