The good, the bad, and the ordinary: Estimating agent value-added using real estate transactions

Despite the prevalence and high cost of real estate agents, there is limited empirical evidence as to the nature or efficacy of their services. In this paper we estimate real estate agents' value-added when either selling or buying homes using data from three large multiple listing services (MLS). We find that homeowners who forgo a conventional real estate agent, but who list their homes on the MLS via a flat fee broker, sell for between 1 and 4 percent more before commission but take longer to sell and are less likely to complete a sale. However, these average effects mask a significant amount of real estate agent heterogeneity. Using a novel aspect of our data, which allows us to identify and track agents over time, we estimate the distributions of real estate agents' fixed effects in both hedonic and time-on-the-market models. We find that exchanging a listing agent in the 25th percentile for one in the 75th would increase the final sales price by 5-6 percent, and a similar exchange for buying agents would lower purchase prices by 4-6 percent. The interquartile range of agent fixed effects from our model of time-on-the-market is 17-25 days. We do not find a significant trade-off between price and time-to-sell, however, as agents who obtain higher prices do not take longer to sell, suggesting that they are not simply setting higher reservation prices. We also show that agents who sell homes for more also appear to pay more for a home when serving as a buyer's agent, indicating that the average agent does not possess exceptional negotiating skills or that such skills are overwhelmed by principal-agent problems. Finally agents do not appear to get better at bargaining; agents do sell faster with experience, but mostly by selling for lower prices.