Dividends and earnings: their effect on firm value

Thesis (Ph. D.)--University of Washington, 1992This dissertation investigates the relation between firm value, dividends, and earnings. It develops a model of firm value which distinguishes between the operating activities of the firm, in which the average return on investment exceeds the marginal return, and the nonoperating activities, in which the average and marginal returns are equal.Firm value is modeled as nonoperating assets plus the expected present value of future operating cash flows less the expected present value of the effective tax cost of future dividends. Differences in the tax treatment of dividends and capital gains between individual and corporate shareholders allows agreement on the level of the dividend payout rate to reduce agency costs. This agreement implies that firm value will also be a concave function of the payout rate. Regressing the market value at the ends of 1980 to 1987 of the common shares of 249 NYSE listed firms on proxies for the model's constructs yields parameter estimates generally in accordance with the predictions of the model, except for the estimate on effective dividend tax cost.Assuming that the ratio of dividends to persistent operating cash flow is intertemporally constant and that operating cash flow is expected to grow at a constant rate implies that current dividends can proxy for both the expected present value of future operating cash flows and the expected present value of the effective tax cost of future dividends. Thus firm value equals nonoperating assets plus a multiple of dividends. This multiple, the dividend response coefficient, is an increasing function of the growth rate and a decreasing function of the dividend payout rate, the discount rate used in the present value calculation, and the effective tax rate on dividends. Tests using the 1980-87 data of the same 249 firms generally support these predictions, except that those relating to the effects of the discount and tax rates on the dividend response coefficient are inconclusive.The model is then used to analyze the effect of dividend and earnings announcements on firm value. The dividend and earnings response coefficients are shown to vary depending on which variable is announced first, and to vary as the rates noted above vary. Tests using quarterly 1981-87 data of 270 NYSE listed firms show marginally significant differences in the magnitude of the response coefficients due to the relative timing of the earnings and dividend announcements. The tests do support the predictions that the response coefficients would be larger the larger the growth rate and the smaller the interest and tax rates. Support for the predictions that the response coefficients would be smaller the larger the payout rate or beta is weak