This is the second part of a study about common stock returns around split events (Part I) and dividend change events (Part II) as revealed in the 1947–1967 experience of the New York Stock Exchange (NYSE). Part I is the subject of a companion article in this issue of the Journal. The evidence about splitting stocks was found in many ways consistent with the efficient capital markets hypothesis, slightly method-dependent and time-dependent to an appreciable degree. By contrast, the results from Part II presented below point to persistent inefficiencies in the market. In almost any way one looks at the stocks' residuals in the months following the selected dividend changes, decreases in particular, they turn out abnormally large. The interpretation is advanced that, on the average, the NYSE under-reacts when dividend changes are announced. The possibility is also recognized that unexplored basic problems with the residual approach could account for the abnormal results.

This is the first part of a study about common stock returns around split events (part I) and dividend change events (part II) as revealed in the 1947–1967 experience of the New York Stock Exchange. Competing estimates of abnormal returns (residuals) are obtained and compared. Trading rules, involving fixed and variable monthly investments, are tested for profitability. Trading triggered by split proposals, dividend increases and, in particular, dividend decreases yields significant residuals and would thus point to market inefficiencies. Some drifting in the samples' average risk and residual behavior is explained in terms of risk measurement and two-factor market model characteristics. Residuals hardly change when risk adjustment is based on variance of returns in lieu of beta. The question is raised that the residual approach may at times prove unduly refined. Yet a multi-method approach to risk-return studies seems advisable because the odd behavior found one way can often be fruitfully explained or confirmed through comparisons with results obtained another way. In part I below outstanding features are: (1) the stocks' average excess returns in the three months following split proposals differ from zero but the anomaly hinges on the results for a rather short time sub-period; (2) numerous pitfalls in measuring and interpreting stocks' residuals are brought into light. Part II is the subject of a companion article in this issue of the Journal.