and

We develop new tests of the dividend signaling hypothesis by focusing on the role of liquidity. We allow for two different types of signaling models: one where current dividends signal firm value and the objective is to prevent current dilution, and the other where commitments to future dividends constitute the signal. We find that the results differ by the sign of the dividend surprise. Signaling models of the commitment type explain the market reaction to negative dividend surprises. Interestingly, this result is significant only for the earlier sub-period in our sample due, perhaps, to the well-documented increase in institutional investors with longer horizons. The market reaction to positive dividend surprises, on the other hand, is shown to be consistent with the over-investment and wealth transfer hypotheses. We show that the failure of the signaling model for these firms could be due to lower costs of dividend increases.