A difficult decision for directors of both public and private limited companies is to determine the appropriate level of dividend to be paid to shareholders, and to decide whether or not to offer non-cash alternatives such as scrip dividends. From the perspective of the theorist, rationalization of the corporate dividend decision (or in its obverse form, the corporate savings function), provides one of the severest difficulties for the development of a unified theory of finance. The 'irrelevance' theory, although logically consistent within its idealized framework, does not explain observed dividend behaviour. As a consequence, a number of partial theories such as those based on 'signalling' and 'agency' models have been developed, but empirical testing of these generally gives disappointing results. However, empirical models of the partial adjustment type work quite well in capturing the time series variation of dividends at both the aggregate and individual company level. Unfortunately, since these models are essentially ad hoc, they do not give any real advice to directors to enable them to arrive at a theoretically justifiable assessment of the appropriate dividend level to set.

Arguments based on these models could lead, in a given instance, to one level of dividend being preferred to all others. Such a result might come about because of value effects relating directly to the investors' personal circumstances (for example, personal tax); or from the dividend level being interpreted as a signal of the directors' future intentions or degree of confidence of earnings growth. The existence of some share price reaction on dividend announcement prompts an analysis of the evidence for both shareholder clienteles and possible interaction of firms' dividend policies with key activities such as internal investment.