Internal Control System, Earnings Quality and the Dynamics of Financial Reporting∗

Using an earnings management model in which managers can manipulate the earnings announcement when the firm’s internal control system fails, this paper proposes a measure of earnings quality, based on Ganuza and Penalva’s (2010) notion of integral precision that can be estimated using either the time series of earnings or the relation between earnings and prices. I show that large scale frauds are likely to be observed when the firms’ internal control system is good. By contrast, it is the high frequency of small frauds that reveals a bad control system. I also find that earnings may unboundedly increase investors’ uncertainty and that a better control system, though it leads to better earnings quality ex ante, also entails higher ex post uncertainty over the right tail of the distribution of earnings. In a dynamic extension in which managers are allowed to restate prior information, a natural link between the firm’s balance sheet and the earnings announcement arises. In this setting, I examine the determinants of restatements and also the validity of some empirical measures of earnings quality. Neither the predictability nor the smoothness of earnings would be valid metrics of earnings quality.