self-reported dwelling valuations—how accurate are they? dmitri romanov larisa fleishman aviad...
TRANSCRIPT
Self-Reported Dwelling Valuations—How Accurate Are They?
Dmitri RomanovLarisa FleishmanAviad Tur-Sinai
Israel Central Bureau of Statistics
There are three kinds of estimates for a There are three kinds of estimates for a dwelling value:dwelling value:
A. Price agreed upon in a sale transactionB. Appraiser's valuationC. Owner’s self-reported valuation elicited by
surveys with a question:
“ “What do you think your home is worth? What do you think your home is worth? That is, what do you think you could get for That is, what do you think you could get for your home if you sold it now?”your home if you sold it now?”
UsesUses of s of self-reported dwelling valuationself-reported dwelling valuations
In calculating housing price index In research of the housing market and household’s
economic behavior In micro-economic empirical analysis, as an indicator of
household’s economic status and well-being
There is an immense importance to test the There is an immense importance to test the extent of accuracy of subjective dwelling extent of accuracy of subjective dwelling valuations, the differences between these valuations, the differences between these valuations and the benchmarks (appraiser’ valuations and the benchmarks (appraiser’ estimates or sale prices) and the factors estimates or sale prices) and the factors associated with these differencesassociated with these differences
Summary of previous researchSummary of previous research
The most referred studies that focused on investigation of accuracy of subjective dwelling valuations: Kain & Quigley (1972), Robin & West (1977), Follain & Malpezzi (1981), Goodman & Ittner (1992); Kiel & Zabel (1999)
On average, owners tend to overestimateoverestimate the value of their homes by around 5%, though the estimates of the bias range from minus 2 percent to 16 percent
The extent of overestimation is not correlated with owner, property or neighborhood characteristics
Thus, the surveys do provide reasonable estimates of Thus, the surveys do provide reasonable estimates of dwelling valuations (with a uniform slight upward bias)dwelling valuations (with a uniform slight upward bias)
Value-added of this studyValue-added of this study
Focus on investigating the accuracy of Focus on investigating the accuracy of subjective valuations across a distribution of subjective valuations across a distribution of dwelling valuesdwelling values
Based on a large dataset (more then 21K Based on a large dataset (more then 21K observations) that contains a rich variety of observations) that contains a rich variety of homeowner, dwelling, neighborhood and homeowner, dwelling, neighborhood and environment characteristics and covers more environment characteristics and covers more than a decade (1997–2008) than a decade (1997–2008)
Examine the interval of time in which “news” Examine the interval of time in which “news” about transaction prices in the respondents’ about transaction prices in the respondents’ residential environment “trickle-down” into residential environment “trickle-down” into individuals’ dwelling evaluations individuals’ dwelling evaluations
This study aims:This study aims:
To examine the extent of accuracy of To examine the extent of accuracy of homeowners’ subjective valuations homeowners’ subjective valuations across a across a distribution of dwelling sale pricesdistribution of dwelling sale prices, while , while exploring:exploring:
whether the valuations of inexpensive and whether the valuations of inexpensive and expensive properties are biased in the same expensive properties are biased in the same direction;direction;
whether the variance of the bias is homoscedastic whether the variance of the bias is homoscedastic across the distribution of property value across the distribution of property value
what factors are associated with the valuation biaswhat factors are associated with the valuation bias
Data constructionData construction
Household Household Expenditure Expenditure
SurveysSurveys
Annual sample of more than 6000 households countrywide. Pooled sample of 72K in 12
years
Sale Sale Transactions Transactions
FilesFiles
612K transactions in 1997-2008
Research Dataset
Owners’ Owners’ dwelling dwelling
valuationsvaluations
Demographic, Demographic, social, economic social, economic indicators of the indicators of the household and household and
its head its head
Characteristics Characteristics
of the propertyof the property
Data on Data on demographic demographic
and and socioeconomic socioeconomic indicators of indicators of
population by population by census tractcensus tract
Prices, Prices, timing and timing and number of number of
transactions transactions by census by census
tracttract
Population Register
&Income Tax
records GISGIS
Data on Data on environment environment and location and location indicators of indicators of
neighborhoodneighborhood
Research populationResearch population
