UFDC Home  Search all Groups  UF Institutional Repository  UF Institutional Repository  UF Theses & Dissertations   Help 
Material Information
Subjects
Notes
Record Information

Full Text 
THE HOUSING MARKET AND REAL ESTATE AGENTS BY SURYAMANI MANTRALA A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA MAY 1992 Copyright by Suryamani Mantrala May 1992 I dedicate this dissertation to my parents Alapati Narasinga Rao and Sugunamani. ACKNOWLEDGEMENTS This dissertation could not have been successfully completed without the support and guidance of Professors Edward Zabel and G. their usefu Maddala. wish to thank them for comments, insights and faith in me which kept me on track whenever was faltering. would also like to thank Professors David Denslow, David Sappington, Lawrence Kenny and Stephen Donald for their invaluable comments and suggestions. am also greatly indebted to the illustrious economics faculty at the University of Florida, for contributing in one way or another to my development in the Ph. D. program. In particular shall cherish my association with Professor Steven Slutsky whose zeal for teaching and research has inspired me immensely. am grateful also to Professor Barton Weitz of the Marketing Department for being on my dissertation committee despite his extremely busy schedule. Many thanks are due to Pam Dampier for helping me with all my wordprocessing needs and to Dian Studstill and Joan Feuston for invaluable office assistance. Finally, could not have endured the Ph. D. program without continual support from my family. thank my parents Narasinga Rao and Sugunamani and Puma Rao and Satyavati and my brother and sisterinlaw Surya Rao and Sundari for their support and encouragement throughout my education. Syamala and brotherinlaw Ram for their confidence in me. also thank my sistersinlaw Annapurna, especially thank my mother I .. .1 S, A, a. Ba! a I aI ~ a lrr A~ a r llr Ir Ilr .1I. a J aa Laa taa Alr.& rnl:u,, J LLr ;= started or successfully concluded my doctorate. Last, but definitely not the least, the very presence of my lovely daughter Vidya kept me going during the final years. It is their pride in me today that is the greatest reward for me. TABLE OF CONTENTS PAGE ACKNOWLEDGEMENTS LIST OF TABLES ... ABSTRACT ...... CHAPTERS INTRODUCTION OPERATION OF THE HOUSING MARKET LITERATURE REVIEW THEORETICAL MODEL . S S U S S U S S S S S C S S C I S S S S S S S S C S S S S U S S S S S S S S S S S S S S U S C C S S S Review of Theoretical Models Model Description and Specifics Introduction . .. ation . . . . . . . .. . . 27 Specification of the Model with a Single Broker Specification of the Model with MLS Brokers EMPIRICAL MODEL AND ESTIMATION Review of an Empirical Model .. Model Description and Specification DATA AND EMPIRICAL RESULTS . Data Empirical Results SUMMARY AND CONCLU SIONS . . . . . 46 . . . . . . 5 6 . . . . . . 5 8 . . . . . . 6 4 APPENDICES MATHEMATICAL APPENDIX ANTITRUST ISSUES AND REAL ESTATE BROKERS .... ..... DEPARTMENT OF HOUSING AND URBAN DEVELOPMENT SETTLEMENT STATEMENT HUD1 FORM RESULTS FROM SEPARATE ESTIMATION OF EQUATIONS (23), (24) AND (25) . . . . . 75 REFERENCES *5 S S a a a S S S U S S S A S S S S S S S S S C 7P7~ B IOGR1APHIICAL SKETCH . . . . . . . . . LIST OF TABLES TABLES PAGE ESTIMATED REDUCED FORM TOBIT EQUATION FOR COMMISSION RATE .. .. ....... ESTIMATED REDUCED FORM PROBIT EQUATION FOR BROKERED SALE DUMMY . . .. 61 ESTIMATED REDUCED FORM EQUATION FOR PRICE .... .. ESTIMATED STRUCTURAL TOBIT EQUATION FOR COMMISSION RATE... ........... ESTIMATED STRUCTURAL PROBIT EQUATION FOR BROKERED SALE DUMMY .. .. .... ESTIMATED STRUCTURAL EQUATION FOR PRICE .... Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy THE HOUSING MARKET AND REAL ESTATE AGENTS BY SURYAMAN MANTRALA MAY 1992 Chairman: Professor Edward Zabel Major Department: Economics This dissertation studies the interaction of buyers, in the housing market. sellers and real estate agents Most of the current work on housing markets has ignored the importance of real estate agents. Buyers sellers have incomplete information about each other availability of houses. "marketplace." Spatial fixity of houses also implies the lack of an organized The main role of a real estate agent is to provide information and bring buyers and sellers together thus making the market. a commission. United States. For these services the realtor is paid Commission rates have been in the neighborhood of 6% across the This has led to charges of noncompetitive commission rates fixing. The present thesis disputes the argument of commission rate fixing. In a competi tive market with incomplete information characterized by a large number of buyers and sellers with little or no market power, a market maker' ability to charge high prices is by the market maker then tend to be very close to the competitive equilibrium prices. This implies here that the observed commission rates are kely to be the competitive rate of return for the real estate agents' services. present a theoretical model which considers the sellers, buyers and brokers of housing in an integrated manner. to be a function of the seller' The optimal commission rate set by brokers is shown decision rule for using a broker in selling his house. deriving outcome examine a variety models consider various implications of incomplete information. then develop an empirical model that simultaneously estimates the commission rates and probabilities of brokered sales of housing, states across the U n 540 housing transactions in fifteen My hypotheses regarding the commission rate are validated by this model. CHAPTER INTRODUCTION Expenditures on housing are a large component of household spending. They are also an important economic indicator in a major industrialized economy like the U Despite the obvious importance of the housing market, an analysis of how this market operates. little attention has been given to The focus in this dissertation is the interaction of various agents n the operation of this market. There are three kinds of agents that we will analyzethe seller of a house, the buyer and the real estate agent. Real estate agents provide a variety of services to both the buyers and the sellers of housing and charge a commission rate for their services , usually from the sellers. These services are mainly those providing information about available houses, prices, financing assistance n arranging matches between buyers and sellers. As producers and sellers of information about the housing market, real estate brokers perform the important function of "making the market" for buyers and sellers of housing. This marketmaking function of brokers has not been given the importance it deserves in the existing literature. When a seller wants to sell a house, he can do so via two channels(i) on his own and (ii) through a real estate broker by listing with a brokerage firm. The two channels of sale differ in the costs involved and may differ in the expected time on the market. the seller sells the house himself he has to ncur the cost of searching for buyers (e.g. advertising), showing the house to prospective buyers and the cost of holding the house until a sale occurs. If the house is sold through a real estate agent, on the other hand, the seller has to pay the broker a percentage of the sale price of the house as a commission. A rational seller will choose the better alternative. A buyer searches for a house which meets his requirements at the least price. this process of search, he too has various options(i) to go directly to individual sellers, (ii) to use the services of a brokerage firm or (iii) to build a house. Usually a buyer would use both alternatives (i) and (ii) for an already existing house since the costs of both alternatives are similarthe time costs of visits to a seller or the broker. Brokers produce and sell information to sellers and offer this information without charge to buyers of houses and aim to maximize returns from their activities. The objective of the present dissertation is to critically evaluate the theoretical and empirical work previously done on the housing market and suggest ways in which the analysis would be enriched when the importance of a real estate agent as a market maker incorporated in the analysis. In doing so an empirical model of the price making behavior in the housing market is developed. The rest of the dissertation is organized as follows. describe the operation of the housing market and the role of real estate agents in the next chapter. Chapter examines the existing literature. In Chapter examine a theoretical model. develop an empirical model and describe estimation procedures in Chapter V. Data sources and empirical results are discussed Chapter Finally, present a summary CHAPTER II OPERATION OF THE HOUSING MARKET The housing market involves interactions among three types of economic agents sellers wanting to sell houses, potential buyers and real estate agents who facilitate transactions by providing information about available houses to buyers and about the pool of buyers to sellers. For the function of providing information the broker charges a commission , usually from the seller. The seller of a house wants to sell his house at a reasonable price with minimum delay. This involves searching for buyers by spending time and money to attract them. A larger number of potential buyers implies a higher probability of getting a better price for any house and a higher probability of attracting a buyer within any time interval. seller may undertake this search himself or employ the services of a real estate agent. These two channels of search differ n the costs involved and the rational seller would choose the better alternative. A buyer's objective is to buy a house with a set of characteristics at a minimum price Thus a buyer searches over the set of available houses in the market. Buyers have the option of getting information from any or all of the following sourcesthrough advertisements, word of mouth or a real estate agent. A buyer may choose a real estate agent along with any of the other alternative sources since in most cases the cost of anhnaaan af +ka ra.,I aene%+a nrht.". b. kavna k.. *MrcA ..t'IIA 4 The main function of real estate agents is matching prospective buyers and sellers. This function assumes great importance in the housing market due to several distinguish ing characteristics of the commodity being traded which makes the existence of a usual marketplace impossible. These characteristics are (i) fixed locations of houses, heterogeneity of houses, (iii) durability of houses, (iv) infrequent transactions among buyers and sellers and (v) complexity arising from the financial and legal dimensions of the transactions. The function of real estate brokers is to collect and maintain information about the prospective supply of and demand for houses. The buyer search is thus expedited by such availability. This information is useful to the seller in determining (i) whether to sell the house and (ii) the initial asking price. buyers and sellers. The broker's information facilitates trade among There are several benefits of involving a real estate broker in the process of buying or selling a house. First, real estate brokers have more expertise and familiarity with general bargaining and specific real estate transactions. may have the potential ability to mediate differences between principals, Secondly, they which may be more difficult if buyers and sellers try to negotiate directly. Finally, the broker may be able to expedite the transaction by providing guidance and information about the settlement process. commissic For all the important functions listed above, , usually from the seller (see Crockett, the real estate agent charges a 1982). Commission rates charged by the estate agents are influenced by the characteristics of the market and the resulting behavior of buyers and sellers. From the perspective of both buyers and sellers, the commission rate represents an opportunity rn through an intermediary. High commission rates are also likely to buyers to negotiate directly with sellers as they hope to find lower prices than those listed with the broker Even though the seller technically pays the commission, the incidence may be on the buyer, implying that buyers have some prospect of obtaining a price between the price received by sellers who list with brokers and that paid by buyers. Thus, the ability of buyers and sellers to transact directly among themselves exerts pressure on the commission rates charged by the broker. When choosing commission rates, real estate agents must strike a balance between two opposing forcesi.e. providing a reasonable opportunity costs to buyers and sellers while ensuring a sufficient return on their own investments. Hence, the existence of potential traders outside the real estate agent' activities implies considerable competitive pressure on the choice of commission rates by marketmaking real estate agents. The operation of the housing market as described above provides an explanation for the empirical observation that commission rates do not differ much from region to region in the United States. These rates tend to vary only in a small interval around six percent with some variation depending on the price of the house. apparent uniformity has lead to concerns about price fixing by real estate brokerages and violations of antitrust regulations (People vs. National Association of Realtors, 1981 Owen 1977 Crockett, 1982; Miller and Shedd, 1979). Several Federal Trade Commission Reports have attributed the observed pricing behavior to the cooperative marketing arrangements by the local multiple listing services (see e.g. , 1984). Some recent studies have disputed this argument of price fixing as there are a large number competitive number competitive Association of Realtors (Crockett, 1982). In particular the characteristics of the housing market suggest that it is the competitive pressures exerted by the buyers and sellers who have the ability to transact among themselves, rather than those exerted by other brokers, that lead to near competitive commission rates. Jud (1983) has also observed that real estate agents lack any monopoly power in influencing the sales prices of houses. The argument that commission rates tend to be competitive, however, does not rule out some variation in commission rates due to differing costs incurred by the real estate agents on different transactions. depending on the costs of acquiring Commission rates may vary within regions information and characteristics of houses. extensive study covering over 7000 housing transactions in the United States in 1975, 1978 and 1979 was conducted by