Given the growth in the importance and popularity of share repurchases, we use an alternative timeseries approach to test various hypotheses on the motives for share repurchases and dividends. Each hypothesis is formulated on the basis of a vector autoregression (VAR) of relevant variables and
characterized as restrictions on the VAR, thus providing a measure of its empirical validity. By investigating both share repurchase and dividend payout policies in the context of a time-series VAR, we account for the dynamic and multi-dimensional nature of the two payout policies. Therefore, our analysis provides a comprehensive understanding of various hypotheses on dividends as well as those on share
repurchases. Our findings can be summarized as follows. First, share repurchases do not contain additional information about future earnings, whereas dividend changes do. Second, share repurchases are significantly related to undervaluation, whereas dividends are not. Third, share repurchases are associated with temporary components of earnings, whereas dividends are not. Fourth, share repurchases and dividends are not perfect substitutes. They are both complements and substitutes to each other.
Keywords: share repurchase, dividends, VAR identification
