Use of Brokers in Experimental Real Estate Markets

Abstract: This paper is an investigation into the use of brokers in experimental real estate markets. We ask whether people will make the optimal decision regarding the use of brokers when selling their home, and whether they are able to price their home optimally when they place it on the market.

Introduction

In this paper we examine the problems a homeowner faces when trying to sell her home. Specifically, we ask the following two questions:

  1. Does a home seller make the optimal decision regarding the use of real estate agents?
  2. Does a home seller price their house optimally when putting it on the market?

Buying or selling a house is typically the largest financial transaction that a consumer will undertake in her lifetime. According to a recent survey by the National Association of Realtors the national median sale price of homes was $208,700 compared with $235,000 when a seller uses an agent (a 13% increase). They also find that 88% of sellers use an agent, 88% of buyers use an agent, and these numbers have been steadily increasing over the past two decades. Thus, the competitiveness of residential real estate agents is important to the welfare of the consumer.

Agents compete with each other on service and price. In the area of service, buyer’s agents can conduct searches on the Multiple Listing Service and help buyers place offers on a property. Seller’s agents can host open houses and other networking events to match buyers and sellers, help the seller set the asking price, and provide a variety of marketing services such as advertisements in radio, print, television, and online media. In the area of price, agents can compete by offering lower fees and rebates.

A 2007 report by the Federal Trade Commission and the Department of Justice on competition in the real estate brokerage industry notes that the data on commission rates are generally propriety information and not made available to the public. The only source of data on commission rates they could find were provided by REAL Trends, an independent real estate research firm based in Colorado. These data show declining average commission rates from 6.1% of selling prices in 1991 to 5.02% in 2005 (a 21% decrease). There was also a steep increase in median home prices over this time period from $153,925 to $230,059 (a 49% increase). The report concludes that while commission rates vary slightly with home prices and market conditions, they do not vary in proportion to changing home prices.

In light of this inflexibility to changing home prices, we use an experimental market to test whether subjects are capable of determining their best course of action under exaggerated market conditions.

 

Literature Review

Bernheim and Meer (2007) study the Stanford area housing market over a 26-year period to find how much value real estate brokers add for the seller. Measurements include list prices, selling prices, and speed of sale for the homes. They found selling prices for brokered homes were 32% higher than non-brokered homes, however when controlling for home and seller characteristics (home with a pool, sq. footage, length of time lived in the house, for example), there is no statistically significant relationship between using a broker and sale prices. Likewise, brokered homes do have significantly higher asking prices than non-brokered homes, but when the same variables are controlled for they find no evidence that brokers affect initial asking price. Brokered homes did sell 34.7% faster than non-brokered homes, an effect that persisted after controlling for home and seller characteristics. They find that hiring a broker increases the probability of sale in the first month by about 25%, and the probability of sale in the second month by about half that figure.

Levitt and Syverson (2005) conduct a similar analysis of the housing market in Illinois, but focus on how information asymmetries between the homeowners and brokers leads to market distortions. Real estate agents are typically better informed about the value of the house and the state of the market than the seller, and they also bear much of the cost of selling the home while making only a small fraction of the purchase price. This results in misaligned incentives – where the agent wants to sell the house as quickly as possible, and may try to convince the seller to accept a substantially lower offer on the house. To illustrate this point, they find that homes owned by real estate agents sell for 3.7% more than other houses, and stay on the market for 9.5 days longer.

Salant (1991) perceived that nonstationarities inherent in the housing market may result in an optimal strategy where the subject initially chooses to sell the house themselves, and later decides to employ an agent if the house stays on the market late into the selling season (buyers with families typically don’t want to move during the school year). He develops a nonstationary search model to determine if a seller should enlist the services of a broker, and in doing so after unsuccessfully attempting to sell the house themselves whether or not to re-price the house. Employing a broker allows the seller of the house to sample potential buyers more rapidly. This would require two distinct pricing models – one for the FSBO phase, and one for the broker phase. The difference between Salant’s two models is the probability that one buyer is sampled during a period – assume that the probability of sampling a buyer in a period is always higher when an agent is employed. Asking price should decline within each phase because the expected value of continued search declines over time, but the price may increase between the phases to offset the added cost of the commission. This increase between phases would never rise above the cost of the commission because then the homeowner would simply use a broker from the outset, charge a higher price, and sample potential buyers more rapidly.

