Prospect Theory Preferences in Real Estate
A Study on Selling Behavior in Residential Real Estate
Author
Supervisor
Vinnie Vermeulen
prof. dr. M.K. Francke
Faculty Economics and Business
University of Amsterdam
A paper submitted for the degree of
Master in Real Estate Finance
Abstract
This study investigates whether selling behavior in real estate is affected by prospective losses and gains. The non-linear utility function that was first recorded by Kahneman and Tversky in 1979 serves as the theoretical framework of this analysis. The DNB Household Survey, which is annually collected by CentERdata, provides us with general household information as well as information on the families’ housing situation. Although our panel regression estimation shows that loss region sellers expect higher transaction values, the coefficient is not significantly different from zero. However, statistical matching analysis does show significant evidence of loss aversion, which corresponds with previous studies by Genesove and Mayer (2001) and Bokhari and Geltner (2011). Our logistic regression shows no conclusive evidence on whether households facing prospective losses have a lower chance of selling compared to households that are expecting gains, which is in contrast with the disposition effect that is found by Shefrin and Statman (1985).
Key words: Behavioral Economics, Prospect Theory, Real Estate Markets JEL classification: G41, R30, R32.
Statement of Originality
This document is written by Vinnie Vermeulen who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Table of Content
1. INTRODUCTION ... 5
2. LITERATURE REVIEW ... 6
2.1ECONOMIC DECISION MAKING ... 6
2.2PROSPECT THEORY AND THE PRICE-VOLUME PUZZLE ... 8
3. DATA AND METHODOLOGY ... 11
3.1DATA DESCRIPTION ... 11 3.1.1 Dependent Variables ... 11 3.1.2 Independent Variables ... 12 3.1.3 Control Variables ... 12 3.2MODEL SPECIFICATION ... 13 4. RESULTS ... 14 4.1DESCRIPTIVE ANALYSIS ... 14 4.2REGRESSION ANALYSIS ... 15 5. CONCLUSION ... 19 REFERENCES ... 20
1. Introduction
A well-recorded phenomenon of the housing market is the positive correlation between prices and transaction volume. This difference in fluctuation of intense trading activity between hot and cold markets is a puzzle that has been questioned by many researchers before (e.g. Stein, 1995). In recent years the co-movement of prices and volume has been given new insight thanks to the field of behavioral economics. Kahneman and Tversky’s (1979) prospect theory states that losses incurred relative to a reference point loom larger than gains. Genesove and Mayer (2001) showed that this reluctance to losing money determines selling behavior in the 1990’s Boston condominium market. Stating that loss region sellers set higher asking prices, negotiate higher selling prices but as a result spent a longer time on the market. Similar results where found by Bokhari and Geltner (2011) when regarding commercial real estate transactions. Since the aforementioned papers are solely based on U.S. transactions, we study the external validity of their results using Dutch household data, which results in the following research question: “Do prospect theory preferences explain selling behavior in Dutch residential real estate?”
The regression analysis suggests that sellers facing prospective losses expect higher transaction prices compared to homeowners facing nominal gains. However, the coefficients are not significantly different from zero. In the subsequent statistical matching analysis we find significant evidence of loss aversion among Dutch homeowners, which corresponds with the previous findings by Genesove and Mayer (2001) and Bokhari and Geltner (2011). Besides that, we find no conclusive evidence of a disposition effect among Dutch homeowners, which strengthens the findings of previous research in this field.
The remainder of this paper starts with an introduction to economic decision-making before switching to Kahneman and Tversky’s (1979) prospect theory and the price-volume puzzle in section 2.2. This is followed by section 3.1 where we discuss the DNB Household Survey. Thereafter we present the model specifications and describe the various variables used in our regression analysis. Section 4.1 starts with a descriptive analysis of our sample, which is followed by a discussion of our regression estimates in section 4.2. The conclusion starts with a brief summary of this paper and presents an answer to our research question.
2. Literature review
This chapter starts with an introduction to the underlying economic reasoning behind decision making under risk as has been explained by the expected utility theory. After discussing the shortcomings of the more classical utility theory we thoroughly explain Kahneman and Tversky’s (1979) prospect theory. Section 2.2 starts with an introduction to the price-volume puzzle that has been identified in previous empirical studies worldwide. Next, we implicate how behavioral economic theory can provide an explanation to he positive correlation between prices and transaction activity. At last, we present the empirical findings of the current literature that used Kahneman and Tversky’s (1979) prospect theory to explain the price-volume puzzle.
