Buying a Car on the Internet

Those who lack information about cars, or have personal characteristics that put them at a disadvantage in negotiations, benefit the most from the information and the low search costs provided by the Internet.

In Cowboys or Cowards: Why Are Internet Car Prices Lower? (NBER Working Paper No. 8667) and Consumer Information and Price Discrimination: Does the Internet Affect the Pricing of New Cars to Women and Minorities? (NBER Working Paper No. 8668) authors Fiona Scott Morton, Jorge Silva-Risso, and Florian Zettelmeyer explore whether people buying new cars through the largest U.S. online car buying service pay lower prices, how much lower those prices are, and whether women and minority buyers disproportionately benefit from transacting business online. They conclude that buyers do gain from using the Internet -- they save approximately 2 percent on average -- and Hispanic and African-American buyers save an additional 1.5 percent by using the Internet.

Using a large dataset of transaction prices for new automobiles and referral data from Autobytel.com, the authors are able to compare the online prices -- paid by 3.1 percent of the dataset of 671,846 -- and the remaining purchase prices, while at the same time controlling for characteristics of the cars and transactions. When the authors further control for who chooses to use Autobytel.com, the estimated causal effect of buying via the Internet increases to over 2 percent. This suggests that the consumers using this service disproportionately dislike bargaining and would have done poorly through the traditional channels; in other words, those who choose to use the Internet have the most to gain from it. The authors conclude that online buying could save Autobytel customers alone as much as $240 million per year.

The second paper examines the prices paid by different demographic groups using the buyer's last name as an indication of ethnicity and the buyer's first name as an indication of gender. The authors also use census data for race and ethnicity data at the neighborhood level.

They find that African-American and Hispanic buyers pay about 1.5 percent more for their cars than the equivalent white buyer, however most of this price premium can be explained with differences in income, education and search costs; they find no evidence of statistical discrimination. Within this dataset, that result holds even for those who did not get financing from the dealer. The result is not driven by minorities purchasing at higher-cost dealerships, but is explained partially by higher search costs faced by minority buyers. However, minority buyers who use the Internet referral service pay the same price as white consumers, despite the fact that dealers likely know the race or ethnicity of the buyer. Women pay about 0.5 percent more than men on average. Still, the authors conclude that those who lack information about cars, or have personal characteristics that put them at a disadvantage in negotiations, benefit the most from the information and the low search costs provided by the Internet.

-- Linda Gorman