【正文】
e and outoftown buyers to the prices paid for parable housing by more knowledgeable, local and repeat homebuyers. Their results indicate that home prices are similar across buyers with different information sets and search Costs. Since these were all brokerassisted transactions, Tarnbull and Sirmans conclude that existing brokerage institutions, such as the MLS, successfully eliminate the potential price effects of asymmetric information and, thereby, improve the efficiency of the housing market. It isnot possible, however, to determine from this study whether price differences exist between brokerassisted and nonbrokerassisted transactions. Although selling prices are not pared, a study by Baryla and Zumpano (1995), for the first time, uses a national sample of brokerassisted and nonbroker sales transactionsto assess the impact of intermediation by the agent on search effort. This study indicates that information asymmetries are present in the residential real estate market and that intermediation by agents does affect buyer search efforts. Firsttime buyers and outoftown buyers search longer than more experienced and local home buyers. Equally important, real estate brokers are able to reduce search time for virtually all classes of consumers, whether firsttime, experienced, local, or outoftown buyers. A followup study by Baryla, Zumpano, and Elder (1995) finds that the mechanism by which agents reduce buyer search duration is an increase in search intensity. Having more market access and housing information than buyers working without brokers, brokerassisted consumers are able to visit more homes during a given period of time. These results suggest that buyers with high information and search costs are more likely to seek out the services of real estate brokers. Whether this implication is true, and what effect such a search decision has on selling price, however, remains to be seen. The next logical step, therefore, is to examine this choice and its consequent effect on home prices. Data and methodology This study uses data from a nationwide survey of home buyers conducted by the 5 National Association of Realtors in 1987. The survey, The Home Buying and Selling Process, was mailed to over 30,000 households. After eliminating inplete or faulty questionnaires, the database totaled 2,495 observations, all of which took place in 1986. Table 1 displays summary statistics from the survey sample, categorized by whether the transaction was agentassisted or nonagentassisted. Table la displays agent usage based on differences in information and search costs among the buyers. Evaluating the impact of the real estate broker on prices paid by home buyers would seem, initially a straightforward process: One estimates a price equation using ordinary least squares, controlling for differences in housing characteristics (as in a hedonic price model) and the impact of buyer characteristics on the price, including a dummy variable defined to indicate whether the buyer used a broker in the transaction. Or one could estimate separate equations for the brokerassisted and nonbrokerassisted buyers. The difficulty is that the price paid by the buyer and the choice of whether to utilize a broker may be linked. A higher price paid by buyers using a broker may not imply that prices are higher f