Determinants of Real Estate Agent Compensation Choice
Total Page:16
File Type:pdf, Size:1020Kb
Determinants of Real Estate Agent Compensation Choice
Lenny V. Zumpano The College of Commerce and Business Administration The University of Alabama [email protected]
Ken H. Johnson College of Business Administration Florida International University [email protected]
Randy I. Anderson College of Business Administration University of Central Florida [email protected] Abstract
This research seeks to determine what factors are decisive when an agent chooses between a 100% payout and the more traditional split-commission arrangement with his firm. Besides the expected positive relationship between income and the 100% payout election, tolerance for risk, a complementary sales force, and experience also influence the choice of compensation arrangement of agents. These finding suggest that compensation arrangements may not always be effective markers of agent productivity and that compensation incentives alone may not elicit greater effort and output.
I. Introduction
The impact of real estate brokerage intermediation has been the subject of much study over the past decade. Most of this research looks at either differential market outcomes
(price and marketing time) across brokerage categories or outcome differences between brokered and non-brokered transactions. Recently, the compensation arrangement between an agent and his firm has come under investigation with conflicting results. In particular, claims are made by some researchers that differences in skill levels among agents can be discerned by the type of compensation arrangement (100% vs. Spilt) between the agent and his firm1, suggesting the possibility that some agents can sell properties at premium prices and/or over shorter time horizons than is the case with other agent-assisted transactions handled by less skilled or motivated agents (Munneke and
Yavas, 2001; Allen, Faircloth, Forgey, and Rutherford, 2003; and Johnson, Zumpano, and Anderson, 2008). Such findings have important efficiency implications, as most studies of the market for brokerage services have assumed no differences in agent skill levels.
2 So far, however, no study addresses the more fundamental, prerequisite question of what factors determine an agent’s choice of compensation arrangement between himself and his firm. If these factors are positively associated with agent productivity and motivation, then one might reasonably expect differences in agent performance to be systematically linked to the types of compensation incentives available.
The goal of this research is to model the choice of compensation arrangement between an agent and his firm.2 The theoretical and empirical literature is examined to help identify important determinants of this choice. Empirically, compensation choice is modeled as a
Probit estimate using the National Association of Realtors® 2001 Membership Survey, which provides a national, cross-section profile of NAR member characteristics.
The next section of this paper examines the relevant literature on agent performance.
Sections on methodology and data, empirical analysis, and concluding comments follow in order.
II. Literature Review
IIa Buyer Search and Market Intermediation
Many researchers have examined the impact of broker intermediation in the marketplace.
There is a large body of research on the search process in the real estate market, much of it focuses on the seller’s side of the market, and is commonly referred to as the pricing and time on market literature3. Other research, however, has begun to look at buyer
3 search and the role of the intermediary. Early studies by Jud (1983) and Jud and Frew
(1986) suggested that intermediation by real estate agents influences search time and increases the demand for housing. In the latter study, Jud and Frew find that agent- assisted buyers have a greater demand for housing than buyers who search without benefit of an agent. They attribute this outcome to the informational impact of agents, offering the analogy of the effect of advertising in markets with imperfect information.
A paper by Baryla and Zumpano (1995) finds that broker intermediation reduces search time for all classes of consumers, regardless of their demographic characteristics.
Zumpano, Elder, and Baryla (1996) examining the decision to use an agent subsequently find that buyers with high opportunity costs (travel, time, and information) of search are the most likely to use agents, which in turn reduces search time. Additionally, Elder,
Zumpano, and Baryla (1999) determine that agents, by reducing within-period search costs, increase buyer search intensity, thereby reducing search duration.4
In a more recent paper, Elder, Zumpano, and Baryla (2000) find that the type of agent has a discernable effect on the search duration of buyers. Specifically, the authors find buyer agents are more effective at reducing search time for their clients than more traditional seller agents or non-agent facilitators. As is the case in most earlier studies, agents do not affect price, no matter the type of agent. While this line of research does generally support the idea that agent actions can affect market outcomes, it does not differentiate agent performance on the basis of compensation.
