Minimum Service Requirements, Limited Brokers and Menuing of Services

Transcription

1 J Real Estate Finan Econ (2012) 45: DOI /s y Minimum Service Requirements, Limited Brokers and Menuing of Services Kimberly R. Goodwin & Ken H. Johnson & Leonard V. Zumpano Published online: 24 August 2010 # Springer Science+Business Media, LLC 2010 Abstract In the past few years, many states have responded to the increasing number of limited service brokers by passing minimum service requirements. Limited service brokers can be viewed as those brokers who are offering their marketing and representative services A La Carte as opposed to the more traditional full-services brokers offering of a Table D hôte (one size fits all) for their services. Supporters claim the legislation is necessary to protect consumers who are otherwise hurt by limited service brokers, but critics assert that this legislation is anticompetitive and not necessary. This study provides empirical evidence that sellers using limited service brokers experience a trade-off between a higher selling price and longer marketing spans with accompanying lower probabilities of finding a buyer during a given marketing period. Keywords Limited service broker. Minimum service requirements. Selling price. Time on market Introduction For the past few years, minimum service requirements (MSRs) have garnered the attention of the media and federal regulators. While real estate brokers have been K. R. Goodwin Department of Finance, Real Estate, and Legal Studies, The University of Southern Mississippi, 118 College Dr #5072, Hattiesburg, MS , USA kimberly.goodwin@usm.edu K. H. Johnson Department of Finance and Real Estate, Florida International University, SW 8th Street, Miami, FL 33199, USA L. V. Zumpano (*) Department of Economics, Finance, and Legal Studies, The University of Alabama, P.O. Box , Tuscaloosa, AL , USA lzumpano@cba.ua.edu

2 472 K.R. Goodwin et al. under scrutiny in the past for non-competitive behavior and price collusion, the debate has been rekindled as a response to the increasing use of non-traditional brokerage agreements. 1 The years between 2002 and 2006 witnessed an increase in the popularity of limited service brokers, as accelerating home prices drove up commission costs encouraging sellers to find alternative ways to market their properties. Many states, however, responded to this trend by passing legislation that precisely dictates the minimum level of service that all brokers, both full-service and otherwise, must provide. The Department of Justice and the Federal Trade Commission are both investigating these states MSR legislation with the claim that the legislation is anti-competitive and does not promote unrestricted trade. Advocates for the legislation claim that they are acting to protect the consumers, and MSRs insure that home sellers are receiving an adequate amount of information and service to provide protection in their dealings with prospective buyers as well as effective marketing of the property. For example, it is argued that minimum service requirements are especially useful for inexperienced sellers who may not fully understand the intricacies of the real estate market and how to maneuver through the process of negotiation and closing. Additionally, there is anecdotal evidence regarding the need for minimum service requirements. Numerous brokers working with buyers have reported that sellers using limited service brokers often ask them for advice about price negotiations, closing, and for explanations of contractual terms. 2 Providing such advice to these sellers, however, could easily violate the contractual and/or agency relationship a broker may have with the buyer. Proponents of MSRs say these stories illustrate how sellers choosing limited service brokers to list their property may not be getting the information and service they need. If a listing broker does not review purchase documents with the seller, then a seller may be signing away important rights, making needless concessions, and getting far less than market value for a property. Additionally, if true, one might initially suspect that selling price would be lower and time on the market longer for homes listed by limited service brokers. Critics assert that minimum service legislation reduces competition, limits product offerings, and overall, hurts consumers by keeping brokerage fees high. Buyers and sellers may have to pay for services they do not want or need. Transaction prices may be higher, and brokers offering these non-traditional brokerage arrangements may be forced from the market entirely, further limiting consumer choices. While advocates and critics of minimum service requirements have been quite vocal about their concerns, neither side has anything more than anecdotal evidence to support their case. To date there is no conclusive empirical evidence making the case for or against minimum service requirements. While others have investigated subsets of limited service brokers, this paper makes the first examination of the economic impact of limited service brokers on selling price, time on market, and the probability of sale. In particular, limited service brokers are defined herein as brokers who offer their marketing and representative services A La Carte (where sellers pick 1 Though not technically correct, the term broker is used in this work to represent both the brokerage company and the salespeople within that company. The term agent is specifically avoided so as to avoid confusion over being an agent and agency duties. 2 These anecdotes are reported in many news stories, including the February 2006 edition of Realtor magazine in an article entitled Is Real Estate Anticompetitive?