Carney. He uses an economic search model and various brokerage cost assumptions to derive brokerage pricing implications and argues that relative search cost differences mply that commission rates will be lower on (i) sales of higherpriced homes, (ii) sales of new relative to existing homes and (iii) noncoop relative to coop sales (Carney, 1982). Carney's findings are consistent with argument that commission rates are kept in check due to the competitive pressures because these variations are caused by genuine cost differentials rather than the power a real estate agent possesses as a market maker. The above explanation of the features of the rea estate market is similar in spirit to the analysis of securities exchanges using the specialist system provided by Bradfield and Zabel (1979) and Zabel (1981). The former introduces explicit price making behavior in competitive markets. The specialist n the securities exchange is a price maker (akin to our real estate agent). The existence of such a price making agent challenges the notion that all agents in a competitive market must be price takers. In an extension of this model, Zabel (1981) shows that even though a securities exchange specialist possesses some monopoly power in choosing the spread between the ask and the bid price of stock, this power is restricted by the prospect of traders trading among themselves within the spread. sales. This prospect induces the specialist to reduce the spread to avoid losing Similarly real estate agents' fees are limited by the competition among various brokers and also because sellers and buyers can deal directly with each other. Despite the similarities between the securities market and the market for real estate, importance differences also exist. The specialist guarantees buy or sell orders since he satisfies excess demand by trading for his own account (i.e. from the inventories he holds or by selling short). Also , in the securities market buyers and sellers do not have to search for each other and the commodity being traded (shares in corporations) is homogeneous much different. In the housing market the informational and availability aspects are 'S One of the important features of this market is the heterogeneity of the commodity being traded. The characteristics of houses are usually difficult to assess. Another uncertainty is introduced due to the difficulty buyers and sellers. n determining the availability of Thus it is essential to search for information about housing charac teristics and buyers and sellers. n this securities market. In most cases there is no buyer or seller of last resort Real estate agents only provide intermediation between buyers and sellers because holding of inventories in this market is costly and sellers have to bear the cost of holding a house while awaiting a sale. 8 Thus, in the real estate markets while competition leads to some uniformity in commission rates across regions, characterist of regional markets affect the decisions of buyers and sellers in regard to trading outside the market or with real estate agents. One of the important characteristics of the market which determines whether buyers and sellers trade among themselves or through real estate agents is whether the market is active or inactive. In a market with a large number of buyers, ., an active market, the likelihood of a seller finding an interested buyer is high and thus the cost of searching for buyers would be relatively low. estate agent would be low. To Thus the probability that a seller would list with a real a buyer the active market would thus indicate a situation where negotiating directly with the seller or seeking the services of a broker may be equally attractive. an inactive market, however, number of buyers is small compared to the number houses available thus requiring sellers to expend considerable amount of time and money in finding a prospective buyer Thus the likelihood of sellers engaging the services of a broker in such a market will be relatively high. Since housing is an infrequently traded commodity and buyers and sellers do not have the advantage of extensive experience, they might revise their decision about trading with the help of real estate agents. Initially, a seller may attempt to sell a house without intermediation but, as the burden of holding a house increases, may later shift to a brokerage listing. Thus a seller' behavior has a dynamic element attached to it. Similarly, buyers too may need to revise their decision to approach a broker after they have entered the market. prior work n this area reveal that either the relationship between agents (i.e. sellers, buyers and brokers) has been completely ignored, or modeled very inadequately. next chapter In the examine and review existing literature on theoretical as well as empirical issues in this area. CHAPTER II LITERATURE REVIEW In this chapter give a broad overview of research on the housing market. In the next chapter and in Chapter V consider some of the studies most relevant to my research in more detail. A recent survey of economic models of the housing market by Smith, Fallis (1988) suggests a shortage of analytica Rosen and and empirical work incorporating the extremely important role played by real estate agents in the market for housing. literature on models of the housing markets is divided according to which of the several special features of housing are emphasized. heterogeneity, durability, These special characteristics of housing are: spatial fixity and the extensive government involvement in the housing sector. almost impossible to build a realistic economic model incorporating all these features, each strand of the literature deals with one or the other However , models are now becoming more general and thus many of these strands overlap. Some of the earliest economic models started out ignoring most of the special characteristics of the housing market (Muth 1960; Olsen 1969) by assuming the existence of an unobserved commodity called housing service which takes care of the heterogene ity feature. Further, these studies dismiss any problems involving durability of housing   ~ Deaton and Muellbauer, 1980). These models also ignored intertemporal and spatial issues. Questions of interest were: How does an increase in income affect the price of housing services? How do increases in prices of building materials affect housing output and prices Early empirical models mainly study the estimation of price and income elasticities of demand for housing and the production function on the supply side. Leeuw 1971 , Quigley 1979 and Mayo survey demand side issues; supply side issues are discussed by Muth 1960, de Leeuw and Ekanem 1971 1983 and Bruce Smith 1983. Arnott and Griesen McDonald 1981 and Edelstien 1983 survey the literature on production functions.) Starting from these simple models several modifications are introduced to make analysis more realistic to study some of the special features of housing. introducing durability in these basic models makes the distinction between housing stock and housing services important. Durability is introduced into the simple models of housing explicitly by assuming that the market adjusts in a stockflow manner. These models assume that the shortrun supply of housing stock is perfectly inelastic. Then it is the demand for housing that determines the equilibrium price in this market. (See Duesenberry 1958; Muth 1960; Lawrence Smith 1969; and Olsen 1969 for a discussion of the stockflow models of housing.) macroeconomic models of the housing Similar stockflow models were also used in market. These are surveyed by Grebler and Maisel (1963), Fair (1972) and Fromm (1973). explanation for tenure choice, choice betw Durability of housing also presents an 'een renting and owning a house, as markets for housing stock (i.e. ownership) and for housing services (rental market) now 1981 analyses must now consider discrete choice of owning or renting as well as the continuous choice of quantity. Empirical analyses including durability mainly involve estimating expected user cost of housing which are used to test hypotheses about tenure choice (e.g., Diamond 1980). services Douglas The durability issue also focuses on the production process for housing ., owning a house implies home production of housing services and renting implies buying these services in the market (Weiss 1978). Spatial fixity another important attribute of housing which has received extensive attention in the urban and regional economics literature as it has important implications in studying effects of race and racial discrimination in urban housing markets. analysis of housing demand in this case includes location n the utility function. Utility maximizing households must choose location as well as quantity of housing. involving location choice issues are surveyed by McDonald (1979), The mode Wheaton (1979) and Henderson (1985). Heterogeneity of housing deters formation of an organized commodity market since accurate pricecharacteristics information is not readily available in this market. Information gathering involves expensive search on the part of buyers and sellers of housing, thus making real estate agents central to this market. However adequate attention has not been paid to this information generating function of real estate agents in the housing market. The surveys by Smith, Rosen and Fallis (1988) and by Muth and Goodman (1989) seem to suggest that studies which deal with the economics of housing markets do not has become customary to distinguish a housing market which deals with the demand for and supply of housing and a brokerage market which deals exclusively with the demand for and supply of brokerage services with little concern for the interaction of these markets. Zumpano and Hooks (1988) survey theoretical and empirical models on the market for real estate brokerage services. They present and critically evaluate past studies which attempt to explain the structure and performance of this market. the research has sought to market. Much of justify or challenge certain popular observations about this It has been charged that this market is inefficient, that it is monopolistic and most importantly that local multiple listing services lead to little competition and hence too high commission rates. Even this literature surveyed here does not provide any formal treatment of the alternative methods of search for individual buyers and sellers in the market for real estate. One of the firMt theoretical models of real estate broker behavior was developed by Yinger (1981). The primary goal of this study was to promote an understanding of the market for rea estate broker services under uncertainty. He formulates the problem as one of broker's search for buyers and sellers in order to maximize his income. Income is derived by the broker as a commission successfully making a match succeeding making a sale). Yinger derives equilibrium amounts of search commission rate and the expected number of matches through partial equilibrium analysis and suggests comparative static exercises for changes in the equilibrium values. looks at the effects of MLS on the variables of interest and studies the welfare and policy Among other conclusions, considering the entry of brokers, Yinger argues that entry shifts the supply of listings downwards and raises commission comparing the system with and without MLS, he concludes that MLS increases the average housing price, decreases resources devoted to broker search and increases brokers' income but that the effect on the commission rate is ambiguous. Finally, concludes that there may be substantial welfare gains from government intervention in this market. Recent work by DeBrock (1988, 1989) is also a theoretical analysis of the market for broker services. As in Yinger' study, DeBrock does not allow direct sales between buyers and sellers. In contrast to Yinger, DeBrock considers interaction among brokers to be a noncooperative game. He considers a twostage game in which prices are set in the first period and marketing effort occurs in the second period. He demonstrates that under MLS the BertrandNash outcome is that the prices of houses are bid up (possibly to the monopoly level) and not down to cost (this conclusion is exactly the opposite of the usual BertrandNash outcome). leads to an unambiguous He also shows that MLS increase in commissions and an increase in the pricing of houses with a decrease in the return to the house owner. From the point of view of efficiency, he states that the MLS system may attract too much or too little cobrokerage efforts thus leading to lower than maximum industry profits. general notion that MLS is a beneficial of view of increased information, is inco Thus he concludes that the institution in the housing market, from the point rrect in that other effects negate this positive one. In contrast to Yinger DeBrock, Wu and Colwell (1986) introduce direct equilibrium) and simultaneously (general equilibrium). equilibrium in the housing market, an For example, considering partial increase in the commission rate has no effect on total housing supply but reduces the quantity and price of housing. In the general equilibrium analysis an increase search listings decreases commission rate but the effect on the price of housing is inconclusive. Similarly, increase in the cost of search for buyers decreases price of housing with an inconclusive effect on the commission rate. n their analyst , they conclude that an MLS brokerage system wil increase the price housing, , like Yinger , they too conclude effect MLS commission rates is ambiguous. MLS each broker wil Contrary to Yinger, Wu and Colwell conclude that with devote more resources to search than an independent broker. Most of the claims by Yinger, DeBrock and Wu and Colwell, some conflicting, have not been subjected to empirical testing. However, whatever empirical evidence does exist in this area does not corroborate any of the claims made by them (Jud, 1982). 