A New Way to Gauge the Housing Market
Ross finance professor finds a way to measure public sentiment on housing using news content analysis.
ANN ARBOR, Mich. — "Animal spirits," or public sentiment, has long been suspected to play an important role in housing prices, especially during the boom and bust from 2000 to 2011.
But there wasn't a good way to measure sentiment or how big a role it played — until now. New research by U-M Ross Professor Cindy Soo has uncovered a way to measure this sentiment (dubbed animal spirits in academic circles) across local housing markets.
Soo, who recently joined the Ross finance faculty, ran a text analysis of newspaper stories on housing in the markets covered by the Case-Shiller home price index. From there, she built a sentiment index that forecasts the boom and bust trend of housing prices by a two-year lead. That is, the sentiment peaked in 2004 before actual prices peaked in 2006. The same pattern held with the bust.
At a time when the fundamentals didn't account for the price swings, Soo's sentiment index explains 70 percent of the additional variation in national house price movements beyond what the observable numbers could explain. It also corresponds to patterns of homebuyer expectations in surveys by Case-Shiller and the U-M's Survey of Consumers, which also peaked in 2004.
"The newspaper medium is a channel for people's voices, and it reflects and amplifies their beliefs," says Soo. "Since we can measure it and show that it plays an important role in economic outcomes, it probably deserves greater attention. It might be a useful indicator for policy makers."
Soo combined prior work from other studies in behavioral finance and computer science to develop this index and empirical test. Text analyses that captured the qualitative tone of news coverage had been used in the past to measure stock market sentiment.
The housing bubble, and its burst, made for a good laboratory since the market fundamentals of supply and demand alone didn't explain the prices after 2000. Sentiment played an economically important role in the housing bubble, her research suggests.
Monitoring a sentiment index based on a text analysis of news coverage could be a way to help predict a market, especially when it starts to exceed or dip below what the observable numbers can explain. It's faster and less costly than a survey and can measure sentiment in both local markets, and a larger, national aggregate.
"People are starting to pay more attention to the role sentiment plays in markets, and we're coming up with ways to measure that empirically," Soo says. "It might be important in some cases to monitor newspapers and news coverage in particular."
For more information, contact:
Terry Kosdrosky, (734) 936-2502, firstname.lastname@example.org