3 Minimum Service Requirements, Limited Brokers and Menuing of Services 473 from a menu of desired or needed services) versus the more traditional and common full-service brokers who offer their services as a Table D hôte (where sellers are offered an all inclusive set of fixed service whether desired or not). Development of Limited Service Brokerage and Minimum Service Requirements Traditionally, real estate brokers have performed almost every service related to the sale of a home. They would take responsibility for all marketing efforts including listing the home on the MLS, pricing properties, placing advertisements, and holding open houses. Under this traditional, full-service brokerage arrangement, the real estate broker may also provide advice on staging the property, home inspections, and help with document preparation and contract closing. As compensation for these services, the broker would typically earn a percentage fee contingent upon the sale and closing of the property. It has become increasingly common, however, for some sellers to request some but not all of these services. Experienced home sellers may only want to have their home listed on the MLS but take care of any other marketing responsibilities on their own. Particularly in a strong housing market, many sellers may not feel the need to have the assistance of a full-service broker in marketing their homes. As home prices rose during the previous decade, driving up the dollar cost of brokerage fees, sellers have increasingly sought out less expensive and alternative arrangements with brokers. As the demand for more non-traditional brokerage agreements increased, the market for limited service brokers has also increased. While limited service brokers can be widely viewed as brokers that offer their marketing and representative services in some reduced form from that of full-service brokers, this definition is very unmanageable and difficult to analyze. More specifically, there is no one uniformly accepted definition of limited service brokerage: (a) services can be provided contingent upon success at either a fixed or percentage fee, (b) services can be provided not contingent upon success at either a fixed or percentage fee, (c) fees for services can be reduced, (d) fees can be itemized but not reduced, or some combination of these enumerated possibilities can all be viewed as forms of limited service brokers. All this makes it very difficult to define limited service brokerage, resulting in difficulty in analyzing limited service brokers impact on market outcomes and requiring a consensus of findings among numerous studies before the issue can be settled. One form of limited service brokerage is when brokers work under a fee-forservice or A La Carte menu-driven business model. Here the client can choose to purchase only the services he or she requires of the broker. Fortunately, this branch of limited service brokers is easily identifiable and thus very well suited for analysis. Concerns have been voiced that many consumers are drawn to the itemized fees presented in this model without fully understanding the complexities and legal requirements of real estate transactions. They may also underestimate the time and effort it takes to effectively market a property under this arrangement. Consequently, these sellers may not receive the best prices for their homes and take longer to sell, all else being equal.

4 474 K.R. Goodwin et al. The general response of the full-service residential brokerage industry to these changes has been advocacy of MSRs. State associations of Realtors have lobbied their state legislatures to mandate minimum service requirements for real estate brokers. Supporters say they provide critical consumer protection in the real estate transaction, which is growing in complexity and typically involves people s most valuable asset. A total of 21 states currently have laws requiring real estate brokers to perform a baseline of services, which usually includes presenting offers and assisting clients with negotiations. Both the Federal Trade Commission (FTC) and the U.S. Department of Justice (DOJ) have challenged the state minimum service laws in letters they have written to the states of Texas, Oklahoma, and Alabama. 3 In these letters, the FTC and DOJ assert that MSRs hurt consumers in two main ways. First, consumers who desire to maintain primary responsibility for marketing and negotiating the sale of their property will be forced to purchase a higher level of product service than they require, paying for services neither wanted or needed. Second, to the extent that limited service brokers provide a competitive constraint on the pricing of real estate professionals offering only a full-service option, all consumers involved in real estate transactions will be subject to higher fees. Additionally, the FTC and DOJ argue that there is nothing but anecdotal evidence of consumers being harmed by non-traditional brokerage agreements being offered by limited service brokers. Without any empirical evidence, the FTC and DOJ assert there is no basis for any argument in favor of consumer protection. If there was evidence proving harm to consumers, the FTC and DOJ suggest that the need for consumer protection could be met by disclosure requirements. Rather than reducing competitiveness in the real estate industry by passing MSRs, states could require real estate professionals offering non-traditional brokerage agreements to specifically delineate in writing those services the clients and customers will not receive and what other services may be available. The next section of this paper provides an overview of the varying state s minimum service requirements. Sections on the relevant literature, data, methodology, and empirical results then follow in order. The last section presents the conclusions and implications of this study. Overview of State Minimum Service Requirements Limited service brokers are commonly described as real estate brokerage firms that will place an owner s property in the local MLS, but provide no other services. In point of fact, the limited service brokerage industry is not so homogeneous or monolithic. Instead, limited service brokers generally allow consumers to pick needed and desired services while rejecting others resulting in a general reduction in fees as not all consumers will pick the full basket of services traditionally offered by full-service brokers. 3 The letter from the DOJ and FTC to the Alabama Senate can be found here: public/comments/ htm. Comments by the DOJ and FTC to other states can be found here:

