Projection Bias in the Car and Housing Markets

Featured in print Digest

Projection bias causes consumers in the car and housing markets to make decisions that are overly influenced by the weather at the time of the decision.

Weather clouds people's judgment when it comes to buying cars and homes, according to Projection Bias in the Car and Housing Markets (NBER Working Paper No. 18212). If it's warm or sunny, they're more likely to buy a convertible. After a snowstorm, they're more likely to buy a four-wheel drive vehicle and, when it's cold, a black car or truck. Buyers pay more for a home with a swimming pool when it's hot than when it's cold.

These findings suggest a phenomenon known as projection bias: the tendency of individuals to exaggerate how much their future taste will be like it is today. According to authors Meghan Busse, Devin Pope, Jaren Pope, and Jorge Silva-Risso: "Many of the most important decisions that we make in life involve predicting our future preferences. Projection bias may limit our ability to make these predictions accurately." They show that "projection bias causes consumers in the car and housing markets to make decisions that are overly influenced by the weather at the time of the decision."

For example, when the authors look at data from roughly 20 percent of all new car dealerships in the United States over a period of eight years, they find that the percentage of convertibles purchased peaks in April in seven out of those eight years. According to conventional economic theory, that's rational: buyers get to enjoy the car all summer before cool weather sets in. But the authors also find considerable variation in the purchase of convertibles at other times of the year, depending on weather. For example, an abnormally warm week in November in Chicago results in a significant increase in the percentage of convertibles sold there. And when a clear sky gives way to a completely cloudy one, convertible sales fall.

Similarly, a snowstorm increases 4-wheel drive vehicle purchases by almost one percentage point. The authors caution that overall, sales drop during snowy times because people typically don't buy cars immediately after a snowstorm, but the sales of four-wheel drives drop less than sales of other types of vehicles, increasing their share of all vehicles sold.

On the other hand, a 20-degree rise in temperature is associated with a 0.26 percentage point decrease in sales of black vehicles (a 2.1 percent change relative to the norm). Weather changes that affect sales volume also affect sales prices, but not much. For example, a 20-degree temperature rise during the week boosts the price of a used convertible by an average of $79.60, which is modest compared with the average transaction price of $22,222.

Weather patterns also affect sales volumes and prices of certain types of homes. When the authors study data on more than 4 million single-family homes across the United States that were sold at least twice between 1998 and 2008, they find that homes with swimming pools that sold in the hottest months of the year (June, July, and August) commanded a price that on average was 0.22 percentage points more than would be expected. By contrast, homes with pools that went under contract in the coldest months of the year (December, January, and February) sold for an average 0.18 percentage points less than would be expected. The average sale price of such houses was $398,000, so this represents a swing of about $1,600 between the warm and cold months.

The authors point out that buyers who purchase houses in mid-summer may under-estimate the delay between a contract and a closing, when the owner can actually move in. They note that: "The houses that we identify as selling in August are houses that will close in October -- meaning that the buyers of those houses will move in just at the point in the year in which swimming pool season is the farthest away."

Temperature also makes a difference: the authors find that houses with pools that went under contract in a month where the average daily high was more than 90 degrees sold for 0.37 percentage points more than when these same houses went under contract in a month whose average temperature was below 90. There is also similar though smaller seasonal variation in the prices of homes with central air-conditioning.

--Laurent Belsie