A working paper floated by Professor Ronnie Sadka (Finance) and three colleagues made headlines this past summer, including this one from CBS聽MoneyWatch鈥斺��.鈥澛燭he title alluded to the standard assumption that CEOs downplay bad news when providing guidance to the markets about recent earnings and forecasts. They accentuate the positive, in other words. But Sadka and his fellow scholars have turned light on a different pattern.
Their research indicates that corporate executives often downplay聽the good news鈥攑otentially for personal gain. Even when sales are strong, they might exude something less than optimism and then buy shares of their company鈥檚 stock at a price reflecting low expectations, according to the paper聽聽by the National Bureau of Economic Research and now forthcoming from the聽Journal of Financial Economics. When the good news about earnings finally comes out, the executives are able to sell at a higher price.
The transactions may seem questionable, but Sadka stresses that the finding is 鈥渘ot a smoking gun鈥濃�攁t least not yet.
The most innovative aspect of the study was not the findings, however. It was how the team collected the data鈥斅爄ncluding tens of millions of cell phones and tablets. Seemingly random information鈥攖he number of Google searches for driving directions to Target, for example鈥攈elped researchers construct estimates of retail sales performance in real time.
In an interview with the聽Carroll Connection, Sadka鈥攚ho chairs the Carroll School鈥檚 Finance Department and is the Seidner Family Faculty Fellow鈥攕poke about the novel use of datasets in the paper, titled 鈥淲hat Do Measures of Real-time Corporate Sales Tell Us about Earnings and Post-Announcement Returns?鈥� And, he discussed possible next steps for researchers as well as regulators.
CC: What made you and your colleagues look into this question in the first place?
Sadka: Our motivation actually was to try to understand how we can use new datasets that fall under the heading of Big Data. We looked at these data in the framework of retail shops, trying to understand how people consume. And then we used the information to predict quarterly corporate earnings before the earnings are announced.
That鈥檚 what we really wanted to do. What we found is that these data are indeed really useful. We were able to predict corporate earnings and revenues fairly well. That鈥檚 the innovation here. We鈥檙e using novel datasets, most of them proprietary, drawn in part from millions of mobile devices. With this information, we can better understand what managers and insiders in the company know at any given point, and in particular, at the end of a fiscal quarter as well as just before the quarter鈥檚 earnings are announced.
CC:聽And your basic finding is that managers know one thing and say another. They actually play down the good news about earnings in the information they provide to investors and analysts. Was that a surprise to you and your team?
Sadka: Absolutely. The usual assumption in the industry and in the literature is that when managers have good news, they typically release it quickly to the market. When they have bad news, they tend to withhold it, maybe in the hope that things will change. Overall, I think that鈥檚 a sensible hypothesis. 聽
What we found is the opposite鈥攎anagers deciding to downplay the good information. The question is why.
I think there could be several hypotheses. There could be a desire to 鈥渟mooth earnings,鈥� for example. Investors are often looking for smooth and steady earnings because less volatility in earnings makes them more confident in the firm. As a result, they鈥檙e going to be happier with the firm, which would result in a higher stock price. That鈥檚 one hypothesis, but we didn鈥檛 find much evidence to support it.
Another possibility is that managers are concerned about the risk of litigation. They might say: 鈥淲ell, it鈥檚 good news, but let鈥檚 not commit to it because we don鈥檛 know.聽I only have information for half the quarter.聽I don鈥檛 know what鈥檚 going to happen in the next half. Let鈥檚 not over-promise, because if we communicate something very positive to the market and it ends up not being good, we could be facing some litigation risk.鈥�
I don鈥檛 think we can completely rule out this possible explanation. But we also found that managers downplay bad news as well.
CC:聽In other words, if they were so concerned about litigation risk, presumably they wouldn鈥檛 be doing that, either?
Sadka: Right.
CC: So where does that leave you?
