Pieces to the Puzzle: New Evidence on the Interpretation of Dividends: A Study on the Information Content of Dividends Hypothesis using the Novel Approach of High-Frequency Data and Systematic Risk

Christian Nygaard & Andreas Zarp-Karsholt

Student thesis: Master thesis


Using intra-day equity prices and recently developed variance estimators of high frequency data, the informational content of dividend announcements is examined. This is done by estimating changes in systematic risk (beta) on the day of the dividend announcement hereby enabling an examination of investors’ reaction to an unexpected change in the dividend level. The main high frequency estimator applied is the realised (co)variance estimator used to construct realised beta values but due to potential biases in this model additional estimators of realised (multivariate) kernels, bi-power (co)variance, and the Hayashi-Yoshida estimator accompany it for additional testing. Using a sample of all S&P 500 firms in the period 2000-2012, the daily beta estimates are regressed on event day dummies to model the dynamics of beta. It is found that beta increases on the announcement day in the order of 0.21 when a firm decreases the dividend level by at least 20%. After a series of robustness tests, it is found that the estimated change in beta range from 0.13 to 0.37 following announcements of a dividend decrease. These results are both statistically and economically significant. For a dividend increase no change in beta is found on the announcement day and all robustness tests consistently confirm this. In light of these results, it is concluded that no relationship exist between dividend increases and beta while beta exhibits an inverse relationship w.r.t dividend decreases. This is partially in support of the Information Content of Dividends Hypothesis which postulates an inverse relationship between unexpected dividend changes and beta. The application of beta is due to its positive association with the expected return of a stock which is difficult to observe in practice. The inverse correlation is then hypothesised to be related to the variability of a firms future cash flows. As the variability of the cash flows increase, the firm decreases dividend payments to create more room to manoeuvre financially. Due to the positive relationship between beta and expected return, investors should be aware of dividend decreases as this can affect financial portfolios’ risk/return composition. The missing link between dividend increases and beta can potentially be explained by three factors; (i) the model presented by Lintner (1956) in the sense that the market might see a dividend increase as a move towards the long term dividend target of the firm. Hence all the information related to a dividend increase is already incorporated in the market at the time when the firm first decides to pay out dividends. (ii) The use of actual dividend changes and not unexpected changes due to data availability issues and (iii) if the postulated inverse relationship exists for dividend increases then managers can lower the firm’s cost of capital by increasing the dividend. Assuming a rational market, investors anticipate such sub-optimisation by managers and refrain from decreasing the beta value of the stock in the wake of a dividend increase. These conclusions can however only be discussed qualitatively as further research is required to allow for a better understanding of the asymmetric findings of this paper.

EducationsMSc in Applied Economics and Finance, (Graduate Programme) Final Thesis
Publication date2013
Number of pages153