Households:
that live in dwellings they own (71% of the survey that live in dwellings they own (71% of the survey sample);sample);
whose dwellings were geo-referenced at the level of whose dwellings were geo-referenced at the level of structure or census tract;structure or census tract;
that provided valuation of their dwellings (less of 15% of that provided valuation of their dwellings (less of 15% of item non-response); item non-response);
that live in Israel’s sixty largest cities (assumed of having that live in Israel’s sixty largest cities (assumed of having regular housing market conditions)regular housing market conditions)
After merging the records on these households with an After merging the records on these households with an average price of transactions in the census tract in the average price of transactions in the census tract in the three-months window preceding the survey interview, three-months window preceding the survey interview, working file contained 21,238 observationsworking file contained 21,238 observations
Correlation between subjective valuation and average price of sales transactions in census tract, by deciles of valuation and period of time between valuation and sales transactions
0.000
0.050
0.100
0.150
0.200
0.250
1 2 3 4 5 6 7 8 9 10
subjective valuation deciles
coef
fici
ent
of
corr
elat
ion
12 months after 9 months after 6 months after 3 months after concurrent
3 months before 6 months before 9 months before 12 months before
Subjective valuation vs. average dwelling price in census tracts, by percentiles of dwelling-price distribution, 1997–2008
10
11
12
13
14
15
0 10 20 30 40 50 60 70 80 90 100
percentiles of dwelling price distribution
Ln
of
sub
ject
ive
valu
atio
n
and
d
wel
lin
g
pri
ce
subjective valuation average dwelling price
Average bias of subjective valuation and its standard deviation
-50
0
50
100
150
200
250
300
350
0 10 20 30 40 50 60 70 80 90 100percentiles of dwelling price distribution
per
cen
t
average bias standard deviation of bias
Average bias in subjective valuation by selected percentiles of annual dwelling-price distribution, 1997–2008
-10
0
10
20
30
40
50
60
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
perc
ent
percentiles 11-20 percentiles 45-54 percentiles 81-90
Average price of dwelling, by selected percentiles of annual dwelling-price distribution, 1997–2008 (in current prices, index 1997=100)
60
80
100
120
140
160
180
1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
Years
Ind
ex
Percentiles 11-20 Percentiles 45-54 Percentiles 81-90
Main results (1)Main results (1) Homeowners tend to overestimate the value of their
dwellings by 27 percent, with the median bias of 23 percent
The valuations of inexpensive and costly dwellings are biased in different directions: estimates reported by people who live in the dwellings that are placed in the first eight deciles of the price distribution are upward-biased, whereas those who live in the most expensive dwellings tend to understate the value of their homes
In relative terms, subjective valuations bias is much higher among the owners of inexpensive dwellings
Variance of valuation bias is heteroscedastic in the lowest 20 and the top 10 percentiles of the price distribution
The modelThe model
Dependent variableDependent variable: bias in subjective dwelling valuation relative to the average sale-transaction price of dwellings in the census tract in the three months preceding the survey
‘Household’’ - indicators of an owner-occupier household (number of persons, average income per capita in household, and size of the mortgage loan)
‘‘Personal’’ - homeowner characteristics (sex, age, marital status, origin etc.)
‘‘Asset’- dwelling indicators (whether it is a stand-alone house or an apartment in a condominium building, number of rooms, age of building etc.)
‘‘AvgTract’ - indicators of the census tract in which the dwelling is located (socio-demographic characteristics of residents)
‘Area’ – environmental characteristics of the dwelling’s nearest surroundings
‘dUSD’ - the change in the exchange rate (NIS/dollar) in the three months preceding the survey date
‘NumTransactCT’’ - the number of sales transactions in the census tract three months preceding the individual’s subjective valuation
‘‘Year’ - fixed effect for the year in which the subjective valuation of dwelling i was given
‘ ‘CT’ - fixed effect for census tract in which the valuated dwelling i was located
ijttjttijt
iiijtijt
uYeartCTNumTransacdUSDAreaAvgTract
AssetPersonalHouseholdiceCTAvgValuation
87654
321
)ln(
)Prln(
Models estimated in the studyModels estimated in the study
Subjective valuation, a hedonic price model with addition Subjective valuation, a hedonic price model with addition of average price and the number of transactions in census of average price and the number of transactions in census tracttract On the full sample (21,23821,238 obs.) On the sub-sample of dwellings purchased in 12 months
preceding the survey (738 obs.)