4 IIb Incentives and Agent Performance
Recently, three papers (Munneke and Yavas (2001), Allen, Faircloth, Forgey, and
Rutherford (2003), and Johnson, Zumpano, and Anderson (2008)) appeared which try to assess whether different agent compensation arrangements could serve as proxies for differences in skill/motivation levels of agents as reflected by differences in time on the market and selling price of properties market by this classification of agent. However, their empirical findings are not reconcilable. Munneke and Yavas find that full-payout,
REMAX agents, presumed to be more skilled than their split-commission counterparts, have no long-term impact on selling price or property marketing time. Allen et al., on the other hand, using a similar construct but different data, find that residential properties marketed by 100% agents are sold more quickly and at a premium relative to homes sold by “less productive” agents. Johnson et al. find that 100% agents reduced search time, but have no effect on selling price.
In all three studies the way in which the agent is compensated is singled out as the appropriate productivity marker or signal. Productive agents are differentiated from their less productive counterparts by the way they are compensated by their firm. The 100% payout election, such as that offered by REMAX, is deemed to attract more productive agents because they will earn more money than if they share their commissions with their broker-owners.
5 In the case of the Munneke and Yavas and the Allen, et al., neither study actually links compensation plans to agent output or actual compensation differences among agents.
Instead they look at whether selling prices and marketing times are related in a systematic way to agent compensation structure. Both papers assumed that all 100% agents work for REMAX firms and all split-commission agents are employed by other companies.
Segregating the sample by firm rather than at the agent level may, therefore, create specification problems because some split–commission companies also have 100% commission agents on their payroll.5
More importantly, the papers cited above either do not model the choice of compensation plans or employ one dimensional, self-selection criteria that makes the choice of compensation arrangement a simple financial decision. Munneke and Yavas, for example, consider it a simple adverse selection problem. Agents base their decision as to which type of compensation arrangement to choose using a simple break-even analysis.
In this setting, a highly skilled or motivated agent maximizes his earning by choosing a
100% commission arrangement as his expected production is such that his gross income from his total closed volume, less his periodic payments to the firm, will exceed what he is expected to earn with a split-commission plan.
In point of fact, there are many other factors that could influence agent productivity besides compensation arrangements, including, an agent’s risk tolerance, years of work experience, education, current and other income levels, the location and size of the market, and the size of the firm where the agent works. Until the link between
6 compensation arrangements and productivity is proven, nothing definitive can be said about how alternative compensation plans impact agent performance and market outcomes.
IIc Human Capital, Earnings Models, and Agency Theory
A number of studies have examined the determinants of real estate agent earnings6.
Numerous papers by, among others, Follain, Lutes, and Meir (1987), Glower and
Hendershott (1988), Crellin, Frew, and Jud (1988), find education, experience, size of firm, and the number of hours worked per week, all have a positive impact on agent income. Abelson, Kacmar, and Jackofsky (1990) and Sirmans and Swicegood (1997) also find a positive relationship between non-pecuniary factors such as job satisfaction and agent earnings. None of these papers, however, directly examine the link between income and agent performance.
Additional insights into the linkages between agent compensation choice and agent characteristics can be gleaned from the vast amount of research on the moral hazard problem that arises when agent actions cannot be monitored. Commonly referred to as the principal-agent problem, much of this literature examines agent performance and the use of incentive compatible contracts, such as percentage commissions, to better align principal’s and agent’s interest. Some of this research indicates that neither flat fees nor percentage commissions perfectly align the interests of both principal and agent.7 For example, Holmstrom (1979) shows that when an outcome is a function of both effort and
7 a random state of nature, basing an agent’s compensation on outcomes alone is a second- best solution8.
Other research [Frederickson (1992) and Stewart (1999)] examines the efficacy of alternative payment arrangements, such as relative performance measures and contests, as ways to either improve risk sharing between the principal and agent without reducing agent effort, or alternatively, provide greater agent incentives without imposing additional risks on the agent. In a real estate brokerage context, when the variance of outcomes is high due to factors beyond the control of the agent, such as changes in mortgage interest rates or the state of the economy, risk-averse agents would be encouraged to seek improved risk-sharing arrangements with their companies.