1983; Carney, More perspective on these models is presented in the next section which provides a model directly incorporating interaction among buyers, sellers and rea estate agents. On the empirical side, there have been only two studies which have explicitly included the importance of real estate agents in the housing market. The main emphas of the study by Carney (1982) is to derive implications for brokerage pricing under various cost assumptions. He describes a search model showing a relation between the search and the expected maximum price for a house, of the model. but he does not provide a formal analysis He argues that the maximum price that a seller can obtain increases at a also argues that commission rates will decrease with mean home price, increase with variance in home prices and decrease with the cost of search. Empirical evidence is presented to corroborate the above observations and it is found that commission rates are lower on (i) higher priced houses, (ii) on sales of new relative to existing homes and (iii) on nonMLS (or to MLS sales. coop as he refers to them) relative To obtain these results, he regressed commission rates on prices of houses and on dummy variables for households (as a proxy for existing versus new homes) and for MLS sales. The second empirical study which appears in the current stream of literature is one by Jud (1983). He develops a model of demand for broker services by both sellers and buyers of housing. The basic idea that the seller' demand for brokerage services is a negative function of commission rates and a positive function of transaction costs and the price of the house. The demand for broker services by buyers is similarly a negative function of existing stock of market information and a positive function of the opportunity cost of time. The probability that a seller will decide to employ a broker and the probability that a buyer will choose to buy with the help of a broker are then estimated. Both transaction costs and the sale price were found to be significant in predicting the probability of seller employing a broker The results on the probability of buyers transacting through brokers were more ambiguous than those on seller probabilities with income being taken as a measure of buyer' opportunity cost of time. As to market information , new buyers were considered as dummy variables and they are expected to have less i information than experienced buyers, and resident buyers who buy the house Another finding is that listing with brokers was not statistically significant in explaining variations in prices, thus suggesting that brokers do not possess any special monopolistic advantage which enables them to extract more for a house (thus getting a higher commission) than if it were sold by the seller himself. CHAPTER IV THEORETICAL MODEL In this chapter present an integrated model of seller, buyer and broker behavior which captures the essential features of the housing market as described in the previous sections Before settingup the model, however, it is appropriate to examine existing work in this area. As mentioned earlier, there are very few studies incorporating the importance of real estate brokers in the housing market. Review of Theoretical Models In a pioneering paper Yinger (1981) studied the "search and match" behavior of real estate brokers in the housing market. In this model , brokers face three types of uncertainty about (i) the number of buyers in the market (ii) available istings and (iii) matches between buyers and sellers. In formulating the problem, he makes the following assumptions: Brokers try to influence the average price of housing but not the distribution of housing prices. The average price becomes a choice variable in that the probability that a buyer will accept the asking price is a function of the average price. Brokers consider each time period to be homogeneous in that the number of buyers and sellers in the system is the same in each time period. 19 According to Yinger both of these assumptions imply that there are no bargains or overpriced houses in some time periods and each buyer and seller encounters the same prospects in each period. The broker's Traditionally, income comes from the commission payments made for his services. the commission is a percentage of the sale price of the house and in Yinger' model the commission rate is also a choice variable. Clearly, for such a broker, the larger the number of customers, the larger is the expected income. However, the broker has to incur costs to attract customers. Thus the objective of the broker is to maximize the expected income net of the costs of search. The choice variables are, then, the commission rate, the average value of the house and units of search levels for buyers sellers of housing. Yinger abstracts from dynamic considerations, resorting to common demand supply static analyses focussing on market clearing choice variables. Characteristics of demand and supply functions derive from probabilistic analysis of interaction of buyers and sellers and various assumptions about demand and supply behavior. For example, the probability 6 that a buyer will accept the asking price is a diminishing function of the average price of housing. However, the paper does not distinguish buyers and sellers who arrange trades without using the broker's intermedia tion. Thus , all sellers and buyers use the ntermediation services of a broker. In a market without a multiple listing service (MLS), expected income Y a single broker maximizes where Y=(cVa)NP8[1 Sp P1  TBPBNB with respect to the commission rate c, the average price of housing V, units of search for ~TBSBILSL attracting a buyer in the system, re PL is the probability of attracting a listing, is the price of a unit of search for buyers, N, is the number of listings rL is the price of a unit of search for listings and T, is the showing cost per buyer. The term n brackets [1(1 6P)NL] is the probability that the broker can match a buyer with a sting in the system. Yinger then examines the equality first order conditions to determine the characteris tics of the system. He does not explicitly consider circumstances causing competitive or noncompetitive outcomes, but does make the following distinction. If the market for broker services is competitive and it is not costly for sellers of housing to find buyers, the (aPL/c) is negative infinity and the broker accepts the marketdetermined commission. Also, if the housing market is competitive, and it is not costly for buyers to find houses, then , (6/V) is negative nfinity and the broker has no influence on the price of housing. Moreover , he also claims that high search costs for the seller of housing can generate market power for brokers but does not demonstrate how this result occurs. In the event the market is competitive, outcomes are determined by equality of demand and supply functions for matches in conjunction with first order conditions. difficult summarize outcomes since Yinger derives results under a variety assumptions. downward variables rg, rL For example, sloping with and o. with simplified assumptions, respect to the commission Considering entry of brokers, the supply curve of listings is independent of the Yinger argues that entry shifts the supply of listings downward and raises the equilibrium commission. Yinger next considers an MLS system where a broker makes matches with his own listings or other br6i~er's listings which commissions are then shared. Without 21 the commission rate remains the same is ambiguous, but if brokers have market power, MLS will increase V. With or without entry, fewer resources are devoted to search with an MLS than without one and, taking into account other considerations (e.g. an increase in the average housing price), MLS boosts brokers' income. Hence, brokers who set up an MLS have a strong economic incentive to keep other brokers out of it. With additional assumptions he draws some conclusions about the share of a commission a broker receives matching another brokers listing. example, more heavily influenced by experience and prestige than P8, then established brokers prefer a lower value of this share than new brokers. With regard to public policy Yinger argues that his analysis suggests that the optimal level of broker search is less than the level generated by a system with no MLS or even an unregulated MLS. Citing arguments by Stigler (1961) that competition should lead to price variation, a finding Owen and Grunfest (1977) about uniformity of commission rates in California, and price fixing cases brought against real estate brokers, Yinger also claims that this evidence implies that brokers have considerable market power. Hence, because MLS reduces search relative to the nonMLS system, government should encourage an MLS system, laws to prevent market power. Ying accompanied by vigorous enforcement of antitrust jer concludes by stating that his paper suggests that the welfare gain from government intervention in the market for real estate broker service could be substantial. In contrast to Yinger, Wu and Colwell (1986) introduce direct exchanges between sellers and buyers as well as transactions with real estate brokers as intermediaries. housing. The real estate brokerage market determines the equilibrium commission rate. Equilibrium in these two markets residually determines equilibrium in the market of direct transaction. Circumstances which determine whether buyers and sellers trade in the brokerage market or the market of direct transactions are not considered. In analyzing the brokerage market Yinger and Wu and Colwell make a number of similar assumptions. Major difference in assumptions concern characteristics of housing housing prices choice variables. Colwell assume houses homogeneous with sellers' asking prices specified by some probability distribution. Brokers do not choose the average price of a house or the commission rate. choice variables are units of search for listings and buyers. notation The only In the Wu and Colwe , without ML;, the ith brokers profit function becomes = (cV  o)NP,[1 8)1411]  rbSbl  tPbN,  rs~ Most variables have the same definition as n the Yinger model. Here 6 is the probability of matching a buyer with a seller and is equivalent to 6PL in the Yinger equation. A major difference is the interpretation of V which is now a parameter given to the broker. V is the expected selling price per housing unit with broker intermediation. Here, The expected selling price is related to an expected price determined in buyer search for low housing prices. According to Wu and Colwell, V may or may not equal this expected search price depending on whether buyers only search in the brokerage market, search only in the direct transaction market or search in both. determination of V is not explained. A similar function. If V does not equal this expected price the r problem arises in the ownerseller objective The objective function iri is maximized with respect to Sbi and Si where the 23 In the market for direct transactions the jth ownerseller maximizes the net price of a house by choosing the optimal search for buyers in the objective function  u')PbiN,b  rbSb/  td,/N, where the primes onh ariables distinguish ownerseller variables from broker's variables. In particular, V is the expected selling price per housing unit with ownerseller transac tions and, once more, equality first order conditions equate appropriate marginal returns and marginal costs. The buyer's optimization problem is a key element in the analysis since, as noted, this problem determines an expected minimum price which either equals V orV or is somehow related to these parameters. The buyer uses Stigler (1961) search with recall. The buyer visits a number of houses for sale and then chooses to purchase house lowest observed price. The optimal number of visits determined by equating the marginal return to search and of a marginal cost to search. That is, the buyer equates the reduction in the expected minimum price by one more search to the cost of visiting a house, obtaining an optimal sample size, houses to visit. the number of The details of the problem are well known and need not be repeated. nonoptimality of Stigler optimal sample search is also well known , being dominated by sequential search in which a buyer visits houses until a price is observed which equals or is less than a reservation price determined in the analysis of search. However, correcting the Wu and Colwell model to introduce sequential search would not change qualitative outcomes in the model. The parameters V and V would now be related to an expected minimum price determined by sequential search. Wu and Colwell then complete their analysis by an examination of partial general (V' of housing. In the general equilibrium analysis an increase in the cost of search for istings decreases commission , but the effect on the price of housing inconclusive. Similarly, an increase in the cost of search for buyers decreases the price of housing with an inconclusive effect on the commission rate. In an analysis similar to Yinger's, Wu and Colwell consider the system with MLS. With MLS, the partial equilibrium price of housing is higher and the commission rate is lower. However, in the general equilibrium analysis the effect on the commission rate is inconclusive In analyzing the ratio of the split between the brokers, extreme assumptions, they argue that the split would be close to, using various but higher than, than the equal split observed in most real estate markets. In comparing search levels, Wu and Colwel conclude , contrary to Yinger' results , that, under MLS each broker will devote more resources to search than an independent broker. They also conclude that with MLS the price of housing is higher and the evel of search by ownerseller is greater. Wu and Colwell refrain from drawing conclusions about public policy issues. The papers by DeBrock (1988, 1989) are similar to Yinger's study in that direct sales between buyers and sellers are not allowed. But, whereas Yinger uses competitive marginal analysis and comparative statics to obtain outcomes, DeBrock considers interactions among brokers to be a noncooperative game. DeBrock also explicitly allows dynamic elements of the trading process by considering probabilistic features of the time it takes to sell a house. We consider outcomes in his second paper which is a revision and extension of the first paper. As do other authors, DeBrock considers the brokerage system with and without MLS, 25 Real estate agents buy houses from sellers at the price P and then resell houses to buyers at the full price p = (1 +pP where p, is the commission rate. Hence pi gives the commission income on the sale of a house. Each agent also invests an amount of marketing expenditure m, per house at time t n the system without MLS, m, is assumed fixed, so that an agent's only decision variable is the commission rate pi which is chosen to maximize discounted expected profit on sales. Without MLS the objective function for the ith agent is then  f~p,rF'.Jq,(p,,p,)h(t n)e~d  m7q,(p/.pl) where ci is the fixed closing cost per sale, qi(pi, e) is buyers' demand for houses listed with agent , which depends on own price p, and the vector of prices of rival sellers 9, h(t, m) is the density function of time to sale where h diminishes with increases in marketing