Do Dividend Changes Really Signal? : Evidence from Sweden
Abstract: The purpose of the studies in this thesis is to clarify and hopefully shed some light on one important issue in dividend policy: the market reaction to announcements of cash dividends (changes) on the Swedish market.The first study, Joint Dividend and Earnings Announcements, Firm Size and Tax Law Changes, examines stock price reaction to joint dividend and earnings announcements. We found a statistically significant interaction effect for firms whose earnings and dividends indicate positive unanticipated news. The results also show that the hypothesis that small firms exhibit larger stock returns than large firms, also known as the small firm effect, was on the other hand, rejected. Finally, the tax law change in 1991 was accompanied by significant changes in stock return.The second study, Market Response to Dividend Initiations and Omissions, examines stock price reactions to dividend initiations and omissions and compares them to regular dividend increases and reductions. Signaling theory states that initiations (omissions) will produce larger positive (negative) reactions than regular dividend increases (reductions). However, signaling theory fails to fully predict the relationship for dividend reductions and omissions over the 2-day announcement period. One potential explanation of this inconsistency with theory is that firms that omit their dividend are viewed by investors as firms that accept the existence of problems and attempt to rectify them. We also found a U-shaped relationship between firms increasing and initiating their dividends over the 2-day announcement period return. This dampening effect could reflect that: (1) the market actually expected higher dividends from the large dividend increasing and initiating firms or, (2) simply that the signal issued by management was not believed by the investors.The third study, Annual Dividends and Subsequent Quarterly Earnings, examines the relationship between dividend announcements and subsequent changes in quarterly earnings per share on the StSE. Contrary to previous studies, the results are not consistent with the hypothesis that dividend signals are followed by unanticipated changes in earnings per share in the subsequent quarters in a direction consistent with the signal. Possible explanations of the lack of effects are: (1) not all firms set dividend policy in accordance with their expectations of future performance, (2) managers are unable to predict the future accurately, and (3) dividend policy is backward rather than forward looking.The forth study, Annual Dividend Cuts and Surrounding Annual Earnings Performance, examines annual earnings per share (unadjusted and industry-adjusted) for 112 Swedish firms whose annual dividends declined after three or more consecutive years of growth. The results are in contrast with signaling theory but consistent with recent empirical results, i.e., firms' earnings drop prior to a dividend cut (reduction or omission) and increases afterwards. Further, an examination of dividends and dividend yields reveal that many firms experience additional financial problems following the first dividend reduction as they continue to reduce the dividend.The fifth and final study, Cash Dividends in Sweden: The Management's View, reviews the underlying aspects of dividend policies pursued by Swedish firms. The empirical evidence has been gathered from a survey based on a questionnaire of the firms listed on the Stockholm Stock Exchange in 1996. The results show that dividends are not residually determined. Instead, they support the standing that managing directors of Swedish firms pursue a policy of active management (where the respondents among other things agreed that, on average, dividend payments provided a "signaling device" of future firm prospects). These results are also supported by earlier studies on the Australian and Irish stock market, respectively. Further, the list-, size- and industry a firm belongs to seems to have little or no influence on the dividend policy, with the exception of the decisions of the managing directors of investment firms. They, on the other hand, differ significantly in their opinions on several statements from the managing directors of other industries, probably due to the investment firm rebate and to differences in taxation.In sum, there is not much evidence in this thesis to support the claim that dividend changes in Swedish firms inform us about future earnings changes, despite managing director consensus that, on average, dividend payments provide a "signaling device" of future firm prospects. Instead, dividends seem to be better described as lagging earnings rather than leading earnings. However, on average, we find that dividend changes are associated with abnormal returns at the announcement. This suggests that the market does treat dividend changes as having informational content. This implies that if firms are sending a signal, it is not about future earnings or the market does not receive it as such. Why firms would burn money to send a signal that is not received is, indeed, a puzzle.
This dissertation MIGHT be available in PDF-format. Check this page to see if it is available for download.