2.1 Economic Decision Making
Maximizing future expectations is common ground for many economic theories. In 1738, Dutch-born Daniel Bernoilli offered a resolution to the St. Petersburg paradox by imposing a logarithmic cardinal utility function similar to the one shown in Figure 1. The concavity indicates the decision-makers risk aversion against large but uncertain outcomes.
Figure 1: Risk Averse Utility Function
In 2002, Daniel Kahneman was the first psychologist to receive the Nobel Prize in Economics. Together with Amos Tversky they presented their seminal paper in 1979 that connected economic behavior with psychological reasoning. Their quantitative analysis consisted of various hypothetical choice problems where one needs to choose between certain and
7
uncertain monetary payouts. They find that respondents favor certain outcomes over very uncertain probabilities. This certainty effect violates the assumption in expected utility theory, which states that utilities are weighted by their probabilities and would yield a linear relationship of evaluated probabilities whereas Kahneman and Tversky (1979) state that people evaluate probabilities nonlinearly. Another finding from their questionnaire was that people evaluate outcomes as either losses or gains measured against some relative reference point rather than outcomes reflecting future wealth. These anomalies from the assumptions of expected utility theory are the starting point for Kahneman and Tversky’s (1979) value function and are the essence of prospect theory.
Figure 2: Value Function
As is indicated by Figure 2, the utility curve over gains is concave indicating risk aversion against substantial payoffs. However, the convex utility curve for losses reflects risk-seeking behavior regarding choices with certain losses. Where the negative domain transforms into a positive domain the utility function adjusts its curvature. In this case, the origin is where the S-shaped curve is steepest. This contradicts with the utility function presented by Markowitz (1952), which showed a very shallow turning point. Kahneman and Tversky (1979) argue that the center of the figure depicts a neutral reference point that people use to measure the value of losses and gains. Another important characteristic of this figure is that the function is steeper in
the negative domain than in the positive domain. This translates to a dislike of losses that is stronger than the experienced happiness associated with a similar monetary gain.
2.2 Prospect Theory and the Price-Volume Puzzle
John Muth (1961) argued that agents form expectations based on all available information in the market, which is also known as rational expectations. Using the assumptions underlying Muth’s theorem, Lucas (1978) demonstrates that transaction volume and prices should be uncorrelated. However, previous empirical studies contradict with what was previously thought. Using data on U.S. home sales, Stein (1995) shows that a 10 percent decrease in value reduces volume by more than 1.6 million units. Taking into account that the yearly transaction volume ranges from 3 to 4 million the implications of this correlation is of major importance. Stein (1995) reasons that the down-payment serves as a form of wealth accumulation with surging house prices. Accumulated personal wealth enables them to move up the housing ladder by allowing them to meet the down payment on a larger property. The increase in transaction volume, which is a result of increased house prices, would implicate a positive correlation between volume and prices. Although credit-constraint theory seems convincing, Leung, Garion & Leong (2002) argue that it does not explain the lead-lag relationship between both variables. Ortalo‐Magné and Rady (2006) used Stein’s (1995) reasoning to develop a steady-state equilibrium model where endowment shocks lead to overreacting house prices and exhibit a positive correlation between volume and prices. De Wit, Englund and Francke (2013) also show a positive correlation between the rate of sale and price changes when investigating transaction data from the Dutch brokerage association NVM between 1985-2007. However, this period is characterized with LTV rations above 100 percent meaning that down payments where not a necessity and resulted in less credit constraint households (DNB, 2015). This line of reasoning questions whether Stein’s (1995) theory is able to fully capture the price-volume puzzle.