expenditure m, r is the discount rate and Fi is a fixed cost. Seeking a noncooperative Nash solution prices, DeBrock asserts one derive common non cooperative equilibrium prices pi. In the MLS system with joint marketing DeBrock poses the problem as a twostage game. In stage one an agent chooses the price p, and n stage two the marketing level He assumes an equal split in the event an agent, who is not the listing agent, sells the house. Later in the paper he considers nonequal splits. The objective function is now modified to take into account the split and probabilities of a listing being sold by rival agents. For the ith agent the objective function becomes Vnats H(t, m))K' erdt  mq, P)q,(K1)h(t,m)(1 Here (1  H(t,m))KI1 is the probability that the K1 rivals will not sell an agent listing by time t where H is the distribution function associated with the density h. Here F includes fixed cost of membership in the MLS. To obtain a noncooperative solution to this problem and to compare the outcome with the system without MLS, DeBrock (1989) assumes all agents are identical, density function h is exponential and the demands qi are quadratic. He then shows that the equilibrium commission rate is larger in the MLS system, where M represents the MLS system and the non MLS (independent) system. Moreover, MLS agents' profits increase relative to the system without MLS. DeBrock also compares the MLS noncooperative system to the MLS collusive system where agents jointly maximize industry profit. Outcomes depend on the parameter values. According to DeBrock, there exist a wide range of parameter values under which an MLS would price above the collusive price level. Moreover , MLS may lead to overmarketing or undermarketing relative to the collusive outcome. In considering the choice of a split variable he argues that the outcome is also parameter dependent. Finally, DeBrock briefly considers cases where the supply function of houses is upward sloping and a case in which houses are supplied by a monopolist. Again, results are parameter dependent. However, to consider a case of monopolistic supply of houses which contradicts facts in the real world does (2p P)qh(t, m)(l I~r~ m))Kle'~dt Model Description and Specification Introduction The model to be developed focuses on the interaction between ownersellers and real estate brokers. Yinger and DeBrock ignore the possibility that sellers may attempt to market their own houses, and Wu and Colwell only assume the separate existence of ownerseller and real estate brokerage markets without explicitly considering the interac tion of these markets. Rather than distinguishing the housing market, the brokerage market and market of direct transactions (as in Wu and Colwell's paper), which has become common in the literature on housing, the emphasis here is that there is a housing market in which transactions are arranged either by direct interactions between sellers and buyers or by brokers acting as intermediaries. In this perspective the housing market incorporates strong purely competitive features. There exists a large stock of houses, of various ages and characteristics , individually owned, a fraction of which are available for resale. home building industry consists of many builders with little impediment to entry. Hence, housing prices, both new and old, tend to be competitively priced. literature, as discussed in Chapter III, The earlier empirical which ignored marketing of houses, emphasized the competitive structure of the housing market. Given the strong competitive feature of the housing market, methods by which houses are bought and sold would affect housing prices mainly through'the spread between ask and bid prices, that is, between the prices paid by buyers and the prices received by sellers. In transactions with brokers the 28 in the housing market, then, is whether the commission rates are determined competitive ly or noncompetitively, as attested by numerous antitrust suits against brokerage firms. issue not directly studied Yinger Colwell , though Yinger nevertheless, concludes that the evidence implies that brokers have considerable market power and encourages vigorous enforcement of antitrust laws to prevent market power. Using game theoretic analysis of brokerage firms, DeBrock qualifies his results, but does emphasize noncompetitive outcomes among brokerage firms and specifies conditions under which the equilibrium price that arises under MLS would be higher than the price that a monopolistic (cooperative) brokerage market would charge, even though the individual brokers act noncooperatively under MLS. Specification of the Model with a Sina. Broker Specification of the model with identical sellers The housing market is characterized by several classes of houses with various characteristics. Housing prices reflect only the class houses region differences such as land prices, labor costs and differential market activity, generally affecting regional demand and supply of houses. However, since the emphasis here is the spread between ask and bid price as determined by interaction among sellers, buyers and brokers , I assume, as does DeBrock, that there is a single class of houses with a given competitively determined price P. A second major assumption is the heterogen ity of houses, even though the houses have the same listing price. Houses have different characteristics due to location, state of repair, arrangement of rooms, exterior building material, landscaping and other  w As do other authors, assume that brokers receive a commission p for their services , as is customary in the real world. do not attempt to explain this method of payment, other than to note that the use of a commission rate has some support in the principleagent literature (e.g., Harris and Raviv , 1979). Since sellers cannot directly monitor the broker' activities, payment linked to success provides an incentive for brokers to consummate sales. assume as does Yinger explicitly and DeBrock implicitly that the number of sellers and buyers (M, N) remains the same in each period. Hence sellers and buyers consummating sales in a period are replaced next period to maintain (M, N) constant. Initially it is assumed that sellers have identical costs even though they may be ng houses with differing characteristics and there exists a single broker Hence. if one seller decides to list (or not) with the broker, the others will also do the same. Thus wish to compare a situation where all sellers list with the broker or none list with the broker . Thus, when sellers do not list with the broker, all buyers only visit ownersellers, and when all list with the broker all buyers only visit the broker. The object then is to determine a broker commission rate which will induce all sellers to list with the broker. assume both sellers and brokers wish to maximize the expected return over an ndefinite horizon. First, the random process for an ownerseller is considered. Suppose an indefinite horizon is divided nto discrete time periods, a week. Potential buyers visit the house randomly during the week. per week. Let 1 be the expected arrival rate of buyers Moreover, suppose a fraction p of these buyers offer to purchase the house. Then the expected number of offers per week is ip. situations involving random arrival rates (Parzen, 1960 and Ross, 1972). If P(k) is the probability that k buyers arrive and make offers during the week, then P(k) = e ~ WP)k = 0,1,2,... Then P(0) = eX'P is the probability that no offers are made during the week. Since 1P(O) is the probability that one or more offers will be made during the week and since the ownerseller would accept any offer, the relevant probabilities for the ownerseller are P(O) and 1  P(0). In this framework , P(0) represents a failure and 1  P(0) a success for the ownerseller Then , the probability for a first success in period t would be [P(O)'.(1 P(O))] = [e( eap)] = (eP_l)el t which is the probability of failure in the first t1 periods times the probability of success in period t. In a more general framework, F(t,m,M,N) would represent the probability of a first success n period t, indicating the dependence of F on marketing expense m and (M,N) the number of sellers and buyers in the market. It would be reasonable to assume that the sum of these probabilities, over any finite number of initial periods, would increase with an increase in m and N , and decrease with an increase in M (and, of course, the sum of these changes would go to zero as the number of initial periods goes to infinity). In Appendix A it is shown that, in the Poisson process, this outcome is obtained if d(ap)/dm > 0, d(Xp)/dN > 0 and d(lp)/dM < 0, which are reasonable assumptions. In formulating the seller' of a house, the expected time of problem of maximizing the expected return in the sale sale, or, equivalently, the expected time of a first offer, r)~p~(l = (LP" Using the properties of infinite sequences,  1) tept it easily follows that = 1/(1  e^ .1 However , in a general framework, such a convenient expression would not be available. In the Poisson case, given the assumptions d(Ap)/dm > 0, d(Xp)/dN > 0 and d(Xp)/dM < 0, it then follows that the expected time on the market E(t) diminishes as m and N increase and increases as M increases. Given the assumption about F(t,m,M,N) in the general case, the expected time on the market has the same properties as in the Poisson case. Ownerseller's Problem The ownerseller'. problem, then , is to maximize the net expected value of the house over an indefinite horizon. The sale price of the house is assumed to be fixed by the competitive forces in this market. The ownerseller also ncurs various costs. These costs are: Marketing costs per period to attract buyers, e.g., advertising in the local newspaper. The outcome described by E(t) is analogous to a similar result obtained in the continuous time exponential process where the distribution function G(t) the density function g(t) = XpeA"" for t Here, the expected value, S =,1 ;ay, I eX" and E(t) equals  A Holding costs per period. house as an asset. Maintenance expense, interest and user cost of Holding costs strongly depend on whether the owner lives in the house, the house is rented or it is vacant during time on the market. Showing cost per prospective buyer. to the house by an interested buyer. This is the opportunity cost of a visit The expected cost per period is As. Closing costs. Fixed costs. These are the costs of preparation of the house for sale. For the ownerseller the net expected value is then given by = E p'1[P where p is the discount factor and ot  c(m + h + As)oF(t,m,M,N) = (1 fit)/(1 p)(3t1). When the seller decides to sell the house through a broker, the costs that need to be incurred are different. The advertising and other related costs are now borne by the broker. Therefore m need not be spent by the seller. The seller also need not incur showing costs, since now that is the responsibility of the broker. However, the holding cost until the sale is finalised is still the responsibility of the seller. Similarly, the closing costs and the fixed costs are also the obligation of the seller. For the services of the broker, the seller needs to pay him pP upon completion of the sale, commission rate. where p is the The net expected value to the seller in this case is then p)P  cho ] Fb(t, m, M,N) where Fb(t,mb,M,N) is the probability of a first success in period t if the property were sold by the broker. ownerseller. Here, we distinguish between the marketing cost for the broker and the Thus the property will be listed with the broker if Vb p)P r '1iP  c ho ] F(t mb, M, N)  c (m+h+Als) o,] F(t, m, M, N) = F then the constraint reduces to Broker' problem The broker considers (M,N), the numbers of sellers and buyers n the market, given the assumed renewal properties, to be the same for each house there is a probability of no offers n each period. n each period (i.e., In each period, a failure), Pb(O), and a probability of a success, 1Pb(0), allowing for the possibility that probabilities for the broker and ownerseller may differ. Then the process of selling houses n each period is described by the binomial distribution. Hence, the expected number of houses sold each period equals M times the probability of success or M(1Pb(O)), recalling that Pb(O) depends on (M,N). In the general framework, note that (1Pb(0)) = Fb(1,mb,M,N), the probability of a = E P"[(1 , .e., p t'[pP 1 (m+As)ot]F 8"'[(1 In selling houses, the broker incurs various costs. These costs are Marketing expenditures per period to generate listings and to attract buyers. Showing costs per prospective buyer per house. The expected showing cost per house for the broker is AbSb where 1b is the number of expected buyers visiting the broker per period per house. Db Fixed costs in establishing a real estate agency. The objective of the broker is to maximize the net expected profit from selling houses, subject to the constraint that any seller would use the broker only if Vb problem can be written as V= )"p[pP  .bsbI MF(1 ,mM, N) E_ Pt m subject to E p[(1 SPt'[P p)P  c(m +h +Xs) oj]F(t, m,M, I') where the choice variables are the commission rate and the marketing rate (p,mJ. constraint (9) does allow some variation in p depending on the choice of mb. An increase in mb would decrease the expected time to sale and would allow an increase in p. Nevertheless , the constraint (9) limits the choice of the commission rate. Hence, the constraint on p, imposed by a seller's option of selling directly to buyers, induces a competitive outcome in this market, even with a single real estate agency, since a seller D,  c ha,] Fb(tl mb' M, N) 35 maker for a security, the specialist, chooses the spread, the difference between the ask and bid prices. However, the traders are not obligated to use the services of the specialist. Thus, if the spread is sufficiently large, traders may arrange mutually beneficial exchanges among themselves inside the reduce the spread to avoid losing sales. This prospect induces the specialist to In the real estate market, with the constraint on the choice of the commission rate, sellers are not able to arrange mutually beneficial exchanges with buyers inside the "spread", the difference between the ask price of Pand the bid price (1p)P. The specialist would welcome a rule which requires all traders to arrange exchanges at the ask and bid prices he calls, for then he could exploit his monopoly power, as a single pricemaker, in the choice of the spread. Similarly, the real estate broker would Welcome a similar rule as the intermediator in the real estate market. The broker's problem illustrates another feature of the housing market. The broker interacts with various sellers and buyers, market infrequently. week by week, whereas a seller enters the Some implications of this feature are discussed n the next section. Specification of the model with heterogeneous sellers Here , once again it is assumed that there is a single broker and a single class of houses. However, now suppose sellers are not all identical with respect to costs. Differences among sellers arise for several reasons. purchase a house depends on its characteristics. Sc First, the willingness of buyers to >me houses are more "salable" than others. For sellers with less desirable houses the probability of an offer p would be lower, the expected time to sale higher and the net expected value of the house would be lower. Second , the cost of showing the house to prospective buyers may also be different. 