Experimental research on search tasks suggest that subjects use heuristics and not an optimal stopping rules when deciding when to end the search. Schunk (2009) finds, in trying to explain why sellers tend to stop prematurely, that decision heuristics subjects use in search tasks are related to loss aversion, and not individual risk attitudes. Utility functions are evaluated for each subject on the gain and loss domains in the manner introduced by Abdellaoui (2001). After their evaluation they participated in an unlimited number of practice search tasks before moving on to a set of 10 or 11 payment-relevant search tasks. One such task involved purchasing an item with a value of €500, and they are told about the distribution of buyer values which was truncated normal distribution. Subjects could choose to accept or reject each price draw, or choose to accept a previously rejected offer. The authors allow at the end of the paper for the possibility that results could change if sellers weren’t informed about the distribution of buyer values.

A number of studies can be found in the literature on a middleman’s role in bargaining models and search markets with incomplete information. Biglaiser (1993) shows that a middleman improves welfare in a general bargaining model when adverse selection is present. Rubinstein and Wolinsky (1987) describe the role of a middleman in a time-consuming bilateral searching model as an intermediary that extra surplus in return for shortening both the seller and buyer’s waiting time before there is a transaction. Carroll (1989) compared the efficiency of markets with middlemen depending on the method of surplus extraction; a fixed-commission and a fixed fee. Levitt and Syverson (2005) and Mitchell (2011) argue that middlemen are beneficial for the fact that they have market information unknown to the sellers, but that the sellers can receive a better price if they know that information.

There is some research that investigates the cost and benefit rather than the role of the broker. Shy (2009) investigates in the conflict of interest introduced by the brokers. Since the selling and buying brokers split the commissions with their respective agencies, he argues that the 6 percent commission is a just incentive for the brokers to sell at a high price.

 

Model

We construct a model as an extension of the one developed in Salant (1991). In our model, the seller has a value for a house and moves through a sequence of periods attempting to sell the house in each period. In each period she decides whether to use buyer’s and seller’s agents, only a buyer’s agent, or no agents. She then sets an asking price for the house. She then encounters a number of buyers that are drawn from a Poisson distribution, and each buyer is assigned a reservation price for the house that is drawn from a truncated Normal distribution. If the highest reservation price of all the buyers is greater than or equal to her asking price, the house sells. If the highest reservation price of all the buyers is less than the asking price, the house does not sell and she advances to the next period. The seller has an endowment of cash in the first period, and is forced to pay a holding cost from the endowment in each period she does not sell the house. If she does sell the house the remaining endowment is added to her earnings, but the endowment is always zero in the last period.

The discounted value of the home to the seller is , for where are constants determined by the seller’s stage in the mortgage repayment. However, for our purposes, given that we are only concerned with the change in the value of the home over a short time period we can keep this value constant without fear of losing salient information. That is to say .

A seller’s objective is then to maximize their expected profits at each time period :

Where is the fraction of the price retained by the seller, is the asking price at , is the value in the current time period, is the endowment in the current time period, is the mean of the Poisson distribution that draws the number of buyers the seller encounters, and is the distribution of buyer reservation prices. This problem can be solved recursively by first maximizing the expected profits in the final period and then working backward to the initial period. From this the seller extracts the asking price at each time period that will maximize expected profits.

Now if we allow them to choose between hiring a buyer’s and a seller’s agent, hiring only a buyer’s agent, or selling the house on their own, the seller’s objective is to maximize expected profits over three possible decisions:

 

Experimental Design

We first ran a pilot session with 18 subjects to estimate how long it would take subjects to converge to the optimal decisions. We brought them into the laboratory, seated them at computer terminals, and gave them the task of selling an unknown item. Their decisions affected only their own earnings, there was no communication or cooperation between subjects; they interacted solely with the software. We will refer to the three treatments as NA, in which the optimal decision was to use no agents; SA, in which the optimal decision was to use only a buyer’s agent; and DA in which the optimal decision was to use both a buyer’s agent and a seller’s agent. Subjects were placed into 3 groups of 6, and each group only experienced a single treatment.

