Abstract This paper solves a model that links earnings quality to the equity risk premium in an infinite‐horizon consumption capital asset pricing model (CAPM) economy. In the model, risk‐averse traders hold diversified portfolios consisting of risk‐free bonds and shares of many risky firms. When constructing their portfolios, traders rely on noisy reported earnings and dividend payments for information about the risky firms. The main new element of the model is an explicit representation of earnings quality that includes hidden accrual errors that reverse in subsequent periods. The model demonstrates that earnings quality magnifies fundamental risk. Absent fundamental risk, poor earnings quality cannot affect the equity risk premium. Moreover, only the systematic (undiversified) component of earnings‐quality risk contributes to the equity risk premium. In contrast, all components of earnings‐quality risk affect earnings capitalization factors. The model ties together consumption CAPM and accounting‐based valuation research into one price formula linking earnings quality to the equity risk premium and earnings capitalization factors.

Abstract In this paper I show that the aggregation of operating and financial income imposes three conditions on earnings‐based value functions. These three conditions provide a shortcut way to identify dividend irrelevant value functions. For example, consider any value function V t of book value b t , earnings x t , and dividends d t . The aggregation conditions imply that V t must be of the form V t = (1 − k)bt + k [f xt − dt]. f is the permanent earnings capitalization factor and undetermined weight k may be any function of Δ t ≡[φx t − d t ] − b t . The Ohlson 1995 model is the special case when k is constant. But generally k does not have to be constant to maintain dividend irrelevancy. Whenk varies with Δ t , V t is nonlinear in earnings. Hence, this result specifies how V t may be nonlinear in earnings in settings with limited liability or production or abandonment options and still be dividend‐irrelevant. An even more remarkable feature of this result is that it holds whether accounting is clean surplus or not. One must conclude that accounting‐based valuation properly builds from accounting aggregation and Δ t , and not from the clean surplus relation and abnormal earnings as many now believe.