Sadka: Well, what we found is that there鈥檚 some correlation with the downplaying of the stock and the buying activity of insiders. 聽
Just to explain: Our data enables us to proxy for [find indirect measurements of] corporate sales during the fiscal quarter. We call that 鈥渨ithin-quarter sales.鈥� But it also allows us to proxy intra-quarter sales, in particular, over the period elapsing from quarter-end to the date of earnings announcement. That鈥檚 typically several weeks into the following quarter. We call this 鈥減ost-quarter sales.鈥�
Anyhow, we found that when the post-quarter sales are high鈥攚hen things are going well during the current quarter鈥攎anagers on average downplay this information. They do that in the earnings guidance they issue along with the announcement of the previous quarter鈥檚 earnings. And when that happens, there鈥檚 more buying activity by insiders over the next 20 days [the period in which corporate managers are legally permitted to trade their stocks in the company]. That information is available to the public because insiders are required to disclose this type of trading activity to regulators.
That鈥檚 the troubling result.聽It suggests that there鈥檚 some incentive for managers to downplay post-quarter sales information鈥攆or their personal gain.聽
Here鈥檚 an incentive for them. If you see things are going well, downplay it, then buy the stock. Then when investors understand the true value [that the earnings are good after all], the price will go up. It鈥檚 a mechanism by which the managers would be able to buy their own shares at a lower price. 聽
Again, we need to put things in perspective. This is not a smoking gun.
CC: It鈥檚 not?
Sadka: I don鈥檛 think so鈥攂ecause we don鈥檛 know what managers actually know. We have this information, unstructured Big Data. Maybe managers don鈥檛 look at it. Maybe they do not have such granular information. I鈥檓 assuming they have, but maybe they don鈥檛 use it on a daily basis. It could be like your retirement account. You鈥檙e not looking at your investment retirement account every single day. So maybe they don鈥檛 look every day or every week to see what鈥檚 happening in their stores. And even if they do, we do not know that they use this information when deciding to buy/sell their company鈥檚 stock.
CC: It would be a coincidence?
Sadka: Again, I鈥檓 happy to think about alternative explanations, but it鈥檚 hard for me to say 鈥渢his is proof鈥� [that managers are tamping down expectations for personal gain]. I鈥檓 not sure what the other factors are. There could be some other correlated variable that I鈥檓 not aware of. I think the importance of this result is to say: Hey, there could be some reasons for managers to withhold good information. I think then it鈥檚 up to the regulators to decide whether they want to pursue it. They probably have more information鈥攁ctual information, not proxies. Again, we have proxies.
Also, we need to be careful in drawing general conclusions. We have 50 firms in our sample [among them, Costco, Walmart, CVS, and Home Depot]. Of them, we have a smaller fraction of companies that actually issue earnings guidance. At the end of the day, we鈥檙e not talking about many, many firms. What I think is interesting is that at least we highlight some mechanism鈥攖he incentives that could happen in this environment. Whether it鈥檚 actually what鈥檚 going on is something that deserves more investigation.聽
CC: When you were聽, the anchor said you were leveling a 鈥渟trong allegation.鈥� Is that how you see it?
Sadka: No. We鈥檙e reporting the results as we find them. All of us on our team have been in academia for a long time. We try not to overstate or understate the results鈥攚e report them as they are.聽People can draw their own conclusions. They can offer alternative explanations. We rejected some alternative hypotheses, but there might be some we haven鈥檛 considered.
And even if this is going on, I think it鈥檚 going to be harder and harder to get away with it, given the increasing availability of this kind of data. There鈥檚 an electronic digital footprint on almost anything we do now.
CC: Where does the research go from here?
Sadka: I think there鈥檚 a lot of avenues for further research.
As I mentioned, we looked at a limited number of firms. So one possibility would be to expand the universe of firms but also to look continuously throughout the year to see how corporate sales information comes out in the market.
It鈥檚 part of a larger research agenda that I have been working on over the past several years鈥攗sing unique, novel Big Data or new data to better understand how financial markets work. The asset management industry is undergoing major changes. Technology is slowly finding its way into this industry, which traditionally has been very conservative. Many portfolio managers still invest in a traditional way, using traditional information like past prices, trading volume, and analyst forecasts. With more data available, things are going to change dramatically in the way portfolio managers invest.
I suspect that in five to 10 years, a significant part of the market will be using different information based on large sets of unstructured data. It鈥檚 an exciting time to be in this business.
Sadka鈥檚 co-authors are Kenneth A. Froot (Harvard Business School), Namho Kang (University of Connecticut), and Gideon Ozik (EDHEC, an international business school with campuses primarily in France).
William Bole is senior writer and editor at the Carroll School of Management.