Three Patterns in Need of a Unified Theory

Abstract

ow earnings affect stock price, how earnings are formed from underlying events and management actions, and how pay affects performance are questions that have been prominent in the research agendas of many accounting scholars. Among the results of this xtensive research effort are three striking empirical regularities: 1 stock returns are an S-shaped unction of earnings; 2 the distribution of earnings is lumpy around benchmarks; and 3 pay for op executives is an increasing convex function of stock price. To my mind, these findings suggest big unanswered question in accounting: Why are stock prices, earnings, and pay packages elated in these ways? Because price, earnings, and pay are interrelated, a good theory ought to mply relationships among these variables that are consistent with the patterns schematized in igure 1. We have some explanations for these regularities in isolation and even a few explanaions that address two of them simultaneously, but no model I know persuasively comprehends hem all. Some elements in Figure 1 bear emphasizing: First, the “divots” in the distribution of firms’ eported earnings lie just above a variety of salient benchmarks. Second, at least one divot lies in he region of steepest slope in the S-shaped relationship between the unexpected component of arnings and the contemporaneous price response. Finally, compensation contracts are smooth unctions of stock price. To illustrate some inadequacies of partial explanations, consider the argument that caps and oors in managers’ compensation contracts lead them to distort reported earnings in the neighborood of the kinks in the pay schedule. While kinks may indeed induce divots, this line of reasonng begs the theoretical question of why kinked compensation contracts are optimal and the mpirical question of whether kinks are prevalent enough to drive the divot. On the latter point, tock-based compensation is widespread and obviously imposes no cap. Another difficulty that