Dividend Displacement, Market Value, Regular and Special Dividends, and Share Buybacks in the Uk

Prior research, especially in the UK, has established a positive relationship, in crosssectional regression analyses, between market value and dividends. This result holds after controlling for the effects of earnings, book value, major media advertising expenditures, research and development expenditures, capital expenditures, capital contributions and ‘other information’. Papers with this result include, for example, Rees, 1997, Akbar and Stark, 2003, Shah, Stark and Akbar, 2009, Dedman, Mouselli, Shen and Stark, 2009. all in the UK, and Hand and Landsman, 2005, in the USA. The result appears to be at odds with theoretical work on the relationship between accounting numbers and market value by, for example, Ohlson (1995) which suggests that the overall impact of dividends, following Miller and Modigliani (1961), ought to reduce market value on a one-to-one basis (dividend displacement). Dividend displacement implies that, in a regression involving market value as the dependent variable, and containing book value and measures of dividends within the set of independent variables, the negative of the estimated coefficient of book value plus the estimated coefficients of the measures of dividends ought to be statistically indistinguishable from minus one. Such a result is strongly rejected by prior empiricism, with the estimated coefficient for dividends substantially higher than that for book value. One possibility for this result is that dividends are capturing the role of permanent earnings in the regression (Giner and Rees, 1999). This conjecture is supported by the observation that the estimated coefficient for earnings is very low and sometimes statistically indistinguishable from zero in some of the studies referred to above (for example, Akbar and Stark, 2003, Dedman, Mouselli, Shen and Stark, 2009, and Shah, Stark and Akbar, 2009), consistent with this coefficient capturing the impact of transitory earnings only. In order to test this conjecture more fully, we refine the research design of the previous studies. First, we observe that the definition of dividends used in the studies referred to above aggregate regular and special dividends. Hence, we estimate the impact of separating dividends between those that are regular and those that are special. Second, prior studies ignore share buybacks as a method of returning assets to shareholders. As a consequence, we investigate the impact of share buybacks on market value. Third, if the conjecture above is correct, we would expect that the estimated coefficient of regular dividends would exceed those for special and share buybacks, on the grounds that the latter are transitory and, as a consequence, less likely to capture permanent earnings. Indeed, the coefficients of special dividends and share buybacks might be, when combined with the coefficient of book value, consistent with dividend displacement. Fourth, prior studies have not separated out profitable from loss-making forms, which Jiang and Stark (2009) suggest could lead