5 Minimum Service Requirements, Limited Brokers and Menuing of Services 475 Not surprisingly, minimum service requirements vary from one state to the other, with some states requiring limited service brokers to perform more services than is the case in other states. Currently, there are 21 states with minimum service requirements. 4,5 Of these, 8 states allow consumers to waive these extra services. The remaining 13 states with minimum service requirements mandate that all licensed brokers, both full-service and limited service, render certain services to consumers beyond listing a property in the MLS. Table 1 summarizes the requirements for those states with mandatory minimum service requirements. The variance in minimum service requirements runs the gamut from very extensive, where all brokers are required to help promote the best interests of clients and negotiate for their sellers, to, at most, accepting delivery of and presenting offers of purchase to their sellers. Even in states with no MSRs, all brokers, both fullservice and limited service, are commonly required by state laws or commission regulations to provide other services to sellers. For example, most states require that all licensees provide estimated closing statements to their sellers when presented with written offers from buyers. Review of the Literature Although hedonic modeling has been around for nearly 100 years, Rosen (1974) is considered to have laid the basic foundation for current hedonic pricing models in real estate. The first papers to link price and marketing time of properties are by Cubbin (1974) and Miller (1978). Haurin (1988); Geltner et al. (1988); Larsen and Park (1989); Asabere et al. (1993); Mantrala and Zabel (1995); Forgey et al. (1996); Haag et al. (2000); and Knight (2002), are among the many works that investigate either property price, property marketing spans, or both. Sirmans et al. (2005) provide a good review of the hedonic pricing literature and the often conjoined determination of property marketing spans. More recent research has looked at how broker compensation influences selling price and marketing time Munneke and Yavas (2001); Allen et al. (2003); Johnson et al. (2008). Other areas of investigation include, but are not limited to, the effect of different types of listing contracts on selling price and duration Rutherford et al. (2001) and whether individual broker characteristics can significantly affect the transaction outcomes of price and marketing span (Turnbull and Dombrow, 2007). Papers explicitly attempting to model the impact of broker intermediation on selling price and time on the market include Zumpano et al. (1996), and Elder et al. (2000), Rutherford et al. (2005) and Huang and Rutherford 4 The District of Columbia (D.C.) is reported as a state in this present study. The district has its own licensing requirements, is represented by its own Association of Realtors within NAR, and is at the doorstep of the Federal government; therefore, it seems reasonable to account for this area independently in this study. 5 While there is no mandate that MSR only cover services to sellers, Alabama s, the sampled state, MSRs are aimed exclusively at sellers. Anecdotally, this seems to be the case in most states. Omission of buyers from MSRs could possibly lead to some inequities. However, this issue is beyond the scope of this present study.

6 476 K.R. Goodwin et al. Table 1 Summary table of non-waivable minimum service requirements by state AL DC IA ID IL IN KS MO OR TX UT WA WV Accept delivery/present offers X X X X X X X X X X X X X Answer questions X X X X X X X Assist communications/negotiations X X X X X Provide access to listed properties X Accept deposits X X X X Promote best interests of client X X X Disclose all adverse material facts X X X X Use reasonable skill/care/good faith X X X X Provide advice to client X X X Continuous good faith effort to find X buyer unless otherwise agreed Present written contract with all terms and conditions X This Table summarizes the minimum service requirements for states with non-waivable minimum service requirements as noted by the Department of Justice. Delaware, Florida, Nevada, New Mexico, Ohio, Pennsylvania, Tennessee, and Wisconsin have waivable minimum service requirements and are not include in the table. (2007) and are worthy of additional coverage as this present work is essentially a brokerage intermediation paper. That is, it seeks to determine if there are different transaction outcomes (price, time, and likelihood of a transaction) that can be indentified between full-service broker and limited service brokers. Zumpano et al. (1996) examinesellingpriceandtimeonthemarketfora national sample of residential real estate transactions with and without broker intermediation. They find that after adjusting for selection bias the presence of a real estate broker has no appreciable, independent influence on selling price of listed property compared to FSBO transactions. However, brokers do reduce buyer search time. In an extension of this research, the same authors (2000) find that the use of buyer brokers also has no discernable effect on selling price; but they were more effective at reducing buyer search time than traditional selling brokers or non-agent facilitators. Rutherford et al. (2005) empirically test the notion that asymmetric information causes homeowners to be at a disadvantage when it comes to price setting and negotiating. The study finds that broker-owned homes do not sell any faster than client-owned homes, but broker-owned homes sell at a 4.5% price premium compared to client-owned homes. 6 Huang and Rutherford (2007) look at differences in selling price and marketing time between properties marketed by Realtors and non-realtors. The results show that properties marketed by non- Realtors sell at lower prices, take longer to 6 While the authors do not directly control for marketing time in their hedonic pricing models, they do include a control variable for the impact of unsold properties.

