We empirically examine the effects of intensified scrutiny over non-GAAP reporting on the quality of non-GAAP earnings exclusions. We find that, on average, exclusions are of higher quality (i.e., more transitory) following intervention by the Securities and Exchange Commission (SEC) into non-GAAP reporting. We further find that firms that stopped releasing non-GAAP earnings numbers after the SEC intervention had lower quality exclusions in the pre-intervention period. These results are consistent with the SEC's objectives of improving the quality of non-GAAP earnings figures. However, when we decompose total exclusions into special items and other exclusions, we find evidence that the quality of special items has decreased in the post-intervention period, which suggests that managers adapted to the new disclosure environment by shifting more recurring expenses into special items. This suggests that there may be unintended consequences arising from the heightened scrutiny over non-GAAP reporting.

The frequency of special items has increased dramatically over time, offering a convenient conduit for the inappropriate classification of past, present, and future recurring expenses as nonrecurring. Identifying this misclassification is especially important in light of the pervasive use of non–generally accepted accounting principles earnings in recent periods because special items offer camouflage for excluded recurring expenses. Building on prior research, we propose a method for identifying the predicted level of special items, attributing any excess to opportunism, and demonstrate the importance of this partitioning for financial-statement users. In particular, we provide evidence that the opportunistic portion of special items is associated with lower future earnings, cash flows, and returns. We conclude that this portion of special items is more likely to contain opportunistically misclassified recurring expenses that should have been recognized as such in prior, current, or future periods. Thus, we provide a meaningful partition of special items that should be useful to investors, analysts, creditors, auditors, and regulators because each of these parties must assess the implications of special items. This paper was accepted by Suraj Srinivasan, accounting.

Prior research documents a negative link between risk and executive holding of stock, but a corresponding positive link for options. We find a similar negative relation for non-executive holding of stock. Our finding is consistent with the view that non-executives not only face significant incentives to reduce risk when they hold stock, but they are also able to affect corporate risk. While endogeneity cannot be ruled out fully, the results of a battery of tests suggest that it plays a limited role. A second robust result is that the documented relation becomes more negative as option-based executive compensation increases. Overall, corporate risk is related to the incentives created by stock and options held by both executives and non-executives, as well as interactions among those incentives. JEL Classifications: G30.

Capitalizing on the disclosure mandated by FAS 157, I examine the equity market’s perception of the reliability of internally generated fair value estimates. For the sample of S&P 1,500 financial institutions for the first three quarters of 2008, I document a significantly positive association between stock price and fair values measured using unadjusted market prices (FAS 157 Level 1), other observable inputs (Level 2), and significant unobservable inputs (Level 3), with valuation coefficients generally increasing in the observability of the measurement inputs. Using the reconciliation of the change in Level 3 net assets, I then directly examine the periodic re-measurement of the fair value estimates and document a significantly positive association between Level 3 net gains and quarterly returns. This result manifests even among observations with thin capital cushion, poorer information environment, and weaker corporate governance. Collectively, the findings are consistent with the conjecture that investors perceive the management-provided, mark-to-model, fair value estimates sufficiently reliable to use in firm valuation and do not discard them as “markings-to-myth.”

No abstract