Download - Why Do Vacant Houses Sell for Less
-
8/6/2019 Why Do Vacant Houses Sell for Less
1/25
2011 V39 1: pp. 1943
DOI: 10.1111/j.1540-6229.2010.00285.x
REAL ESTATE
ECONOMICS
Why Do Vacant Houses Sell for Less:Holding Costs, Bargaining Poweror Stigma?
Geoffrey K. Turnbull
and Velma Zahirovic-Herbert
This article introduces Nash bargaining into a search model to identify variouschannels through which vacancy affects selling price and liquidity in the resalemarket for houses. The model shows the various vacancy effects in the formof greater seller holding cost, lower seller bargaining power and unobservednegative attributes or stigma. We use a 20-year data series on house trans-actions to test for these effects in a simultaneous model of price and liquidity,using the long data series to allow for variation across market phases. The ro-bust vacancy effects on price and liquidity across all market phases primarilyreflect greater seller holding cost and diminished bargaining power. Repeat-edly, vacant houses also exhibit significant stigma effects in the rising marketbut not in stable or declining market phases. At the same time, vacant housesenjoy stronger shopping externality effects from surrounding houses for salethan do their occupied counterparts.
There is a generally held view within the real estate brokerage industry that
vacant houses will experience longer marketing periods, lower selling prices
or both. Real estate brokersand real estate scholarsoffer two explanations
for the vacant house discount. First, empty houses do not show nearly as
well as those that are occupied, which reduces their aesthetic and emotional
appeal for prospective buyers. Second, vacancy signals a more motivated seller,thereby weakening seller bargaining power. Although vacancy has not been the
primary topic of rigorous empirical study, there is abundant empirical evidence
that vacant homes tend to sell for less. This evidence is typically a by-product
of studies focusing on other unrelated questions.1 Vacancy takes a central roleDepartment of Economics, Georgia State University, Atlanta, GA 30302-3992 [email protected].
Department of Housing and Consumer Economics, University of Georgia, Griffin,GA 30223 or [email protected].
1
See, as examples, Turnbull, Sirmans and Benjamin (1990), Asabere and Huffman(1993), Sirmans, Turnbull and Dombrow (1995), Forgey, Rutherford and Springer(1996), Springer (1996) and others. In a different vein, Zuehlke (1987) provides ev-idence of positive duration dependence for vacant homes, suggesting that owners ofvacant dwellings are more likely to reduce reservation prices as their homes remainunsold.
C 2010 American Real Estate and Urban Economics Association
-
8/6/2019 Why Do Vacant Houses Sell for Less
2/25
20 Turnbull and Zahirovic-Herbert
in Harding, Knight and Sirmans (2003); they rely on the logical link between
weak bargaining power of sellers and vacancy to motivate their vacancy-priceresults as a test of bargaining power effects.
This article focuses on the relationship between vacant houses, selling prices
and liquidity in the resale market for houses. It introduces Nash bargaining
into a simple search model to highlight the various channels through which
vacancy affects selling price and selling time. The theoretical model examines
the consequences of vacancy as indicating greater seller holding cost, weaker
seller bargaining power or unobserved negative house attributes or stigma.
Embedding the Nash bargaining approach within a standard search frameworkemphasizes features glossed over in rationales for the vacancy price effect
offered by both practitioners and researchers. This approach emphasizes the
difference between a seller with weaker bargaining power and a seller motivated
to accept a lower price because of higher holding costs. Although it turns out
that both tend to reduce the sellers reservation price (hence expected selling
price), the systematic changes in bargaining power associated with the various
phases of the housing market cycle alter the expected vacancy price discount
over market phases while stable holding cost will not. Further, because higher
holding costs and weaker bargaining power both prompt the seller to lower the
reservation price, they also increase the probability of an earlier sale. Therefore,if the vacant house price discount arises from motivated or weak sellers, the
model suggests that vacant houses will sell more quickly on average, rather
than more slowly as in the generally held view of professionals.
This study uses two decades of house transactions data to test the various
vacant house effects in a simultaneous model of price and liquidity. The long
sample period allows us to study possible variation across declining, stable
and rising market phases. The results show robust vacancy effects on price and
liquidity across all market phases that primarily reflect greater seller holdingcost and diminished bargaining power. At the same time, vacant houses that
are no longer fresh listings tend to enjoy stronger shopping externality effects
from surrounding houses for sale than do their occupied counterparts. The
estimates show no vacant house stigma or negative unobserved attribute or
vacant-related atypicality effects on price and selling time in declining or
trough market phases; there is, however, a significant negative stigma effect
associated with repeatedly vacant houses in rising markets.
This article adds to the existing empirical work that often treats vacancy as a
tangential feature. For example, Forgey, Rutherford and Springer (1996) focus
on the impact of marketability factors and search effort and liquidity when
they estimate a two-stage least squares model of house prices and days on
market. They use house sales data for Arlington, Texas, over 19911993. In
-
8/6/2019 Why Do Vacant Houses Sell for Less
3/25
Why Do Vacant Houses Sell for Less 21
this study, vacancy is one of the marketability variables. In the first stage, the
dependent variable is the log of selling time. The first-stage estimates showthat vacant homes have longer marketing times. The results also show that days
on market depends on the sellers search effort, market conditions, physical
characteristics of the property, the size of the brokerage firm and listing price.
They use the predicted values and residuals from the first-stage estimates to
create the expected days on market variable and the relative difference of the
actual and expected selling time, using these variables in the second-stage price
estimation. The results show that lower selling prices are associated with vacant
houses. Furthermore, the findings of the second stage indicate that increases in
expected selling time result in higher sales prices, supporting the notion thatthe expected sales price rises as a seller more thoroughly searches the market
for the highest offer. The second-stage estimates also indicate that deviations
from the expected time on market are inversely related to selling prices, sup-
porting the notion that buyers pay a premium for more liquid properties.
Similarly, Springer (1996) estimates impacts of seller motivations on selling
prices and marketing times using data for single-family homes sold in Ar-
lington, Texas, again over 19991993. His results show price discounts for
houses with sellers who are eager, motivated or anxious; houses with sellers
who have relocated; foreclosures and vacant houses. However, only foreclo-sure houses show the reduced marketing time expected for properties with
motivated sellers. Sirmans, Turnbull and Dombrow (1995) use Baton Rouge,
Louisiana, data for sales during 19851991 to provide evidence that owners of
vacant houses set lower reservation prices to reflect holding-period costs that
are higher than those for owner-occupied houses because the full cost of carry-
ing the vacant home has no offsetting benefits from either occupancy or rental
income.
