Don’t blame Wall Street analysts, rank-and-file investors and journalists for failing to read corporate earnings reports from beginning to end.
For one thing, the quarterly and annual reports churned out by publicly traded companies are forbiddingly long. Microsoft’s most recent annual report, known as a 10-K, was 130 pages; Apple’s was 96 pages; Amazon’s was 89 pages.
For another, their prose is often dreary. Filled with legal language aimed at satisfying securities regulators without alarming investors and driving down share prices, many earnings reports seem designed to be mind-numbingly boring.
It is the financial figures within these pages that are critically important. Is a company generating profits and cash, is it growing, is it borrowing too much or investing too little? Answers to these questions are often revealed in the numbers, but I have always thought it best to avoid spending much time on the words.
There is evidence that most other people have made the same judgment, yet it turns out that we have been making a mistake. The turgid language in these corporate reports is actually sprinkled with important clues about major problems – and there is a way to get an inkling about them without having to read every word.
Those are the main insights of Lazy Prices, a fascinating research paper featured in the November issue of The NBER Digest, a publication of the National Bureau of Economic Research. It found that corporate reports are repetitious, but when the language in the current text varies a great deal from previous versions, it frequently signals trouble that will become evident several months later.
Three economists – Lauren Cohen and Christopher Malloy of the Harvard Business School, and Quoc Nguyen of DePaul University – downloaded every quarterly and annual corporate report of every publicly traded U.S. company from 1995 to 2014.
They then sifted through thousands of reports, using a text-analysis program.
“We filtered out the reports that made a lot of wording changes over the previous year’s version,” Mr. Cohen said in an interview. “It turned out that when there are a lot of changes, there’s a good chance that something important is going on, and most of the time, it’s negative.”
Among the reports that included many such changes, the researchers found a high probability that the companies’ share prices would decline several months after the reports appeared.
Textual changes in the “risk factors” section were most likely to predict subsequent moves in share prices. (Corporate earnings reports are rigidly formatted to comply with Securities and Exchange Commission requirements and are divided into predictable sections.) When the researchers refined the analysis further, differentiating between textual changes that were positive and negative, the results were even more striking.
The vast majority, 86 per cent, of reports with substantial wording changes were primarily negative in tone. But the minority that contained primarily positive language generally correlated with increases in share prices later.
That brings up another important point: The stock market rarely responded to the subtle hints in the reports immediately. In fact, it typically took several months for whatever good or bad news was embedded in the reports to be widely understood – and to move the stock market.
This delay means that there is a profit opportunity for those able to exploit it, the researchers said.
When they constructed a hypothetical portfolio that shorted – or bet against – the companies with primarily negative word changes in the reports, while buying shares of the companies with positive changes, the portfolio outperformed the stock market by 22 percentage points a year.
What is perhaps most remarkable is that they selected these investments without actually reading any of the reports or analyzing the stocks or the companies.
“Changes to the 10-Ks predict future earnings, profitability, future news announcements, and even future firm-level bankruptcies,” the researchers said. That was the case even when the reports contained no outright announcements about such matters.
As a result, they said: “We find no announcement effect associated with these changes – with returns only accruing when the information is later revealed through news, events, or earnings – suggesting that investors are inattentive to these simple changes across the universe of public firms.”
Investors are apparently still being inattentive. An excellent Heard on The Street column in The Wall Street Journal in June, 2017, by Justin Lahart reported on an earlier version of the Lazy Prices findings. In that column, Mr. Lahart said explicitly that investors might profit by following the researchers’ methods. And in August, the open-source Quantopian investing site created an algorithm replicating the researchers’ methods, and found that the results remained robust: A new model portfolio outperformed the market.
In other words, word changes in corporate reports still appear to be signaling subsequent changes in stock prices and few, if any people, have noticed. That suggests that scarcely anyone has been reading corporate reports thoroughly – and that scarcely anyone has read the Lazy Prices paper, either.
To be fair, the clues unearthed by the researchers’ approach are fuzzy and imprecise at best.
In November, for example, I asked Mr. Cohen how I might make use of his insights.
He suggested two things: “First, always download the previous version of a corporate report as well as the current version, so you can compare the language. Focus on the differences from year to year. Second, focus on one section, the ‘risk factors section.'”
I did just that, looking mainly at the latest Apple corporate report, posted on its website on Nov. 5. There was plenty of new language describing the risks that Apple faced – risks that Apple executives discussed in November and disclosed more directly in a letter to shareholders on Jan. 2.
These risks included the possibility of an economic slowdown in a place like China, the possibility of slowing iPhone sales, the threat of a trade war and the potential impact of unfavorable foreign exchange rates. All of these risks have actually materialized and Apple shares have sunk, pulling down the overall stock market.
But based on the language changes in the corporate report alone, was there a specific reason for investors to bid down Apple stock, or for a journalist to conclude that Apple was in serious trouble?
I did not think so then, and I still don’t. The linguistic clues were another piece of evidence that needed to be evaluated and, perhaps, bolstered by other research.
Clues like that have real value, though. The Lazy Prices research will, at the very least, lead me to read these reports a little differently, in the hope of using them to unearth deeper truths. What’s more, until the Lazy Prices research is widely understood, it may be possible for some investors to profit from it.