After observing that most subjects could converge after about 15 contracts, we ran a second session with 36 subjects. We divided them into 6 groups of 6, and each group experienced two of the three treatments so we could have all the different treatment orders. We gave them 45-minute blocks in each of the treatments to complete as many contracts as they could in the allowed time, and paid them at the end of the experiment based on the average profits of 5 randomly selected contracts.

 

Group

Block 1 (45 min)

Block 2 (45 min)

1

NA

SA

2

NA

DA

3

SA

NA

4

SA

DA

5

DA

NA

6

DA

SA

 

Each period consisted of three stages, each lasting 15 seconds. In the first stage, the subjects must choose whether to use a buyer’s agent and a seller’s agent, only a buyer’s agent, or sell the item themselves (these three choices are listed as radio buttons so only one selection is possible). When they have made their choices they click ‘Submit’ and move to the second stage. In the second stage they enter an asking price, and if they chose to use a buyer’s agent and a seller’s agent the seller’s agent provides a recommendation for what price to set. This price is the asking price that maximizes profit to the seller in the current period. When they have entered a price they click ‘Submit’ and continue to the third stage. In the third stage they are told whether they sold the item or not and how many buyers they encountered. If a contract is formed the period ends and the subject proceeds to the next period.

Parameter Selection

We had the same 10 house values for each of our three treatments. Each time a subject sold their house we drew a new house value at random. We initially planned to vary only the buyer flow rates between treatments, but the commission rate has too great of an impact on earnings so we weren’t able to achieve an adequate effect size in the DA treatment without increasing the variance of the distribution of reservation prices and the holding costs as well.

In the NA treatment, our buyer flow rates were 6 when the seller didn’t use agents, 9 when the seller used both agents, and 6.82 when the seller used only a buyer’s agent. The distribution of buyer reservation prices had a mean equal to the house value, and a standard deviation equal to the house value divided by 10. In the SA treatment, our buyer flow rates were 2 when the seller didn’t use agents, 9 when the seller used both agents, and 8.15 when the seller used only a buyer’s agent. The distribution of buyer reservation prices was the same as the NA treatment: a mean equal to the house value, and a standard deviation equal to the house value divided by 10. In the DA treatment, our buyer flow rates were 1 when the seller didn’t use agents, 9 when the seller used both agents, and 2 when the seller used only a buyer’s agent. The distribution of buyer reservation prices had a mean equal to the house value, and a standard deviation equal to the house value divided by 8. We made the holding costs higher in both the SA and DA treatments, because with the higher flow rates they would sell earlier in a period and therefore be able to collect a larger remaining endowment.

The table of all parameters is displayed in Appendix 1.

Subjects’ Information

We provided subjects with visual information about the market they were in. This entailed showing them distributions of contract prices and the stage in which the unit was sold, as well as the percent of units that sold overall. These were 100 contracts drawn randomly from our Monte Carlo simulations, in which our simulated sellers were always pricing optimally. While this information is conceivably available to real-world sellers, we also broke the distributions apart by the seller’s decision regarding agents, which is perhaps a less realistic assumption. These images are displayed in the screenshots in Appendix 2.

Results

We first see how often a subject made the each decision in a period, and classify each period by the most frequently made decision in that period. We then sum those classifications over the entire treatment to see how frequently a subject made each decision over the treatment.

Discussion

Conclusion

This paper serves as a preliminary attempt to evaluate the behavior of sellers in real estate markets.

Treatment

Value

?s

?r

?rs

Mean

St. Dev.

Endow.

