Feng Gu and Baruch Lev
Another earnings season is over, and the controversy about non-GAAP earnings—those alternative earnings numbers disclosed by companies—rages on. “Wishful thinking! Kool-Aid! Outright deception!” say the detractors. “Should have been banned by the SEC long ago!” “Not so quick,” counter the supporters: GAAP earnings—those statutory performance measures calculated according to generally accepted accounting rules—no longer reflect true enterprise performance, and therefore it is incumbent upon corporate managers to set the record straight by disclosing alternative, more meaningful performance figures. Over 90% of S&P 500 companies concur: They routinely provide various non-GAAP measures of earnings, sales, and other indicators. But this widespread reporting of non-GAAP measures seems only to fire up the detractors. And thus, the controversy continues.
We wish to inform the non-GAAP debate with certain salient facts—empirical evidence—sorely missing in much of the debate. You’ll be surprised by our findings. We collected a sample of 265 S&P 500 companies reporting fourth-quarter 2016 earnings, 212 of those also disclosed non-GAAP earnings. Based on this sample, we address the main criticisms of non-GAAP earnings.
First, a few stats: Non-GAAP earnings (sample median earnings per share: $0.92) are generally higher than GAAP ones (median EPS: $0.74), because they exclude various expenses. The most frequent expenses excluded by our sample companies were various tax-related items, restructuring charges, acquisition-related gains/losses, and asset impairment charges. These items share two characteristics: They don’t involve cash disbursements and they are one-time (transitory) charges, and hence don’t inform about future enterprise performance. Some companies—Boeing, Comcast, and Corning, for example—buck the trend by reporting lower non-GAAP earnings than GAAP numbers.
And now to the objections:
First and foremost, is the information conveyed by non-GAAP earnings relevant to investors? Absolutely. We correlated both the percentage changes in GAAP and non-GAAP earnings (recent quarter relative to same quarter a year earlier) across the sample companies, with the three-day share price reaction to the earnings releases (from the day before the release through the day after). We focus on the earnings change (growth) because it represents the surprise—that is, the new information—in the earnings release. If the just-released earnings growth is relevant to investors, they will act upon it, triggering a stock price change. Surprisingly, the change in GAAP earnings is uncorrelated with the surrounding share-price change (namely, was ignored by investors), whereas the change in non-GAAP earnings was positively—and statistically significantly—correlated with the price change around the release. Examples: Archer Daniels Midland and Air Products & Chemicals reported a quarterly GAAP EPS decrease, but a non-GAAP increase, and their share prices rose on the earnings announcements. Lesson 1: Non-GAAP earnings provide relevant information to investors.
Second question: Are non-GAAP earnings inconsistent from period to period, as alleged by detractors who claim that managers just pick and choose every quarter the items they exclude from GAAP earnings to fit their nefarious intentions. This issue can be examined by computing the time-series correlation of earnings; namely, the persistence over time of GAAP and non-GAAP earnings. Picking and choosing every quarter different items to exclude from GAAP earnings will result in erratic, non-persistent series of non-GAAP earnings. Another surprise: The time-series correlation, or persistence of non-GAAP earnings, is substantially higher than that of their GAAP brethren. Why? Non-GAAP earnings eliminate various one-time items (restructuring costs, impairment charges), which by their erratic occurrence detract from the consistency of GAAP earnings. Lesson 2: The pick-and-choose allegation of non-GAAP earnings is inconsistent with the data.
Third question: Are non-GAAP earnings intended to mislead investors? This is a hard one, because it deals with an unknown executives’ intent to manipulate. We sidestepped this difficulty by searching for prima facie cases of trickery where the GAAP quarterly earnings missed the (Zacks) consensus, but the non-GAAP earnings beat (surpassed) the consensus. By no means a full proof trickery, but enough to raise eyebrows. We found six! such cases, just 3% of the sample companies providing non-GAAP earnings. This hardly qualifies for “widespread manipulation” by non-GAAP earnings.
What to make of all this evidence? Banning non-GAAP earnings is silly—they obviously provide valuable information to investors. Additional SEC regulations enhancing the consistency and composition of non-GAAP earnings aren’t necessary. These measures are more consistent than GAAP earnings. However, investors should carefully consider suspect cases of non-GAAP earnings, where legitimate, recurring expenses are excluded from GAAP earnings, or a trend of decreasing GAAP earnings alongside increasing non-GAAP earnings is manifest. Finally, the major reason for the proliferation of non-GAAP earnings should be seriously considered by the SEC and FASB: As consistently demonstrated by research, in recent decades GAAP earnings lost much of their usefulness and relevance to investors. It’s GAAP earnings, rather than non-GAAP earnings, that deserve the full attention of accounting policymakers.