In the closest antecedent to this study, Harding, Knight and Sirmans (2003)examine bargaining power in two separate markets (Baton Rouge, Louisiana
and Modesto, California) using vacancy as a proxy for weak seller bargaining
power. Previous research finds evidence that weak buyers pay higher prices and
weak sellers receive lower prices for their homes (Genesove and Mayer 1997,
Arnold 1999, Miceli, Sirmans and Yavas 2001, Anglin, Rutherford and Springer
2003, Harding, Rosenthal and Sirmans 2003). The main argument underlying
the Harding, Knight and Sirmans (2003) study is that sellers of vacant homes are
clearly at a disadvantage relative to other sellers, a disadvantage that presumably
weakens their bargaining power. The disadvantages come from two sources
identified at the outset: vacant homes are less appealing to potential buyers
and the higher holding cost makes their sellers more impatient to negotiate
a sale. Recognizing that price and marketing time are jointly determined in
search markets, they use instrumental variables for the selling time variable in
-
8/6/2019 Why Do Vacant Houses Sell for Less
4/25
22 Turnbull and Zahirovic-Herbert
the price equation. The results offer clear evidence that vacant houses sell for
lower prices, attributed to weaker seller bargaining power.
This article is organized as follows. The next section presents a simple frame-
work that integrates a Nash bargaining solution between buyer and seller into
a model of seller search behavior. The framework is simple yet sufficiently
flexible for analyzing the various channels through which vacant house status
can affect expected selling price and liquidity. The framework is useful in that
it suggests empirical proxies for measuring the relative importance of these
various channels. The empirical analysis is explained in the third and fourth
sections. The third section describes the identification problem that applies toall simultaneous models of selling price and liquidity in the course of explaining
the identification method used here. The fourth section discusses the empirical
results for the 20-year sample and studies more closely how different market
phases and degree of listing staleness affect the conclusions. The fifth section
concludes.
A Search Model with Bargaining
We assume that the selling price is determined under Nash bargaining, so
that the buyer and seller split the surplus from the transaction according to
their relative bargaining power. Suppose the sellers reservation price is r and
the buyers willingness to pay is b v, where v is a parametric shift variable
defined below. The selling price of the house under Nash bargaining is therefore
p arg max{(b v p)1 (p r) },
where the parameter reflects the sellers bargaining power or negotiating skill
relative to the buyer. Solving for p yields
p = (b v) + (1 )r. (1)
Clearly, the stronger the sellers bargaining power, the larger is and the closer
the ultimate selling price is to the buyers reservation price, b v. The weaker
the sellers bargaining power, the smaller is and the closer the ultimate selling
price is to the sellers reservation price, r. It is reasonable to assume (0, 1)
so that both the buyer and seller enjoy positive net benefits from the transaction.
The bargaining solution is easy to integrate into the simplest seller search
model. Consider a particular house that is listed for sale. The probability of
a potential buyer arriving to visit this house during a unit of time is . The
population of buyers is ordered by their willingness to pay, b, summarized
in the distribution function F(b; x). The distribution of buyers is in general a
-
8/6/2019 Why Do Vacant Houses Sell for Less
5/25
-
8/6/2019 Why Do Vacant Houses Sell for Less
6/25
24 Turnbull and Zahirovic-Herbert
E[p] =
br +v [(b v) + (1 )r
] dF(b)
1 F(r + v)(5)
which yields the expected selling price comparative statics
E[p]
> 0;
E[p]
c< 0;
E[p]
v< 0;
E[p]
> 0. (6)
Substituting the equilibrium reservation price r into (2) and differentiating
yields the comparative statics on the probability of sale at a given point in time
asq
< 0;
q
c> 0;
E[p]
v< 0;
and
q
=
[1 F(r + v)]2
f(r + v)
c
f(r + v)
[1 F(r + v)]
.
The equilibrium liquidity or expected selling time equals the inverse of theinstantaneous probability of sale, 1/q, so the expected selling time effects
take the opposite signs as the q effects. The following comparative statics
immediately follow from the above
E[T]
> 0;
E[T]
c< 0;
E[T]
v= 0;
and
E[T]
0)
lowers the sellers optimal reservation price and the expected selling price,
while shortening the expected time to sell. A decrease in buyers willingness
to pay (dv > 0), from whatever source, elicits a countervailing increase in
seller reservation price, which yields a negative effect on selling price and a
zero net effect on the expected marketing time. Recall from (6) that the seller
responds to a greater buyer arrival rate by increasing the reservation price.
Therefore, the effect of a higher buyer arrival rate on expected marketing time
(d > 0) depends on the relative size of the countervailing effects from the
-
8/6/2019 Why Do Vacant Houses Sell for Less
7/25
Why Do Vacant Houses Sell for Less 25
higher resultant reservation price (which tends to lengthen the expected selling
time) and the direct effect of the increased arrival rate (which tends to shortenthe expected selling time). The last inequality conditions in (7) reveal that a
larger c or smaller lead to a larger value of c/ henceE[T]/ > 0.
This counterintuitive result can arise because high seller search costs or weak
bargaining power are situations in which a faster buyer arrival rate prompts the
seller to set a much higher reservation price that outweighs the effect of the
greater arrival rate on the probability of sale; in these situations, the net effect
of a faster buyer arrival rate is a longer expected marketing time even though
expected selling price rises. Similarly, a faster buyer arrival rate when seller
search costs are low or seller bargaining power is high leads to a shorter expectedmarketing time coupled with a higher expected selling price. Because seller
bargaining power is likely to vary across the declining and rising market phases
(that is, after all, what defines a buyers market and a sellers market), the
model predicts that the buyer arrival rate effect on selling time is likely to vary
as well, with a more rapid buyer arrival rate decreasing selling time in rising
markets (when is large) and possibly increasing selling time in declining
markets (when is sufficiently small).
So what do these results imply for vacant house effects on observed prices
and liquidity? The higher holding cost associated with a vacant house (dc > 0)yields a lower seller reservation price and therefore lower selling price and
shorter expected selling time (Turnbull, Sirmans and Benjamin 1990). If, at the
same time, vacancy results in lower seller bargaining power (d < 0), as in
Harding, Knight and Sirmans (2003), then the Nash bargaining model implies
lower selling price for a given seller reservation price. The implications for
seller search behavior include a lower reservation price, hence lower selling
price and shorter time on the market. The prediction is identical to the effect
of higher holding cost on price and selling time, but, as noted earlier, we
expect bargaining power to wax and wane over the market cycle, althoughwe do not expect holding costs to exhibit the same systematic pattern. When
controlling for the influence of vacant house characteristics that reduce buyers
willingness to pay and other vacant house effects identified below, the vacant
house dummy variables in the price and liquidity equations will pick up the
combined effects of holding cost and bargaining power. Only by examining
changes in the dummy variable parameter estimates over the phases of the
housing market cycle can we expect to ascertain whether the dominant vacancy
effect is from higher holding costs or weaker bargaining power.
Vacancy might also signal the presence of an unobservable factor, possibly
atypicality, that reduces buyer willingness to pay for the house. The notion
here is that vacant houses have undesirable characteristics that are observed by
sellers and buyers but are not reported in the data (condition, architecture, etc.).