Other insight explaining the positive correlation between volume and prices was presented by Wheaton (1990). In his search model households belong to one of three occupancy states. The first category consists of households living in a property that satisfies their needs. In the second category you find households that have found a suited property but have not sold their former house. The latter became undesired after a change in household composition. The third occupancy group solely consists of households living in a non-matched dwelling. The unmatched households are looking for vacant properties that were listed by the households from category two. Once unmatched households have found a suited property they immediately move and list the unsuited property on the market. After completing the sale both buyer and seller
9
move up one category. In Wheaton’s (1990) model housing supply exceeds the amount of households, which explains the natural vacancy rate. In this case supply reacts slowly to vacancy and prices similar to the stock-flow model. The vacancy rate increases when a change in demographics increases the amount of unmatched properties. This results in a decrease in both volume and prices. Recent contributions to the search theorem by Genosove and Han (2011) investigate variations in market liquidity and matching quality by using a random-matching model.1
An interpretation that gained ground over the last decades uses Kahneman and Tversky’s (1979) prospect theory to give more insight on the price-volume puzzle. Genesove and Mayer (2001) show that the reluctance to losses determines sellers’ behavior in the Boston condominium market. When the housing market peaked and prices start slipping, part of the dwelling stock will have a value that is below of what the current owner paid. Using the purchase price as reference point, Genesove and Mayer (2001) find significant evidence that the aversion to prospective losses will make homeowners set higher list prices and negotiate higher transaction prices, with the result that the dwellings spend a longer time on the market. Bokhari and Geltner (2011) extend this research to the market for commercial real estate. One would expect that on this market actors act professional and show less emotion when managing their portfolio. Using sales data that covers 90 percent of all U.S. commercial property transactions greater than 5 million dollars between January 2001 and December 2009 they find that more experienced sellers have an even greater reluctance to making losses compared to less sophisticated sellers. Another notable finding is that after the start of the financial crisis in 2007 there is less loss aversion. Bokhari and Geltner (2011) suggest that a major downturn wakens investors and enforces rational behavior.
De Wit, Englund and Francke (2013) put forward that the various explanations of the price-volume puzzle should not be seen as contradictory, but rather as complementary. This study focuses on the least documented explanation, namely behavioral economics. Similar to previous research conducted by Genesove and Mayer (2001) and Bokhari and Geltner (2011) we expect that that prospect theory preferences explain pricing behavior in residential real estate. Therefore the first hypothesis tested in this research is as follows:
Hypothesis 1: Loss aversion determines selling behavior in the Dutch housing market.
Shefrin and Statman (1985) state that “investors tend to sell winners too early and ride losers too long", which is also known as the disposition effect. Since this phenomenon might bias the results found by Genesove and Mayer (2001) and Bokhari and Geltner (2011) we investigate whether homeowners show similar behavior when choosing to sell a property. Therefore the second hypothesis tested is:
Hypothesis 2: The disposition effect determines the likelihood of selling in the Dutch housing market.
11
3. Data and Methodology
This chapter with a brief description of the DNB Household Survey and is followed by an overview of the variables used in our analysis. At last, we present the setup of our regression models, which owe a lot to the empirical setup of respectively Genesove and Mayer (2001) and Bokhari and Geltner (2011).
3.1 Data Description
This article examines how prospected gains and losses affect selling behavior of Dutch homeowners. In order to answer this question we retrieve economic data from the DNB Household Survey, which is annually collected by CentERdata using a questionnaire among Dutch households. The survey covers general household information, the respondents’ occupation, current housing situation, income, assets and economic and psychological concepts. Since its introduction in 1993 the size of the survey increased significantly by adding more questions each year.
On January 1st, 2002, the Euro was put into circulation replacing the Dutch Guilders. The
2003 questionnaire, which is collected after this introduction, marks the starting point of this research and makes sure all values are reported in the same monetary unit. Whereas the most recent wave, which is collected between April and October 2017, marks the last year investigated in this study. Eventually, this fifteen-year period yields a well-balanced timespan of both stable periods and crises.
The questionnaire allows us to identify the family member that is most involved in the household’s financial decision-making process. We regard their answers as the most economically sound. In line with this reasoning answers from non-decision-making family members are neglected. The final sample used to conduct the subsequent regression analysis consists of an unbalanced panel of 1849 respondents.
3.1.1 Dependent Variables
Both empirical studies by Genesove and Mayer (2001) and Bokhari and Geltner (2011) use U.S. market based data to study prospect theory preferences. In contrast with these papers, the Household Survey restricts us with the respondents perceived transaction price at the time of carrying out the survey. Although we assume that the most financially informed family member has a realistic value of its house, the realized transaction price might deviate from the ex ante expectation.
