Working Paper: NBER ID: w2029
Authors: Robert McDonald; Naomi Soderstrom
Abstract: The most widely accepted empirical dividend model is that proposed by Lintner, who argued that firms smooth dividends over time. Many theoretical dividend models, however, either predict that dividends should be highly variable, or at least offer no support for the smoothing hypothesis. We use a switching regression model to test the Lintner model against an alternative which allows dividend behavior to differ depending upon whether or not firms are issuing shares. We reject the Lintner model, finding no evidence of dividend smoothing when firms are not issuing shares, and a high negative dividend growth rate when firms are issuing shares. This description of dividend behavior suggests the existence of a financing hierarchy in that the marginal source of finance differs over time. To further explore the financing hierarchy, we estimate logit models which explain the decisions by firms to change dividends, and to issue or repurchase shares. The results are consistent with the existence of a financing hierarchy.
Keywords: Dividends; Equity Issuance; Financing Hierarchy
JEL Codes: G32; G35
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Equity Issuance (G24) | Lower Dividends (G35) |
| Not Issuing Shares (G24) | No Dividend Smoothing (G35) |
| Equity Issues (D63) | Sharp Negative Dividend Growth Rate (G35) |
| Lagged Stock Returns (G17) | Dividend Changes (G35) |