Who You Going to Call? Performance of Realtors and Non-realtors in a MLS Setting

In this study we are not looking at for-sale-by-owner properties or broker properties that were not listed on the MLS, but are examining whether brokers listing on a MLS provide a benefit to clients if they join NAR and obtain the REALTOR membership.

See:http://​www.​realtor.​com/​Basics/​AllAbout/​Realtors/​Why.​asp?​poe=​realtor

http://​www.​realtor.​org/​PublicAffairsWeb​.​nsf/​Pages/​RealtorBrand?​OpenDocument

http://​realtytimes.​com/​rtapages/​20050826_​realtorbrand.​htm

The initial data sample had 125,923 single family listings. Due to missing values and extreme variables that were considered to be data entry problems, the final data set has 116,596 single family listings.

These groups closely follow categories used by NAR. “The 2005 NAR Member Profile shows that those REALTORS^® working in real estate for two years or less earned a median income of only $12,850 in 2004. By contrast, the more experienced NAR members who have been in business for six to 10  years earned a median income of $58,700 in 2004. And those with at least 26 years of experience earned $92,600, up 37 percent from two years earlier.” Quote from: REALTOR Magazine Online: “Rookies Work More, Earn Less,” by Haley M. Hwang, January 3, 2006. We elected to roughly follow their categories, dropping down to 5 years for experienced agents and not differentiating those with 26 or more years of experience. Models were also estimated with experience and experienced squared instead of the dummy variables. The coefficients for non-realtor listings did not change.

Days-on-the market may be larger than our measure. We do not have repeated listing information and thus our calculation of DOM may be less than the actual time on the market. This problem is common in the empirical studies we are aware of.

We classified 13,379 listing agents in this process. We expect there might be some classification errors given that any point in the year a broker might elect to later join NAR or to drop their NAR membership. We did not repeat the classification each month.

Licensed Broker identifies an actual broker not a listing or selling agent.

Otherwise, we use broker as a generic term throughout the paper to represent a salesperson, an agent or an actual Broker.

The Heckman selection model corrects for selectivity bias by adjusting the conditional error terms using the Inverse Mills Ratio (IMR) so that the conditional error terms will have zero means.

In an attempt to see if our results were sensitive to the estimation procedure, we estimate a multivariate probit for sold and non-realtor jointly, and then calculate the IMR for the non-realtor model and include this IMR and the non-realtor variable in the selection equation of a Heckman model with selling price as the dependent variable. In addition, we estimate the Heckman model with the selection equation conditioned on the choice of a realtor or non-realtor without including the non-realtor IMR above. In both cases, the results for the non-realtor coefficient remain approximately the same.

We split the sample into three relatively equal groups and estimated the selling price models. The highest priced (LP > $199,999) group had a coefficient of −3.2% significant at 10%, the middle group had a coefficient of −1.84%, significant at 5% and the lowest priced (LP < 130,000) group had a coefficient of +0.72%, that was not statistically significant. These results indicate that for lower priced properties a REALTOR or non-realtor agent obtain similar prices, with non-realtors obtaining increasingly lower prices across the middle and highest priced groups.

In an attempt to examine other possible reasons for the discount, we examine whether non-realtor listing agents systematically misprice their listings compared to REALTORS. The results indicate that there is no observable systematic mispricing by non-realtors compared to REALTORS. We also examined whether there was a Large Firm effect by creating a size dummy variable where 1 = firms with at least 2% of the listings in the market to examine if agents associated with large firms might be systematically related to the non-realtor variable. The results for the non-realtor coefficient do not change when we include the large firm dummy variable. An anonymous referee suggested that professional designations might impact the results. We agree that a study on levels of professional designations attained through education and the impact on price, time on the market and probability of sale would be interesting. While we do not have the data to examine this issue in our dataset, recent research by Ford and Rutherford (2003), in a technical report for the Real Estate Center at Texas A&amp;M using 59,599 sales during 2002 indicate that if a listing agent had at least one of the certifications (CRS, GRI, E-Pro), this was associated with a statistically significant and marginally higher sales price of approximately 0.54%. Using the estimate from their results, that would still leave approximately 2.5% as a premium for REALTORs compared to non-realtors.