Still other researchers have criticized principal-agent theory that is based solely on economic or financial incentives. Two papers [Frey (1997) and Fehr and Falk (2002)] present work that suggests that there are also intrinsic, non-pecuniary incentives and rewards that can influence work effort and performance; a finding consistent with some of the real estate agent earnings research cited above. In some cases, economic incentives complement intrinsic incentives such as job satisfaction, social approval, and worker morale, while in other cases financial incentives (or penalties) can “crowd out” such intrinsic incentives, thereby reducing performance. Financial incentives may also be counterproductive if they lead to undesirable behavior (sabotage of co-worker performance) by employees. These authors point to studies showing a weak relationship between financial incentives and worker performance as support for their hypotheses.
8 III. The Model and Methodology
IIIa The Variables
This study seeks to determine whether any considerations besides productivity may cause an agent to choose a full-payout compensation arrangement versus a split-commission arrangement with his firm. To date, there does not appear to be any studies that explicitly model how and why an agent, when given the choice, chooses a specific compensation arrangment. The previous literature review does, however, provide insights into some of the factors that may influence this choice. For organizational purposes three broad, non- mutually exclusive, categories (job characteristics, agent characteristics, and risk tolerance) of explanatory variables are created.
IIIa1 Job Characteristic
Job-related characteristics include SalesStaffOffice, which represents the number of other agents in the agent’s office. This study hypothesizes that the larger the sales force, the greater the marketing impact of the agent’s firm and the greater the probability of closings, both co-op and the typically more lucrative in-house sale, which should make it easier for an agent to meet and exceed the periodic fees associated with a 100% commission structure. Therefore, SalesStaffOffice is expected to be positive and significantly related to choosing a 100 % contract. Similar synergies may be present as the firm itself gets larger spatially, increasing its market area which may allow agents to reduce the risk associated with the 100% contract structure. To capture this effect, the
9 total number of offices (Offices) for a given firm is specified. In the interest of brevity,
Exhibit 1 provides definitions and abbreviations for these, as well as, all other variables employed herein.
IIIa2 Agent Characteristics
Experience, represents the total number of years an agent has practiced in residential real estate. It is conjectured that this variable will have a positive relationship with an agent’s choice to become a 100% agent. With job tenure, agents gain experience and knowledge, which are critical to an agent’s ability to generate the listings and sales necessary to cover the fixed costs associated with full-payout compensation arrangements. However, it also seems logical that there are diminishing returns associated with experience; therefore,
ExperienceSqrd is specified in the Probit estimation to control for the decreasing marginal impact of agent experience on choice of compensation arrangement. If these suppositions are correct, Experience will sign positive and significant while
ExperienceSqrd will sign negative and significant.
The agent earnings and human capital literature suggests, in addition to experience, that agent-related compensation decision parameters may also include education (Education) and agent age (AGE), both of which are included in the model. If education improves agent performance it should be positively associated with the choice of the full-payout compensation arrangement. If knowledge increases with age, agent age should be positively related to the 100% commission choice. On the other hand, older agents may
10 be more or less risk-averse than younger agents due to their personal wealth level making it difficult a priori to determine the sign of AGE.
Finally, a number of studies have argued that full-payout compensation arrangements are positively linked to agent productivity and work effort. Accordingly, HoursWorked, representing the average number of hours worked per week by the agent, is entered as a proxy for agent effort in the compensation choice model. A priori, it seems reasonable to assume a positive relationship between this proxy and the choice of compensation arrangement.
IIIa3 Risk Tolerance Vatiables
Risk tolerance should play a role in the choice of compensation package. With the traditional split-commission arrangements, an agent is assured of some income, even at low levels of productivity. In contrast, full-payout agents will receive no income until their commission production exceeds their payments to the firm. Although this form of operating leverage can prove very profitable for highly risk tolerant agents seeking to maximize their incomes, more risk-averse agents may opt out for less risky income sharing arrangements.
Risk preference can be proxied by such things as marital status, gender, how often an agent changes jobs, and both agent and other income noted in the model and exhibits as
Married, Male, YearsCurrentFirm, Income, and IncomeDifference, respectively. The
11 potential marginal impact of each of these risk tolerance variables on agent compensation choice is discussed in turn.
