Dividend clientele, new insights, and new questions: the Brazilian case

this paper analyzes the dividend clientele effect and the signaling hypothesis in the brazilian stock market between 1996 and 2000. during this period, the dividend tax was zero and the capital gains tax varied between zero and 10%. brazilian firms face two information regimes, which allow us to test the signaling hypothesis. from a sample of 394 observations studied, 39% show a first ex-dividend day stock price higher than the price on the last cum-dividend day. the market price is higher for unanticipated dividends but, even with pre-announced dividends, stock prices are higher than the expected level, which is inconsistent with the clientele hypothesis. we also find evidence of a positive abnormal volume around the unanticipated dividend date, which is consistent with the signaling hypothesis, but no abnormal trading volume around pre-announced dividend dates. our findings are inconsistent with the clientele hypothesis but provide support for the signaling hypothesis.