The bargaining model implies that a sellers reservation price entirely counters
-
8/6/2019 Why Do Vacant Houses Sell for Less
8/25
26 Turnbull and Zahirovic-Herbert
the reduction in the implicit value of such unobservable factors to buyers. An
undesirable feature therefore lowers the sellers reservation price (dv > 0),which commensurately lowers the expected selling price. The comparative
static prediction (7) reveals that the combination of the reduction in buyer
willingness to pay for the vacant house and the sellers matching reduction in
reservation price induces no change in expected liquidity. An empirical test
for this effect requires controlling for Haurins atypicality that is not related to
vacancy with atypicality variables in the price and liquidity equations. To pick
up effects of unobservable factors specific to vacant houses, we identify houses
that are vacant more than once during the 20-year span for which we have
data. If there are unmeasured attributes associated with vacant houses, thenhouses that are repeatedly vacant when sold are most likely to exhibit these
undesirable characteristics. In this case, we expect repeatedly vacant houses to
have negative price and zero liquidity effects.
Finally, we need to account for the possibility that vacant houses are subject to
different shopping externality and competition effects arising from neighbor-
ing houses that are also for sale. A greater number of surrounding houses for
sale has several effects (Turnbull and Dombrow 2006). The potential shopping
externality is introduced through visits to this house by buyers that have been
either attracted to visit this neighborhood in general or some other specifichouse in the neighborhood. Thus, in our search model, the shopping externality
expresses itself (if present) through > 0 with a greater number of surround-
ing listings. The potential competition effect can also be seen through this
term. If an increase in the number of houses for sale simply dilutes the number
of potential buyers visiting each individual house in the neighborhood, then
< 0 in the search model. But if the greater number of surrounding listings
reduces potential buyers willingness to pay for a given house, then v > 0 as
well. Under the assumption that holding cost is higher and bargaining power
lower for vacant houses, the shopping externality increases the expected salesprice and decreases the expected selling time. Similarly, in such cases the spa-
tial competition effect decreases expected sales price and increases expected
selling time.
The Identification Problem
Search and matching models of the housing market envision price and selling
time as jointly determined outcomes; different market or property character-
istics typically lead to combined price and liquidity effects.3 This approach
3Lippman and McCall (1978) provide a seminal influence on search models of hous-ing markets. See Arnott (1989), Haurin (1988), Krainer (2001), Williams (1995)and Wheaton (1990) for a variety of approaches grounded in search or matchingenvironments.
-
8/6/2019 Why Do Vacant Houses Sell for Less
9/25
Why Do Vacant Houses Sell for Less 27
motivates the Fisher et al. (2003) notion that relatively slow seller reaction to
changing conditions explains the observed relationship between prices and liq-uidity as well as the Genesove and Mayer (1997) notion that equity-constrained
sellers are more reluctant to lower their reservation prices, instead incurring
longer selling times to await the arrival of high-bidding buyers. In general,
search theory implies that empirical hedonic price analysis should take into
account simultaneous selling time or liquidity effects whenever possible. Most
attempts to do so, however, have been hampered by the theoretical implication
that both price and liquidity are simultaneously determined by virtually identi-
cal factors and therefore represent an underidentified simultaneous system.
Many empirical studies have long used log-linear regression models to sepa-
rately estimate selling time determinants (Belkin, Hempel and McLeavey 1976,
Miller 1978, Kang and Gardner 1989, Asabere, Huffman and Mehdian 1993).
A growing number explicitly recognize that selling price and selling time are
simultaneously determined using simultaneous or two-stage models (Sirmans,
Turnbull and Benjamin 1991, Yavas and Yang 1995, Forgey, Rutherford and
Springer 1996, Huang and Palmquist 2001, Rutherford, Springer and Yavas
2001, Knight 2002, Turnbull and Dombrow 2006, Turnbull, Dombrow and Sir-
mans 2006, Zahirovic-Herbert and Turnbull 2008). These papers offer a variety
of innovative ways of dealing with the identification problem, which typicallyhinge upon a variety of rationales for why some property characteristics only
affect selling price although others only affect selling time. Still, there is as
yet no generally accepted empirical framework for dealing with endogenous
price and liquidity in a systems context. In this article, we follow the method
proposed by Zahirovic-Herbert and Turnbull (2008), using variables captur-
ing neighborhood market conditions to identify separate price and liquidity
equations.
To understand the intuition of this method, recall that the search model impliesthat expected price, E[p], and selling time or liquidity, E[T], are simultane-
ously determined. Thus, for a house with characteristics vector X and neighbor-
hood market conditions summarized in the vector C, the relationship between
expected price and selling time is implicitly defined by the surface
(E[p], E[T], X, C) = 0.
We can express the realized price and selling time surface by separate functions
with jointly distributed stochastic errors p and T.p = p(T , X, C) + p (8)
T = T(p, X, C) + T. (9)
-
8/6/2019 Why Do Vacant Houses Sell for Less
10/25
28 Turnbull and Zahirovic-Herbert
The vector C captures the localized competition and shopping externality ef-
fects that turn out to be essential for identifying Equations (8) and (9). Turnbulland Dombrow (2006) measure neighborhood competition from nearby houses
for sale as long each competing listed house overlaps with the period that
this house is on the market, inversely weighted by the distance between the
houses to reflect the assumption that nearby houses will have stronger effects
on the sale of this house than houses that are farther away. The days on mar-
ket or selling time is s(i) l(i) + 1, where l(i) and s(i) are the listing date
and sales date for house i. Denoting the listing date and sales date for house
j by l(j) and s(j), the overlapping time on the market for these two houses
is min[s(i), s(j )] max [l(i), l(j )] + 1. The straight-line distance in miles be-tween houses i and j is D(i, j). The measured competition for house i is
C(i) =
j
(1 D(i, j ))2{min[s(i), s(j )] max[l(i), l(j )]}, (10)
where the summation is taken over all competing houses j, that is, houses for
sale within one mile and 20% larger or smaller in living area of house i.
The other neighborhood market condition variables in (8) and (9) are con-
structed following the same approach as that taken for the neighborhood com-petition variable (10). It turns out to be useful to also define another variable,
listing density, as the measure of competing overlapping listings per day on the
market
L(i) =
j
(1 D(i, j ))2{min[s(i), s(j )] max[l(i), l(j )]}
s(i) l(i) + 1. (11)
The price and selling time Equations (8) and (9) are functions of the same
predetermined variables and so do not appear to be identified. Note that re-
gressing sales price on the right-hand side variables yields the estimated effectof competition on price as the partial derivative p/C holding selling time
constant. Changing competition while holding selling time constant, however,
is simply a change in listing density (11). Therefore, p/C = p/L so
that the price function (8) can be rewritten as a function of the listing density
(11) instead of competition (10), which means that the system of equations for
price and selling time can be expressed as
p = p(T , X, L) + p (12)
T = T(p, X, C) + T. (13)
The separate L and C variables make it possible to identify both equations
in the estimation (Zahirovic-Herbert and Turnbull 2008). As important, this
-
8/6/2019 Why Do Vacant Houses Sell for Less
11/25
Why Do Vacant Houses Sell for Less 29
approach also explicitly introduces empirical controls for the neighborhood
market conditions thatwhen neglectedjustify the need to correct spatialcorrelation in housing price models. This approach models the spatial compe-
tition effects directly and therefore obviates the usual rationale for applying
spatial estimation methods.