To check whether the aforementioned papers were possibly biased by the disposition effect we measure the likelihood of moving using a dummy variable that is zero when the respondent indicated that household wants stay in their current accommodation and one if they plan to move in the distant future. To construct this variable we make use of question WO53 from the Household Survey. As a robustness check we constructed a dummy variable measuring the likelihood of selling using question WOD52L.
3.1.2 Independent Variables
In accordance with prospect theory preferences the households’ expected losses and gains are evaluated against a neutral reference point. Genesove and Mayer (2001) argue that the original purchase price serves as well-known focal point for homeowners. A study by Shafir, Diamond and Tversky (1997) shows that people observe gains in nominal rather than real terms. In line with the findings of the latter studies, the households expected gain or loss is calculated in nominal terms by subtracting the purchase price from the appraised value at the time of completing the survey. This is scaled back to a percentage change by dividing the gain or loss by the original purchase price. This information is transformed to absolute values and used to construct our two variables of interest using namely Profit and Loss. If the respondent is facing a prospective loss, its percentage gain is truncated at zero.
Where Genesove and Mayer (2001) and Bokhari and Geltner (2011) make use of hedonic valuation to assess the appraised value, we make use of the WOZ-value, which is estimated by the local municipality and is the basis for calculating property tax in the Netherlands.
3.1.3 Control Variables
In 1997, Genesove and Mayer showed that financially constrained homeowners set higher asking prices in order to limit the equity loss and fulfill the down payment requirement on their new house. The households’ current loan-to-value ratio is calculated by dividing the current outstanding loan balance by the properties appraised value. We also control for the respondents holding period since we expect that the reference point blurs as the years pass by and therefore limit the effect of loss aversion in real estate transactions.
The control variables used to test the second hypothesis originate from Engelhardt’s (2003) study estimating the difference in mobility among homeowners facing either losses or gains. Besides the loan-to-value ratio this includes variables household income, wealth and age.
13
3.2 Model Specification
To assess the effect of prospect theory preferences in residential real estate transactions we use the following regression model, which is based on the comparable studies by Genesove and Mayer (2001) and controls for the non-linear relationship as proposed by the value function shown in Figure 2. SPit=αit+ β1PFit+ β2PFit2+ β 3LSit+ β4LSit 2+β 5LVit+ β6HPit+ β7+t 15 t=2 YR+εit SP = Selling price; PF = Profit; PF2 = Profit squared; LS = Loss; LS2 = Loss squared; LV = Loan-to-value; HP = Holding period; YR = Time fixed effects.
Our logit regression estimates the likelihood of moving and is based on the setup by Engelhardt (2003) and is structured in the following way:
WMit=αit+ β1PFit+ β2PFit2+ β3LSit+ β4LSit2+ β5LVit+ β6HIit+ β7HWit+ β8AGit+ β9AGit2+ β10+t 15 t=2 YR+εit WM = Willingness to move; PF = Profit; PF2 = Profit squared; LS = Loss; LS2 = Loss squared; LV = Loan-to-value; HI = Household income; HW = Household wealth; AG = Age; AG2 = Age squared; YR = Time fixed effects.
4. Results
This chapter starts with a descriptive analysis of the sample. This is followed by section 4.2 where we present the model estimates that assess the effect of proposed gains and losses on the properties expected selling price. Thereafter we present the outcomes of our logit regression, which investigates the likelihood of moving among homeowners.
4.1 Descriptive analysis
The descriptive statistics given in Table 1 are based on survey data from 2003 to 2017 and illustrate the respondents’ personal and financial characteristics. The average age is close to 53 and shows that the sample is skewed to the older fraction of the Dutch population, which had an average age ranging from 38 in 2000 and 42 in 2017 (CBS, 2018b). Variables household income and aggregate wealth are measured in euros and equal 28236 and 66262.
