The Effect of Internet Car Shopping on Prices and Discrimination

National Bureau Of Economic Research Inc, Cambridge MA

Investigators

Abstract

Internet referral services (IRS's) are web-based businesses that provide information about the characteristics and market prices of new cars to consumers, and allow them to request a price quote from an affiliated dealer in the region. If the quote is satisfactory, the consumer can buy the car from that dealer without further negotiation. IRS's sell subscriptions to one or two dealers in a region in exchange for giving the dealer a stream of customer leads. If the dealer does not sell to a sufficient proportion of the leads, the IRS may terminate the dealer and contract with another dealer in the area. Our dataset combines referral requests from Autobytel.com (the leading IRS) in 1999 with new car transaction data from J.D. Power and Associates that contain prices, car characteristics, and customer characteristics. We compare the prices paid by consumers who did use Autobytel.com and those who did not. We conjecture that online shoppers pay lower prices because the IRS "bargains" on behalf of its customers by being willing to move the group's purchases to another dealer. Also, the web site provides information about current market prices, and this may also reduce overcharging. In addition, the IRS may be contracting with more efficient dealerships that sell at lower prices to all consumers. We find that IRS users pay about 2% less for a given car than an "offline" consumer. However, in that analysis we do not formally distinguish between a causal role for the referral service and a selection effect. If customers who use the Internet are more educated, for example, we expect them to achieve lower than average prices regardless of the Internet. Alternatively, the Internet could be educating customers who were previously not informed so they pay lower prices. This is obviously a very important distinction both for social welfare and industry structure. We therefore to use two different econometric techniques to establish what the causal effect of an IRS is on the price that customers pay for new cars. A second goal of the expanded research program is to create an index of competition in the local dealer market to see if the IRS is "creating competition" in markets where there is little existing competition. The second project examines whether there is race and gender discrimination in new car prices. Since we have individual-level information on gender as well as on whether a consumer is Asian or Hispanic, we test if female, Asian, and Hispanic buyers pay above or below average for a new car. To analyze the prices paid by African-Americans we have no individual-level data and thus rely on the probability that a customer is black according to census block information. The final aspect of the project is the most exciting: to integrate the race and ethnic discrimination results with the use of the Internet. We plan to find out a) if minorities and women who use the Internet pay prices that are closer to those paid by whites and men, respectively, and b) if the Internet is causing the change. Our results have very important policy implications. If use of the Internet can reduce the effects of racial discrimination then the so-called "Digital Divide" is of even greater importance and concern. Our results may demonstrate the benefits of Internet usage to minority consumers, and inform the choice of policies to help disadvantaged groups get Internet access.