The Empirical Analysis
The Data
We use a sample comprising broker-assisted housing transactions completedbetween October 1984 and April 2005. The sample period ends 3 months before
hurricane Katrina to avoid the influence of that event on property markets. The
data are drawn from the multiple listing service (MLS) sales reports for Baton
Rouge, Louisiana, a medium-size urban area that has been the subject for much
academic housing market research. Our data cover two decades during which
the local housing market experienced a downturn (19841987) followed by
an extended market trough (19881991) and then a modestly rising market
(19922005). Therefore, this sample also allows us to investigate the extent to
which the vacancy effect on price varies over the market cycle.
We restrict our attention to detached single-family houses sold within a con-
tiguous region in the urban area. There is evidence that the prices of houses in
new subdivisions diverge significantly from the broader market until the new
development reaches a critical mass (Sirmans, Turnbull and Dombrow 1997);
we avoid this potential pricing bias from new development by including in our
sample only those houses that are at least 4 years old.4 To avoid erroneous data
entries and outlier influence on selling time estimates, we exclude from the
sample houses that take fewer than 14 or more than 400 days to sell, houses
that sell for less than $40,000 or more than $320,000, houses with unusuallysmall (less than 300 square feet) or large (greater than 4,500 square feet) living
area and similarly for the area under roof net of living area (110 and 4,000,
respectively).5 The resultant data set comprises 27,630 transactions.
Table 1 summarizes the means and standard deviations of the variables used in
the empirical models for the full sample and for vacant and occupied subsam-
ples. The sales price (Price), days on the market prior to sale (DOM), number
4This has the added advantage of eliminating from the sample houses that are vacantbecause they are newly built.
5Note that houses sold within 14 days of listing are nonetheless included in the con-struction of the New Listing Density and New Competition variables explained in theprevious section.
-
8/6/2019 Why Do Vacant Houses Sell for Less
12/25
30 Turnbull and Zahirovic-Herbert
Table1
Variabledefinitionsandsamplesu
mmarystatistics.
Fullsample
Vacantsubsample
Occupiedsubsample
Differenc
ein
Variablename
Definition
Mean
Std.dev.
Mean
Std.dev.
Mean
Std.dev.
meansT-test
Price
Sellingpriceofhouse
106,183.60
50,426.67
97,571.79
46,665.8311
0,001.00
5,1549.80
19.77
DOM
Daysonthemarketpriortosale
86.20
71.14
94.52
72.41
82.52
70.26
12.82
Bedrooms
Numberofbedrooms
3.32
0.60
3.29
0.60
3.34
0.60
7.3
Bathrooms
Numberofbathrooms
2.01
0.47
1.97
0.48
2.03
0.47
9.51
Livingarea
Squarefeetoflivingarea
1,929.83
586.59
1,874.25
585.77
1,954.47
585.28
10.5
Netarea
Squarefeetofotherare
a
697.31
317.33
660.31
309.74
713.72
319.27
13.1
Listingdensity
Competinglistingswei
ghtedby
days
2.46
2.064
2.42
2.05
2.48
2.07
2.18
Competition
Competinglistings
211.00
280.07
233.13
300.99
201.19
269.71
8.4
Newlistingdensity
Competingnewlistingsweighted
bydays
1.13
1.12
1.13
1.13
1.13
1.11
0.1
Newcompetition
Competingnewlistings
118.39
197.00
131.83
208.96
112.42
191.17
7.31
Vacantlisting
density
Competingvacantlistings
weightedbydays
1.01
1.10
1.06
1.4
0.99
1.08
5.1
Vacantcompe
tition
Competingvacantlistings
87.80
139.31
103.79
157.19
80.72
129.98
11.84
Larger
Positivedeviationsfrom
local
meanlivingarea
0.11
0.19
0.10
0.19
0.11
0.19
3.93
Smaller
Negativedeviationsfro
mlocal
meanlivingarea
0.07
0.11
0.08
0.11
0.07
0.10
7.06
Vacant
Vacanthousedummyv
ariable
0.31
0.46
1
0
0
0
Repeatsale
Soldmorethanoncedu
ringthe
sampleperiod
0.47
0.50
0.42
0.49
0.48
0.50
9.14
Twicevacant
Dummyindicatinghou
sesoldas
vacanttwotimesseq
uentially
0.04
0.20
0.14
0.34
0
0
36.54
Repeatvacant
Dummyindicatinghou
sesoldas
vacantmorethanonce
0.10
0.30
0.20
0.40
0.05
0.23
30.78
Observations
27,630
8,486
19,144
-
8/6/2019 Why Do Vacant Houses Sell for Less
13/25
Why Do Vacant Houses Sell for Less 31
of bedrooms (Bedrooms), number of bathrooms (Bathrooms), a set of dummy
variables for the age category of the house (Age i) and living area (Living Area)are drawn directly from the MLS report for each sale. The Net Area variable is
calculated as the difference between the total square footage under roof less the
square footage of living area, and it captures the size of utility rooms, garages,
covered porches, carports, etc. The house characteristics also include location,
captured by dummy variables for the set of 84 census tracts. Finally, to control
for broad housing market conditions, all of the models include year and season
fixed effects.
The models also include neighborhood market condition variables suggestedby Turnbull and Dombrow (2006). The variable Listing Density measures the
intensity of competition from other houses for sale per day on the market
(11); Competition measures the total competition from other houses over the
entire marketing time for a given house (10). These variables explained above
not only provide the means of identifying the two separate price and liquidity
Equations (12) and (13), they also convey important insights into the nature
of spatial competition. As explained earlier, a greater number of competing
houses for sale surrounding a given house increases competition for buyers,
but at the same time can lead to shopping externality effects as the greater
concentration of listings draws more potential buyers to the neighborhood. Thesign of the coefficients on the listing density and competition variables therefore
reveal the relative strength of the spatial competition and shopping externality
effects.
The variables New Listing Density and New Competition are defined anal-
ogously to the listing density and competition variables, except that they
only include newly listed houses in their calculation (i.e., houses within
their first 14 days of listing). Vacant Listing Density and Vacant Competi-
tion are constructed similarly except that they only include competing vacanthouses. The coefficients on these variables reflect the offsetting or add-on
spatial competition or shopping externality affects of newly listed and vacant
houses relative to occupied listings that have been on the market longer than
2 weeks.
These neighborhood market condition variables are based on all applicable
competing house sales (within 20% of the living area and within one mile of
the sold house), which include all relevant competing houses listed before and
after our sample period that overlap with the sample period. We use the Stata
algorithm explained in Zahirovic-Herbert (2008) to construct these variables.
