The CNBC Effect

According to Michael Morris, a professor at Columbia's Graduate School of Business, certain words used by the financial media to describe daily stock-market movements can subtly influence the way investors interpret the news. His research found that "agent" metaphors — verbs like leapt or climbed that imply purposeful actions — convey an expectation to the audience that today's trends will persist tomorrow. The opposite tends to be true when journalists employ "object" metaphors — words that don't imply the market is moving by its own volition, such as rolled or dropped — or no metaphors at all.

"This is one of the things we do as humans," says Morris, who will present his findings at the annual meeting of the Academy of Management next week. "We anthropomorphize things. It has consequences in the stock market because metaphors carry implications, in the sense that the event being described was purposeful, goal-oriented and therefore likely to continue."

It's this expectation of continuity in the stock market that's conveyed by journalists' word choices that can affect buy and sell decisions. A phrase like "The Nasdaq vaulted higher..." makes a bigger impression than "The Nasdaq fell..." While Morris doesn't think journalists make conscious choices to use agent metaphors especially for up markets and object metaphors (or no metaphors at all) for down markets, there's something about the way the human mind works that leads to this behavior — even on the part of ostensibly objective members of the media.

SmartMoney.com spoke with Morris about how subtle differences in word choice can affect the way people interpret news, and, ultimately, the way they invest.

SmartMoney.com: What were you and your colleagues trying to investigate in this study?

Michael Morris: This program of research came out of our interest in the way people in the stock-market commentary sector use anthropomorphic and "agentic" words. This is one of the things we do as humans — we anthropomorphize things. It has consequences in the stock market because metaphors carry implications, in the sense that the event being described was purposeful, goal-oriented and therefore likely to continue.

One study wanted to confirm the idea that ordinary investors are affected by how a price change is described to them, whether it's described in agentic terms or using physical-object metaphors or no metaphors whatsoever. In the stock-trading experiment in our laboratory, people who heard about an up day in agentic language thought it would continue. So that might lead someone like that to buy a stock. Same thing with a down day. An agentic description of a down day led them to believe that would continue.

SM: Can you explain the different categories of descriptive language?

MM: Purely nonmetaphorical language is words like decreased, increased, rose, fell. Object metaphors [are] skidded, rolled — things an object can do that don't imply that it's self-propelled and purposive. And the agentic words describe something with internal force, or internal volition, like soared, vaulted, leapt, climbed.

SM: What differences did you notice in the news about a positive movement in an index vs. a negative one?

MM: What's particularly interesting is that the choice of language is not random. We tend to use these agentic metaphors to describe an uptrend rather than a downtrend. That might lead people to take uptrends more seriously, and think downtrends wouldn't continue. We looked at a CNBC show called "Business Central," the transcripts of which are on its web site. We used a computer program that finds references to the movement of the Nasdaq, the S&P 500 index and the Dow, and we tabulated the proportion of metaphoric content. We found that when a given index had been up for the day it was more likely to be described with agentic metaphors, and when it was down for the day it was more likely to be described with nonagentic metaphors or object words, like "the Nasdaq rolled off a cliff today." When we put people in the role of commentators, they did the same thing. They launched into agentic language when describing an up day, and non-metaphorical language when describing a down day.

Our hypothesis comes from what some other researchers have shown, which is that people are responsive to trajectories. Trajectories going upward correspond to animate objects, as opposed to downtrends that correspond more to inanimate objects. This is potentially biasing people to think the trends continue. They're more likely to encode an up day as an action, to think about the market as something that's acting. When you see financial headlines, you see that the "Nasdaq climbed," "the Dow vaulted upward," or the "market crawled upward."

Certainly, some days downtrends are being described in agentic language, like the Nasdaq dove, or the Nasdaq searched for a bottom. It's just not as frequent. For the naive investor who's trying to make sense of the newspaper, the mental concepts that get triggered by agentic descriptions of uptrends may lead them to think that an uptrend is a meaningful signal for tomorrow, while the downtrend isn't a meaningful signal about tomorrow.

MM: The students we asked to do this — they didn't have the same pithy phrases stock commentators used. But they said things like, "the market went crazy in the afternoon." We think it's more likely that stock commentators are not aware they do this. I'm sure they're aware they use some metaphors. But the more concrete the language is, the more attention-grabbing it is. They're not aware of the asymmetry, that they're using more agentic metaphors — what your high school English teacher would call action verbs — like soared, vaulted and jumped, for an up day as opposed to a down day.

