【正文】 point of view, for a pany in this type of tax regime to pay any dividends. If equilibrium conditions are assumed, the argument also suggests that rational investors should demand a higher riskadjusted return, on a pretax basis, from high dividend yield stocks. This will pensate for their relatively unfavourable personal tax treatment. Trueman (1986) model the interaction of dividend policy, investment decisions and financing decisions in the context of a corporate tax regime with varying investor personal tax rates. Their model implies that shareholders in different personal tax brackets will not agree on what constitutes the optimal investment/dividend policy. Investors in high personal tax brackets would prefer the firm to invest more whilst those in lower tax brackets will prefer higher payouts and lower investment. This problem will be reduced if clienteles of investors cluster around firms with different dividend payout policies. The possibility of a dividend clientele effect was first suggested by Miller and Modigliani (1961). It provides one potential explanation of panies39。s creditors but not the shareholders. Rozeff (1982) suggests that in the absence of taxes, it is possible for a firm to have an optimal dividend policy due to the existence of agency costs. He argues that dividend payout ratios could be conditioned by a tradeoff between the flotation costs of raising external finance and the benefit of reduced agency costs, realised when the firm increases its dividend payout ratio. Easterbrook (1984) also suggests that dividend payments may serve to reduce agency costs. Management can change the risk of the firm by changing both the profile of its real investment projects and the relative balance of debt and equity. By maintaining a constant payment of dividends it avoids a build up in the balance of equity funds and simultaneously forces the firm to seek external finance. The raising of external finance will cause periodic reviews of the firm39。 their presence therefore eases the burden on existing shareholders. Shareholders will also benefit from the adjustment of leverage ratios which will acpany the raising of external finance. This argument also provides a potential explanation of why panies pay dividends and raise external finance at the same time. The final paper in this volume, by Allen, examines some of the pecularities of Japanese pany dividend policies. Japanese dividend policies have a number of interesting aspects. The standard policy is to pay a dividend which is 10 per cent of the par value of the stock. At first glance, this suggests that dividend policy cannot act as a signalling device in Japan. However, the payment of special and memorial dividends is mon and there is an emphasis on stock splits, both of which could be used as signalling mechanisms. On the other hand, many Japanese panies are members of groups with stable systems of crossshareholdings. The group banks hold equity in group panies as well as providing loans. As group panies have many points of contact and channels of information, they are unlikely to require information signals from dividend payments. Furthermore, the fact that major lenders are also shareholders means that conflicts of interest giving rise to agency costs are also likely to be reduced. Dividend policy should have less of a role in reducing agency costs in Japan. Indeed, it is qui