We include the relative house size variables Larger and Smaller to capture
atypicality effects unrelated to vacancy. These variables measure the extent to
-
8/6/2019 Why Do Vacant Houses Sell for Less
14/25
32 Turnbull and Zahirovic-Herbert
which a given house is either larger or smaller than the average living area
in the surrounding neighborhood. Following Turnbull, Dombrow and Sirmans(2006), indexing all houses within a one-half-mile radius of house i by J, the
standardized measure of the relative house size is
Localsizei =
Livingareai j J
Livingareaj /Nj
j J
Livingareaj /Nj,
where Nj is the number of surrounding houses in the neighborhood J. To allow
for asymmetric relative house size effects on sales price, we define the relativesize variables Largeri and Smalleri as the absolute values of the positive and
negative values ofLocalsizei, respectively:
Largeri = |Localsize| for Localsizei > 0
= 0 otherwise;
Smalleri = |Localsize| for Localsizei < 0
= 0 otherwise.
Using the absolute value means that the variable Smalleris always nonnegative,a point to note when interpreting the empirical results.
Vacant is a dummy variable indicating an unoccupied property. When controls
for spatial competition/shopping externalities and unobserved atypicality or
undesirable attributes are included in the model, the Vacant coefficient should
primarily pick up the combined effects of higher seller holding costs and
lower seller bargaining power (recall that these two effects are observationally
equivalent in the search model). Nonetheless, to the extent that seller search
or holding costs do not vary across market phases, observed differences inthe Vacancy coefficients across the market phases reveal differences in seller
bargaining power. Table 1 shows that 31% of transactions are for vacant houses
in the sample. This proportion varies significantly over the different phases
of the local housing market cycle, with 38% vacant in the declining market
(19851988), about 35% vacant in the market trough (19881991) and 29%
vacant in the rising market (19922005).
Repeat Sale is a dummy variable for houses that sell more than once during
the sample period. Approximately 47% of the transactions involve houses that
sold more than once during the sample period. The variable Twice Vacant
is a dummy variable indicating that the house has, at some time during the
20-year sample period, been vacant two times sequentially. Repeat Vacant,
on the other hand, is a dummy variable indicating that the house (whether
-
8/6/2019 Why Do Vacant Houses Sell for Less
15/25
Why Do Vacant Houses Sell for Less 33
currently vacant or not) has been vacant more than once during the 20-year
period, not necessarily sequentially. Although 10% of the sample compriseshouses that are sold vacant more than once during the sample period, only
4% are vacant two or more times in sequence. The Twice Vacant and Repeat
Vacant variables are included to identify houses that might have unobserved
(in the data) characteristics or atypicality not related to relative size, features
that may explain their multiple vacant status whether or not they are vacant in
the current transaction. Negative attributes of these repeatedly vacant houses
are associated with a lower buyer willingness to pay (dv > 0 in the search
model) and therefore imply negative coefficients on these variables in the price
equation and insignificant liquidity equation coefficients.
Full Sample Analysis
Table 2 reports the model estimates for the full sample. As indicated in the
first two columns, the base model specifies the natural log of sales price as
a function of the selling time, house characteristics, location and time period
dummy variables (not reported) and the set of listing density variables capturing
neighborhood housing market conditions. Liquidity or selling time is, in turn,
a function of the log of the sales price, house characteristics, location and time
dummy variables and the set of competition variables as a different measure ofneighborhood housing market conditions. The coefficients on these variables
follow expectations.
Both equations include a dummy variable for houses that are vacant during the
listing period. The price equation estimates follow popular notions as well as
what has been typically found to date: vacancy leads to lower selling price. In-
terestingly, the Vacantcoefficient in the liquidity equation indicates that vacant
houses sell more quickly on average than do their occupied counterparts, ceteris
paribus. These coefficients are significant and robust across specifications. Thesign pattern is consistent with higher holding cost or lower seller bargaining
power; the model in the first two columns, however, does not control for the
other channels through which vacancy might affect price and liquidity.
Together, the insignificant coefficient on Listing Density in the price equa-
tion and the significantly positive coefficient on Competition in the liquidity
equation indicate the presence of both spatial competition and shopping ex-
ternalities (Turnbull and Dombrow 2006). The sign pattern on New Listing
Density, New Competition, Vacant Listing Density and Vacant Competition re-
veal that new listings in the neighborhood have add-on shopping externality
effects while vacant listings increase the spatial competition effects on price
and selling time. The interaction variable, Vacant Listing Density, is signif-
icantly positive in the price equation, which means that vacant houses enjoy
-
8/6/2019 Why Do Vacant Houses Sell for Less
16/25
34 Turnbull and Zahirovic-Herbert
Table2
Tw
o-stageleastsquaresparameterestimates.
(1)
(2)
(3)
Price
DOM
Pric
e
DOM
Price
DOM
Variables
Equation
Equation
Equation
Equation
Equation
Equation
Ln
Price
139
.
2
138
.
9
107
.
4
(7.
03)
(7.
01)
(7.
31)
DOM(103)a
0
.
257
0
.
257
0
.
263
(0.
024)
(0
.
024)
(0.
023)
Bedrooms
0
.
0103
3
.
742
0
.
0105
3
.
698
0
.
0126
2
.
455
(0.
0023)
(0.
77)
(0
.
0023)
(0.
77)
(0.
0022)
(0.
74)
Bathrooms
0.
0324
5.
137
0
.
0324
5.
120
0.
0366
4.
560
(0.
0028)
(0.
97)
(0
.
0028)
(0.
97)
(0.
0026)
(0.
94)
Livingarea(103)
0.
435
0.
0796
0
.
435
0.
0795
0.
607
0.
0872
(0.
0029)
(0.
0032)
(0
.
0029)
(0.
0032)
(0.
0044)
(0.
0047)
Netarea(103)
0.
144
0.
0184
0
.
145
0.
0183
0.
127
0.
0125
(0.
0037)
(0.
0016)
(0
.
0037)
(0.
0016)
(0.
0035)
(0.
0015)
Vacant(103
)
66
.
8
4
.
139
66
.
3
4
.
550
65
.
7
2
.
706
(2.
9)
(0.
99)
(3
.
0)
(1.
02)
(2.
8)
(0.
99)
Listing
density(103)
0.
551
0
.
508
2
.
72
(1.
3)
(1
.
3)
(1.
2)
Compet
ition
0.
0188
0.
0188
0.
0301
(0.
0046)
(0.
0046)
(0.
0045)
New
list
ingdensity(103)
6.
05
6
.
04
6.
19
(1.
9)
(1
.
9)
(1.
8)
Newcompetition
0.
233
0.
233
0.
227
(0.
0053)
(0.
0053)
(0.
0051)
Vacant
list
ing
density(103)
0
.
472
0
.
433
0.
684
(1.
7)
(1
.
7)
(1.
6)
-
8/6/2019 Why Do Vacant Houses Sell for Less
17/25
Why Do Vacant Houses Sell for Less 35
Table2
continued
(1)
(2)
(3)
Price
DOM
P
rice
DOM
Price
DOM
Variables
Equation
Equation
E
quation
Equation
Equation
Equation
Vacantcompet
ition
0
.
0188
0
.
0187
0
.
0226
(0.
0057)
(0.
0057)
(0.
0054)
Vacant
listing
density(103)
4.32
4.
33
3.
87
(1.9)
(1.
9)
(1.
8)
Vacantcom
pet
ition
0
.
00620
0
.
00630
0
.
00820
(0.