Married agents, especially those with children, might be expected to be more risk-averse than single agents as more people are dependent upon their earnings. As a result, they may prefer the risk-sharing attribute of the more traditional split-commission agent arrangement. If this reasoning is sound, Married should enter as a negative predictor of compensation choice.
Psychology studies have shown that in areas such as finance, men are more overconfident than women. In particular, they are more overconfident about their abilities, knowledge, and future prospects. Empirical support for the contention that males tend to be more overconfident than their female counterparts is provided by Barber and Odean (2001) who examined the stock trading records of over 35 thousand households finding the turnover rate for men was nearly one and a half times greater than that of female stock traders. Such overconfidence could easily transfer to compensation choices as well resulting in Male having a positive affect on compensation choice.
Additionally, a causal relationship is expected between agent income and the choice of compensation plan. If productive agents are more likely to choose full-payout plans, then income earned as an agent should be positively and significantly related to this choice.
Interest here, however, is in the degree to which past income proxies anticipated future income earning ability, and, therefore, an agent’s ex ante estimate of the likelihood of
12 covering the periodic fee associated with 100% payouts. Thus, the higher the agent’s last year’s income, the lower the risk of not being able to meet the periodic fee. Accordingly, increases in Income should be associated with a greater likelihood of becoming a 100% agent. For similar reasons, the total household income of an agent, another risk proxy, should also be positively related to the choice of a 100% commission payout. As outside incomes increase, agents become less dependent upon commission income. Thus,
IncomeDifference is included as a variable representing the difference between an agent’s annual income from their residential practice and their gross annual reported income, which should sign positive and significant in the Probit estimation.
Non-pecuniary factors, such as job satisfaction and social approval can reasonably be expected to influence compensation arrangement choice. Such incentives, however, are difficult to proxy given the database, although years with the current firm
(YearsCurrentFirm) may reflect the attractiveness of the work environment, separate and apart from productivity. As an alternative explanation, the amount of time spent with a given firm could also act as a proxy for willingness to bear the risk of becoming a 100% agent. In particular, it seems plausible that non-risk takers will be more likely to remain at the same firm while risk takers, seeking ever better employment opportunities, will change jobs more frequently and, hence, have shorter job tenure. Therefore, all else being equal, it is reasonable to expect YearsCurrentFirm to be inversely related to becoming a 100% agent.
IIIb The Model
13 The dependent variable of this study is agent compensation choice (CC). It is a categorical variable coded as either a 1 or a 0. Agents that choose to pay their firm a periodic fee in exchange for retaining 100% of their generated commissions, receive a 1 for identification purposes. Agents that choose the more traditional split-commission arrangement with their firm are coded with a 0 designation. Split-commission agents differ from their full-payout counterparts in that they do not pay a periodic fee to their firm. Instead, they split their generated commissions with their firm on a negotiated percentage basis with 50/50, 60/40, and 70/30 sharing arrangements being commonly agreed upon splits. 100% agents, on the other hand, pay their firm a periodic fee in exchange for all of their earned commissions.
Following the arguments outlined above the following Probit estimation is specified to model an agent’s compensation choice:
CC = f (YearsCurrentFirm, Experience, ExperienceSqrd, HoursWorked, Income, IncomeDifference, Male, SalesStaffOffice,Age, Married, Education, Offices) (1)
IIIc The Data
The data for this study comes from the National Association of Realtors® 2001
Membership Survey. The 2001 survey is chosen because it appears to be more representative of a typical agent force participating in a constrained market place such as the one faced by agents today as opposed to the makeup of the agent population during the recently past, rapidly escalating property market. Said another way, the typical agent
14 force seems more likely to look like that of 2001 than that of recent years making it sensible to use prior data.
The Membership Survey is a national survey of NAR members that contains questions on business activity, technology use, office information, agent production, agent income, and other agent demographic information. The raw data set consists of 7,440 responses.