36 selling differ widely depending on whether the owner lives in the house, rents it or leaves it vacant during the time that it is up for sale. Differential costs among sellers would imply a distribution of net expected values V,, where ..., M distinguishes different sellers. Thus, assuming the broker knows the distribution over Vi the problem is to choose the commission rate and marketing rate (p,mb) to capture enough of the sellers with low V to ensure a viable operation. With a choice of (p,m) made by the broker, a fraction of the M sellers would list their houses with the broker and the remainder would choose to sell themselves. With respect to the number of buyers N, then , a fraction would only visit owner sellers , a fraction would only visit the broker and a fraction would visit both. Ownerseller Problem To be more specific write the problem in terms of the Poisson distribution. Now, in a situation where the sellers are heterogeneous in costs, the expected net value of the house for the ith seller would be given by  i rpt[P Here, I.i, seller 1 )eIplt the expected number of prospective buyers per period who visit the ith owner , depends on some fraction of M (possibly all M) and a fraction of N who visit the ownerseller houses. seller The probability of an interested buyer making an offer to the ith , P,, depends on housing characteristics. As in the earlier case, the objective of the seller is to maximize the net expected value of the house. Hence, the owner will choose to list with the broker only if D,  c(m, + h, + ~.ls3 al(e~lPI L KI E t'1[(1 1 a p)P c h, t] (e Pb 1)e9li where, distinguish broker probabilities from owner probabilities and the expression on the left equals Vbi. Broker's problem Unlike the situation with homogeneous sellers, now only a part of the total number of sellers list with the brokers and similarly not all the buyers in the market will seek the help of the broker. Thus Snow the broker's prospects are (Nb,Mb) where Nb is the number of buyers going to the broke and Mb is the number of sellers listing with the broker and both of these numbers are functions of the commission rate p. expected net value which the broker seeks to maximize is a = p '[ pP The broker will then choose (p, is an equality. Hence, the *AebPh)  bS]Mb(1 mb) such that, for one or more owners, the inequality (11) That is, Pt'[(1 1pr[p 1 p)P chiotJ(elAb c(mL +hL+XL 1)9AbPbt ALPLt s) oJ,](LLP" where L specifies an owner for which the equality is satisfied. expected values greater than VL Hence, owners with net will become ownersellers and those with net expected =VL  c (mi h,+ ~, rs3 o,] (eL 9~l)e~IP~' _ 11  Db values less than V, will list with the broker. Thus, the broker' optimization problem can be written as ,sb]JM,(1e*P ) 1 ri p)P EKI E P''[(1 1. In this problem the choice of (p, hL,o (eLPb l("" o(mL+ hL +XrSdua(eLdPL mb) also selects L, taking into account the distribution of net expected values Vi. Whether broker survive, given fixed alternative opportunities for employment, depends on the distribution of V for ownersellers if the distribution has large weight on high V, too low for survival. then the broker may be forced to choose a p In small communities, there may be no broker or a broker may need to attempt to cover more than one community. The number of sellers satisfying V, may be too small to support a p for the broker to survive, , to find a p such that V is greater than the opportunity cost for the broker. The problem for the broker is now more realistic and much more difficult for the broker to solve, but the competitive feature of the previous problem remains. In principle, it is not difficult to specify the firstorder equality conditions and to examine the trade off between the choice variables (p, mb) but not much is gained by this exercise. Rather, assuming viability, it is more useful to examine insights about information, time on the = = p[pP 1 Db 39 A broker whose income depends on being well informed about the housing market will tend to be better informed than an individual seller about the state of the market, desirable characteristics of houses and market price P. Even if wellinformed about (M,N) and the arrival properties of prospective buyers, the seller may misjudge the desirability of his house and overestimate pi and, hence, underestimate expected time on the market and overestimate the return V,. Over time, as the ownerseller becomes better informed about p, and with no success in the sale of the house, he may decide to list with the broker. A similar situation arises about estimating the state of the market, whether it is strong or weak. The ownerseller here , may underestimate M and/or overestimate N. Again the seller gains information over time if unable to sell the house then , may decide to list with the broker. The seller may also overestimate the price of the house R This overestimation will also lead to the same process of updating information and then revising the choice of the channel of sale. Of course , the seller may also underestimate P and be rewarded with a quick sale (which is part of the motivation of buyers' visits to ownerseller houses). All the analysis presented n this section assumes the existence of only one broker. In the next section relax this assumption and analyse the model with more than one broker having access to a common pool of information, such as the multiple listing service. Specification of the Model with MLS Brokers In the previous sections, it was assumed that there was onlv one broker orovidinn institutional arrangement among brokers, .e. the multiple listing service (MLS), where two or more brokers agree to poo first shows h information about listings of available houses. "own" listings to prospective buyers. A broker This ensures the broker the full commission from the seller if a sale is However if the buyer rejects these house showing the buyer other listings in the h made to a prospective buyer from this list. ies, then the broker has the opportunity of ILS. If the buyer selects a house listed by another broker, the selling broker is paid a prearranged proportion of the commission by the listing broker. In presenting the case involving more than one broker, will assume that there are two identical brokers who are a part of an MLS (this assumption is for ease of notation alone and the logic can be easily extended to more than two brokers). Second , I first analyze the case of identical sellers. As noted earlier, in this case sellers have identical costs even though they may have houses with different character istics. Again, if one seller decides to list with a broker, the others wi do the same and buyers only visit brokers when sellers decide to use brokerage services. Third , I assume that the two identical brokers act as Nash competitors. In a Nash equilibrium the two competitors wi choose a common commission rate and marketing expense (p, m) and agree upon a split rate 0 < 1 where OpP is the commission earned by a broker when selling a listing of the other broker. would be indifferent to listing with either broker and, on the a* would list with one broker and half with the other broker. Sit In this situation a seller average, half of the sellers nilarly, half of the buyers would visit one broker and half with the other. Specification of the model with identical sellers Ownerseller' problem The ownerseller' problem here remains the same as in the previous case with identical sellers. As in equation (4), the net expected value for the seller is given by = B pt'[p  (m+h+X s)o,]F(t,m,M,N) In selling the house through a broker, broker has the listing, nce it is a matter of indifference which the net expected value to the seller is analogous to the result in equation (5) = p''[(1 p)P  hO]Fb(t, mb.M,N) where Fb(t, mb, M, N) is the probability of a first success in period t if the property were sold by a broker. Here, Fb need not be identical to Fb in equation (5) since total marketing expense by the two brokers may exceed the marketing expense of a single broker Broker' Again, a seller will list with a broker if Vb problem n the previous case (M, is the same in each period. the number of sellers and buyers in the market, Moreover, since on the average half the sellers list with each broker, the probability of any house being sold by any broker is Fb(1 mb, M, N)/2. Hence , the expected number of houses sold each period by a broker from hi own listings is M/2.Fb(1 m, M , N)/2. However, under the MLS system, a broker may also sell the other broker's listings so that the expected number of houses sold, listed with the M/2.Fb(1 m, M houses period or, in total , MFb(1 m, M, N) is the expected total number of houses Moreover. sold by both brokers, in the MLS system, case a broke of the receives single broker. income from three sources. Income is received from selling his own listings, from selling the other broker' listings, and from the other broker' sales of his listings. net expected value to be maximized by a broker is  XbSb ,m, M, N)  yt1 .m + tz[0PP 1 + Pt'[(1 fp1i 1 bSb] 2 , mb,IM, N) , mb, N) 6)p PM2 2 subject to the constraint that Vb n equation (18) the first term gives the expected return on the broker' own listings, seco nd term gives discounted marketing expense per period mb and Q is the fixed cost where indicates an added harge for establishing th MLS system. fourth term gives the expected value from selling the other broke 's listings. final term gives expected rev nue fro m the other broker' sales of the given broker' listings. immediately obvious relevant. from equation What a broker expected to gain from th 18) that the choice of the split rate 0 is sales of the other broker' listings is offset by the rev rse operation by that broker ence why not ch oose which is the common choice in an MLS system? Whatever the choice of 0, equation (18) reduces = p [pp = rE t'[pP .Fb(,m, M,N)  P'1 b  Db,. Thus, when the brokers maximize equation (19) subject to Vb _ V, the problem and hence the solution is analogou to the previous case with identical sellers and a single broker . Once again, previous case with a single broker, the constraint Vb limits the choice of th common commission rate chosen by th brokers. next section consider case with heterogeneou sellers and examine implication of the existence of two identical MLS brokers. Specification of the model with heteroq eneo us sellers Once again, re th lers differ from each other n the costs that they face ng the house. Thus differential costs mply a distribution of the net expected values where distinguishes the selle marketing expenditure brokers must now choose th so as to maximize commission rate net expected earnings while capturing enough of the low Vi sellers to en ure viability of operation. As in the previous case , only a fraction of the total sellers will list with broker (either of th two brokers) and similarly a fraction of th total buyers would vi brokers. Ownerseller' problem Here net expected of th house for the ith selle is given by the following = E P''[p  (m,+h,+ X ao Fi(t, m, M, N) Di  P9  h;.o] aFb(t, mb,M,N) g Hence , as previously, the ith seller chooses to list with the broker only if Vbi Broker' Problem In thi case , only a fraction of th total sellers will list with brokers. Thus of that number Mb, on an ave rage each broker expects to have Mb/2 list with him. Then each broker' expected net valu will be Mb f13 tl[pP 1PM1pp 1 Fb(l ,mb, Mb, N) where again the choice of the split rate 6 is E P relevant. D, brokers wil now choose (p, uch that fo one or more owne rs Vbi Again, each of th two brokers will satisfy above constraint and will be limited in the choice of the commission rate. Hence , the general outcomes are similar to ones obtained in the discussion heterogeneous sellers and a single broker . However, the choice (p m,) may differ in the case more than broker Since broker now has fewer listings, a Nash equilibrium may dictate a decision to increase total number of listings with brokers by lowering the commission rate. investigate this prospect would require more detailed information about the distribution of the net expected values V i to weigh the gain listings against th Moreover n revenue isting. , in the real world it would be difficult to argue that brokerage agencies are i identical. Brokers acquire reputation whether ustifi d or not. As noted earlier = E P'[(1 arrangements title search companies. Howeve as noted in Appendix argument is that these arrangements tend to increase, rather than decrease, closing costs and are subject to antitrust proceeding Reputation seem to be acquired through marketing costs. n all previous models have assumed that marketing costs for brokers are a fixed amount per period, independently of the number of listings. While there is some truth n this assumption, since some costs are independent of the number of listings, advertisements n local newspapers, part of marketing cost may be more directly associated with listings and potential buyers. Brokers display varying degrees of creativity and costs) in newspaper rtisements to attract listings and potential buyers. Some brokers feature open houses expense) more than others. Some emphasize canvassing neighborhoods by mailing residents information about their local sales record and market conditions n the local area (including asking prices of houses for sale n the neighborhood). Some brokers are more solicitou than others, n direct dealings with sellers and buyers. All attempt, in one way o another , to use marketing cost to enhance the reputation leading to some variation n marketing costs across brokers. some monopolistic competitive markets , attempts brokers differentiate their products may lead to excessive marketing expense as they compete to attract listings and buyers from a common pool. n the next chapter present an mpirical model and its estimation taking consideration