7 Minimum Service Requirements, Limited Brokers and Menuing of Services 477 sell, and are less likely to sell at all during a given marketing period. In four different hedonic model specifications, they find that time on market negatively and significantly influences selling price. None of the studies cited here, however, while suggestive, assess the impact of minimum service requirements and limited service brokerage on transaction outcomes. A recent study by Levitt and Syverson (2008) comes close to accessing the influence of limited service brokers on transaction outcomes. Specifically, they looks at the impact of flat-fee brokers (an incomplete subset of limited service brokers) on selling price, marketing spans, and the probability of sale using data from three locations, Sacramento and Santa Cruz County in California and Cook County, Illinois. The authors find that in Cook County and Sacramento homes listed by flat-fee brokers take longer to sell and have a lower probability of sale than is the case for traditional full-service brokers. In Santa Cruz County, however, homes listed with flat-fee brokers have a higher probability of sale. Pricing effects are mixed, at best. In Sacramento, the sign on the flat-fee broker variable is positive and significant, while in Santa Cruz and Cook County flat-fee brokers sign as a positive and significant influence on price only in their most parsimonious models. In their fully specified model, which controls for housing characteristics, there is no evidence of statistical influence. Besides these mixed results, there is some question as to whether using flat-fee brokers is a good proxy to represent limited service brokers. While some limited service brokers are paid using a flatfee compensation structure, not all flat-fee brokers are limited service brokers. Many flat-fee brokers offer full service for a reduced fee. For these full-service brokersthisisapricingandrisksharingstrategy to gain market share. This present study takes a narrower view of limited service brokers and excludes flat-fee, fullservice brokers in favor of brokers that offer a menu of marketing and representative services. Data The data for this study come from Montgomery, Alabama, a medium-sized metro area in the eastern part of the state. Alabama has enacted minimum service requirement legislation, but it is not particularly burdensome for brokers, whether full- or limited service, to comply with. Basically, all licensees in Alabama must present all written offers to their sellers and provide them with estimated closing statements. There is no explicit requirement that brokers must represent or negotiate for home sellers or buyers. All of the data is collected after the state s minimum service requirements were passed into law and spans from January 1, 2006 to July 30, The data collected for this time period contains a total of 11,704 total observations. Of these total observations, 8,228 are sold properties while 3,476 observations are for properties whose listings expired or have a hold or withdrawn status. In all cases, limited service brokers are identified by their firm affiliation. Specifically, four firms stand out by offering their service A La Carte. These four firms all actively market themselves along this business model suggesting that it is the preferred manner with which to market property and significantly

8 478 K.R. Goodwin et al. different from the more traditional and common full-service arrangement of other firms. 7 The MSA from which these properties are chosen is not particularly susceptible to large swings in value seen in many other cities during this time period. As a result, it does not seem necessary to add control variables for market conditions, macroeconomic variability, or seasonal variations. It does, however, seem prudent to remove outliers in the property characteristics. To maintain a reasonably homogeneous sample of properties, certain conditions are imposed on the collected data. Specifically, the properties are chosen from several contiguous areas designated by the MLS, and the composition of properties in these areas is reasonably similar. The data is filtered, however, to include properties that are between 1,000 and 4,000 square feet, contain 1 5 bathrooms, have 2 5 bedrooms, and are less than 90 years old to avoid potentially spurious loading of variability in the estimation of the three investigated transaction outcomes (price, marketing span, and likelihood of a transaction). Thus, the final analyzed sample contains 11,415 properties of which 8,051 of these properties are reported as sold. For the properties in the MLS reported as sold, it is known that there are significant inconsistencies in the way that brokers actually report the number of days marketed. Therefore, for the purpose of this study, the number of days a sold property is on the market is calculated as the difference between the listing date and pending property contract date. While the number of days of marketing for an unsold property is calculated as the difference between the listing date and the date at which the property was removed from marketing activity. To give the reader a better feel for the data, Table 2 defines the variables used in this study, and Table 3 provides the summary statistics for these variables. Methodology Recognizing the simultaneous nature and joint determination of the selling price and marketing time (Miller, 1978), the following generalized model specification is offered: SP ¼ ΩðX i ; L i ; Z i ; LSB i ; DbOM i Þ ð1þ DOM ¼ ΨðX i ; L i ; Z i ; LSB i ; SbP i Þ where X i is a vector of property-specific characteristics, L i is a vector of location control variables, Z i is a vector that accounts for the degree of overpricing (DOP) as ð2þ 7 Similar to the difficulties experienced by Levitt and Syverson, it is likely that some one-off limited service marketing arrangements were made by brokers in traditional full-service firms. However, these numbers are most certainly small and their presence in the pool of full-service brokers only serves to strengthen this present study s results in that but for their appearance in the full-service pool by some limited service brokerage arrangements significance levels would be greater than reported. Additionally, isolating and removing these observations is not feasible due to the nature of the definition of limited service brokers and corresponding paucity of the data.

