The Nonequivalence of the Earnings and Dividends Approaches to Equity Valuation

Accounting theory treats a wide class of equity valuation approaches as equivalent. For example, under clean surplus accounting, the earnings approach is viewed as identical to the discounted dividends approach. Empirical research, however, typically finds that the two valuation approaches do not predict market prices equally well. This paper offers a theoretical explanation for this apparent anomaly: expectations of discounted infinite sums (incomes, cash flows, or dividends) are undefined unless some restrictive probabilistic conditions hold. Without the usual stationarity and ergodicity assumptions, it may still be possible to estimate upper and lower bounds on such sums, but these bounds need not coincide. In such a setting, earnings and discounted dividends yield intervals of justifiable valuations, which intersect but need not coincide. Depending on the extent to which a firm is held by insiders, differences in the valuations that different formulae justify may not show up in market prices. This provides an explanation for two additional empirical puzzles. First, empirical studies detecting little incremental information in dividends over earnings may be predisposed toward this finding. Second, stronger apparent reactions to dividend omissions than to initiations may be an illusion.