However, many of the responses were either incomplete, provide erroneous information such as negative income, or were completed by individuals who do not practice in residential real estate on a daily basis, making the assignment of variability in the model suspect due to the loss of degrees of freedom fostered by these absences. Thus, these incomplete responses are eliminated in order to avoid this potential problem. Next, since the survey also indicates that very few agents are on straight salaries, this choice is excluded from the analysis that follows. Therefore, the observations are limited to those agents that have sufficiently completed the questionnaire, i.e. provided answers to all the necessary variables employed herein.
After making these adjustments to the data set, a final sample of 1,853 observations remain for analysis. Copies of the full sample and the Membership Survey itself are available from the National Association of Realtors. Descriptive statistics for the examined sample are provided in Exhibit 2. Additionally, descriptive statistics split by compensation choice are available in Exhibit 3. A casual review of these split statistics suggests that many of the proposed modeling arguments seem warranted. For example, casual observation (i.e., without statistical validation) suggests that the means for Income
15 and IncomeDifference appear to be higher for 100% agents than those of Split-agents.
The next section discusses the descriptive statistics in additional detail and provides a review of the empirical findings.
IV. Empirical Results:
IVa Analysis of Descriptive Statistics
Examining the basic demographic information in Exhibit 2 reveals that less than half the sample are male, while over three-quarters of the sample are married. Additionally, the average sample respondent is nearly 45 years of age and has almost 15 years of education, which translates into the average sample agent having some college or an associate’s degree. The mean years of work experience is slightly over 11, with slightly under 6 of those years being at the respondent’s current firm. The average real estate derived income is slightly more than $123,000 per agent per year. And,
IncomeDifference reveals that the mean difference between what an average agent produces in commission income and their total household income is slightly over
$87,000.
Next, some casual empirical results of the descriptive statistics are examined by dividing the sample into those respondents that are full-payout agents and those who contract as split-commission agents. Note that in Exhibit 3 a value of 1 in the column for compensation choice indicates that the summary statistic refers to the subsample of full- payout agents, while 0 refers to the subsample of agents on a commission split.
16 Contrary to expectations from reasoning provided in Barber and Odean (2001), it seems that males make up slightly less than half the 100% subsample. Additionally, as mentioned above, the income variables are noticeably higher for the 100% subsample.
Finally, there are a number of variables that appear, at least anecdotally, to have no impact on the choice of compensation method. In particular, an agent’s marital status, age, and education are virtually the same for 100% and split-commission agents.
While these casual statistics are suggestive, they are not statistically verifiable; therefore, it is necessary to model these indicators in the presence of one another in order to determine their actual impact on agent compensation choice. The next subsection discusses the results for the specified compensation choice model.
IVb Binary Probit Regression Results
The estimation of compensation choice is formally reported in Exhibit 4. Experience and
ExperienceSqrd, both sign as hypothesized as does YearsCurrentFirm. Thus, it appears that an understanding of the business that tends to comes with experience makes it more likely for an agent to become a 100%er, while job satisfaction (or simply the familiar environment of the more common split-commission arrangement) makes it less likely for an agent to become a 100%er.
17 Not surprisingly, Income and IncomeDifference are both positively related to CC. The higher the agent’s real estate derived income the more likely the 100% commission compensation structure is chosen. Furthermore, the larger an agent’s total income, as indicated by the difference between an agent’s total income and his real estate derived income, the less dependent the agent is on commission income, and the greater the probability that the agent chooses the 100% commission option.
It was anticipated that male agents having, on average, a greater propensity to accept risk would be more likely to chose the 100% payout commission structure. The empirical analysis confirms this reasoning with Male being a significant and positive predictor in the choice of the 100% compensation arrangement. Interestingly, an agent’s age has a negative affect on the probability of choosing to become a 100% agent. Thus, it appears that younger agents, on average, are more willing to take on the additional risk of operating as a full-payout agent.