9 Minimum Service Requirements, Limited Brokers and Menuing of Services 479 Table 2 Variable legend Variable Variable description LSB A dummy variable equal to 1 if the seller uses a limited service broker SP The home's selling price as reported in the MLS DOM Days of marketing until sold or otherwise removed from the market HBATH A dummy variable equal to 1 if the home has at least one half bath BATH1 A dummy variable equal to 1 if the home has one bathroom, otherwise 0 BATH2 A dummy variable equal to 1 if the home has two bathrooms, otherwise 0 BATH3 A dummy variable equal to 1 if the home has three bathrooms, otherwise 0 BATH4 A dummy variable equal to 1 if the home has four bathrooms, otherwise 0 BATH5 A dummy variable equal to 1 if the home has five bathrooms, otherwise 0 BED2 A dummy variable equal to 1 if the home has two bedrooms, otherwise 0 BED3 A dummy variable equal to 1 if the home has three bedrooms, otherwise 0 BED4 A dummy variable equal to 1 if the home has four bedrooms, otherwise 0 BED5 A dummy variable equal to 1 if the home has five bedrooms, otherwise 0 FIREPLACE A dummy variable equal to 1 if the home has a fireplace, otherwise 0 DRIVEWAY A dummy variable equal to 1 if a driveway is the best available parking at the home, otherwise 0 CARPORT A dummy variable equal to 1 if a carport is the best available parking at the home, otherwise 0 GARAGE A dummy variable equal to 1 if a garage is the best available parking at the home, otherwise 0 SSHOWER A dummy variable equal to 1 if the home has a separate shower, otherwise 0 GDNTUB A dummy variable equal to 1 if the home has a garden tub, otherwise 0 POOL A dummy variable equal to 1 if the home has a pool, otherwise 0 SQFT The amount of heated and cooled area in the home AGE The number of years since the home was built D022 A dummy variable equal to 1 if the home is in the zip code ending in 022, otherwise 0 D024 A dummy variable equal to 1 if the home is in the zip code ending in 024, otherwise 0 D025 A dummy variable equal to 1 if the home is in the zip code ending in 025, otherwise 0 D054 A dummy variable equal to 1 if the home is in the zip code ending in 054, otherwise 0 D064 A dummy variable equal to 1 if the home is in the zip code ending in 064, otherwise 0 D066 A dummy variable equal to 1 if the home is in the zip code ending in 066, otherwise 0 D067 A dummy variable equal to 1 if the home is in the zip code ending in 067, otherwise 0 D080 A dummy variable equal to 1 if the home is in the zip code ending in 080, otherwise 0 D092 A dummy variable equal to 1 if the home is in the zip code ending in 092, otherwise 0 D093 A dummy variable equal to 1 if the home is in the zip code ending in 093, otherwise 0 D104 A dummy variable equal to 1 if the home is in the zip code ending in 104, otherwise 0 D105 A dummy variable equal to 1 if the home is in the zip code ending in 105, otherwise 0 D106 A dummy variable equal to 1 if the home is in the zip code ending in 106, otherwise 0 D107 A dummy variable equal to 1 if the home is in the zip code ending in 107, otherwise 0 D109 A dummy variable equal to 1 if the home is in the zip code ending in 109, otherwise 0 D110 A dummy variable equal to 1 if the home is in the zip code ending in 110, otherwise 0 D111 A dummy variable equal to 1 if the home is in the zip code ending in 111, otherwise 0 D116 A dummy variable equal to 1 if the home is in the zip code ending in 116, otherwise 0 D117 A dummy variable equal to 1 if the home is in the zip code ending in 117, otherwise 0 FEDFUNDS The monthly average Federal Funds rate on the list date DOP The percentage deviation of list price from the predicted list price in a hedonic model

10 480 K.R. Goodwin et al. Table 3 Summary statistics for full and sub-samples Full sample Limited service brokers Full service brokers Mean Median Standard deviation Mean Median Standard deviation Mean Median Standard deviation SP 175, , , , ,950 66, , , ,082 LP 193, , , , , , , , ,337 DOM DOP 1.29% 0.86% 1.85% 1.30% 0.87% 2.82% 1.28% 0.86% 1.79% HBATH BATH BATH BATH BATH BATH BED BED BED BED FIREPLACE DRIVEWAY CARPORT GARAGE SSHOWER GDNTUB POOL SQFT 1, , , , , , AGE N 11,415 8,051 3,364 This Table provides summary statistics for the variables of interest in this study. Variable descriptions are located in Table 2.