Where some property related financials needed to be reported in thousands others were not. Since manual inspection of the data showed that not all respondents carried out these instructions correctly we winsorized the property related variables noted in euros at the 5-95 percent level. When regarding the families’ housing characteristics it is shown that the average purchase price is slightly above 140 thousand euros and the average appraised value is 72 percent higher. The average outstanding mortgage balance is 110066 euros and the current loan-to-value ratio is equal to 53 percent, which indicates that the average household is probably not financially constraint. This can be explained by both price increases in the Dutch owner occupied sector and the accumulated amortization in the average 17 year old holding period. According to statistics presented by CBS (2018a) the average transaction price in the Netherlands ranged from 172050 in 2000 and 263295 euros in 2017. Between 2003 and 2017 respondents reported an average expected selling price of 260253 euros. This shows that the average respondent either lives in a house that is more expensive than the national average or has too high expectations of the value of their house. We need to note that the average percentage loss and gain give a slightly distorted image in this table. Since the variable reports a zero when the counterpart is positive both averages are pushed downwards. The overrepresentation of respondents facing gains intensifies this effect for the variable Loss.
15
Table 1: Descriptive statistics
Variable Count Mean Std.
Age 27947 52.86 15.27 Household income 27947 28235.46 28520.76 Wealth 16281 66261.71 155155.03 Purchase price 11049 140665.94 78088.58 Appraised value 9912 241514.73 90998.88 Mortgage balance 7783 110066.14 76843.25 Mortgage costs 7410 5517.58 3715.67 Loan-to-value 6446 53.15 36.91 Selling price 10119 260253.48 91086.53 Profit 8566 101.79 108.85 Loss 8566 2.10 5.94 Holding period 11513 16.97 12.34 Willingness to move 14991 0.39 0.49 Willingness to sell 10981 0.18 0.39 4.2 Regression analysis
The regression coefficients from our selling price regressions are reported in Table 2. Column (1) contains model estimates based on a linear relationship between both gains and losses on the expected selling price. The estimates reported in column (2) control for the non-linear relationship between losses and gains on selling prices, which correspond with prospect theory preferences.
When comparing coefficient estimates on the variables profit and loss in column (1) we find that higher profits indicate a significant increase in the expected selling price, although the effect itself is fairly small. The regression coefficients shown in column (2) indicate that prospective losses result in an expected selling price that is higher than that of respondents facing a potential gain. Besides variable Loss2 suggests a marginally decreasing effect of loss aversion,
which corresponds with the findings in previous empirical studies by Genesove and Mayer (2001) and Bokhari and Geltner (2011). However, none of these variables are significantly different from zero. We expect that the difference in results is caused by the fact that we measure the respondents’ intentions and the latter studies investigated human behavior using realized transaction. We do find significant evidence that an increase in the LTV-ratio decreases the properties’ expected selling price. These findings do not support the evidence found by Genesove and Mayer (1997), which showed that financially constrained homeowners realize higher transaction prices.
Table 2. Regression estimation of the relation between expected losses and gains on selling prices
(1) (2)
Variables Log selling price
Profit 0.0002*** 0.0001 (0.00004) (0.000102) Profit2 0.0000001 (0.0000002) Loss 0.0002 0.0008 (0.0004) (0.0008) Loss2 -0.00002 (0.00002) Loan-to-value -0.0004*** -0.0004*** (0.00008) (0.00009) Holding period -0.002*** -0.002*** (0.0005) (0.0006) Constant 12.361*** 12.361*** (0.013) (0.013) Observations 5,006 5,006 R-squared 0.215 0.216
Time fixed effects Yes Yes
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
To give more insight in the difference between comparable properties we conducted a statistical matching analysis, which results can be found in Table 3 below. The sample is subdivided between loss and gain region sellers and the subsamples are matched on the properties’ appraised value, the LTV-ratio, the holding period and the year of survey using propensity score matching (PSM). When regarding analysis (1) we find that households in the loss region set an average selling price of 228698 euros, which is 17784 euros higher compared to matched properties of those respondents facing gains and is significant at the 99,9 percent level. This outcome corresponds with the previous findings from Genesove and Mayer (2001) and Bokhari and Geltner (2011) and confirms presence of loss aversion in the Dutch residential housing market there is no conclusive evidence to reject hypothesis one.