theoretical issues discussed in the present chapter CHAPTER V EMPIRICAL MODEL AND ESTIMATION In this chapter, first examine Jud (1983) model n detail and then present an alternate approach to th mpirical estimation. Review o an Empirica Jud stud the role of real estate b rokers in the housing market and develops an empirical model of demand for broker services by both buyers and sellers. Seller' demand for broker serve is presented as function commission rates transaction costs for the seller (ts) and sales price of th property (p). Formally, = D,(c, t,, p) where < 0, (aD,/at,) > 0, and A buyer's demand for broker services depend nt on his preexisting stock of market information ( and opportunity cost of his time (tb) b = Db( where aDb/ai) < 0 and (aDjatb) assumes that horn sellers are price takers (given the prevailing commission and sale price alone. Since the dependent variable D  is a dichotomus variable, ordinary least squares estimation methods are not appropriate. Therefore , he estimates the D, equation as a probability function. described as logistic, prob(Listing) probability that a seller lists with a broker is = 1/{1 +exp[(a++p3t+pp)]} n[ prob(Listing) 1 prob(Listing) =D,=a + pt+ Pip which he estimates using conditional logit techniques. The variables he used are as follows Listing = one if the house was sold through a broker = zero erwise; Trans Cost = one if the seller moved out of th county = zero , otherwise; Price = the sale price of the property (in $ 000s The dichotomous variable "Listing" was used to construct the dependent variable (D). It was found that both transaction cost and sale price variables were statistically significant (at the 1 with a broker. As level n predicting th transaction costs probability that a seller will list his property increase probability of a seller using a broker increases. As th price of the property increases, probability of selling through a broker increases. Jud then estimates an equation for demand for broker services by buyers, also Broker = one , if the buyer consulted a real estate broker to aid him zero n his housing search, other New Buyer = one, if the buyer n ever owned a h ome zero other = one , if buye lived previously in the same county, zero, otherwise; Black = one, if the buyer was black zero erwise; New = one ,if buyer purchased a new h ome, zero other = buyer annual income in $1 O00s The dichotomous variable Broker was used to construct the logistic dependent variable. The variables "New Buy r" and "Res" were used to provide measures of the buyer existing knowledge of the housing market. New buy ers are mo re likely to consult brokers since they hav very little informati on and similar y resident buyers have more information so their need to consult a broker is less. Estimated coefficients of both these variables had the expected sign, but only th variable es" was statistically significant. variable "Black" was included to ncor porate racial differences but was found to be insignificant. "Inc" was also not gnifican "New" was with its negative coefficient which was expected since brokers are less active new h ome market. It sri thna vanraeoarlr tt nrr'a ni I, ;!. h rn arrthaehrl 4kr'M. in,,k A..A A I r~\ ' r n n ~ n rk ntk h I ~k kni rFn rnl i variables were statistically significant except the listing variable. This result seems to suggest that the brokers do not possess any special monopolistic advantage which enables them to extract more for for a house han if it were sold by the owner. agrees with ou intuition about the market. present an empirical model of th housing market with buyers, sellers and real estate agents. There are several shortcomings of Jud' study which hope to resolve in this dissertation. First, he assumes that sellers are price takers given that the commission rates are fixed. However, as argued in the previous section , the ability of sellers to deal outside the market does influence the commission rates. sellers are not entirely passive price takers. Moreover evidence about th fixity of commission rates does indicate some variation n rates (see Carn , 1982) as outlined n Chapter As noted there , Carney finds that commission rates are low r on (i) higher priced houses, (ii) on sales of new relative to existing homes i) on nonMLS (or, coop as he refers to them) relative to MLS sales. Model Description and Specification n the present study include the estimation of commission rates and explicitly consider the simultaneous interaction between the three types of agents, ., sellers, buyers and real estate brokers. commission rates are included n the model then Jud' estimation procedure no longer correct. This is because now there is a simultaneity in the system. Commission rates are affected by buyer and seller behavior which in turn VI%~~~kl tjI+4 t1 I flf rn v,,.I t at~ an v 'r :r t. I t.n *l .:i Znna .a. jil t.AArj~ ArA ACIAA)Arl k~l(kA AAMMlnnln~ rA(AA * simultaneou equation model. discuss my model and estimation procedure in the rest of this section. The main interest n this dissertation is in how commission rates affect and are affected by the behavior of buyers and sellers and their estate brokers. ability to deal outside the real Clearly these effects need to be determined simultaneously. The commission rate that a real estate agen can charge depends on several variables including the housing price, seller probabilities of housing characteristics seeking the serv are not available housing characteristics of a broker in the data in the the best market environment and sale of the house. can do is Since assume the mix of housing characteristics is constant. housing market consists of new houses as as existing ones. Usually, th new properties are adv ertised and sold by the builders and developers of such properties. Thus brok ers face competition from new housing construction therefore consider the number of housing permits issued as an i ndex of construction activity capture measure competition from new construction. However, new construction may also provide some means ure of whether the housing market is active Finally native seller probabilities o so that th outcome of using housing p listing the property with a broker ermits may be mixed. influence the commission rate that can be charged by th broker How ever these probabilities are also determined by commission rates along with th characteristics of th sellers and buyers (to be described next). A seller's nf lietinn th decision to make the sale through a broker nrnnof'r t lA/ith a hrnlter\ r4onQnrt  "A', (and hence A  the probability  AL .. ni I11 If L0 r1aELr tnI ggtyt '' 0 lrr fY n I o nT TB costs. These are all transaction costs involved n completing the sale which could be viewed as opportunity costs of time). transaction costs are kely to be highly correlated with the price of the house, higher the sale price of the house, more likely it is that the seller income is high (which in turn mplies that his opportunity cost would be high). Thus only consid er the sale price of th house in the actual estimation of this model. The market condition determining the probability of a house being listed with a broker relate to the numbers of buyers compared to the numbers of sellers (as explained n the previous section which in turn influence th expected time to In the estimation use the un employm ent rate as an i ndex of the market activity. n areas with high un employment, the number of buyers is likely to be less than the number of sellers, making it more probable for sellers to list with brokers. The probability of a sale taking place through a broker would also depend on th opportunity costs of the buyer because what is observed is the actual sal through a broker and not just the seller' probability of listing with a broker rese nt paper the opportunity cost faced by the buyer are m measured as th cash paym ent by th buyer at the time of , including the down payment and other costs. Finally, price house was assumed previous section exogenously determined by the market upply of and d mand for housing. However, while empirically studying the housing market, interesting to consider the effect that brokers may have on the price of houses. price of a house would also depend on the geography location of th property and th une employment n that area. Thus key point in this model is that th commission rates and the probability brokered sales are interdependent variables hence must determined simultaneously. There are two opposing fo rces which are at play here. First, under the conditions described abov market is ve ry active, it is not costly fo the seller to search for buye rs him self and undertake to sell the house directly, and the demand fo services of a broker This feature tends to dampen the commission rate. However , this dampening effect encourages seller to mploy a broker and hence increase the probability of a brokered Therefore estimating commission rates and the probability of brokered sales separately w would yield biased results. Also , the price of th house must be estimated simultan eously n this system because am i interested effect of a brokered sale on th price of the house. problem can then be appropriately formulated as th following system of equations COM + P12(B) + P13( 11 (HF') Cc21 + 1P1(COM) + P31(COM) + Y22() + p23( + M2(B*) + Y23( 4 + Y34(G2) '2 (24) 35(03) The reduced forms for the abov system of quations are COM B' P K2i It3 + a12n(HP) + 22(HP) + "32(HP + 7ni(D) + X23(D) + %733(0) C4U +714U + ,U + 15(G2) + "is(G + 25(G2) + "xz(G2) + "16(G3) + X26(G3) + i3(G3) + w3 The variables n the above system of equations are defined as follows sale price of the property (in $ probability that th ,000s) property would be sold through a broker (which is a latent variable) if the house is sold thro ugh a broker zero otherwise cash down payment plu other costs paid by th buyer (in $s) COM commission rate employment rate number of housing permits by if the house is located zero issuing location eastern U otherwise if the house is located western U zero, I G otherwise  G3 specifies location in the Central U The simultaneous system of equation and (25) involves the unobserved dependent variable B ( probability of a brokered sale) instead we observe another variable B* (where B = 1 ifB zero rwise we must use a two stage limited depend reduced form equation (26) nt variable estimation method. by tobit since first stage consists of estimating commission rate variable COM is not observed for all cases ., it is not observed for houses sold by own rsellers rall ira rl frrm nis latifnn 1/07\ bK th" a nrnhit KIll rm thth one ~in~a ;t a~fllall\r th~ nmkak;l;br dependent variables into th structural equations and once again estimating equation (23) by tobit, (24) by probit ML method and (25) by OLS (see Maddala, 1983 for a description of twostage estimation methods simultan eous equations involving unobserved dependent variables). Using th rank condition for identification of our system of equation find that the above system of equation is identified. above system equation for the results to be cons istent with theoretical model , the hypoth eses regarding structural coefficients would be as follows: (the coefficient of B n equation (23) should be close to zero and P13, coefficient of P must be negative received because as commission would be higher r high priced houses, making the "absolute" amount brokers willing to accept a lower "rate" of commission. n equation the structural coefficient of the commission rate variable must be negative and p23 must be positive. n equation 24), p31 coefficient of the variable COM will give commission rate is on th an indication of whether th seller or on the buyer and 32 would be posit incidence of the ive if brokers are able to obtain higher prices than do own rsellers. The othe variables are expected to hav the following The sign of y,, is is not certain , as argued earlier, because on the one hand the number of housing permits issued indicates competition to brokers fr om new construction. On the other hand the new construction may be a m measure of an activ or inactiv market. The higher the level of new construction easier it would be for a seller to make a direct sale with a buyer. Thus. the effect of HP on the commission rate will be uncertain. Turning to equation (24), Snn n~t.,., ai~ +knrtln;~ an.%,'n4 .ntnn+4 "I 1  n nrA 4kl :n :tlnr an Itrr Irst* a knt TJ 1~  I rr tnh nl nnr r^ f positive influence on the likelihood of a brokered sale. If unemployment n an area is high, the number of buyers in that area is likely to be low . Thus it is more important for the sellers to seek the n equation (24), services of a broker. the sign of y Thus , the sign of y is likely to be negative, is likely to be positive. because in areas with high unemployment leve ere are kely to be few buyers relative to sellers) and thus the prices of houses are lower . The signs and y will give impact of geographical location of the property on the prices of houses (taking Central U.S. The model presented here differs from Jud's as a base). following ways. First, he ignored commission rates in his estimation procedure, which takes account of the nterrelation betw een agents in th housing market. Secondly, by including the index of market activity (i.e. unempioym ent rate allow for differences n seller behavior in decision about sale through brokers. n the next chapter discuss data sources and results from the estimation procedure described in this section. also discuss seven ral alternative data sources which might be better suited to the analy presented here. CHAPTER VI DATA AND EMPIRICAL RESULTS Data In estimating the system of equation described n Chapter V used data from 540 housing transaction in cities states United States which took place early months of 1989. The data on six of th variables COM P, B, D, G, and G3 were obtained from actual Settlement Statement forms HUD1 (or RESPA forms) which are signed by the parties involved ( buyers, sellers and lenders at the time of closing of a deal) One observation about this data set is that it covers only low and medium priced properties This is because obtained th data from the U Department of Housing and Urban Development regional offices and these offices only deal in transactions financed through federal and Veterans Administration loan These federal loan have an upper limit and thus the data that could obtain is only for low and medium priced houses. Thus the results apply only to moderately priced properties. For the ndex of market activity , I have used unemployment rates. my study com from the Bureau of Labor Statistics. data on unemployment rates that These use i figures are annual averages 1989 which have been