0048)
(0.
0048)
(0.
0046)
Tw
icevacant
(103)
7.
28
2.
307
8
.
88
2.
589
(6.
4)
(2.
15)
(6.
1)
(2.
07)
Repeatvacan
t(103)
4.
82
0.
353
6.
14
0.
243
(4.
4)
(1.
48)
(4.
2)
(1.
42)
Repeatsa
le(103)
5.
57
1
.
794
3.
70
1
.
422
(2.
2)
(0.
72)
(2.
0)
(0.
69)
Larger
0
.
491
28
.
24
(0.
0089)
(4.
84)
Sma
ller
0.
193
102
.
8
(0.
013)
(4.
45)
Constant
11
.14
1,5
81
11
.
14
1,5
78
10
.
83
1,1
77
(0.014)
(77
.
6)
(0.
014)
(77
.
4)
(0.
015)
(78
.
4)
Observations
27,63
0
27,6
30
27,6
30
27,6
30
27,6
30
27,630
R2
0.86
0.
43
0.
86
0.
43
0.
88
0.
48
Notes:Whitesrobuststandarderrorestimatesinparenthesis.
Dummyvariablesfor84censustractareas,eighthouseagerangecategoriesand
seasonandyearsoldarenotreportedinthistable.
indicatessignificanceatthe10%level;indicatessignifi
canceatthe5%level;indica
tes
significanceatthe1%level.
a(103)usedonlyinthepriceequations.
-
8/6/2019 Why Do Vacant Houses Sell for Less
18/25
36 Turnbull and Zahirovic-Herbert
additional shopping externalities from surrounding listings than do occupied
houses. The interaction variable Vacant Competition is not significant in theselling time equation, which by itself indicates no add-on competition effects
from surrounding vacant houses.
Columns (3) and (4) in Table 2 report the results when Twice Vacant, Repeat
Vacant and Repeat Sale are added to the model. As already noted, the addition
of these variables does not dramatically change the coefficient estimates on
the variables already discussed. The Repeat Sale coefficients are significantly
positive in the price equation and negative in the liquidity equation; the vacancy-
related variables, however, are not significant. The insignificant coefficients onthese variables in the liquidity equations are consistent with dv 0 in the
search model. The insignificant coefficients in the price equations, however,
indicate that they are not picking up any vacant attribute effects that reduce
buyer willingness to pay. Overall, the results are consistent with dv = 0 for
vacant houses in the search model.
Columns (5) and (6) report the estimates when the general atypicality variables
Larger and Smallerare included in the model. The coefficients on these neigh-
borhood relative house size variables are significant and indicate that larger
houses in mixed neighborhoods sell at a discount while smaller houses inmixed neighborhoods sell at a premium, both relative to homogeneous neigh-
borhoods (Turnbull, Dombrow and Sirmans 2006). More importantly for our
purposes, their inclusion in the model does not significantly alter the Vacant
effects found earlier.
When taken together, the results reported in Table 2 indicate that our Vacant
coefficient estimates are picking up primarily seller holding cost and bargaining
power effects on price and liquidity. There is no evidence of the vacant house
effect being driven by atypicality or unobserved negative attributes. The inter-action term, however, provides evidence that vacant houses enjoy an enhanced
shopping externality effect from surrounding listings.
Market Cycle and Stale Listings
We expect seller bargaining power to vary over the housing market cycle. One
advantage of this data set is that it encompasses periods covering different
market phases. Based on a residential constant quality price index, the market
phases are as follows: declining market over 19841987, market trough over
19881991 and modestly rising market over 19922005. Table 3 reports the
parameter estimates of central interest for the separate phases. The first two
columns present the full sample estimates for comparison purposes. Overall,
the Vacant effects on price and selling time are robust across market phases
-
8/6/2019 Why Do Vacant Houses Sell for Less
19/25
Why Do Vacant Houses Sell for Less 37
Table3
Ke
yparameterestimatesacrossm
arketphases.a
FullSample
RisingMarket
MarketTrough
FallingMarket
Variables
Ln
Price
D
OM
Ln
Price
DO
M
Ln
Price
DOM
Ln
Price
DOM
Vacant
0
.
0663
4.
550
0
.
0624
1
.
352
0
.
0847
30
.
23
0
.
0691
20.17
(0.
0030)
(1.
02)
(0.
0036)
(1
.
05)
(0.
0067)
(4.
39)
(0.
0079)
(5.08)
Vacant
Listing
Density
0.
00433
0.
00862
0.
00489
0.
00478
(0.
0019)
(0.
0025)
(0.
0039)
(0.
0040)
Vacant
Com
pet
ition
0.
00630
0
.
0176
0
.
00949
0.0143
(0.
0048)
(0
.
0065)
(0.
014)
(0.013)
Repeat
Vacan
t
0.
00482
0.
353
0
.
000134
0
.
639
0.
0143
4.
958
0.
0155
0.387
(0.
0044)
(1.
48)
(0.
0050)
(1
.
49)
(0.
012)
(5.
77)
(0.
014)
(7.16)
Tw
ice
Vacant
0
.
00728
2.
307
0
.
0157
2
.
016
0
.
00257
1.
123
0.
00921
5.201
(0.
0064)
(2.
15)
(0.
0075)
(2
.
24)
(0.
016)
(8.
01)
(0.
018)
(9.20)
Observations
27,6
30
27,6
30
19,8
23
1
9,8
23
4,8
34
4,834
2,8
18
2,8
18
Notes:White
srobuststandarderrorestimatesinparenthesis.
indicatessignificanceatthe10%level;ind
icatessignificanceatthe5%level;
indicatess
ignificanceatthe1%level.
aKeyparameterestimatesfrommodel(2)in
Table2estimatedontheindicatedsubsamples.
-
8/6/2019 Why Do Vacant Houses Sell for Less
20/25
38 Turnbull and Zahirovic-Herbert
(with the exception of the insignificant liquidity effect in the rising market).6
Somewhat surprisingly, there is no strong evidence of systematic variation inseller bargaining power for vacant houses relative to occupied houses over the
market cycle. The results are consistent with the notion that vacancy does not
reduce seller bargaining power per se, but rather increases sellers willingness
to sell in response to the relatively higher cost of holding vacant houses on the
market.
The Repeat Vacant and Twice Vacant estimates are also robust across the dif-
ferent market phases. The Vacant Listing Density and Vacant Competition
interaction terms, however, yield different results for the rising market thanthe declining and trough phases. The latter two show no additional shopping
externality or competition effects for vacant houses. In the rising market, how-
ever, the coefficients on these variables are both significantly positive. This last
result is puzzling because the positive price effect indicates a stronger shopping
externality on vacant houses from surrounding listings while the positive sell-
ing time effect indicates stronger competition from surrounding listings. These
results, in particular, deserve additional scrutiny.