In analysis (2) we replaced the appraised value with property characteristics that are frequently used in hedonic valuation model as a robustness check (De Wit, Englund & Francke, 2013; Genesove & Mayer, 1997). Although the difference suggests that loss region sellers expect a higher value than respondents anticipating gains, the result is not significantly different from
17
zero. A shortcoming of the second analysis is that properties with very comparable characteristics can be matched across completely different regions. Since the current average house price in Amsterdam is almost two times higher than in the Northern Netherlands (CBS, 2018a) this would result in biased estimates.
Table 3: Matching analysis testing the difference between expected losses and gains on selling prices
Variables Loss Gain Difference Std. T-stat
(1) Appraised value Loan-to-value Holding period Year 228698.154 210914.224 17783.931 3867.705 4.60 (2) Region Urbanization House type Construction year Number of rooms Garage Garden size Loan-to-value Holding period Year 232939.439 228642.541 4296.898 4752.131 0.90
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Where prospect theory provides suggestive evidence of human behavior in real estate transactions it does not address the setting in which homeowners decide to sell their asset. The disposition effect that is found by Shefrin and Statman (1985) and Odean (1998) might indicate the previous results by Genesove and Mayer (2001) and Bokhari and Geltner (2011) are biased when similar behavior is found in real estate.
The regression estimates of our logistic regression are reported in Table 3. In columns (1) and (2) we use the willingness to move as an endogenous regressor whereas the likelihood of selling is used as a dependent variable in columns (3) and (4). The sign of the coefficients in column (1) indicate that expected losses decrease the respondents’ chance of moving and prospective gains increase the chance of moving. However both estimates are not significantly different from zero, which is also the case for the coefficients in column (2). Across both specifications we find significant evidence for a convex relationship between age and the likelihood of moving. The other control variables used in this regression are not significantly different from zero. Similar to columns (1) and (2) the regression coefficients in column (3) do
not show significant evidence of different behavior between respondents facing prospective losses or gains. However in column (4) we find that expected profits have a positive, but diminishing effect on the likelihood of selling. For both estimations we find significant evidence for a u-shaped relationship between age and the likelihood of selling their property, which is consistent with the estimates found in column (1) and (2). The positive relationship between accumulated wealth and the likelihood of selling can be explained by the fact that more funds enable households to fulfill the down payment on a larger property. However, we do not find conclusive evidence that previous research is biased by the presence of a disposition effect and hereby reject hypothesis two.
Table 3. Logistic regression estimation of the relation between expected losses and gains on the likelihood of moving
(1) (2) (3) (4)
Variables Move Sell
Profit 0.0002 0.004 0.002 0.010** (0.001) (0.003) (0.001) (0.004) Profit2 -0.00001 -0.00002** (0.000008) (0.00001) Loss -0.0166 0.00679 -0.00186 0.0251 (0.014) (0.030) (0.019) (0.042) Loss2 -0.0007 -0.0006 (0.0009) (0.001) Loan-to-value 0.002 0.003 0.005 0.006 (0.003) (0.003) (0.004) (0.004)
Log household income 0.160 0.163 0.181 0.179
(0.133) (0.134) (0.203) (0.204) Log wealth -0.080 -0.083 0.348*** 0.353*** (0.076) (0.076) (0.113) (0.114) Age -0.174*** -0.185*** -0.160** -0.185*** (0.050) (0.051) (0.066) (0.067) Age2 0.001** 0.001** 0.001** 0.002*** (0.0005) (0.0005) (0.0006) (0.0006) Constant 4.357** 4.472** -5.129* -4.960* (1.886) (1.894) (2.670) (2.685) Observations 3,234 3,234 1,770 1,770
Time fixed effects Yes Yes Yes Yes
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
19
5. Conclusion
This paper uses Kahneman and Tversky’s (1979) prospect theory preferences to explain pricing behavior in the Dutch residential market. Where previous research by Genesove and Mayer (2001) and Bokhari and Geltner (2011) used realized transaction prices the DNB Household Survey restricted us with the respondents’ expectations about their property. Our initial panel regression suggested that those facing losses expect higher transaction values compared to respondents facing gains. However, these results were not significantly different from zero. We argue that the difference in results is caused by the fact that human intentions measured in the questionnaire are not exactly similar to how people act on the real estate market. In the subsequent statistical matching analysis we find significant evidence of loss aversion among Dutch homeowners, which confirms the previous findings by Genesove and Mayer (2001) and Bokhari and Geltner (2011). At last, by making use of a logistic regression we showed no significant evidence of the presence of the disposition effect in residential real estate, which does not affirm the results by Genesove and Mayer (2001) and Bokhari and Geltner (2011) are biased.