11 Minimum Service Requirements, Limited Brokers and Menuing of Services 481 in Anglin et al. (2003), among others, and a market-swing control for market rates (FEDFUNDS). 8 The vector LSB i is a dummy variable defining whether or not the property is marketed by a limited service broker. Due to the simultaneous nature of the model, SP is the dependent variable from Eq. 1, while DOM is the dependent variable from Eq. 2. Obviously, DbOM in Eq. 1 is the vector containing the first-stage estimates of days of marketing that are then loaded into the second-stage pricing equation. This process is simply reversed for the duration estimate. The operational form of the models can be easily deduced from Tables 3, 4, and 5. 9 Recent studies, Rutherford et al. (2005) and Huang and Rutherford (2007), and Levitt and Syverson (2008) have begun to incorporate information from the distribution of sold and unsold properties into the determination of property price and marketing time. 10 These studies have the benefit of utilizing information from the entire set of offered properties rather than only concentrating on the subset of transacted properties, which typically is the case in previous studies. These studies recognize that information contained in unsold property data is important enough to be included in the analysis of price and time on market relationships. The current manner in which the unsold property information is included, however, diverges considerably. The output of the probability models constructed by Rutherford et al. (2005) and Huang and Rutherford (2007) is intended for use only as orthogonal proxy vectors for unobservable variables associated with properties that did not sell, i.e. the Inverse Mills Ratios. While Levitt and Syverson (2008) seem to implicitly argue that if a variable is worth specifying in pricing and duration estimates, it is reasonable to specify the same variables in the probability of a transaction model. Thus, none of the three works employ economic intuition into the specification of their likelihood of a transaction models. As a result, making any general inferences about property attributes that significantly influence the probability of sale is not easily accomplished. A collateral contribution of this present work is its economic intuition behind the probability model employed. The arguments for inclusion into the model follow. Zuehlke (1987), an oft-cited work in the literature utilizing simultaneous determination of property price and time on market, points out that property characteristics important to price determination may not be important in the determination of probability of sale due to marginal valuation differences between buyers and sellers over these property characteristics. Therefore, if buyers and sellers unobservable marginal characteristic values differ, a seemingly straightforward equilibrium process between price, probability of sale, and marketing span may become delinked. Anglin (2006) further argues that the relationship between property hazard function and property duration estimates may break down during periods of instability in the market. Given the likelihood that marginal values may 8 While the sample market is well known for its stability, specifying a market-swing control (FEDFUNDS) captures any variability caused by market condition to which rates are historically tied. 9 The validity of the instrumental variables is always a concern with 2SLS modeling. In order to address this issue, the Sargan test is performed. The results indicate that our instruments are uncorrelated to our residuals and are therefore satisfactory instruments. 10 Rutherford, Springer, and Yavas (2007) do not jointly estimate property price and time on market while examining the impact of broker-owned condominiums on these two metrics. However, the authors do incorporate information regarding unsold properties into their reported hedonic estimations. In addition, a similar work in the commercial literature by Fisher, Gatzlaff, Geltner, and Haurin (2004) investigates the determinants of transactional frequency.

12 482 K.R. Goodwin et al. Table 4 2SLS regression dependent variable is LnSP First-stage regression Second-stage regression Dependent variable: LnDOM t-statistic Dependent variable: LnSP t-statistic LSB b a 1.67 HBATH c c BED a a 1.74 BED c 2.97 BED c 3.80 BATH c BATH c BATH c BATH c 7.15 LNSQFT c LNAGE c c GDNTUB a c 6.34 SEPSHOWER c c POOL c 8.02 FIREPLACE c CARPORT a c 7.01 GARAGE c FEDFUNDS c DOP b 2.38 LNDOM c 7.58 CONSTANT c c N 8,051 8,051 R-squared This Table provides a 2SLS regression using robust standard error. Ln(DOM) is estimated in the first stage regression and entered into the second stage regression on Ln(SP). Variable descriptions are provided in Table 2. Location control variables are included but are unreported in the table. a, b, c denote significance at the 10%, 5%, and 1% levels respectively. Results for the variable of interest are highlighted. differ between buyers and sellers for at least some property characteristics and the likelihood of market instability, modeling the probability of sale for a given transaction may provide additional valuable information concerning property marketability. Drawing on the arguments offered in Zuehlke (1987) and Anglin (2006), the current study contends that explicitly modeling the probability of sale offers information not contained in property price or time on market models and is important and distinctive enough to warrant its inclusion as a third separate metric of transaction outcomes. However, without a well-established literature for modeling this particular metric of property marketability, the current study attempts to exploit the relationship between hazard functions and marketing time estimates as well as the relationship between buyers and sellers marginal valuations of property attributes, among other arguments, for assistance in constructing a parsimonious model of a successful marketing effort.

13 Minimum Service Requirements, Limited Brokers and Menuing of Services 483 Table 5 2SLS regression dependent variable is LnDOM First-stage regression Second-stage regression Dependent variable: LnSP t-statistic Dependent variable: LnDOM t-statistic LSB c 3.24 HBATH c 9.20 BED a 1.81 BED c 3.54 BED c 3.98 BATH c 6.69 BATH c BATH c BATH c 5.65 LNSQFT c LNAGE c GDNTUB b 2.10 SEPSHOWER c 8.86 POOL c 5.65 FIREPLACE c 9.90 CARPORT c 7.54 GARAGE c FEDFUNDS c c DOP c LNSP a 1.79 CONSTANT c c N 8,020 8,020 R-squared This Table provides a 2SLS regression using robust standard error. Ln(SP) is estimated in the first stage regression and entered into the second stage regression on Ln(DOM). Variable descriptions are provided in Table 2. Location control variables are included but are unreported in the table. a, b, c denote significance at the 10%, 5%, and 1% levels respectively. Results for the variable of interest are highlighted. Hazard functions and estimates of time on market have an inverse relationship by definition. Specifically, 1/λ is the estimate of marketing time, where λ is the hazard function. Assuming market conditions remain relatively constant through the sample period, DOM should have an inverse and statistically significant relationship with successfully marketed properties. Seller motivation can also be expected to impact those properties that do sell during a given marketing period. Glower et al. (1998) and Anglin, Rutherford, and Springer (2003) investigate the impact of seller motivation on property price and selling time. Anglin, Rutherford, and Springer, in particular, adopt the degree of overpricing (DOP) as a proxy for seller motivation. Since past studies have found a positive relationship between property marketing time and DOP, the a priori expectation is that a higher degree of overpricing should decrease the probability of selling the property during a given marketing period.

