The academic literature has numerous definitions of earnings quality, yet few of these have theoretical support. For example, while researchers have advanced informal arguments that suggest smoother earnings are “better,” few formal arguments have been made to show when, or if, reporting smoother earnings implies the earnings are more informative or of “higher quality.” This is our research objective.
More specifically, we first characterize equilibrium disclosure strategies for a manager who has better information about the long run value of the firm. We then analyze when, if ever, the manager will be prompted to report higher quality earnings, where we define higher quality earnings as earnings closer to the long run value of the firm.
We build a multi-period model where earnings are composed of a permanent portion, which repeats each period, and a transitory portion that is independently distributed over periods. Investors value the firm using reported earnings to infer both the level of the permanent earnings and the precision of the reported earnings. We contribute to the theoretical literature in two ways. First, we characterize equilibrium reporting strategies when managers distinguish between the permanent and transitory components of earnings for the first period, but not in subsequent periods. Second, we use the model's structure to define higher quality earnings in what we regard to be a natural fashion: as reported earnings that are closer to the permanent earnings component. We show that how managers employ discretion depends on the type of information they observe.
If the news is “good”, the manager smoothes transitory earnings, reporting earnings that are closer to the permanent level that he privately observes than he would report if he had no discretion. If the news is bad, then the manager introduces noise into the report, lowering the quality of earnings below the level that would obtain if he had no discretion. Furthermore, we show that our results on earnings quality fail to hold in two natural benchmark cases. If the manager is perfectly informed and knows both first and second period cash flows, a fully separating equilibrium obtains so that investor have the same earnings quality as in the no discretion regime. If the manager is relatively uninformed and cannot distinguish the permanent from the transitory cash flow components (as, for example, in Kirschenheiter and Melumad, 2002), then no increase in quality can be shown to hold.
Our result, that a manager with good news will report higher quality earnings via smoothing while a manager with bad news will report lower quality earnings, is intuitively compelling. This immediately suggests two potentially testable empirical hypotheses, that smoother earnings have higher quality and that positive earnings surprises have higher quality than negative ones. However, our analysis also provides results that, while perhaps less intuitive, offer direction to empirical research. We describe three such areas.
