Internet Car Retailing

Summary of working paper 7961
Featured in print Digest

Buyers who use an Internet referral service save about 2 percent on their average new car purchase.

Though Internet shopping has been extolled for its convenience and its ability to deliver vast quantities of information to shoppers, the ongoing debate over how clicks will affect bricks in traditional retail channels has remained a largely speculative one. In Internet Car Retailing (NBER Working Paper No. 7961), Fiona Scott Morton, Florian Zettlemeyer, and Jorge Silva Risso find that buyers who use an Internet referral service save about 2 percent on their average new car purchase.

Car buyers using an Internet referral service submit a purchase request specifying the kind of car they want and the time frame in which they expect to make their purchase. The referral sites have contracts with one or two car dealerships in each geographic area. They send the consumer requests to their affiliated dealerships in the consumer's area. In return, the dealerships pay either an annual fixed fee for referrals or a combination of annual fee and a fee for each referral. The Internet sales person at the dealership then contacts the potential buyer with a non-binding price for the car in question. In theory, a buyer needs to visit the dealership only to pick up the car. Referral sites control quality using consumer satisfaction surveys and by tracking the number of referrals that result in sales. Dealerships with conversion rates that the referral service deems too low, or who generate large numbers of consumer complaints, may be dropped from the referral network. Emphasizing the conversion rate gives dealerships an incentive to offer an attractive initial price to customers referred by the Internet service. is a major online auto referral service. The authors merge the 2 million purchase requests in 1999 with a J.D. Power and Associates sample of every new car purchase from 1101 California car dealerships from January 1, 1999 to February 28, 2000. The J.D. Power data contain "customer information, the make, model and trim level of the car, financing, trade-in information, dealer-added extras, and the profitability of the car and the customer to the dealership." The car is considered an Internet purchase if the name and address on the request match the name and address in the J.D. Power sample. By this standard, referrals constituted 2.9 percent of the 360,255 purchases in the J.D. Power sample.

After defining different "cars" as every combination of make and model, body type, number of doors, trim level, 2 or 4 wheel drive, transmission type, number of cylinders, displacement, and model year, the authors drop "cars" with very few observations (fewer than 500 observations and less than 1.5 percent of their market segment). This reduced the number of observations by 46 percent to 195,772, and resulted in a dataset containing 204 different kinds of "cars" sold through 810 dealers.

The dealerships that have contracts with sold an average of 767 cars over the time period studied. Non-Autobytel dealers sold 346. Thirty-three percent of the sample transactions went through dealerships that have contracts with People using the referral service were less likely to finance their cars through the dealer, to buy insurance through the dealer, or to buy a repair contract from the dealer. They also tended to live in areas with higher average income.

The authors find that the average Autobytel customer pays 2% less for his or her car; approximately one quarter of this price difference is due to purchasing at Autobytel franchise dealerships, which have lower prices than average. The remainder of the price difference is due either to savvy buyers choosing to use Autobytel, or Autobytel improving the bargaining position of standard buyers.

For auto retailers, the question of whether the Internet makes it easier for previously naïve buyers to educate themselves, or whether those who have always been savvy car buyers have migrated to the referral services is crucial; the first story would be expected to reduce their profits, while the second would not. The authors restrict the sample to less sophisticated buyers who obtain financing from the dealer and find that the Autobytel price discount remains for buyers referred via the Internet. In addition, the more cars a dealership sold through, the smaller was the observed spread in the prices consumers paid at that dealership. They conclude that these results suggest that Internet referrals increase buyer information and bargaining clout.

Although Internet car buyers pay less for their cars, the authors note it is possible that participating in an Internet referral service can be profitable for individual dealers. Using data from one Midwestern dealership, they estimate that the cost of a traditional sale could be as much as $600 higher than the cost of selling to an Internet referral. Given that the average Internet buyer pays only an estimated $300 to $500 less for his car, it is possible for an individual dealership's profitability to increase despite the lower purchase price.

-- Linda Gorman