14 484 K.R. Goodwin et al. Finally, Zuehlke (1987) provides a convincing argument that buyers and sellers value certain proxies for property quality marginally different. When in agreement (as is the assumption for hedonic modeling), no effect should be witnessed in the likelihood that a property will transact but these proxy qualities will have shadow pricing effects on property price. However, when buyers and sellers are in disagreement over attribute values, the likelihood of a transaction will be influenced. Unfortunately, the coefficient estimates now only reflect relative movements in the marginal values of buyers and sellers and need to be interpreted across time. This will hold until the prices for these attributes once again reach equilibrium. It is in this manner that shadow prices may become delinked from probability effects. Therefore, the use of property quality proxies as controls in any probability of sale model seems logical. However, since unobservable differences in buyers and sellers marginal valuations of property attributes will vary across markets and time, it is difficult to identify a set of quality proxies valid for all property markets. Accordingly, and in order to enhance the exposition of the current work, property characteristics from the hedonic and duration estimates are included in the probability of a transaction model. Following the same line of reasoning, including location control variables seems justified. It is as reasonable to believe that if buyers and sellers can differ in their valuations of property attributes, they can also be reasonably expected to disagree over the value of location. Thus, the probability of a transaction within a given marketing period can be written as: ProbðSOLDÞ i ¼ pðdom i ; DOP i ; X i ; L i ; LSB i Þ ð3þ where DOM i is a vector controlling for marketing time, DOP i is a vector controlling for seller motivation estimated from the seller s degree of overpricing, X i is a vector of property characteristics (i.e. proxies of quality), and L i is a vector of location controls. The vector LSB i is a dummy variable defining whether or not the property is marketed by a limited service broker offering marketing and representative services A La Carte. Finally, SOLD i is the dependent variable representing whether the property sold during its marketing period. Empirical Results The results from the 2SLS regression of sales price is reported in Table 4. The results from the first stage regression also include control variables for the degree of overpricing and credit cost as proxied by the Federal Funds rate. Location control variables are included as instrumental variables because they have been shown to significantly influence time on market both empirically and in practice. In the second stage, the coefficients for property specific variables generally behave as expected. The presence of a half bath and a greater number of bathrooms increases the selling price. More square footage results in a higher selling price, and an older home results in a lower selling price. Better parking amenities (carport is better than a driveway and garage is better than both) provide a higher selling price. In addition, the presence of a garden tub, separate shower, pool, or fireplace

15 Minimum Service Requirements, Limited Brokers and Menuing of Services 485 also increases the selling price. The variable of interest in this study, LSB, somewhat surprisingly increases the selling price of property, on average. 11 This result is not what one would expect to find or at least not what anyone has anecdotally suggested to this point. One of the criticisms of limited service brokers is that they do not work to get the highest possible price for the seller. The results here indicate that employing the services of a limited service broker, on average, actually increases the selling price by approximately 1.9%. The 2SLS regression for time on market in Table 5 also provides some interesting results. As expected, properties in certain locations are shown to sell faster or slower than other areas. Results are omitted for reasons of exposition. An increase in the Federal Funds rate appears to decrease marketing spans on average. This may be explained by the fact that buyers may be more inclined to close a deal before rates rise again and increase their cost of borrowing. The DOP performs as expected with higher degrees of over pricing being associated with extended marketing times. Of particular interest in this study, the presence of a limited service broker significantly increases the time on market by approximately 22% (e ). Overall, Tables 4 and 5 seems to suggest a tradeoff when hiring a limited service broker versus hiring a full-service broker. More specifically, the presence of a limited service broker appears to produce a tradeoff between the two transaction outcome of property price and property marketing time with an expected increase of 1.9% in price being given up for an extension in marketing time of approximately 8 days. 12 This trade seems reasonable but it does not reflect other costs associated with failing to transact. The influence of unsold properties adds to the analysis. Specifically, Table 6 reports the results for a Logit model where the dependent variable is a dummy that equals 1 if the property sells during its marketing period and 0 otherwise. Consistent with earlier arguments DOM significantly decreases the likelihood of a transaction, while DOP has the correct sign but is not statistically significant. Buyers and sellers appear to be in general agreement over the marginal value of the property attributes as only property size, property age, the presence of a garden tub, half-bath, and two baths influence the likelihood of a transaction. Interestingly, however, the presence of a limited service broker significantly decreases the probability of a transaction during a given marketing period. Specifically, the odds ratio of.572 indicates an increase likelihood of not selling due to the presence of a limited service broker of 1.75 times greater than with other brokers. Thus, the tradeoff for hiring a limited service broker is far greater than an additional 8 days of marketing as suggested in the Summary Statistic exhibit. 11 Some might question the level of significance associated with LSB in Table 4, however if one considers that a Type II error (beta error), which is essentially a false negative is what is most important here, acceptance at the 10% level seems reasonable. More specifically, while there is strong evidence to support longer marketing spans and lower probabilities of a transaction, it seems unwise to not consider weaker but otherwise convincing evidence that a choice is being made by consumers. Failure to recognize the possibility of this tradeoff only supports one side of the argument, while recognizing this tradeoff suggests something entirely different than what has been proposed by the two competing factions is occurring. 12 This interpretive value comes from exponential of the LSB coefficient in Table 4 and adding 22% to the mean marketing time for all properties from Table 3.