disaggregated at the county level. Since housing transactions data is at a city level, cities and counties are related by the information available in statistical abstracts of each individual state. The problem with this process available. Also some cities are part of two (or more) counties. Fortunately there were very few uch cities. Construction activity n the area of interest is considered as another index of market activity. measure construction activity by using the number of housing permits issued n that yea n the particular area. This was obtained from Current Construction Reports published by the U Departm nt of Comm erce, Bureau of the Census (the particular report was titled Housing Units Authorized by Building Permits). to date numbers from the November 1989 used the year issue. The main problem with the above study is th unavailability of any organized form of data for the type of problem n the real estate market that have have described in this pape . Due to this problem had to collect the data from several different sources and combine them. This makes the data ve ry noisy because, the different data were available at different levels of disaggregation. For instance the RESPA data from the U Department of Housing and Urban Development (for th variables COM , B, P and D) is actual transaction data is disaggregated at th level. data unemployment rate were obtained from stat offices of Bureau of Labor Statistics and these were mainly at county levels although th data was available fo some of the larger cities Thus for the smaller cities had to relate the RESPA data to the relevant county. This is obviously an approximation, which has some bearing on the validity and the quality of the results. Similarly the data on housing permits is published by the ThiQ tnnn in an annrvimatinn fr th0 QmaIllo I tnr\nc permit issuing offices. hinQ anrI ritiio AAhir, Sh mor nft hHao a One of the sources of data which would have been useful is data from brokerage houses. individual This would have made it possible to obtain data on a wider range of prices of housing, the corresponding comm issio n rates, probable time onthemarket as well as some housing characteristics as the age of the house, the size of the house etc. Unfortunately, the real estate brokerage industry has antitrust allegations which has lead to extra cautiou spoke with several brokers at been particularly vulnerable to behavior by brokerage houses. several large brokerage houses and also with the research departments of the National Association of Realtors and Florida Association of Realtors as well as Gainesville Board of Realtors and each time was told that there is no available source information commission data as it could interpreted as non competitive price fixing by the FTC. Another , appropriate method of data collection is to undertake a survey of house owne rs (those who have sold houses as well as those who have bought hou ses directly or through brokers) and brokers. This method w would actually be the best for the type of empirical work undertaken 1983 (see n this study FTC Staff Report 1983) but . This type was survey was unable to obtain this done for the FTC in survey data for use in the present research as my communication with th organizer of this urvey revealed that those data tapes were destroyed. Empirica results The results from the two stage estimation of the simultaneous equations system (usina LIMDEP package for estimation of limited dependent variables) defined in Chaoter coefficients for equations (26), (27) and (28) respectively Tables IV, V and VI present the corresponding structural coeffici ents. The estimated structural coefficients presented n tables IV , V and VI give some conflicting evidence about the hypoth eses presented earlier R2s are not high. However, this result is not uncommon with large samples and particularly with noisy data have available. The coefficients of th two i important variables, B and COM have the following signs. The coefficient of B equation ( 23) gives the expected result it is found to satisfy the hypothesis that the coefficient is not significantly different from zero. other words , the probability of sell rs and buyers of housing seeking brokerage services does not hav a significant effect on commission rates. This is consistent with the theory presented in the previous sections. How ever coefficient of COM in the equation (24) has a positive sign (although not significant). positive sign if significant would have implied that commission rates hav a positive effect on the decision of sellers and buyers) n choosing to employ a broker (it was expected that coefficient of COM in equation 24) would be negative because higher th commission rate the lowe was the probability of a brokered sale). coefficien of P in equation (23) has the expected negative n equation (24), In equation the sign of the coefficient of P is positive as expected. the coefficient of D which is indicative of buyer' opportunity cost and thu expected to be positive, turns out to be negative but it is not significant. Finally, coefficient of th unemloym nt variable expected positive significant. Next. the results from the estimation of the rice of housina (i t. equation (251 are negative coefficient, suggesting that the ncidence of the commission is not on the buyer of the housing. The unemployment rate has as expected and it is also significant. Finally a negative influence on the price of housing both G2 and G3 are significantly positive suggesting that n the present sample, prices of ho uses in both the Eastern U Western U are s significantly higher than in the ntral U.S. , the results provide only a partial corroboration of the hypoth eses in this paper. But this is expected given that many of th variables have used are proxies and the variables that are not proxies are subject to large measurement errors. The estimation limited dependent variable models with m easurement errors is extremely complex, (especially with a particularly noisy data set like mine) (see Stapleton and Young, 1984) and hence did not attempt that procedure. How ever feel that there is much research to be done n this area and an important part of my future research agenda would be to obtain cleaner data possibly from primary survey ys) and study the housing market in a much greater detail. TABLE ESTIMATED REDUCED FORM TOBIT EQUATION FOR COMMISSION RATE VARIABLE DESCRIPTION COEFFICIENT ONE 4.90 (26.45) DOWN PAYMENT 0.29E01 (1.71) UNEMPLOYMENT 0.25E03 (0.02) HOUSING PERMITS 0.16E03 (2.27) GEOGRAPHIC LOCATION (EAST) 0.96E03 (0.29) GEOGRAPHIC LOCATION (WEST) 0.94E01 (0.38) [tratios are provided in parentheses] TABLE II ESTIMATED REDUCED FORM PROBIT EQUATION FOR BROKERED SALE DUMMY VARIABLE DESCRIPTION COEFFICIENT ONE 1.26 __________________________________(9.00) DOWN PAYMENT 0.26E02 (0.29) UNEMPLOYMENT 0.16E01 (1.48) HOUSING PERMITS 0.25E04 (0.68) GEOGRAPHIC LOCATION (EAST) 0.47 (2.31) TABLE ESTIMATED REDUCED FORM EQUATION FOR PRICE VARIABLE DESCRIPTION COEFFICIENT ONE 56.72 (40.79) DOWN PAYMENT 1.32 (10.17) UNEMPLOYMENT 0.53 (5.29) HOUSING PERMITS 0.12E02 (2.27) GEOGRAPHIC LOCATION (EAST) 0.12 (4.93) GEOGRAPHIC LOCATION (WEST) [tstatistics presented in parentheses.] TABLE IV ESTIMATED STRUCTURAL TOBIT EQUATION FOR COMMISSION RATE VARIABLE DESCRIPTION COEFFICIENT ONE 5.93 (7.91) BROKERED SALE DUMMY (FITTED) 0.64E03 (0.51) PRICE 0.18E01 (1.54) HOUSING PERMITS 0.14E03* (1.96) [tratios are provided n parentheses. * indicates significance at level.] 63 TABLE V ESTIMATED STRUCTURAL FORM PROBIT EQUATION FOR BROKERED SALE DUMMY VARIABLE DESCRIPTION COEFFICIENT ONE 1.64 (1.13) COMMISSION RATE (FITTED) 0.32 (1.45) PRICE 0.19E01 (1.43) DOWN PAYMENT 0.20E01 (1.01) UNEMPLOYMENT 0.14E01 * (12.90) [tratios are provided in parentheses. indicates significance at 5% level.] TABLE VI ESTIMATED STRUCTURAL EQUATION FOR PRICE VARIABLE DESCRIPTION COEFFICIENT ONE 163.05 (3.31) COMMISSION RATE (FITTED) 92.47* (8.46) BROKERED SALE DUMMY (FITTED) 533.09* (6.74) UNEMPLOYMENT 9.23* (7.14) GEOGRAPHIC LOCATION (EAST) 250.4* (6.74) GEOGRAPHIC LOCATION (WEST) 36.98* (8.86) CHAPTER VII SUMMARY AND CONCLUSIONS n the present study, attempt to provide an understanding of the housing market re buyers, sellers and real estate agents interact with each other while making optimal decisions for themselves. The housing market has several distinct characteristics which makes this market different from most other markets. spatial fixity, durability and the extensive These features are heterogeneity, involvement of the government. These attributes and imperfect information among buyers and sellers provide an opportunity for real estate agents to serve n this paper as marketmakers arranging transaction develop a theoretical model of the between buyers and sellers. interaction among sellers, buyers and real estate agents. Taking nto account an uncertain date of sale and various periodic costs the seller must decide whether to list the house with an agent or to attempt to sell the house without an agent' intermediation. rs may visit ownersellers and/o use the services of an agent. estate agents must ch oose a commission rate with the understanding that buy rs and sellers may arrange transactions directly without ntermediation. mode explicitly considers dependent random processes involving decision of buy rs to visit houses and make offers. n this un certain environment sellers must determine the expected gain by attempting to attract buyers directly or by listing estate agents. Real estate agents maximize expected returns under the constraint that they choose commission rates which are The model efficiently favorable to attract listings by sellers. initially considers a market with a single real estate agent and then extends results to introduce numerous agents with a multiple listing service. In either case the outcomes are strongly influenced by the competitive pressures imposed by sellers with the option of arranging trades without real estate agent intermediaries. In the remainder of the paper taking into account outcomes n the theoretical model , I develop an empirical model which simultaneously estimates the commission rate and sellers' and buyers' interaction with a broker which is indicated by brokered sales and nonbrokered sales use a two stage simultaneous model which estimates system of equations consisting of commission rates and the probability of brokered sales. n the first stage, the reduced form equation are estimated with the commission rate equation determined by tobit (because commission rate is not observed for all cases) and the probability of brokered sale by probit procedures. Then the reduced form parameters are substituted the structural form and once again, the commission rate equation estimated by the Tobit procedure and the probability of brokered sales is estimated by probit. use data consisting of 540 housing transactions from fifteen states across the U.S. in the estimation. Although the results from the estimation procedure are mixed and somewhat capturing disappointing, essence the procedure itself consumer an important estate agent first step behavior in correctly the housing markets. Also one of the main contribution of the present pape is to develop an mnnrnflrh uwhirh vnlieitflw rnncirr intorrntinne hotfA/oon hi i/oror anr4 collore nrf nrinr APPENDIX A MATHEMATICAL APPENDIX Let F(t, m, = (eP  1)eApt Then F(t, m, *= increases with increase nm or N and decreases with an increase n M if d(Xp)/dm > 0, d(Xp)/dN > 0 and d(Xlp)/dM Proof: Note that d( t, m,M,N) d(Xp) = eXP(e teAt^ = e pt[ep(i Next, use mathematical induction to v rify that z d(Xp) dF(t,m,M,N) Cr e1 Pt(eAP(1 d(Ap) t) +t) = T'rp We hav ume = (T1) (Ti)Xp Then + T] = = Te (Note that TAp < 1/2 for efficiently large. Tip < Tr2 ence = 1/T for T sufficiently large * CO) Finally, T dIE F(t,m,M,N) = TeT /fgd~p} = S,[dm(p dm + e d[t) F(t, m, M, N)] pi d( IP if d( p) dN I~=Ft ,M ) <0; if d(p) dM d[~ ~ t, m, M, N)1 APPENDIX B ANTITRUST ISSUES AND REAL ESTATE BROKERS The real estate brokerage industry has endured frequent antitrust litigations. Most of these cases have involved fixing of commission rates by loca boards of realtors in order to restrain trade and competition but other issues hav also arisen. An examination of the antitrust implication of brokerage activity led to three main concerns. are the actions relating to barriers to entry into th brokerage industry which might lead to limited competition. Seco nd are th xing of commission rates by th local real estate boards. Finally, there are concern about the tieins among brokers and other real estate conveyancing services. Here look at some landmark cases dealing with the abov issues and note the udgments pronounced in each case. majority antitrust violation private industry considered under the stipulations of th erma n Antitrust Act of 1860. This act outlaws three types of practices, of monopoly power namely, (i) possession and willfu n the relevant market acquisition and maintenance ntent to fix prices and eliminate competition and ii) collusion among firm with the ntent to monopolize. Apart from the Sheman Act there are two other acts which govern antitrust ues Clayton Act 1914) and the RobinsonPatman Act (1936). The former prohibits m ergers which would lead to Barriers to Entry Under the Sherman Act (as wel as under the other two acts) it is illegal to restrict entry to an industry in order to monopolize it. of the main con cerns that this issue creates real estate brokerage industry has been that brokers who are not members of the local board of realtors are not allowed access to the multiple listing service n the area. One of the most important cases concerned with barriers to entry is Grille Vs. Board of Realtors of Plainsfield Area (9 Super 202, 21 8219, 219 A.2d 635, 1966). The case n question was that a N ew J ersey brok r who was not a membe of the Plainsfield Areas Board of Realtors complained that the Board of Realtors prohibited non members from using th Board' multiple listing service. This case was a bemnchmark case for all cases in the future. The verdict in the case was that it is illegal to restrict th use of MLS only to board members (unless membership to a board is open to anyone who is interested under th statutes of the Sherman Act. There have been some cases re th judg eme nt was n favor of the defending board of realtors. However circumstances in these cases were peculia (see Miller Shedd (1979), a discussion of these cases). Based on the Grille case and a vast majority of the cases , it is now well accepted that the use of any MLS information should be open to any licenced brokers n the area whether the broker belongs to the board of realtors or not. Commission