When trying to make language colorful, commentators take from different categories for up days and down days. That may have a subtle influence on what they're communicating. It's possible they're aware they use the metaphors, but they're not aware of how they potentially bias their audience. The way metaphors work is that they draw from one domain to describe another domain. Sometimes we convey the wrong features when we describe something.

Another one I've heard is "the Dow stumbled today." It's sort of like a person, but it's an action caused by an external force. Or the "Nasdaq was flirting with the 5000 mark," as if the index is batting its eyelashes. That connotes that it's getting close, but just hovering around there.

SM: Why do you think agentic metaphors are used more to describe ascent as opposed to descent?

MM: Think about if you're hiking, and you see something on the trail. It's either a rock or a toad. If it starts moving up, it's a toad. If it moves down, it's probably a rock. Evolution built into our brains that things that move up are alive. Things moving down don't provide that cue.

SM: One part of your paper compares people making sense of stock charts to frogs in a famous scientific experiment.

MM: There's a famous paper that revolutionized neuroscience.... In the study they had frogs in the lab, and they tried to determine how frogs are so good at catching flies. They thought maybe they flick their tongues at anything black. So the scientists put black objects in front of the frogs. They didn't do anything. Same thing with anything that buzzes. They thought maybe the frogs flick their tongues at something with the trajectory of a fly. When the frog was exposed to that stimulus, a video of a line moving on that trajectory, the frog flicked its tongue. It's wired to respond to a fly-like trajectory by protruding its tongue. It's an automatic behavior.

Other studies have shown that people are also responsive to the trajectory of things. Just like the frog will occasionally stick out its tongue at something and it's not really a fly, people may be wired to perceive upward-moving things as animate. In some cases they'll be exposed to upward moving things that are not animate. These automatic behaviors we've acquired are generally functional for us, but sometimes they're not. If someone sees a stock moving upward, they'll tend to think it will keep going.... That's a case where you may be ill-served by one of these mental habits going on in your brain. We can't say this is wired into the brain. But that's our interpretation. The distinction between animate and inanimate is one of the first things children learn. We are ready to distinguish things like that, and we have certain rules for doing it.

SM: How can you extend the implications of this research to investors and their stock-market activities?

MM: Some people have noticed that when the stock market became so overvalued in the late '90s, that was the time two things were happening. One was that more investors were making their own decisions as opposed to using a financial adviser. And the other was that the stock-market commentary sector, which previously was in a small section of the newspaper and a small part of the news, grew. Now you had 24/7 news like CNBC. The greater influence of the media at that time had something to do with the market becoming overvalued. One reason might be this phenomenon — the tendency to interpret uptrends as meaningful as regards to future uptrends.

If you look at empirical research of investors, they don't beat the market. The more frequently they trade the worse they do. The actual evidence suggests that most ordinary investors, particularly ones trying to be strategic, do the worst. The ones who are most active as opposed to passive...those investors have the worst overall returns on their portfolio at the end of two years. People tend to buy high and sell low. There are probably a lot of things about our psychology that don't serve us well in this context. If you think after an up day it's likely to continue, you're likely to buy because you want those gains.

SM: So you're saying people want to detect a pattern that's just not there — similar to how some people believe that, in sports, a player's chance of success is greater immediately after a hit or made shot?

MM: There's a tendency that in a basketball game, if someone hits two shots in a row, we say they've got "the hot hand." It's an agentic quality that explains the outcome and we expect them to make the next one. But if you flip a coin, you see it doesn't really work out that way. We tend to not really understand randomness very well. When we see a trend, we interpret that trend and expect it to go on. Basketball coaches of the world really believe there is such a thing as a hot hand and are not convinced of these statistics.

This is similar in a sense to day-to-day trends in the market. If we believe economists, the market is a "random walk." Today's trend tells us nothing about tomorrow's trend. We have a hard time with that. There are as many reasons why there would be a correction as there are for why there would be a continuation. The book, "A Random Walk Down Wall Street," says there's no memory in the stock market. Today's activity has no effect on tomorrow's activity. There's a field called technical analysis, but there is no scientific basis for it. No one has been able to prove there's any predictability whatsoever. But there are things that seem intuitively compelling to us because that's what is comfortable for us to think.