16 486 K.R. Goodwin et al. Table 6 Logit model dependent variable is SOLD property Dependent variable: SOLD Odds ratio z-statistic LSB c LNDOM c DOP HBATH c BED BED BED a BATH c BATH BATH BATH LNSQFT c LNAGE c GDNTUB c SEPSHOWER POOL FIREPLACE CARPORT GARAGE CONSTANT c 7.13 N 11,415 Pseudo R-squared This table provides a Logit model for probability of a sale resulting from matching a buyer and seller in a given marketing period. Robust standard error is used. The dependent variable sold is a dummy variable that equals 1 if the property is listed as sold in the MLS and 0 otherwise. An unsold property may be listed as withdrawn, expired, or on hold. Variable descriptions are provided in Table 2. Location control variables are included but are unreported in the table. a, b, c denote significance at the 10%, 5%, and 1% levels respectively. Results for the variable of interest are highlighted. Specifically the model reveals that LSB dramatically lower the likelihood of a transaction during a given marketing period. Overall, this study provides evidence that the use of limited service brokers has a significant economic impact on property price, marketing spans, and the likelihood of a transaction during a given marketing period. Contrary to expectations, sellers employing limited service brokers actually get higher selling prices, on average. As far as dollar value is concerned, sellers are actually benefiting from the use of limited service brokers. As already noted however, there are also significant drawbacks to using limited service brokers. Sellers choosing to use a limited service broker are going to have their homes on the market for longer periods and face significantly lower likelihoods of a successful marketing effort. What may explain these results? One would normally expect that high income sellers (with expensive homes) and high opportunity costs would choose fullservice brokers. Evidence from prior studies has shown that high income sellers

17 Minimum Service Requirements, Limited Brokers and Menuing of Services 487 are more likely to hire brokers rather than try to sell their homes themselves. 13 However, when they have the option to choose a limited service broker instead of marketing the property themselves, the outcome may be different. For the same reasons, these sellers may also be more likely to choose full-service brokers over limited service brokers. On the other hand, sellers making less money and owning less expensive homes might prefer limited service brokers because commission costs would be lower. Limited service brokers are typically less expensive not because their fee are necessarily lower but because their fees are itemized and most sellers opting to employ limited service brokers do not choose the entire menu. 14 Lower income home sellers would likely also have lower opportunity costs and may be more willing to do some of the marketing themselves. If this type of selection bias is present, it could be expected that the selling prices associated with limited service brokers would be lower, not higher than for full-service brokers. Unfortunately, there is no available information about the seller that would allow us to test directly for the presence of selection bias. However, a comparison of selling prices and the choice of broker (Table 7) reveals that there are a higher proportion of lower priced homes listed with limited service brokers. Consequently, it seems that selection bias is not the explanation for the higher selling prices being associated with limited service brokers. Limited service brokers may also appeal to owners of very expensive homes in desirable neighborhoods who are unwilling to pay the very high dollar costs that would be generated by full-service commissions. 15 If this type of selection bias is present, the higher selling prices associated with the use of limited service brokers could be an artifact, reflecting the types of sellers that are most likely to use limited service brokers. Owners with high reservations prices and under no pressure to sell quickly may find limited service brokerage arrangements preferable to the use of fullservice brokers. If true, this could explain why homes listed with limited service brokers sell at higher prices, an outcome that may have absolutely nothing to do with the intermediation of different types of brokers. These sellers are simply willing to wait longer to preserve their asking price or even accept marketing failure. In effect, sellers employing limited service brokers may price their homes higher than the current market will pay, but they are willing to let the market catch up. Table 7 provides evidence to this argument in the form of decreased seller motivation. Specifically, DOP, which is inversely 13 See Zumpano, Elder and Baryla (1996) 14 In actuality, given that full-service commission are market driven to a certain market rate, the average fee from limited service brokers will be less if only one seller takes less than the full amount of marketing and representative services from a limited service broker. Thus, limited service brokers are less expensive than their full-service counterparts because of itemization and not due to brokers (such as discount brokers) discounting their fees. This is all speculation of course as listing side commissions are not available in the data for this study. 15 This might be the case for homeowners of modest means who have benefited from very rapid house appreciation.