STEVE INSKEEP, HOST:
Several times a year the CEO of almost any major corporation will get on a conference call, and tell analysts how their company's performed in the past quarter, and how they expect to perform in the future.
UNIDENTIFIED MAN: Good morning, everyone. I'd like to thank you all for joining us today to discuss the results of our fourth quarter and fiscal year 2011. As we have shared on our...
INSKEEP: For the most part, these earnings calls - as they're referred to - are incredibly boring.
UNIDENTIFIED MAN: It's not like you can walk down the street and identify parcels, and turn them into revenue in six or 12 months. So...
INSKEEP: But new research suggests that there may be valuable information for investors in the tone of the CEO's voice. NPR science correspondent Shankar Vedantam regularly brings us research findings. Shankar, welcome back.
SHANKAR VEDANTAM, BYLINE: Thanks for having me, Steve.
INSKEEP: OK. What kind of information can we get - not from what they say, but the tone in which the executives say it?
VEDANTAM: Well, just for some bit of context, so these earnings reports are ways for a company to communicate to investors how they're doing.
VEDANTAM: And so they're legally required – the executives are – to tell the truth. But they're also very vigilant to how what they say can affect the stock price. And so, these are very carefully worded statements and question-and-answer sessions.
Now, what these researchers have come up with...
VEDANTAM: ...is the idea that beyond the words that the CEOs are using, the way they communicate, the tone of their voices, might communicate some very important information.
Analysts do this all the time. They're trying to pick up on the cues, the emotional cues that the CEOs are communicating. What's different about this research is that a software program did this, in this case. This was a program developed by a company in Israel that used vast amounts of police interrogations to tell how people sounded when they were lying, and when they were excited, or when they had something to hide. And what these researchers - this is Bill Mayew and Mohan Venkatachalam, at Duke University - have found is that their software program is able to do something that the analysts are not able to do.
The analysts seem very good at picking up when CEOs are positive, when they're excited. But for some reason, the analysts don't pick up when the CEOs are reflecting cognitive dissonance, when the CEOs have something to hide. And the software program seems to be much better at picking up on that negative information.
Here's Bill Mayew.
BILL MAYEW: When analysts hear positive emotion content, they react favorably in terms of their stock price recommendations. On the negative side, we actually don't see such an effect. We see analysts, that if they hear negative emotion in voice, they don't appear to immediately downgrade the stock.
VEDANTAM: Steve, I should also say that there are companies that actually employ former CIA agents to listen to these earnings calls, to pick up on where the CEOs are lying. So in some ways, if you can't afford your own private spook, you know, you can buy this off-the-shelf software.
(SOUNDBITE OF LAUGHTER)
INSKEEP: You could buy it, but it's interesting to note that the human beings, even the experts here, are not showing the skepticism that they really ought to be showing.
VEDANTAM: So that's the question. It's not quite clear why the analysts' recommendations, in terms of negative information, is not good as their positive information. There could be two explanations. One, the analysts stand to profit from positive recommendations...
VEDANTAM: ...much more than they stand to profit from negative recommendations. But it's also possible that, you know, when I hear you speak, I'm much more likely to confirm when you're speaking positively. You know, I'd say Steve, you sound like you're having a good day. But if you sound gloomy and droopy to me, I'm unlikely to say Steve, you look like you're miserable today. We are much more cautious when it comes to judging other people's negative affect, than we are to judging their positive affect.
INSKEEP: So could we use this software to actually bet on the stock market?
VEDANTAM: Well, it turns out this is not a very good program to bet on the stock market, because it doesn't have the specificity to tell us how an individual stock is going to perform. But what Bill Mayew did tell me was that it turns out to be a surprisingly good way to predict when a company is going to restate financial returns...
INSKEEP: In other words, predicting that the company is being too rosy at the moment, that things are going to get worse.
VEDANTAM: Yeah. So, you know, that could be fraud, and it could also just be that the company was too optimistic. So one of the things to keep in mind, of course, is that this is not a foolproof method. But what Bill Mayew told me was that when you look at accounting techniques to detect fraud, turns out to be roughly the same as the ability of these voice techniques to detect fraud.
INSKEEP: OK. So when this guy, Bill Mayew, told you that this voice software works so well, did his tone suggest that he believed it or not?
(SOUNDBITE OF LAUGHTER)
VEDANTAM: Well, I have to say, Steve, that I'm not very good at picking up on affect. I'm a left hemisphere kind of guy.
(SOUNDBITE OF LAUGHTER)
VEDANTAM: And I listen too much to what people say, and too little to what their tone is so...
INSKEEP: OK. Well, thanks for the reporting anyway, Shankar.
VEDANTAM: Thanks so much, Steve.
INSKEEP: NPR's Shankar Vedantam.
NPR transcripts are created on a rush deadline by Verb8tm, Inc., an NPR contractor, and produced using a proprietary transcription process developed with NPR. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.