To better understand these results, we next consider the possibility that vacant
house effects might vary according to listing staleness. Table 4 reports the keyparameter estimates for the complete model when the sample is partitioned
according to how fresh or stale the listing is when sold. In each case, we
partition the data into thirds according to days on the market. The lowest third
represents the fresh listing sample, the next third the average listing sample
and the highest third the stale listings sample. The partitions (in terms of days
on the market), of course, vary across the full, rising, trough and declining
market phases. Note that the number of observations in each data partition is
not precisely one third of the total because of clustering along this dimension.
The full sample results reveal some interesting differences across listing stale-
ness. The Vacantprice effect is weaker and the liquidity effect stronger for stale
than for fresh or average listings. Similar patterns emerge for the three market
phases as well. These patterns are consistent with the simple search theory:
those sellers with more modest holding costs do not reduce their reservation
prices by as much as their counterparts with higher holding costs, and as a
consequence they do not sell at as great a discount and wait longer on average
to sell their vacant houses.
6To examine the robustness of these patterns, we also estimated a model includinginteractive year and vacant dummy variables. The interactive effects in the price equationshow no systematic variation in vacant effect on price over the sample. The interactiveeffects in the selling time equation reveal no systematic variation in vacant effect onselling time beyond that already identified across market phases.
-
8/6/2019 Why Do Vacant Houses Sell for Less
21/25
Why Do Vacant Houses Sell for Less 39
Table4
Ke
yparameterestimatesacrossm
arketphasesandlistingstaleness.a
FullSample
RisingMarket
MarketTrough
FallingMarket
Listing
Variables
Sta
leness
Ln
Price
DOM
Ln
Price
D
OM
Ln
Price
DOM
Ln
Price
DOM
Vacant
Fresh
0
.
00377
0.
140
0
.
0723
0.
852
0
.
0947
2.
682
0
.
0906
5
.
642
(0.
011)
(0.
58)
(0.
0058)
(0.
30)
(0.
011)
(1.
05)
(0.
013)
(2.
01)
Average
0
.
00390
0
.
184
0
.
0659
0.
576
0
.
0910
11.
23
0
.
0506
2
.
242
(0.
010)
(1.
00)
(0.
0059)
(0.
54)
(0.
011)
(2.
94)
(0.
014)
(2.
43)
Stale
0
.
0192
1
.
703
0
.
0510
5.
802
0
.
0627
20.
35
0
.
0552
9
.
697
(0.
011)
(3.
85)
(0.
0062)
(2.
06)
(0.
012)
(6.
35)
(0.
015)
(6.
83)
Vacant
Listing
Fresh
0.
000666
0.
00488
0.
00400
0.
00467
Density
(0.
0031)
(0.
0040)
(0.
0064)
(0.
0061)
Average
0.
00590
0.
0103
0.
00707
0
.
00225
(0.
0030)
(0.
0041)
(0.
0062)
(0.
0070)
Stale
0.
00641
0.
00852
0.
00468
0.
00480
(0.
0032)
(0.
0044)
(0.
0070)
(0.
0078)
Vacant
Fresh
0.
00699
0.
0143
0.
00483
0.
0190
Compet
itio
n
(0.
0054)
(0.
0068)
(0.
013)
(0.
016)
Average
0.
00704
0.
00723
0.
00114
0
.
0143
(0.
0044)
(0.
0056)
(0.
014)
(0.
010)
Stale
0
.
000466
0.
0127
0.
0145
0
.
000796
(0.
0061)
(0.
0086)
(0.
015)
(0.
014)
-
8/6/2019 Why Do Vacant Houses Sell for Less
22/25
40 Turnbull and Zahirovic-Herbert
Table4
continued
FullSample
RisingMarket
MarketTrough
FallingMarket
Listing
Variables
sta
leness
Ln
Price
DOM
Ln
Price
DOM
Ln
Price
DOM
Ln
Price
DOM
Repeat
Vacan
t
Fresh
0.
00737
0.
0799
0.
00162
0.
133
0
.
000431
0.981
0.
0152
0.649
(0.
0068)
(0.
35)
(0.
0076)
(
0.
36)
(0.
018)
(1.22)
(0.
020)
(2.32)
Average
0.
00303
0.
291
0.
00575
0.
0193
0.
0112
5.015
0
.
0000134
4.877
(0.
0069)
(0.
69)
(0.
0077)
(
0.
65)
(0.
020)
(3.63)
(0.
026)
(3.74)
Stale
0.
00805
0.
248
0
.
00569
3.
558
0.
0239
7.015
0.
0344
7.731
(0.
0081)
(2.
87)
(0.
0091)
(
2.
99)
(0.
020)
(9.06)
(0.
025)
(11.5)
Tw
ice
Vacant
Fresh
0
.
00377
0.
140
0
.
00621
0.
375
0
.
00806
0.786
0.
0177
3.100
(0.
011)
(0.
58)
(0.
013)
(
0.
62)
(0.
028)
(1.85)
(0.
028)
(3.21)
Average
0
.
00390
0
.
184
0
.
0206
0.
0292
0
.
00998
5.955
0.
0176
4.073
(0.
010)
(1.
00)
(0.
012)
(
1.
00)
(0.
027)
(4.95)
(0.
033)
(4.68)
Stale
0
.
0192
1
.
703
0
.
0207
1.
909
0
.
0217
8.578
0
.
0134
4.628
(0.
011)
(3.
85)
(0.
012)
(
4.
06)
(0.
027)
(12.0)
(0.
032)
(14.3)
Observations
Fresh
9,1
44
9,1
44
6,6
82
6,6
82
1,6
22
1,622
938
938
Average
9,2
07
9,2
07
6,5
17
6,5
17
1,5
95
1,595
949
949
Stale
9,2
79
9,2
79
6,6
24
6,6
24
1,6
17
1,617
931
931
Notes:White
srobuststandarderrorestimatesinparenthesis.
indicatessignificanceatthe10%level;ind
icatessignificanceatthe5%level;
indicatess
ignificanceatthe1%level.
aKeyparameterestimatesfrommodel(2)inTable2estimatedontheindicatedsubsamples.
-
8/6/2019 Why Do Vacant Houses Sell for Less
23/25
Why Do Vacant Houses Sell for Less 41
Twice Vacant does not affect prices or liquidity in the fresh or average listings
in the full sample, but it does lead to significantly lower prices for stale listings.Coupled with the insignificant effect on selling time, this result is consistent
with a negative stigma effect for stale vacant listings (dv > 0 in the search
model). Looking at the estimates across the different market phases, we see
that these full sample results appear to be driven by the vacant house stigma
effects for average and stale vacant houses in the rising market. No vacant
stigma or atypicality effects are evident in the trough or declining market
phases.
Turning to the interactive terms, the significantly positive coefficients in theprice equations for average and stale listings and the insignificant coefficient
for fresh listing in the full sample indicate that average and stale vacant houses
garner stronger shopping externalities from surrounding houses on the market
while fresh houses do not. The market phase estimates, however, show that this
conclusion for the full sample is primarily driven by the rising market phase.
There is no evidence of differential shopping externality or spatial competition
effects for vacant houses in the trough or declining market phases.
