【正文】 vidends. 3. Informational asymmetry and signalling models Deviations from the Miller and Modigliani (1961) dividend irrelevance proposition is obtainable only when the assumptions underlying the setting of Miller and Modigliani are violated. The taxclientele hypothesis uses the market imperfection of differential taxation of dividends and capital gains to explain the dividend puzzle. Bhattacharyya (1979) develops another explanation for the dividend policy based on asymmetric information. Managers have private knowledge about the distributional support of the project cash flow and they signal this knowledge to the market through their choice of dividends. In the signalling equilibrium higher value of the support is signalled by higher dividend. In other words, the better the news, the higher is the dividend. Heinkel (1978) considers a set up where different firms have different returngenerating abilities. This information is transmitted to the market by means of dividends, or equivalently, from investing at less than the first best level. In the equilibrium of Heinkel’s model, the firm with less productivity invests up to its first best level and declares no dividend, while the firm with higher productivity invests less than its first best level of investment, and declares the difference between the amount raised and the amount invested as the dividend. The firm with higher productivity acts in this way in order to distinguish itself from the firm with less productivity. Dividends are still irrelevant in the sense that both firm types could raise an extra X dollars with a new issue to pay an extra X dollars as a dividend with no signalling effect. The signalling cost in this model es from reduced investment from first best level. In contrast, the signalling cost in Bhattacharyya (1979) es from taxation and nonsymmetric cost of raising funds in the capital market. Bhattacharyya and Heinkel’s work was followed by a number of other papers which posited that dividends are used by managers to transmit information to the capital market. Notable works in signalling paradigm of dividend policy are those of Miller and Rock (1985), John andWilliams (1985) andWilliams (1988). These signalling models typically characterize the informational asymmetry by bestowing the manager or the insider with information about some aspect of the future cash flow. In the signalling equilibriums obtained in these models, the higher the expected cash flow, the higher is the dividend. In Miller and Rock (1985), the signalling cost is the opportunity cost of less than first best investment. In John and Williams (1985), and Williams (1988), the differential taxation of dividends visavis capital gains sustains the signalling equilibriums. In these papers dividends sustain a fully separating equilibrium. By contrast, Kumar (1988) demonstrates that dividends could also sustain a semiseparating equilibrium where the manager has private information about the productivity of the firm. Venezia (1991) set up a rational equilibrium expectation model. Bayesian investors expect that dividends will be proportional to cash flows. Managers have advance noisy information about the future cash flow. The investors observe the dividend and update their belief about the cash flow. Under these circumstances, Venezia show that the optimal dividend is proportional to the cash flow. Brennan and Thakor (1990) focus on a different question pared to the other signalling type papers on dividend policy. Most dividend policy papers model the dividend decision, as a decision about the amount to be distri