In an interesting article, Fama and French (2001a) document a dramatic decline in the propensity to pay dividends over the last two decades. They find that while 66.5% of listed firms paid dividends in 1978, only 20.8% did so in 1999. Part of this decline in dividend paying propensity is explained by the changing characteristics of listed firms. New lists over the last two decades tend to be smaller firms with more growth opportunities, less history of profitability, and more distant payoffs. Such characteristics make firms less likely to be dividend paying. Even after controlling for changing characteristics, however, the propensity of firms to pay dividends still declined over the last two decades, a phenomenon that Fama and French call “disappearing dividends.”
The Fama and French (2001a) findings are striking and demand an explanation. Why did dividends lose their popularity between the 1970s and the 1990s? DeAngelo, DeAngelo, and Skinner (2004) show that the dollar supply of dividends, which is concentrated among large payers, has not declined over the period that dividends have disappeared. Thus, the declining propensity to pay dividends mainly reflects the decreasing interest in attaining dividend paying status by smaller firms who initiate and pay small dollar dividends. The puzzle is why the desire of such firms to attain dividend paying status lost popularity between the 1970s and the 1990s.
In a series of two articles, Baker and Wurgler (2004a) and Baker and Wurgler (2004b) – henceforth BW – propose a “catering” theory to explain disappearing dividends. The catering theory hypothesizes that investors lump stocks into dividend paying and non dividend paying categories. The demand for each category of stocks is time varying and is driven by transient fads or sentiment for dividend paying stocks. BW argue that when pro-dividend payer sentiment is high, firms cater to the sentiment by becoming more likely to be dividend paying. Likewise when the fascination for dividend stocks is low, fewer firms become dividend paying.
BW report empirical evidence consistent with the catering hypothesis. They show that a proxy for catering, the valuation differential between payers and non-payers, is related to the time series variation in the propensity to pay dividends. More-over, catering is economically significant, accounting for a large 30% of the variation in the propensity to pay dividends. BW interpret this evidence in a behavioral finance framework, suggesting that sentiment driven fads are important determinants of firms’ dividend policies.
Our paper investigates the role of a decidedly old-fashioned variable, viz., risk, in explaining the changing propensity to pay dividends. Briefly, we report three major findings. First, risk is an economically and statistically significant determinant of the propensity to pay dividends. It has a marginal effect comparable to or better than that of other determinants of dividend paying status. Second, risk is related to disappearing dividends: it can explain up to about 40% of the Fama and French (2001a) disappearing dividends phenomenon. Finally, catering becomes statistically and economically insignificant once we control for risk. The results are robust across a range of specifications, samples, and tests.