Hold Cost

NA

1600

6

9

6.82

1600

160

66

11

NA

6800

6

9

6.82

6800

680

270

45

NA

1800

6

9

6.82

1800

180

72

12

NA

7800

6

9

6.82

7800

780

312

52

NA

8200

6

9

6.82

8200

820

330

55

NA

2200

6

9

6.82

2200

220

90

15

NA

2400

6

9

6.82

2400

240

96

16

NA

9000

6

9

6.82

9000

900

360

60

NA

2000

6

9

6.82

2000

200

78

13

NA

9200

6

9

6.82

9200

920

366

61

SA

1600

2

9

8.15

1600

160

96

16

SA

6800

2

9

8.15

6800

680

408

68

SA

1800

2

9

8.15

1800

180

108

18

SA

7800

2

9

8.15

7800

780

468

78

SA

8200

2

9

8.15

8200

820

492

82

SA

2200

2

9

8.15

2200

220

132

22

SA

2400

2

9

8.15

2400

240

144

24

SA

9000

2

9

8.15

9000

900

540

90

SA

2000

2

9

8.15

2000

200

120

20

SA

9200

2

9

8.15

9200

920

552

92

DA

1600

1

9

2.00

1600

200

96

16

DA

6800

1

9

2.00

6800

850

408

68

DA

1800

1

9

2.00

1800

225

108

18

DA

7800

1

9

2.00

7800

975

468

78

DA

8200

1

9

2.00

8200

1025

492

82

DA

2200

1

9

2.00

2200

275

132

22

DA

2400

1

9

2.00

2400

300

144

24

DA

9000

1

9

2.00

9000

1125

540

90

DA

2000

1

9

2.00

2000

250

120

20

DA

9200

1

9

2.00

9200

1150

552

92

 

 

 

Bibliography

Bernheim, B. Douglas, and Jonathan Meer. 2007. “How Much Value Do Real Estate Brokers Add? A Case Study.” Working Paper.

Biglaiser, Gary. 1993. “Middlemen as Experts.” The RAND Journal of Economics 212-223.

Caroll, Wayne. 1989. “Fixed-Percentage Commissions and Moral Hazard in Residential Real Estate Brokerage.” Journal of Real Estate Finanace & Economics 349-365.

Levitt, Stephen, and Chad Syverson. 2005. “Market Distortions when Agents are Better Informed: The Value of Information in Real Estate Transactions.” Working Paper.

Mitchell, Tara. 2011. “Middlemen, Bargaining and Price Informaton: Is Knowledge Power?” Working Paper.

Rubinstein, Ariel, and Asher Wolinsky. 1987. “Middlemen.” The Quarterly Journal of Economics 581-594.

Salant, Stephen W. 1991. “For Sale by Owner: When to Use a Broker and How to Price the House.” Journal of Real Estate Finance and Economics 157-173.

Shy, Oz. 2009. “Real Estate Brokers and Commission: Theory and Calibrations.” Federal Reserve Bank of Boston Working Paper.

 

Palmquist, Raymond. “Estimating the Demand for the Characteristics of Housing.” The Review of Economics and Statistics. Vol. 66 No. 3, p. 394-404.

Parsons, George. “An Almost Ideal Demand System for Housing Attributes.” Southern Eco-nomic Journal. Vol. 53 No. 2, p. 347-363.

Bernheim, B. Douglas, and Jonathan Meer. 2007. “How Much Value Do Real Estate Brokers Add? A Case Study.” Working Paper.

Biglaiser, Gary. 1993. “Middlemen as Experts.” The RAND Journal of Economics 212-223.

Levitt, Stephen, and Chad Syverson. 2005. “Market Distortions when Agents are Better Informed: The Value of Information in Real Estate Transactions.” Working Paper.

Mitchell, Tara. 2011. “Middlemen, Bargaining and Price Informaton: Is Knowledge Power?” Working Paper.

Rubinstein, Ariel, and Asher Wolinsky. 1987. “Middlemen.” The Quarterly Journal of Economics 581-594.

Salant, Stephen W. 1991. “For Sale by Owner: When to Use a Broker and How to Price the House.” Journal of Real Estate Finance and Economics 157-173.

Caroll, Wayne. 1989. “Fixed-Percentage Commissions and Moral Hazard in Residential Real Estate Brokerage.” Journal of Real Estate Finanace & Economics 349-365.

Schunk, Daniel. 2009. “Behavioral heterogeneity in dynamic search situations: Theory and experimental evidence.” Journal of Economic Dynamics & Control 1719-1738

Leave a Reply

Your email address will not be published. Required fields are marked *