The evidence suggests that the approximation in expression (2) works well, at least when earnings and book values are measured as they are in the United States, even for horizons as short as four years. This view is consistent with valuation as practiced in investment banking; when valu ations are based on discounted cash flows, the forecasts of cash flows are often hinged on earnings forecasts over a horizon of only five to seven years (Copeland, Koller, and Murrin 1990). The implication is that the prediction of earnings and book value, even over relatively short horizons, is as "worthy" a research objective as the explanation of price. The divi dend conundrum has indeed been skirted effectively without resorting to the circularity involved in explaining price behavior.