Chapter 12 • The dividend decision
provide information about the level of profitability in the years following the change.
Increases in dividends tend to indicate increases in accounting profits in each of the
two years following the increased dividend. This links in with the Fama and French
(1998) finding. The evidence on the informational effect of dividends seems fairly
conclusive.
Mougoue and Rao (2003) found that not all businesses seem to signal future
profitability through dividend policy; only about one-quarter of the businesses that
they examined did so. They noted that the ‘signallers’ tended to be smaller, have a
lower asset-growth rate and have higher capital gearing than the average.
Dewenter and Warther (1998) compared signalling effects in the USA and in Japan.
They concluded that dividends seem to have a greater informational content in
the USA than in Japan. They inferred that this arose from a closer relation between
investors and managers in Japan, which leads to better information flows from the
latter to the former. This might indicate less of an agency problem in Japan. Lack of
information by shareholders tends to be a major cause of agency costs.
Hand and Landsman (2005) (H and L) contradicted the established view on sig-
nalling. They analysed the dividend/share price relationship for a large number of US
businesses between 1984 and 1996. They concluded that businesses do not intention-
ally use dividends to give signals, but that the size of dividends enables investors to
reassess their previous projections of businesses’ future earning and cash flows. H and
L imply that investors tend to underestimate these projections and that dividends can
often indicate this, leading to an increase in the prices of the shares concerned.
If H and L are correct, it is not difficult to see why earlier researchers have misin-
terpreted the role of dividends in this context. It also resolves the enigma as to why
managers do not simply tell investors how things are progressing with their busi-
nesses rather than sending signals that may be difficult to interpret; according to H
and L they do not send such signals.Clientele effect
Elton and Gruber (1970) conducted a rather interesting study to test for the clientele
effect. By looking at the fall in share price when a share goes from cum dividend to ex
dividend (shortly before a dividend payment), they were able to infer an average
marginal income tax rate for any particular business’s shareholders. They found that
lower income tax rates were associated with high-dividend-paying shares and higher
income tax rates with low-dividend shares: that is, they found a clientele effect.
Pettit (1977) gained access to information on the security portfolios of a large num-
ber of clients of a large US stockbroker. He found that low levels of dividends seemed
to be preferred by investors with relatively high marginal income tax rates, by
younger investors, and by those who were less than averagely risk-averse.
However, Lewellen, Stanley, Lease and Schlarbaum (1978), using the same data
but a different approach from that taken by Pettit, reached the dissimilar conclusion
that there is only a very weak clientele effect. Litzenberger and Ramaswamy (1982)
undertook a study that produced results that seemed to support the clientele effect.
Crossland, Dempsey and Moizer (1991), using UK data, also found clear evidence of a
significant clientele effect. Graham and Kumar (2006) looked at the investing beha-
viour of 60,000 US households and found a clear clientele effect.
Allen, Bernardo and Welch (2000) identified an interesting link between the clien-
tele effect and the agency issue. They argue that businesses that pay dividends tend to