Financial Forecasting: Why It’s Getting Cloudy
Companies are reporting earnings that are beating analysts’ estimates by a wide margin. Why isn’t financial forecasting getting more accurate?
If there’s one takeaway from first-quarter earnings so far, it’s that AI hasn’t made financial forecasting any more accurate.
Heading into this week, 84% of S&P 500 companies reporting earnings beat analyst expectations, and by a wide margin. According to financial-data firm FactSet, companies’ reported earnings are 12.3% above estimates in the aggregate, which is five percentage points higher than both the five- and 10-year averages. The margin of inaccuracy—against the backdrop of the SEC’s potential move to reduce earnings reporting from quarterly to twice annually—holds significant implications for investors, in terms of both financial transparency and the ability to gauge company performance. “The differences are huge,” says Peter McDermott, a senior client partner in the Global Corporate Affairs and Investor Relations practice at Korn Ferry.
Estimates are no more than that, of course, and a certain margin of error is to be expected. But the size of the “upside surprises,” as earnings beats are known in industry parlance, goes beyond firms simply underpromising and overdelivering, because these surprises, over time, can affect a company’s ability to maintain strong investor relations. To be sure, given the speed at which current world events are shifting, “analysts likely baked conservatism on the part of firms into their estimates already,” says McDermott.
To derive their estimates, analysts rely on various factors, among them an organization’s historic financial performance, management comments, and industry and market trends—all of them metrics AI is supposed to be better at evaluating and modeling than humans. But the current economic and market volatility “makes it harder for analysts to forecast, even with AI and richer data,” says Chad Astmann, co-head of global investment management at Korn Ferry. Astmann says some forecast misses are hard to predict because they aren’t based on historical models, such as companies reporting changes in the cost structure over the course of the quarter, onetime items, expense timing, and so forth.
Additionally, fewer companies are providing financial guidance in advance. Since the pandemic, the number of firms issuing guidance has steadily declined, even as uncertainty and volatility have increased. Currently, only about 20% of S&P 500 firms issue guidance; during the early 2000s, about 50% did. If firms move to a biannual reporting schedule—which could disadvantage retail and individual investors who rely on signals from the media and analysts to make investment decisions—that figure could fall even further.
Even if analysts are using the most sophisticated AI and financial-modeling tools, their predictions are only as good as the data they are based on, says McDermott. Put another way, in order for AI to help analysts make better predictions, it has to be trained more deeply on the businesses it is evaluating. “AI is definitely not there yet when it comes to cracking the code with equity research,” he says. Astmann agrees, but says it’s only a matter of time. “The mastery of AI usage will improve, and new practices will make their way into financial modeling to sharpen analysis at a rapid clip,” he says. “It’s just early.”
Learn more about Korn Ferry’s Organization Strategy capabilities.
