The paper examines three benchmark earnings concepts: (i) permanent earnings with the cost-of-equity determining the capitalization, (ii) permanent earnings with the risk-free rate determining the capitalization, and (iii) economic earnings (Hicks’ concept). The concepts can be measured empirically using stock prices. The study explains how the three concepts differ in terms of reflecting risk and growth. Critically, (i) and (ii) highlight two cases along a continuum. Case (i) renders growth irrelevant so that risk alone, as reflected by the cost-of-equity, determines the E/P yield ratio. By contrast, for (ii) the risk-free rate determines the E/P ratio because growth cancels out risk. The case is referred to as full cancelling out, FCO for short. The empirical part of the paper compares the earnings concepts to reported earnings using US data, for the period 1976–2015. These evaluations split firms into two categories, industrial and financial. Main findings show that for industrial firms, as an overall average, reported earnings relate closer to (ii) – that is, FCO–than to (i). Though the result is robust across methods, for distinct sub-periods (i) provides the better benchmark. As to reported earnings of financial firms, we hypothesize and find that reported earnings relate closer to (i) than (ii). This conclusion should be expected insofar financial firms rely on approximate fair value accounting, in which case earnings come close to economic earnings, (iii), and such earnings imply an average of (ii).