The model, also, finds that the probability of choosing to be a full-payout agent is positively and significantly related to the size of the sales staff (SalesStaffOffice) in the agent’s office, suggesting that synergies are in play within an office. Specifically, it appears that a large sales staff surrounding an otherwise productive agent leads to a higher probability that this agent chooses a 100% compensation arrangement with their firm. These synergies do not seem to extend to the number of branch offices in an agent’s firm as the number of branch offices does not have a statistical impact on compensation choice. A possible explanation for this outcome comes from a profit
18 maximizing strategy employed by many successful agents. Specifically, it is common for successful agents to establish a “farming” area. These “farms” are areas in which an agent specializes and has significant name recognition and reputation. Obviously, these
“farms” are geographically constrained making additional offices in distant locations irrelevant. Thus, assuming 100% agents (being successful) have a propensity to “farm”, additional branches do not enter into their decision to become a full-payout agent.
Some of the other variables included in the model did not prove statistically significant, despite the apparent importance of these variables in earnings studies. Education and
Married are not important determinants of compensation choice. The reasoning for these outcomes could be couched in the unique nature of the industry. Specifically, it may be such that the demographics of the agents within the industry are significantly different from that of the overall population resulting in non-influential outcomes. Regardless, the explanation for the lack of statistical significance is beyond the scope of this current study.
V. Summary and Conclusions
There has been a great deal of interest in how compensation choice among real estate agents signals differences in agent productivity. Do only the more highly motivated and productive agents choose 100% commission arrangements, based solely on anticipated compensation, or are there other factors that go into this choice? This paper attempts to
19 make the initial investigation into that determination by modeling the choice of compensation scheme between the agent and his firm.
Findings suggest that younger, male agents with significant past commission income and considerable experience in the industry working in a large office tend to favor the 100% compensation arrangement. Other income sources also favorably influence the choice of becoming a full-payout agent. On the other hand, the length of stay at an agent’s current firm lowers the probability of becoming a full-payout agent. Interestingly, the number of hours worked by an agent, his education level, and his being married does not impact compensation arrangement choice.
From an industry standpoint, agents face this choice of compensation arrangement typically once a year. Firm’s regularly struggle with the decision to offer 100% compensation arrangements, the optimal mix of 100ers and split-agents, and the type of agent that is most likely to succeed as a 100%er. Hopefully, the findings in this initial study will facilitate answers to these questions.
From an academic standpoint, this study provides the initial investigation into the determinants of agent compensation choice. The findings suggest that compensation arrangements may not always be effective markers of agent productivity and that compensation incentives alone may not elicit greater effort and output as a number of prior studies suggest and/or implicitly assume.
20 References
Abelson, M.A., K.M. Kacmar, and E.F. Jackofsky. Factors Influencing Real Estate
Brokerage Sales Staff Performance. Journal of Real Estate Research, 1990, 5:2, 265-75.
Allen, M.T., S. Faircloth, F. Forgey, and R.C. Rutherford. Salesperson Compensation and Performance in the Housing Market. Journal of the Academy of Finance, 2003, 1:2
62-71.
Barber, B.M. and T. Odean. Boys Will Be Boys: Gender, Overconfidence, and Common
Stock Investment. The Quarterly Journal of Economics, 2001, 116:1, 261-292.
Baryla, E.A. and L.V. Zumpano. Buyer Search Duration in the Residential Real Estate
Market: The Role of the Real Estate Agent. Journal of Real Estate Research, 1995, 10:1,
1-14.
Benjamin, J.D., G.D. Jud, and D.S. Sirmans. What do We Know about Real Estate
Brokerage? Journal of Real Estate Research, 2000, 20:1/2, 5-30.
Crellin, G.E., J.R. Frew, and G.D. Jud. The Earnings of Realtors: Some Empirical
Evidence. Journal of Real Estate Research, 1988, 3:2, 69-78.
21 Elder, H.W., L.V. Zumpano, and E.A. Baryla. Buyer Search Intensity and the Role of the
Residential Real Estate Broker. Journal of Real Estate Finance and Economics, 1999,
18:3, 351-368.
______. Buyer Agents: Do They Make a Difference? Their Influence on Selling Price and Search Duration. Real Estate Economics, 2000, 28:2, 337-362.
Follain, J.R., T. Lutes and D.A. Meier. Why Do Some Real Estate Salespeople Earn
More Than Others? Journal of Real Estate Research, 1987, 2:3, 73-81.
