The quarterly earnings conference call is a traditional way for public companies to disclose information regarding performance and strategy from the prior quarter. Wall Street analysts and other company watchers dial in, identify themselves, and wait their turn to ask the CEO or other top executives a question.
That's the procedure that HBS Associate Professor Lauren H. Cohen followed to find out more about the odd dealings he'd observed at a company that was the subject of one of his case studies. But he wasn't given the opportunity to ask a question, and the topic wasn't raised by other callers. His curiosity piqued, Cohen went back to examine the public transcript of the call.
"We looked at the people who [the CEO] called on, and it was only those who had strong 'buys' on the company's stock," he recalls.
The exchanges between the analyst and chief executive were sometimes no more than a few pleasantries and in one case, a softball question that the company knew was coming. "That made us wonder if this was true more generally across firms-that they choreograph conference calls when they don't want to talk about something."
“Nearly every firm finds it useful to choreograph or 'cast' a call at some point in its life”
This dance of deception was, in fact, exactly what Cohen and coauthors Dong Lou, of the London School of Economics, and HBS Professor Christopher J. Malloy later found to be occurring. The results of their research, Playing Favorites: How Firms Prevent the Revelation of Bad News, were published in September.
To determine when and why companies engage in this behavior and what it might indicate about their future earnings, the research team examined roughly 70,000 call transcripts from all publicly traded US companies from 2003 to 2011. For each call, Cohen and his coauthors identified the name of the firm and call participants, in addition to matching analysts with the recommendations they gave before the call. Finally, they coded the entire text of each question and answer, classifying the length and tone of each (positive or negative) using a computer algorithm.
Bad News Bears
What they discovered was surprising. This wasn't a small number of companies manipulating their earnings calls. "Instead, it seems that nearly every firm finds it useful to choreograph or 'cast' a call at some point in its life by only calling on analysts it could count on for positive commentary," Cohen says.
The evidence made it clear that firms can't cast calls forever. If they do, analysts eventually drop coverage-the tipping point seems to be four calls in a row.
So what caused the firms to periodically engage in such behavior? What seems to be correlated across the sample, Cohen says, "is that a firm had negative news that it wanted to hide."
The researchers discovered a few variables that made it more likely that a firm would deliberately call on favorable analysts. Not surprisingly, companies that in the past year had "pulled revenues forward" (reported revenues before they were earned) or otherwise manipulated the books to make their earnings look better, were more likely to cast a call. "If you've been engaging in somewhat shady activity in your accounting books, the last thing you'd want is to be called out about it in a public setting," says Cohen.
Also prone to call casting were firms with earnings that just hit analyst expectations or had beaten them by a penny. "Research has shown that firms will play around with their accounting books to hit or barely exceed analyst expectations, since it's perceived to be a very bad negative sign to fall short," Cohen says.
Companies preparing for an equity issuance and companies that engaged in insider trading in the following quarter also tended to call only on positive analysts, for the obvious reason that it would boost the stock's sell price.
Beware Future Earnings
All this manipulation may be for naught. The bad news eventually comes out, resulting in negative returns when it does.
"Firms that have cast their calls in the past have substantially more earnings restatements in the future," says Cohen. "They come clean when fewer people are paying attention, or when the timing is more advantageous."
This trend is so well established that a simulated long-short stock portfolio created by the researchers to take advantage of this fact earned abnormally high returns of up to 101 basis points per month. (In part, the portfolio shorted stocks that had cast their earnings calls a quarter earlier.)
Despite disclosure legislation such as Sarbanes-Oxley designed to ensure a transparent, level playing field, Cohen says that casting calls are a subtle but important way that companies influence the information they reveal to the market.
"What we find in the paper is that markets don't seem to understand this," he says. "This is all perfectly knowable-the transcripts are available 15 minutes after the call is over--yet we still find that returns aren't down until three months later, when the bad news comes out in the next call."
There are a couple of easy, low-cost fixes, he adds-analysts could be required to state their buy/sell recommendation as they joined the call, for example. Or when the transcript was published, it could include the names of everyone who wanted to ask a question and didn't get into the call.
"The question is, how big of an impact is this having, and do people understand that it's occurring?" asks Cohen. "The answer seems to be no. So what should we do about it, if anything? We're still very much at the point of starting that conversation, but we think it's an important one to begin."