A group of researchers at UC San Diego and a Danish university predicts the county’s median home price will increase 8 percent by the end of the year to $810,000.
The prediction was included as part of a working paper that attempts to create a new model for forecasting prices that emphasizes online search activity as the most accurate predictor of price growth.
This story is for subscribers
We offer subscribers exclusive access to our best journalism.
Thank you for your support.
Allan Timmermann, finance professor at the Rady School of Management, worked on the paper with three researchers from Aarhus University in Denmark — Stig Møller, Thomas Pedersen and Christian Shütte.
The paper is still under review at the scholarly journal Management Science but was shared with The San Diego Union-Tribune ahead of publication. Timmermann said a new model for predicting prices is important for metro areas to provide early warnings of a possible bubble for owners, buyers and municipalities. He also said home prices have broader implications for the entire economy.
“Home prices really influence household spending, how optimistic they are for the future, if they choose to save or (are) more willing to consume,” Timmermann said. “It is such a large part of people’s wealth, much more than stocks.”
San Diego metro, which includes all of San Diego County, was not expected to have the largest jump of the 77 metro areas studied. It said Knoxville, Tenn., would have the biggest increase at 15.9 percent. Other fast-climbing metros were Austin (up 13 percent), Nashville (also up 13 percent), Bakersfield (up 11.7 percent) and Stockton-Lodi (up 11.3 percent). The metro areas with the smallest increases were New Orleans and Philadelphia, both up 2.7 percent.
There is no shortage of companies that have price predictors — such as Zillow, Goldman Sachs and Redfin — but the UC San Diego model isn’t proprietary. The general public can see how this model works (The U-T has included the full working paper at the bottom of the online version of this article.)
The 82-page paper contains mathematical formulas, accounts for variables, uncertainty and confidence formulas, and details sources of data, such as the Redfin Housing Demand Index. It used data going back to 2004 to compare online search times to home sales to prove its model works. The team worked on the paper for three years.
The formula starts by using Google Trends (a free website that lets users see what people are searching for on Google) to track keywords related to buying a home, and comparing it to hard data in real time, such as home tours and offers on properties. The paper argues its model using online activity is a surprisingly accurate way to predict prices. They were able to test online activity over three years to find a correlation to price growth that could be repeated.
A running theme of the paper is that online search activity can be a better predictor than the things housing analysts often put more emphasis on — mortgage rates, employment and access to credit. Researchers argue most housing data is focused on supply, such as how many homes are for sale in a given month, but rarely is demand tracked.
Timmermann said an exciting moment for the team was when they realized their formula still worked during 2020 lockdowns. They feared people stuck at home searching for homes with no intention of buying would cause numbers to go haywire but were shocked when it still provided an accurate prediction.
“That was a big surprise,” he said. “I was skeptical the relationship (between price prediction and online search activity) would hold.”
Online search activity is also being applied to other parts of the economy. A team of researchers at Johannes Gutenberg University Mainz in Germany, also using Google Trends, attempted to apply a similar model to the stock market. It wasn’t successful at predicting the stock price, but did show the number of times a company was searched for online was a good indicator of how much its stock would be bought and sold in a given week.
UC San Diego’s forecast says nationwide home prices will increase 5 percent by December. During the same time period, Goldman Sachs says home prices will be up 16 percent; Redfin, 3 percent; Freddie Mac, 5.3 percent; and Zillow, 17.3 percent.
The wide differences are part of the reason Timmermann said a more accurate predictor is needed. He said he wasn’t sure yet what will become of the UC San Diego model, such as an online tool or monthly report, but that it could be automated to continually update.
Download or read the full working paper below:
A group of researchers at UC San Diego and a Danish university have come up with a new model for predicting home prices that puts an emphasis on online search activity.
Get U-T Business in your inbox on Mondays
Get ready for your week with the week’s top business stories from San Diego and California, in your inbox Monday mornings.
You may occasionally receive promotional content from the San Diego Union-Tribune.
Spanish carmaker SEAT says it plans to build Spain’s first electric car battery plant as part of a 10-billion-euro investment in the country’s electric car initiative
A reactor at a nuclear plant in southern Sweden has unexpectedly shut down after suffering a fault in its turbine
Grocers in Hungary will be ordered to sell eggs and potatoes at prices no higher than they were at the end of September
German government blocks planned sale of chip factory to Chinese-owned company
The International Monetary Fund has reached a preliminary agreement with Bangladesh to provide $4.5 billion to help the country to weather disruptions caused by the war in Ukraine and manage risks from climate change
A nationwide strike over the cost of living increases caused by runaway inflation and the war in Ukraine has snarled traffic through much of Belgium and hit companies