Fehr, E. and A. Falk. Psychological Foundations of Incentives. European Economic
Review, 2002, 46:4/5, 687-724.
Frey, B.S. On the Relationship Between Intrinsic and Extrinsic Work Motivation.
International Journal of Industrial Organization, 1997, 15:4, 427-439.
Frederickson, J.R. Relative Performance Information: The Effects of Common
Uncertainty and Contract Type on Broker Effort. The Accounting Review, 1992, 67:4,
647-669.
Glower, M. and P.H. Hendershott. The Determinants of REALTOR Income. Journal of
Real Estate Research, 1988, 3:2, 53-68.
22 Holmstrom, R.G. Moral Hazard and Observablility. Bell Journal of Economics, 1979,
10:1, 74-91.
Johnson, K.H., T.M. Springer, and C.M. Brockman. Price Effects of Non-Traditionally
Broker-Marketed Properties. Journal of Real Estate Finance and Economics, 2005, 31:3,
331-343.
Johnson, K.H., L.V. Zumpano, and R.I. Anderson. Intra-Firm real Estate Brokerage
Compensation Choice and Agent Performance. Forthcoming Journal of Real Estate
Research.
Jud, G.D. Real Estate Brokers and the Market for Residential Housing. American Real
Estate and Urban Economics Association Journal, 1983, 11:1, 69-82.
Jud, G.D. and J. Frew. Real Estate Brokers, Housing Prices, and the Demand for
Housing. Urban Studies, 1986, 23:1, 21-31.
Jud G.D., T.G. Seaks, and D.T. Winkler. Time on the Market: The Impact of Residential
Brokerage. Journal of Real Estate Research, 1996, 12:3, 447 – 458.
Morgan, P.B. and R. Manning. Optimal Search. Econometrica, 1985, 53:4, 23-944.
23 Munneke, H.J. and A. Yavas. Incentives and Performance in Real Estate Brokerage.
Journal of Real Estate Finance and Economics, 2001, 22:1, 5-21.
Rutherford, R.C., T.M. Springer, and A. Yavas. Evidence of Information Asymmetries in the Market for Residential Condominiums. Journal of Real Estate Finance and
Economics, 2007, 35:1, 23-38.
Sirmans, G.S. and P. Swicegood. Determinants of Real Estate Licensee Income.
Journal of Real Estate Research, 1997, 14:2, 137-53.
Stewart, J.. Adverse Selection and Pay Compression. Southern Economic Journal, 1999,
65:4, 885-899.
T.S. Zorn and J.E. Larsen. The Incentive Effects of Flat-Fee and Percentage
Commissions for Real Estate Brokers. American Real Estate and Urban Economics
Association Journal, 1986, 14:1, 24-47.
Zumpano, L.V., H. Elder, and E.A. Baryla. Buying a House and the Decision to Use a
Real Estate Broker. Journal of Real Estate Finance and Economics, 1996, 13:2, 169 –
181.