A potential limitation of this research is the sample is skewed to the older fraction of the Dutch population whereas it is the younger fraction that suffered most from the recent financial crisis. We recommend that this research is repeated in a new setting using a sample of younger households and, when available, uses realized transaction prices. Nevertheless, we are confident that this paper contributes to the current literature on behavioral economics in real estate.
References
Anglin, P. M., Rutherford, R., & Springer, T. M. (2003). The trade-off between the selling price of residential properties and time-on-the-market: The impact of price setting. The Journal of Real
Estate Finance and Economics, 26(1), 95-111.
Bokhari, S., & Geltner, D. (2011). Loss aversion and anchoring in commercial real estate pricing: Empirical evidence and price index implications. Real Estate Economics, 39(4), 635-670.
CBS. (2018a). Existing own homes; Average purchase prices, region. Retrieved from
https://opendata.cbs.nl/statline/#/CBS/en/dataset/83625ENG/table?ts=1530450678098
CBS. (2018b). Population; Key figures. Retrieved from
https://opendata.cbs.nl/statline/#/CBS/en/dataset/37296eng/table?ts=1530450338087
Clayton, J., Miller, N., & Peng, L. (2010). Price-volume correlation in the housing market: causality and co-movements. The Journal of Real Estate Finance and Economics, 40(1), 14-40.
De Wit, E. R., Englund, P., & Francke, M. K. (2013). Price and transaction volume in the Dutch housing market. Regional Science and Urban Economics, 43(2), 220-241.
DNB. (2015). Effects of further reductions in the LTV limit. DNB Occasional Studies, 13(2).
Engelhardt, G. (2003). Nominal loss aversion, housing equity constraints, and household mobility: Evidence from the United States. Journal of Urban Economics, 53(1), 171-195.
Genesove, David, & Han, Lu. (2012). Search and matching in the housing market. Journal of
Urban Economics, 72(1), 31-45.
Genesove, D., & Mayer, C. (1997). Equity and time to sale in the real estate market. The American
Economic Review, 87(3), 255-269.
Genesove, D., & Mayer, C. (2001). Loss aversion and seller behavior : Evidence from the housing market. The Quarterly Journal of Economics, 116(4), 1233-1260.
Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47(2), 263-291.
Krainer, J. (2001). A theory of liquidity in residential real estate markets. Journal of Urban
Economics, 49(1), 32-53.
Leung, C. K. Y., Lau, G. C. K, & Leong, Y. C. F. (2002). Testing alternative theories of the property price-trading volume correlation. The Journal of Real Estate Research, 23(3), 253-263.
Lucas, R. (1978). Asset prices in an exchange economy. Econometrica, 46(6), 1429-1445.
21
Muth, J. (1961). Rational expectations and the theory of price movements. Econometrica, 29(3), 315-335.
Novy‐Marx, R. (2009). Hot and cold markets. Real Estate Economics, 37(1), 1-22.
Odean, T. (1998). Are investors reluctant to realize their losses? The Journal of Finance,53(5), 1775-1798.
Ortalo‐Magné, F., & Rady, S. (2006). Housing market dynamics: On the contribution of income shocks and credit constraints. Review of Economic Studies, 73(2), 459-485.
Shafir, E., Diamond, P., & Tversky, A. (1997). Money illusion. The Quarterly Journal of Economics, 112(2), 341-374.
Shefrin, H., & Statman, M. (1985). The disposition to sell winners too early and ride losers too long: theory and evidence. The Journal of Finance, 40(3), 777-790.
Stein, J. (1995). Prices and trading volume in the housing market a model with down-payment effects. The Quarterly Journal of Economics, 110(2), 379-406.
Wheaton, W. (1990). Vacancy, search, and prices in a housing market matching model. Journal of
Political Economy, 98(6), 1270-1292.
