UNC Kenan-Flagler Business School

Shaping Leaders, Driving Results

UNC Kenan-Flagler Business School

ROI: RESEARCH INSIGHTS FROM KENAN-FLAGLER


It must be true. I read it in the newspaper.

10/8/2013

How columnists have historically influenced the market.

Evidently, people do believe what they read in the newspapers. Diego Garcia, associate professor of finance at UNC Kenan-Flagler, discovered that he can predict the next-day performance of the stock market by reading the newspaper. He analyzed a century’s worth of financial analysts’ columns in The New York Times, noted whether the columnists’ forecasts were gloomy or upbeat, and confirmed that the following day the stock market mirrored the columnists’ predictions quite accurately. The effect was all the more pronounced during a recession.

Either the columnists were remarkably prescient or reader sentiment had a strong influence on market performance. Garcia concluded the latter. 

“It’s difficult to consistently make a lot of money in the stock market,” Garcia said. “Otherwise, we’d all be doing it. But if you’d paid attention to the columnists’ interpretation, you could have made a lot of money.”

Few people did, in either the Great Depression of the 1930s or the Great Recession that began in 2008. That got Garcia thinking, and formally analyzing the data.  He has conducted three research studies examining the effect of reader sentiment on asset prices.

Some background:  Garcia subscribes to online news sources, including The New York Times. In 2008, he got an email that The Times was opening its historical archives that went back to the mid-1800s. He began browsing through the financial pages leading up to the Great Depression. Other researchers had posited a link between printed-word hype and investor action. Garcia’s paper, “Sentiment During Recessions” in The Journal of Finance, is among the first to prove it.

The Times’ archives consist of scanned images of printed pages of the newspaper. By using optical character recognition techniques, Garcia converted data in the images into text that he could search. “This took months,” he said. “I need more computers.” 

He counted the number of positive and negative words in two financial opinion columns appearing daily in The Times from 1905 to 2005. By and large, the columns did not contain any previously unreported information but simply speculated about why the market acted as it did over the recent past and what it might do in the future. He matched each day’s results with the following day’s returns of the Dow Jones Industrial Average.

By covering an entire century of economic cycles, Garcia could gauge the reader-sentiment effect in good times and bad. He found that heightened sensitivity to news media was strongest during economic recessions and that the columnists’ opinions had an immediate effect on market performance. By four days after a decisive opinion column, the effect on the market had reversed. 

Garcia attributes the effect to the reaction of naïve traders, and they would benefit most from the results of his research. Naïve traders tend to react emotionally to pessimistic forecasts and sell immediately. Sophisticated investors, who react more rationally, take advantage of the naïve traders’ willingness to sell and snap up shares at a discount. Enough naïve traders anxious to sell push stock prices down.

Columnists’ opinions also affected trading volume. After particularly optimistic or pessimistic content was published, trading volume rose, Garcia found.

Reader sentiment was particularly strong on Mondays, the implication being that readers have had the weekend to read and digest opinion columns. Because the stock markets are closed on weekends, the newspapers would not contain any previously unreported stock market movement. 

Garcia cites studies by psychologists that show positive and negative emotions elicit different reliance on ways information is processed. He concludes that investors use different decision-making rules in bad times (recessions) than in good times (expansions).

But will columnists’ opinions continue to have such predictive effect? One explanation for the reader-sentiment effect harks back to the beauty contest theory put forth by economist John Maynard Keynes in the 1930s. Keynes used the analogy of a fictional beauty contest in a newspaper, in which readers would win a prize if they selected the photo that would be voted most beautiful by readers. Successful entrants would not necessarily pick the photo they found most beautiful, but the one they thought most people would find most beautiful. 

The Keynesian beauty contest dynamic might come into play, given the changing media landscape.  As the number and types of news media and opinion sources flood the consumer market, it becomes more difficult to determine which sources most investors consume. Fifty to 100 years ago, readers had only a few respected news sources that they relied on routinely.

In the 1970s, financial columns began including a columnist’s byline, and Garcia has conducted another study on the effect individual columnists have on market activity and asset price. He also has examined how journalists write as a function of past market performance, finding that they exaggerate market losses but downplay positive market returns. 

Now that he has the capability of extracting data from the images in the archive, Garcia plans even more studies.

“Historical archives are phenomenal data sets,” he said. “They give us a great opportunity to ask tons of interesting research questions.”  

is an associate professor of finance at UNC Kenan-Flagler.