24 Exhibit 1: Variable Legend
VARIABLE DEFINITION YearsCurrentFirm The number of years the agent has worked with their current firm Experience The number of years the agent has worked in residential real estate The square of the number of years the agent has worked in residential ExperienceSqrd real estate HoursWorked The number of hours worked per week by the agent Last annual income of the agent from their residential real estate Income practice Last other income available to the agent from additional sources, IncomeDifference spouse, investments, etc. Male 1 if the respondent is male, 0 otherwise SalesStaffOffice The number of agents practicing in a given office Age The age of the agent Married 1 if the respondent is married, 0 otherwise Education The education level of the agent with each year completed equal to 1 Offices The number of branch offices in the agent’s firm CC 1 if the agent chooses the 100% compensation plan, 0 otherwise
25 Exhibit 2: Descriptive Statistics – Full Sample
Variable Mean StDev YearsCurrentFirm 5.923 5.384 Experience 11.202 8.807 ExperienceSqrd 203.000 290.420 HoursWorked 41.824 14.725 Income 123281.000 166510.000 IncomeDifference 87482.000 124447.000 Male 0.435 0.496 SalesStaffOffice 31.926 21.277 Age 44.700 11.097 Married 0.751 0.432 Education 14.617 2.603 Offices 51.270 342.820 CC 0.184 0.388 N 1853
26 Exhibit 3: Descriptive Statistics by Compensation Choice Subsamples
Variable CC N Mean StDev YearsCurrentFirm 0 1512 5.907 5.459 1 341 5.997 5.045 Experience 0 1512 10.815 8.831 1 341 12.915 8.503 ExperienceSqrd 0 1512 194.910 292.610 1 341 238.900 278.100 HoursWorked 0 1512 41.247 14.889 1 341 44.384 13.706 Income 0 1512 105794.000148476.000 1 341 200820.000213684.000 IncomeDifference 0 1512 84956.000 119176.000 1 341 98683.000 145206.000 Male 0 1512 0.422 0.494 1 341 0.493 0.501 SalesStaffOffice 0 1512 30.938 21.058 1 341 36.300 21.720 Age 0 1512 44.802 11.253 1 341 44.252 10.377 Married 0 1512 0.753 0.431 1 341 0.742 0.438 Education 0 1512 14.590 2.613 1 341 14.739 2.555 Offices 0 1512 49.600 336.100 1 341 58.700 371.600
27 Exhibit 4: Binary Probit Compensation Choice Model Dependent Variable – CC (Compensation Choice)
Variable Coefficient Standard Error T P Constant -2.0215 0.5019 -4.030 0.001 YearsCurrentFir -0.0547 0.0144 -3.790 0.001 m Experience 0.1217 0.0265 4.590 0.001 ExperienceSqrd -0.0026 0.0008 -3.320 0.001 HoursWorked 0.0036 0.0047 0.760 0.445 Income 0.0001 0.0000 6.280 0.001 IncomeDifference 0.0001 0.0000 2.980 0.003 Male 0.3698 0.1291 2.860 0.004 SalesStaffOffice 0.0080 0.0030 2.690 0.007 Age -0.0184 0.0069 -2.660 0.008 Married -0.1260 0.1473 -0.850 0.393 Education -0.0110 0.0246 -0.440 0.657 Offices 0.0001 0.0002 0.410 0.680 N 1853 Log Likelihood -820.869 Significance Level .0001
28 Endnotes
29 1 100% compensation arrangements are also commonly referred to as full-commission, full-payout, and 100% agents in both the literature and in practice. In a 100% compensation arrangement, the agent pays his firm a periodic fee in exchange for 100% of the earned commissions. This arrangement stands in contrast to the more traditional Split- compensation arrangement where the agent receives a percentage of the earned commission but does not face a periodic fee to his firm.
2 Throughout this study the term “agent” is used to represent the sales force of a firm with one exception. Specifically, in the literature review section, the use of “agent” is referenced when discussing the body of study commonly referred to as the Principal-Agent literature. No where in this study does the term “agent” represent any sort of fiduciary duty
(or lack of duty) from the real estate firm through their sales force to their clients or customers.
3 See Jud, Seaks, and Winkler (1996), Johnson, Springer, and Brockman (2005), and Rutherford, Springer, and Yavas
(2007) for relevant citations and a discussion of this literature.
4 The model used in Elder, Zumpano, and Baryla (1999) to estimate search intensity is based upon optimal search theory models and, in particular, the model employed in Morgan and Manning (1985).
5 The Johnson, Zumpano, and Anderson paper uses intra-firm data. In their estimations, agents are actually classified as full-payout or split-commission agents based upon an actual determination of each agent’s specific compensation arrangement. The firms the agents work for are not used to separate the sample.
6 See Benjamin, Jud, and Sirmans (2000) for a good review of this literature.
7 See Holstrom (1979) and Zorn and Larsen (1986) for good examples of this type of article.
8 Furthermore, Holmstrom shows that in such situations payment contracts based solely on outcomes impose more risks on agents and result in less effort. More specifically, payment schemes that are based upon outcomes and signals that provide more information about an agent’s actions, or the random state of nature, create greater effort incentives than contracts based upon performance alone.