Bias in subjective dwelling valuation, according to the Bias in subjective dwelling valuation, according to the main model:main model: For the full sample with a fixed effect for year (A)(A) For the full sample with a fixed effect for census tract (B)(B) For the sub-sample of dwellings purchased in 12 months
preceding survey (C)(C) For the sub-samples of census tracts in which no fewer than
ten, twenty and thirty observations were found in a census tract in the whole research period
Bias model estimates Bias model estimates Full sample
Sample of dwellings purchased in 12
months preceding survey (C)
Fixed effect for Fixed effect for years (A) years (A)
Fixed effect for Fixed effect for census tract (B)census tract (B)
VariablesVariables Estimate (Std. error)Estimate (Std. error) Estimate (Std. error)Estimate (Std. error) Estimate (Std. error)Estimate (Std. error)
Persons 0.010* (0.002) 0.009* (0.002) 0.023*** (0.012)
Income 0.066* (0.005) 0.067* (0.005) 0.102* (0.027)
Mortgage -0.004* (0.001) -0.003* (0.001) -0.008*** (0.004)
Age 0.004* (0.001) 0.003** (0.001) 0.002 (0.005)
AgeSq -0.00003** (0.00001) -0.00002*** (0.00001) -0.00002 (0.00006)
IsraelBorn 0.025* (0.008) 0.020** (0.008) -0.021 (0.049)
Immigrant1990-0.075* (0.011) -0.052* (0.010) -0.051 (0.050)
ImmEurAmer-0.015*** (0.009) -0.010 (0.008) -0.046 (0.055)
Married0.041* (0.008) 0.045* (0.007) 0.044 (0.034)
* - Significant at .01 level ** - Significant at .05 level *** - Significant at .10 level
Full sample Sample of dwellings purchased in 12
months preceding survey (C)
Fixed effect for Fixed effect for years (A)years (A)
Fixed effect for Fixed effect for census tract (B)census tract (B)
VariablesVariables Estimate (Std. error)Estimate (Std. error) Estimate (Std. error)Estimate (Std. error) Estimate (Std. error)Estimate (Std. error)
Purchased-0.089* (0.011) -0.105* (0.010) -0.098 (0.080)
House0.109* (0.007) 0.143* (0.007) 0.122* (0.030)
Rooms0.172* (0.003) 0.196* (0.003) 0.131* (0.017)
BuildingAge-0.007* (0.001) -0.014* (0.0007) -0.004 (0.003)
BuildingAgeSq0.0001* (0.00001) 0.0001* (0.00001) 0.0001*** (0.00005)
AirConditioner0.046* (0.007) 0.050* (0.007) -0.001 (0.029)
Heating0.056* (0.010) 0.064* (0.011) -0.002 (0.056)
Garden0.072* (0.019) 0.114* (0.018) -0.030 (0.064)
Bought within 12 months 0.027*** (0.015) 0.002 (0.014) -
Price- - 0.163* (0.030)
dUSD0.751* (0.076) 0.712* (0.073) 0.133 (0.335)
NumTrasactCT-0.007* (0.001) -0.006* (0.001) -0.005** (0.002)
MaleCT-0.761* (0.177) 0.189 (0.249) -1.288 (0.873)
Full sampleSample of dwellings
purchased in 12 months preceding
survey (C)Fixed effect for Fixed effect for years (A)years (A)
Fixed effect for Fixed effect for census tract (B)census tract (B)
VariablesVariables Estimate (Std. error)Estimate (Std. error) Estimate (Std. error)Estimate (Std. error) Estimate (Std. error)Estimate (Std. error)
AgeCT0.008* (0.001) 0.002 (0.001) 0.003 (0.004)
Immigrant1990CT0.285* (0.029) 0.217** (0.088) 0.299** (0.139)
ImmEurAmerCT-0.338* (0.064) -0.126 (0.137) -0.080 (0.313)
ImmAsiaAfricaCT-0.292* (0.071) -0.024 (0.149) -0.158 (0.326)
EthiopiaCT0.738* (0.103) 0.169 (0.210) 1.320* (0.442)
IncomeCT-0.083* (0.007) 0.002 (0.015) -0.131* (0.033)
RoadArea0.002** (0.001) -0.001 (0.001) 0.006 (0.004)
Road5mArea-0.009** (0.004) 0.003 (0.004) -0.026 (0.017)
SchoolArea0.003* (0.001) 0.0001 (0.002) 0.002 (0.003)
No’ of Obs.21,238 21,238 738
Mean dependent variable 0.112 0.112 0.108
Mean bias (pct.)26.8 26.8 19.3
Adjusted R2 0.29 0.10 0.33
Main results (2)Main results (2) The interval of three months before the survey date is found
to have the highest correlation between valuations and average transaction prices in a given census tract for most of the dwelling-value distribution
No dependency between the extent of valuation bias and sample size in the census tract
There is a significant and systematic relation between subjective valuation bias and homeowner personal indicators, dwelling properties and census-tract population characteristics
Basic dwelling characteristics (number of rooms and age of building) and the socioeconomic level of the neighborhood exert a larger impact on valuation bias among the owners who bought their dwellings during the twelve months preceding the survey date, relatively to the rest of owners, whereas the bias among the latter is more influenced by dwelling facilities (air conditioning, heating systems, garden)
The observation of a mean (upward) bias in The observation of a mean (upward) bias in subjective valuations relative to actual subjective valuations relative to actual dwelling prices masks quite a bit of variation dwelling prices masks quite a bit of variation across the distribution; the bias may even across the distribution; the bias may even change the sign along the distribution. change the sign along the distribution.
What the reasonable explanations could be assumed?What the reasonable explanations could be assumed?
The bottom lineThe bottom line
What explains the bias sign change?? Does it square with luxury housing as a Veblen good? Is it about a segment of housing market where subjective
valuations do not catch up systematically with a pace of prices rise?
Do owners of expensive dwellings differ from the owners of lower-priced accommodations in the manner of thinking and the considerations that figure in evaluating a dwelling?
Does an upward bias at the lower end of the dwelling-valuation distribution take shape as a consequence of liquidity constraints among owners of inexpensive dwellings that bring about a price level for such dwellings under fair market value?
Is an overvaluation of the least expensive dwellings by their owners akin to an extremely importance of certain utilities of home ownership (stability, local welfare services, community support etc.) for low-income persons?