Do We Know When We’re Headed for a Crash?

You’ve no doubt seen the news about recent stock market losses. It was on the front page of every major newspaper. It was described with words like “carnage” and “bloodbath.”

A new working paper by Yale SOM professors William N. Goetzmann and Robert J. Shiller, with co-author Dasol Kim PhD ’11, demonstrates that such high-profile, negative coverage of markets can increase investors’ estimation of the likelihood of a major crash in the future. They also find that this shift in sentiment about the market likely affects real investment decisions.

Read the study: “Affect, Media, and Earthquakes: Determinants of Crash Beliefs From Investor Surveys” [PDF]

Shiller, who received the Nobel Prize in 2013, has been sending out surveys since 1989 with the same basic question: “What do you think is the probability of a catastrophic stock market crash in the U.S., like that of October 28, 1929, or October 19, 1987, in the next six months?” The new paper exploits this rich, unbroken data series. (Shiller’s survey is sponsored by the International Center for Finance at Yale SOM. View the center's collection of stock market confidence indices.)

While the historical likelihood of a massive crash in any given year is about 2%, investors answering the survey have consistently guessed that the likelihood of such a crash in the next six months is closer to 10%.

“People wildly overestimate the probability of a stock market crash,” says Goetzmann. “So people are walking around terrified that they're going to lose a lot of money in the stock market very soon.”“This has, I think, some fundamental implications for financial theory,” adds Shiller. “People are worried about big, catastrophic events, and that colors their thinking and their actions.”

It Bleeds, It Leads

The researchers examined news coverage of markets and how it corresponds with crash belief data to better understand what was causing investors to have such an exaggerated sense of risk.

From a database of all the articles published in the Wall Street Journal from 1989 to 2015, Goetzmann, Shiller, and Kim selected articles that referred to the stock market, ending up with 133,496 articles of at least 250 words in length. They analyzed the language in each article and ranked them as either positive or negative, and also calculated the overall sentiment for each day. They then compared these measures of news content with surveys that were filled out at the same time. The researchers also considered whether the prominence of positive or negative news—whether it appeared on the front page or in the lead paragraph of stories—had any salience.The overall findings were that news coverage significantly affected individual investors’ estimation of the likelihood of a crash in the future—but, interestingly, it was only negative news that had this effect. Positive stories didn’t move the needle. Bad news in the markets was more likely to get covered, more likely to grab attention, and more likely to affect crash probability assessments.

What might be driving this effect? Investment requires an assessment of probabilities about the future: the likelihood of a recession versus economic growth, the chances that a new product will be a success or that a competitor will get to market first. Over the last few decades, the fields of behavioral economics and behavioral finance have identified a number of consistent psychological biases that affect investors’ thinking as they try to make sense of the future. As an explanation for investors’ overreaction to bad news, Goetzmann, Shiller, and Kim single out the availability heuristic, which posits that people tend to make disproportionate use of easily available information (such as recent events and top-of-mind stories) when estimating the probability of some event occurring in the future.

Consistent with this theory, negative news that appears on the front page and in the lead paragraph of stories appears to have the strongest effect on individual investor sentiment.

Shiller’s surveys distinguish between institutional investors, who work at funds and major institutions, and individual investors, who manage their own money. The data showed that institutional investors, who likely are more sophisticated and have access to more sources of information about market performance, were less likely to be swayed by negative sentiment in the media.“Individual investors really react to yesterday. Institutional investors seem to be more sober,” says Shiller. “While we have many professionals in the stock market, the big decisions as to how much to put in the market are still done a lot by individual or retail investors.”

Earthquakes as Exogenous “Shocks”

The researchers found additional support for the availability bias by looking at events unrelated to markets that could affect investors’ thinking. They identified minor to moderate earthquakes that occurred within 30 miles of the location where surveys were filled out. There’s no economic reason why these minor rumbles—large enough to be noticed but not large enough to do any meaningful damage—should affect investors’ predictions about a stock market crash. But availability bias suggests that if the unsettling news of the earth shaking were at the forefront of respondents’ minds, they might be primed to think that rare events are more likely to occur. Sure enough, individual investors filling out surveys after one of these quakes gave higher probabilities of a market crash.

“This is pretty prima facie evidence that people are not being strictly rational and applying probability,” says Shiller. “If there's a little earthquake in your neighborhood, why on earth would you think the stock market is going to crash? Well, they do. It's something about human nature.”

The last question that the researchers tackle is whether these fluctuating crash predictions actually affect investor behavior. While they can’t draw a direct line between survey results and purchases or sales of stock, they do find that rises in the crash probability predict decreased flows of funds to equity-based mutual funds. In other words, when it seems scary, people put less money into mutual funds.

“Not only is there a fear factor, it’s a fear factor that influences their investment decisions,” Goetzmann explains. “So we think that this fear of a meltdown is a first-order issue in the economy that has to be dealt with, if we're going to get people to make reasonable decisions about their long-term investments.”

Shiller says that the careful study of how people estimate and respond to the risk of a crash shines a light on the complex mix of motivations underlying investor behavior.

“To be a rational investor, you have to have some basis in reality. The stock market in the U.S. lost more than 80% of its value between 1929 and 1932. Is that going to happen again? If so, I’m going to get out,” he says. “But people are not just informed by probabilities. Maybe they stay in the market out of a sense of adventure or machismo… They’re also affected by the story of the times.”