Rate Fixinq The real estate brokers have eni nucar4 f~i j III v uniform commission rates since the uV. suits. Under the Sherman Act it is stated that any contract or conspiracy that results in restriction of interstate trade is illegal. The landmark case about commission rate fixing is the United States Vs. National Association of Real Estate Boards (339 U 1950 n this case the Washington Board of Real Estate required that th members could not accept lower commission than those imposed by the Board. Here, the U . Supreme Court ruled that the Board of Realtors was voilating the stipulation of th Sherman Act and that it is illegal to restrain trade through price fixing, indicated by the mandatory commission rate. Since the ruling of the Suprem Court n 1950 real estate boards nthe U not have any prescribed or mandatory commission rates mentioned in their code of After 1950, commission rates for member realtors were "recommended" by most of th local board of realtors. For example, until 1972 they were recommended by the New York Board and thereafte code of ethics manual of the N merely declared tha ew York Board of R such rates w ere "fair" ealtors all rates w In the 1974 re removed Barasch 1974 for a detailed discu Convevancinq Costs Excessiv cost of conveyancing serve is also subject to antitrust statutes. joint report by the U Department of Housing and Urban Development and the Veteran' Administration 1972) indicated that these costs are too high n many areas in the U. Conveyancing costs are th closing costs of serve Excessive costs were attributed to provided to custom an elaborate rs at the time of system of rebates, kickbacks and referral fees n this industry, (ii) high level of duplication and (iii) other nefficiencies present in this industry. The antitrust concern costs are that brokers receive rebates from title search insurance) companies and hence designate such companies to buyers and reduce competition among firms providing closing services. Despite the system of rebates and tieins, it has been felt that there has not been much attention given to these aspects of closing costs Owen and Grumfest ,1977). They suggest a reform n the egal structure to corporate these antitrust issues. Epley and Parson 1976 present a list of practices of transaction charged to be illegal because of their anticompetitive nature. The discussion presented n this Appendix suggests that the brokerage industry has endured antitrust litigation for half a century and thus that brokers are extremely cautious in dealing with each other as well as with buye rs and sellers of real property. APPENDIX U.S. DEPARTMENT OF HOUSING AND URBAN DEVELOPMENT SETTLEMENT STATEMENT HUD1 FORM A. Settlement Statement U.S. Departmem of Houing end Urlb n Demoeant OMB No. 25020265 (Exo. I i3r 231486) B. Tyn of Loan 1. O FHA 2. O FmHA 3. ] Conv. Ungn.. '"U Nus. Lo rn S. Motn . c"m Nge 4. 0 VA 5. C Cov. ins. C. No This form s fumished to giv you a statement of actual settlement costs. Amounts paid to and by the settlement agent are shown. Items marked "(p.o.c.)" wer paid outside the closing; they are shown henre for informational purpose and are not included in the totals.___ D. Nna ad AdMOre @o4 onfl U. Nam wd AW d nl SE 5l4 F. Nf~ w Ad Lam@4f Ian G Prowlv uLOan n H. kS re( AgM Pi o gImmeiM I. Sstwmnt4 O. J. Summry o narrow's Trnsctlosm K. Summe ot kow's Tnnmesfon 100. Grom Amoum Due PFrm em _r _400. OGre Amou l Due To l841w 101. Contract saMe price 401. Contract saMles pce 102. Personal properly_ 402. Personal popn 103. Settlement chare to boower 105. 4.____ Ad Ummue tow items mMs e i. enmMce Ad mls te s pad by ssll In Sfencl 106. Citytown taxes to 406. Cl town taxes to 107. County taxes to 407. Count taxes to 106. Aslasnments to 4_0. Assessments to 109. 409_ 110. 10._______ 111. 411. 112. 412. 120. Grou Amoust De From onsmse 420. Grow Amman Oue To SMIw 200. Anmues Phid Otr DIen el Of S Son o00. Rdulmns hi Amuam Due To. S6ll 201. Deposit or eanmet money 501. Excess dol fte Ie tnstructions) 202. Princp amount of new a 502. Settlement char s to sllr (line 1400 ____ 203. Existing loIsI) takes to S03. Existing olon(s) tten sub ict to _____ 204. 504. Payoff of firnt mortgage 10m 205. _os. Pay ff second modta loan 206 506. 207 7. 208 . 209 . Ador Ahf u npm id by A_ *domen tr ba a *if 210. City/town taxes Ito 10. Cl/town taxes to 211. County taxe to_ 511. CounIt taxes_ _to 212, AslesMlmens to_ 512. Assessments to 213. 13. 214. 514. 215. 15. 216, _516. 217. 51. 218. 518. 219. 1. L Settlement Charg4es  700. Total $Sallrok'a ComnsrMI be on Pd From Paid From Division of Commission line 700 as follows:orrowr Sler 701. $ to Funds *t Funds at Settlement Settlement 702. S to 703. Commission d at Settlement 704. ___________ 300. Items Pyable In Connetion Wlh Laan 801. Loan Origination Fee % BP I oan Discount % 803. A grasa Fee to 804. Credit Report to 805. Lender's Inspection Fee _ 806. Mortgae Insurance Appication Fee to 807. Assumption Fe 809___ 810. 011. BOO, Iem Required By Lender To Be Paid In Advewne 901. Interest from to IdG 902. Mortgage Insurance Premium for month to 903. Hazard Insurance Premium for e ar to ____ e9t. o _to 1000. Reumyen 1elltod With Ladr 1001. Hazard insurance month$ er month ____ 1002. Mortgage insurance monthsfl cmpr month 1003. City prop taxes month S r month :004. Count propr taxes month r month _____ __ 1005. Annual assessments monthaS T month _____ 100e. monthelr er month 1007. month per month ____ 1008. monthast pr month 1100. Title Chu__ 1101. Settlement or closing fee to_ 1102. Abstract or title search to 1103. Title examination to 1104. Title tnurance binder to 1105. Document preparation to '06. Notary fees to________ 1107. Attomey 's fees to (includes above items numbe:____ 1108. Title insurance to includess above items number: _ 1 109. Lender's coveraGe S_ 1110. Owner's cover S___ 1111. 1112. 1113. 1200. Gewanomet Mrin and Tranufer C lens ___ ,,, 1201. Recording tees: Deed S ; Mortage a ; Reeaes t _ 1202. City/county taxstamps: Deed : Mortgage S 1203. State tax/tamp: Deed S ; Mortga ge ___ 1204. 1205. 1300. Adchlonel SethmeeM CMgee 1301. Survey to APPENDIX D RESULTS FROM SEPARATE ESTIMATION OF EQUATIONS (23), (24) AND (25) Estimated Separate Tobit Equation for Commission Rate VARIABLE DESCRIPTION COEFFICIENT ONE 0.26 (0.77) BROKERED SALE DUMMY 6.6 (24.72) PRICE 0.1 3E01 (3.95) HOUSING PERMITS 0.69E04 (1.55) tratios are provided in parentheses.] Estimated Separate Form Probit Equation for Brokered Sale Dummy VARIABLE DESCRIPTION COEFFICIENT ONE 0.17 (5.11) COMMISSION RATE 0.12 (31.17) PRICE 0.20E02 (4.72) DOWN PAYMENT 0.30E03 (0.21) UNEMPLOYMENT 0.13E02 I c r,\ Estimated Separate Equation for Price VARIABLE DESCRIPTION COEFFICIENT ONE 63.63 (25.27) COMMISSION RATE 3.52 (5.2) BROKERED SALE DUMMY 21.67 (4.61) UNEMPLOYMENT 0.56 (5.22) GEOGRAPHIC LOCATION (EAST) 0.11 (4.05) GEOGRAPHIC LOCATION (WEST) 9.02 (4.43) tratios are provided in parentheses.] REFERENCES Arnott, Richard J. and Ronald E. Gri eson 1983), "The Supply of Urban Housing," in Ronald E. Grieson (ed.) The Lexington Books, 310. Urban Econ omv and Housina Lexington, Barasch Clarence 1974 "How Antitrust Action Affected Real Estate Brokers' Commissions." Real Estate Law Journal ,3, 227240. Bradfield , James and Edward Zabel (1979) "Price Adjustments n a Competitive Market and th Equilibrium, Securities Exchange Growth and Trade >ecialist"1 New Yo in Green Scheinkman Academic P eds.) General ress. Carney Michael (1982), "Costs and Pricing of Home Brokerage rvices" AREUEA Journal , 10, 33 334. Crockett, John H. 1982) "Competitition and Effici ency Transacting: The Case of Residential Real Estate Brokerage" AREUEA Journa ,10,20 9227 Deaton , Angus and John Muellbaue (1980), Economics and Con umer Behavior Cambridge: Cambridge Univ ersity P ress. DeBrock , Larry M. (1988), "Joint Marketing Efforts and Pricing Behavior" October, Faculty Working Paper No. 1500, College of Comm erce and Bu ness Administration ersity of Illinois at Urbana Champaign. 1989), "The Effects of an MLS on the Efficiency of Housing and Brokerag Markets," March , Working Paper, Department of Econom , University of Illinois , Champaign, IL. Diamond , Douglas (1980), "Taxe inflation , Speculation Cost Homeownership," AREUEA Journal, 8. 281298. Duensenberry, James (1958), Business Cycles Econom Growth New York: McGraw Hill. Edelstein Robert H. (1983), "The Productio n Function for Housing and Its Implications for Future Urban Development," in G North American Housing Markets eorg W. Gau and Michae into the Twe A. Goldberg (eds.) ntyFirst Century, Cambridge, Epley Donald R and Cameron Parsons 976),"Real Estate Transactions and the Sherman Act: How to Approach an Antitrust Suit," R al Estate Law Journal '5, 3 Federal Trade Commission (1984), Reports Vols I and II. The Residential Real Estate Brokerage industry, Staff Fromm, Gary (1973), "Econometric Mod prisonn" in R. Bruce Ricks (ed) of the Resid ntial Construction Sector: A Com National Housinq Models, Lexington, Heath Lexington Books, 125155. Grebler Leo and Sherman J. Maisel (1963) "Determinants of Residential Construction A Review of Present Knowledge," n Danie Suits (ed impacts of Monetary Policy, Englewood Cliffs, NJ: Prenti ceHall 475620. Harris and A. Raviv (1979), "Optimal incentive Contracts with perfect information" Journal of Econom Theory 20. 231259. Henderson Vernon (1985), Economic eory and the Cities, Orlando : Academic Press. Donald 1983), "Real Estate Brokers and the Market for Residentia Housing" AREUEA Journal 6982. de Leeuw, Frank (1971), 'The Demand fo Review of Economic Statistics. 5 Housing: (1. 110. eview of CrossSection Evidence," and Nkanta F. Ekanem (1971 "The Supply of Rental Hou ng," American Economic Review ,61(5), 806817. Lippman, S.A and J. McCall (1976), "The Economics of Job A Survey," Economic Inquiry, 155189. Maddala 1983), LimitedDependent and Qualitativ Variables in Econometrics, Econometric Society Monographs, , Cambridge: Cambridge University Press. Mayo, Stephen K. (1981), 'Theory and Estimation in the Economics of Housing Demand," Journal of Urban Econom , 10(1), 95116. McCal (1970), "Economics of information and Job Search," Quarterly Journa Economics McDonald John F . 1Q79Q\ I CI U I  Economic Analysis .I an Urban Housina Market. New York: Miller Norman G. and Peter J. Shedd (1979), Brokerage Industry?," American Business "Do Antitrust Law Law Journal Apply to the Real Estate 7. 313339. Muth Richard F. (1960), 'The Demand for NonFarm Housing," in A. Harberger (ed.) The Demand fo Durable Good s, Chicago rsity of Chicago Press, 2996. and Allen C. Fundamentals of Pure Goodman and App (1989), The ed Econom Economics of Housinhq Markets, 31. Harwood Academic Publishers Olsen , Edgar O. Review, 1969), "Competitive ,59(4) eory of th Housing Market," Am erican Economic 612622. Owen Bruce M. and Grundfest J osep 1977 "Kickbacks , Specialization, Price Fixing, and Efficiency in Residentia 931967. Real Estate Markets." Stanford Law Review Parzen , E. (1960), Models of Probability Theor v and Its Applications New York: Wiley People Vs. National Association of Realtors 981), California Reporter Appendix 174, 728743. Quigley, John M. (1979), "What Have Learn ed About Urban Hou ng Markets?" in Peter Mieszkow Mahlon Straszheim Current Issues Urban Economics, Baltimore: The Johns Hopkin University Press, 391429. Ross Sheldon M. (1972 Introduction to Probability Models N ew York: Academic Press, nate Committee on Banking, sing and Urban Affairs Costs: Report of Department of Housi 1972), nq and Urban Dev Mortgage Settlement opment and Veterans' Administration 92d Cong ress Session. Smith Bruce (1983), "Limited Information , Credit Rationing, and Optimal Government Lending Policy," Am can Econom Review ,73( 305318. Smith ,Lawrence (1969), "A Model o Canadian Housing and Mortgage Makets," Journal of Political Economy 77(5 795816. Kenneth T. R in Economic Models osen and George housing Markets," Fallis (1988), journal of recent Developments Economic Literature. 26. 29 Staoleton. Vn inn SQR4d\ "f'ncn nral N~nrm~I ''' 'I*U . It WUII' *%d II 1 t . I Rearession I VVIVH i IYIIIIWI YI VII W Weiss Yoram 1978), "Capital Renting and Owning a H Gains house, , Discriminatory " Journal of Pub axes Choice Econom between 4556. Wheaton m C. Criticism 1979) "Mono centri n Peter Mieszkowski : Models of Urban Land Use: and Malhom Straszheim (ec Contribution s.) Current issues n Urban Economics, Baltimore, MD: The Johns Hopkin University P ress. Wu. Cunchi and Peter Colwell (1986), "Equilibrium Hous ng and Real Estate Brokerage Markets Under Uncertainty," AREUEA Journal, Yinger John 1981 "A Search Model of Real Estate Broker Beh avior." American Economic Review, 591605. Zabel , Edward (1981), "Competit Price Adjustment Without Market Clearing," Econometrica 12011221. Zumpano, Leonard V A Critica and Donald L. H ooks (1 Reevaluation." AREUEA Journal , 16(1), Real Estate Brokerage Market: 116. BIOGRAPHICAL SKET Suryamani Mantrala was born ew Delhi India in 1961 After graduating from Univers ity of Delh economics with Bachelo taught at the of Arts (Honors) and Master of Arts degrees in rsity of Delh in 19831984. She joined the Unive rsity of nois at Chicago re sh received a Master of Arts degree economy entered th doctor program in economy at the Univ rsity of Florida in 1986 and expects to receive the degree of Doctor of Philosophy in M certify have read this study acceptable standards of scholarly presentation and that opinion and is fully adequate, t conforms to in scope and quality as a dissertation for the degree of Doctor of Philosophy. trc~cef Edward Zabel Chiman Matherly Professor of Economics certify have read this study acceptable standards of scholarly presentation and t in my is fully opinion adequate, conforms in scope and quality, as a dissertation for the degree of Doct r of Philosophy. IJ Maddala Graduate Research Professor of Economics certify that have read th acceptable standards of scholarly study opinion presentation and is fully adequate, conforms in scope and quality, as a dissertation for the degree of Doctor of Philosophy. David Denslow Professo of Economics certify have acceptable standards of quality read study scholarly presentation as a dissertation for the degree of Do or of it in my opinion is fully adequate, philosophy. .4 conforms in scope and Barton Weitz J. C. Penney Professor of Marketing This dissertation was submitted to the Graduate Faculty of the Department of Economics in the College of Business was accepted Philosophy. Administration and to the Graduate School and as partial fulfillment of the requirements for the degree of Doctor of May, 1992 Dean Graduate School UNIVERSITY OF FLORIDA HIIII Il i11 in1 I2 ill lln I 3 1262 08556 7153 