A couple of years ago, the SEC made a big push—through a series of staff oral admonitions and written guidance, as well as one enforcement action—toward requiring issuers to be more transparent and more consistent in the use of non-GAAP financial measures and to avoid altogether non-GAAP measures that were misleading. For example, companies were advised that they needed to present GAAP measures with equal or greater prominence relative to the non-GAAP measures.  (See, e.g., this PubCo post.) And, as this article revealed, according to Audit Analytics, in 2016, over 25% of the companies in the S&P 500 index had shifted their presentations to put GAAP at the top of their quarterly earnings releases and 81% made GAAP numbers most prominent, compared with only 52% for the prior quarterly earnings releases. (See this PubCo post.)  By the end of 2017, the SEC was apparently sufficiently satisfied with the response that the pendulum had swung back, and there was less staff focus and comment on non-GAAP financial measures.  (See this PubCo post.) But is that really the end of the story? How “good” are the numbers that are fed to investors?

Maybe not so good, according to this academic paper, “Evidence of a Positive Trend in Positive Quarterly Earnings Surprise over the Past Two Decades,” to be presented this month at the annual meeting of the American Accounting Association. The study looked at quarterly earnings among the S&P 500 over a 17-year period, focusing in particular on “Street” earnings, defined, in this case, as  “the quarterly earnings per share numbers produced and disseminated by Thompson Reuter[s] I/B/E/S based on the estimates and actuals of the analysts who contribute to that service.”

The paper identified a “sustained upward trend over the past two decades in positive earnings surprise (ES),” defined as the difference between reported and forecast quarterly “Street” earnings.  But not all earnings surprises are alike.  The study found that, since 2002, when SOX was signed into law, there has been a decrease in earnings that just meet or exceed Street forecasts, but significant growth in large earnings surprises.  As described in the press release,

“the professors find a doubling in the proportion of reported earnings that were between 5 to 15 cents per share above analyst forecasts, results in this range increasing from about 12.1% of all earnings surprises in 2000 to about 25.5% in 2016. At the same time, there was a precipitous decline in the proportion of Street earnings reports that barely met analysts’ predictions: thus, between the two periods 2000-2008 and 2009-2016, the average proportion of positive earnings surprises that were squeezed between zero and a penny dropped by about 15% and the proportion between a penny and two cents fell by about 5%. In marked contrast, the pattern of quarterly earnings surprises as measured by GAAP was much more stable over the same 17 years. For example, among the S&P 500, GAAP earnings results that were 5 to 15 cents per share above expectations represented about 10% of earnings surprises in the two years 2000-2001 and about the same percentage in 2015-2016.”

And, of course, as the paper reports, “[m]ost studies show significant stock market reactions to positive Street ES.”

The explanation, the authors suggest, could lie in managed forecasts, managed earnings, or both.  The paper suggests that “analysts increasingly bias their Street expectations downwards to generate a more positive market response for their clients, that is, they engage in strategic pessimism,” especially closer in time to the earnings announcement. If the answer is earnings management, the authors suggest, “it potentially must be of a different form and, possibly, one more acceptable to shareholders’ agents such as auditors, directors, and regulators. Non-GAAP earnings management is one such form.”  According to the press release, “If these monitors [are] focused on small changes around zero earnings surprise…we would expect to observe fewer small positive earnings surprises…and more large earnings surprises.” And that’s what they found.

Also complicit may be the analysts that contribute to and the service that compiles the “Street” data.  Not only do I/B/E/S protocols “guide contributing analysts to forecast earnings per share by normalizing their estimates for discontinued operations, extraordinary charges, and other nonoperating items as decided by a majority of contributing analysts[; they]also adjust reported GAAP earnings on a majority basis using actual amounts of the same adjustments used for normalization.”  That is, while we may assume that actual Street earnings numbers reported by I/B/E/S Thomson Reuters were the same as the actual GAAP numbers, the authors indicate that that’s not necessarily so.  As one of the authors indicated in an email, it’s “more likely the I/B/E/S Thomson Reuters ‘actual earnings’ numbers are non-GAAP actual earnings.” That’s because the Street’s “actual” number includes “adjustments on the theory that normalized earnings are a better reflection of performance for valuation and better gauge of firms’ results versus expectations.”

In conclusion, the authors take the position the explanation for the shift toward large earnings surprises relates to the concept “that I/B/E/S analysts’ focus on Street earnings and that non-GAAP metrics is a form of earnings and expectations management… The participation of firm managers in guiding I/B/E/S analysts to a majority view of Street earnings could further amplify those effects. There are those, nonetheless, who might claim that so far this century the U.S. economy has experienced such an unusual period of economic growth and success that it has taken I/B/E/S analysts and investors increasingly by surprise each quarter with better-than-expected earnings performance for almost two-decades. This view strains credulity, however.”  (See also this article in CFO.com.)

Posted by Cydney Posner