Conclusion
This article introduced Nash bargaining into a simple seller search model
to study the vacancypriceliquidity nexus through greater seller holding cost,
lower seller bargaining power and also unobserved property attributes or stigma
perceived by buyers. The model emphasizes the difference between a seller with
weaker bargaining power and a seller motivated to accept a lower price because
of higher holding costs. Both tend to reduce expected selling price, but the
systematic changes in bargaining power associated with the various phases of
the housing market cycle imply varying vacancy price and liquidity effectsacross the market cycle not expected for holding cost effects. Using 20 years
of house transactions data from Baton Rouge, Louisiana, the estimates show
robust vacancy effects on price and liquidity across all market phases consistent
with greater seller holding cost and diminished bargaining power that does not
vary systematically over the market cycle. At the same time, vacant houses
that are no longer fresh listings tend to enjoy stronger shopping externality
effects than do their occupied counterparts from surrounding houses for sale in
rising markets; vacant houses do not enjoy additional shopping externality or
competitive effects in the trough or declining market phases. The estimates also
show no vacant house stigma or negative unobserved attribute or vacant-related
atypicality effects on price and selling time in declining or trough market phases.
There is, however, a significant negative stigma or attribute effect associated
with repeatedly vacant houses in rising markets.
-
8/6/2019 Why Do Vacant Houses Sell for Less
24/25
42 Turnbull and Zahirovic-Herbert
This article received the 2008 American Real Estate Society Award for Best
Paper in Valuation. We gratefully acknowledge the helpful comments and sug-gestions of session participants and the anonymous referees.
References
Anglin, P., R. Rutherford and T.M. Springer. 2003. The Tradeoff between Selling Priceof Residential Properties and Time-on-the-Market: The Impact of Price Setting. TheJournal of Real Estate Finance and Economics 26: 95111.
Arnold, M. 1999. Search, Bargaining and Optimal Asking Prices.Real Estate Economics27(3): 453482.
Arnott, R. 1989. Housing Vacancies, Thin Markets, and Idiosyncratic Tastes. The Jour-nal of Real Estate Finance and Economics 2: 530.
Asabere, P.K. and F.E. Huffman. 1993. The Impact of Settlement Period on Sales Price.The Journal of Real Estate Finance and Economics 7: 213219.
Asabere P.K., F.E. Huffman and S. Mehdian. 1993. Mispricing and Optimal Time onthe Market. Journal of Real Estate Research 8(1): 149155.
Belkin, J., D.J. Hempel and D.W. McLeavey. 1976. An Empirical Study of Time onMarket Using Multidimensional Segmentation of Housing Markets. AREUEA Journal4: 5775.
Fisher, J., D. Gatzlaff, D. Geltner and D. Haurin. 2003. Controlling for the Impact ofVariable Liquidity in Commercial Real Estate Price Indices. Real Estate Economics 31:
269303.Forgey, F.A., R.C. Rutherford and T.M. Springer. 1996. Search and Liquidity in Single-Family Housing. Real Estate Economics 24: 273292.
Genesove, D. and C.J. Mayer. 1997. Equity and Time to Sale in the Real Estate Market.American Economic Review 87: 255269.
Harding, J.P., J.R. Knight and C.F. Sirmans. 2003. Estimating Bargaining Effects in He-donic Models: Evidence from the Housing Market.Real Estate Economics 31: 601622.
Harding, J.P., S.R. Rosenthal and C.F. Sirmans. 2003. Estimating Bargaining Power inthe Market for Existing Homes. Review of Economics and Statistics 85: 178188.
Haurin, D. 1988. The Duration of Marketing Time of Residential Housing. AREUEAJournal 16: 396410.
Huang, J. and R.B. Palmquist. 2001. Environmental Conditions, Reservation Prices, andTime on the Market for Housing. The Journal of Real Estate Finance and Economics22: 203219.
Kang, H.B. and M.J. Gardner. 1989. Selling Price and Marketing Time in the ResidentialReal Estate Market. Journal of Real Estate Research 4: 2135.
Knight, J.R. 2002. Listing Price, Time on the Market, and Ultimate Selling Price: Causesand Effects of Listing Price Changes. Real Estate Economics 30: 213237.
Krainer, J. 2001. A Theory of Liquidity in Residential Real Estate Markets. Journal ofUrban Economics 49: 3253.
Lippman, S.A. and J.J. McCall. 1976. The Economics of Job Search: A Survey. Eco-nomic Inquiry 14: 155190.
Miceli, T.J., C.F. Sirmans and A. Yavas. 2001. An Experimental Analysis of the Impactof Intermediaries on the Outcome of Bargaining Games. Real Estate Economics 29:251276.
Miller, N.G. 1978. Time on the Market and Selling Price. Real Estate Economics 6:164174.
-
8/6/2019 Why Do Vacant Houses Sell for Less
25/25
Why Do Vacant Houses Sell for Less 43
Mundy, B. 1992. Stigma and Value. Appraisal Journal 60: 714.
Rutherford, R.C., T.M. Springer and A. Yavas. 2001. The Impacts of Contract Type onBroker Performance. Real Estate Economics 29: 389409.
Sirmans, C.F., G.K. Turnbull and J.D. Benjamin. 1991. The Markets for Housing andReal Estate Broker Services. Journal of Housing Economics 1: 207217.
Sirmans, C.F., G.K. Turnbull and J. Dombrow. 1995. Quick House Sales: Seller Mistakeor Luck? Journal of Housing Economics 4: 230243.
. 1997. Residential Development, Risk, and Land Prices. Journal of RegionalScience 37: 613628.
Springer, T.M. 1996. Single-Family Housing Transactions: Seller Motivations, Price,and Marketing Time. The Journal of Real Estate Finance and Economics 13: 237254.
Turnbull, G.K. and J. Dombrow. 2006. Spatial Competition and Shopping Externalities:Evidence from the Housing Market. The Journal of Real Estate Finance and Economics32: 391408.
Turnbull, G.K., C.F. Sirmans and J.D. Benjamin. 1990. Do Corporations Sell Housesfor Less? A Test of Housing Market Efficiency. Applied Economics 66: 555578.
Turnbull G.K., J. Dombrow and C.F. Sirmans. 2006. Big House, Little House: RelativeSize and Value. Real Estate Economics 34: 439456.
Wheaton, W.C. 1990. Vacancy, Search, and Prices in a Housing Market Matching Model.Journal of Political Economy 98: 12701292.
Williams, J.T. 1995. Pricing Real Assets with Costly Search.Review of Financial Studies8: 5590.
Yavas, A. and S. Yang. 1995. The Strategic Role of Listing Price in Marketing RealEstate: Theory and Evidence. Real Estate Economics 23: 347368.
Zahirovic-Herbert, V. 2008. A STATA Program for Calculating Neighborhood MarketConditions. Available at SSRN: http://ssrn.com/abstract=1676839 (accessed October11, 2010).
Zahirovic-Herbert, V. and G.K. Turnbull. 2008. School Quality, House Prices and Liq-uidity. The Journal of Real Estate Finance and Economics 37: 113130.
Zuehlke, T.W. 1987. Duration Dependence in the Housing Market.Review of Economicsand Statistics 69: 701704.