T1 - Modeling the signaling endosome hypothesis
favors the dividend signaling hypothesis over the alternatives.
We therefore adopt a more formal proce-Under the hypothesis that the evolution of dure for testing the differences between the the excess-return relationship is driven by various subregimes. This procedure recog-changes in the tax regime, we would expect nizes the possibility that, in general, a to obtain the following results: change in dividend taxation may shift the entire signaling equilibrium, thereby poten-Test 1: reject equality of coefficients; tially altering the coefficients of all the ex-Test 2: do not reject equality of coefficients; planatory variables included in Table 3." Test 3: reject equality of coefficients.
THE SIGNALING HYPOTHESIS: EVIDENCE FROM THE …
Another potential objection to this analy- sis is that variation in tax rates is only pre- sent across years, and not across firms. Since THETA, tends to decline over the sample period, our effort to measure tax effects could conceivably pick up some spurious trend. We are inclined to doubt this hypoth- esis for two reasons. First, the evidence on tax effects is corroborated by evidence on bond ratings, and bond ratings vary consid- erably across firms. Second, the most plausi- ble explanation for a spurious trend sug- gests that our estimates should be biased against the dividend signaling hypothesis. During the 1980's, domestic firms faced in- creasing levels of competition from abroad. Casual evidence suggests that, as a result, investors became more sensitive to manage- rial inefficiency. Thus, in the absence of signaling, one would have expected to ob- serve an increase in the bang-for-the-buck during the 1980's.