ESTIMATING HEDONIC DEMAND PARAMETERS IN REAL
ESTATE MARKET: THE CASE OF MUGLA
Ercan BALDEMİR
Cüneyt Yenal KESBİÇ
Mustafa İNCİ
ABSTRACT
The affect of each characteristic of a heterogeneous good on its price can be determined
by Hedonic Price Models. This feature is originated from the assumption of the model that the
price of a heterogeneous good is the sum of the prices of the characteristics constituting that good.
Therefore, marginal prices for heterogeneous goods enter into the scene. In this respect, we
examine the marginal effects of different housing attributes on the selling price of houses in the
housing market in Mugla.
178 observations have been obtained through a questionnaire based on face to face
interviews with randomly selected real estate agencies from the urban districts of Mugla. The
analyses have been carried out by linear, logarithmic and logarithmic-linear functional forms,
which are frequently used in Hedonic Price Models. Considering the structure and features of
Mugla, the estimated coefficients are found to be significant in terms of both the characteristics of
housing and its location (its position, whether it is in-site or not, etc.). As expected, the variables
that positively affect the housing price have been found in all linear, logarithmic, and log-linear
models to be central heating, ceramic bathroom floor, location on the street, satellite TV,
hydrophor pump, modular kitchen, sunblind, solar water heating, front side facing south,
1500-2000 meters to the city center, the number of bathrooms, square meter of housing, and elevator.
Key Words: Hedonic theory, house markets, hedonic price model.
Emlak Piyasasında Hedonik Talep Parametrelerinin Tahminlenmesi:
Muğla Örneği
ÖZET
Hedonik Fiyatlandırma Modeli ile heterojen bir malı oluşturan karakteristiklerin her
birinin fiyat üzerindeki etkisi tanımlanabilir. Bu durum, Model‘in, heterojen bir malın fiyatının,
onu oluşturan farklı niteliklerin piyasa fiyatlarının toplamından ibaret olduğunu varsaymasından
ileri gelir. Böylece heterojen mallar için marjinal fiyatlar söz konusu olmaktadır. Bu bağlamda,
çalışmada Muğla Konut Piyasasında konutların sahip olduğu farklı niteliklerin konut satış fiyatı
üzerindeki marjinal etkisi ortaya konmaya çalışılmıştır.
Muğla ili kentsel kesimde merkez ilçelerde emlak bürolarında emlakçılarla yüz yüze
görüşme suretiyle tesadüfi olarak 178 anket yapılmıştır. Analizler hedonik fiyat modelinde
sıklıkla benimsenen doğrusal, logaritmik ve logaritmik doğrusal fonksiyonlar kullanılarak
gerçekleştirilmiştir. Katsayı tahminleri gerek konutun özellikleri, gerekse konumu (konutun yeri,
site içinde olup olmaması vb.) açısından Muğla ilinin yapısı ve özellikleri dikkate alındığında
anlamlı çıkmıştır. Beklendiği gibi, doğrusal, logaritmik ve logaritmik doğrusal modellerin
Associate Professor, Mugla University, F.E.A.S., Department of Business Administration.
Associate Professor, Mugla University, F.E.A.S., Department of Economics.
hepsinde konut satış fiyatını pozitif etkileyen değişkenler; merkezi kalorifer, seramik banyo
döşemesi, konutun sokakta bulunması, uydu sistemi, hidrofor, hazır mutfak, panjur, güneş
enerjisi, güney konumlu konut, şehir merkezine uzaklık 1500–2000 metre, banyo sayısı, konutun
metrekaresi, asansör sayısı olarak bulunmuştur.
Anahtar Kelimeler: Hedonik teori, konut piyasaları, hedonik fiyatlandırma modeli.
1. INTRODUCTION
Housing is a ―unique product‖ with three peculiarities (Harsman and
Quigley, 1991:2-3): (1) Complexity: Housing, as a kind of complicated goods,
can meet a great variety of a family‘s demands and be closely related to the
residents‘ activities such as life, work, amusement, etc.; (2) Fixity: Housing is
directly related to urban land in special location. The movement of housing is
basically impossible under the present technological conditions. This means that
the choice of housing involves consideration of neighborhood relations,
distance to the job site and corresponding public service facilities such as
schools, shopping centers, etc.; (3) Durability: This characteristic affects the
new housing market and stock housing market as well. Different from other
common commodity markets, housing market has a corresponding stock
market. Consumers can carry on replacement among new or old houses, choose
building type, community environment, degree of accessibility, and so on, to
meet individual preferences and get the greatest utility. These characteristics
indicate that influential factors of housing price are very complicated and
closely related to housing characteristics. Therefore, investigating the influence
factors of housing price inside the city from the viewpoint of housing
characteristics is a rational approach. In fact, since housing is a kind of
heterogeneous product, and there are obvious differences between housing
characteristics, scholars often establish hedonic price model to carry on
researches.
2. THE ECONOMIC THEORY OF HEDONIC PRICE MODEL
The term ―hedonic‖ is derived from Latin ―hedonikos‖, meaning
satisfaction. To this end, this concept is used in economics to imply for
enjoyment, satisfaction, pleasure or utility achieved with consumption of goods
or services (Kaul, 2006:4-5).
The term hedonic was first used in correcting price indices for quality
(Cowling and Cubbin, 1972:963). This term was used in an economic sense to
indicate that the index was computed taking into consideration not just the
objective aspects but also the qualitative utility obtained from a product (Kaul,
2006:4-5).
Hedonic Price Model depends on the consumer theory of the classical
economics, implying that each of the characteristics of heterogeneous goods
provides a different level of satisfaction or utility for the consumer. The model
suggests that the characteristics of a good meet different needs of consumers,
and the satisfaction or utility level of the consumers differs with consumption of
each characteristic. That is why this kind of models carries the term ―hedonic‖
in their names with the meaning of enjoyment, satisfaction, pleasure or utility
obtained with consumption of goods and services.
The Hedonic Price Model was first introduced in 1939 by A.T. Court,
an expert of the American automobile industry (Bartik, 1987:81, Goodman,
1998:291). Court regarded automobile price as a function of the automobile‘s
different characteristics, and carried out hedonic price analysis of heterogeneous
goods. His ultimate objective was to structure the price index for the automobile
industry. After that, this method began to expand to other consumer goods, such
as tractors, washing machines, etc. Colwell and Dilmore believed that Haas was
one of the first users of the hedonic price model (the first hedonic model on
agriculture) (Colwell and Dilmore, 1999:620). Haas (1922) used the concept of
―hedonics‖, and set up a simple hedonic price model for farmland, taking the
distance to the city center and the city size as two important characteristic
variables. On the other hand, Ridker (1967) was one of the earliest scholars to
apply hedonic price theory to analyze the housing market. He calculated the
impact of improving environmental quality (such as the elimination of air
pollution) on housing price.
The theoretical foundation of the hedonic price model is generally
regarded as hedonic price theory. American researcher Lancaster (1966) first
came up with a new consumer theory. The theory was expanded from the
consumer theory of classical economics, also known as Lancaster preference
theory. From the product heterogeneity, Lancaster (1966) analyzed the basic
elements that formed the product, and argued that the demand for the product
was not based on the product itself, but on its characteristics. Heterogeneous
goods (especially such as housing) have a number of incorporated
characteristics, and the goods are sold as the gathering of inherent
characteristics. All of these characteristics are variables of the utility function of
the consumer. Therefore, the utility level depends on the quantity of different
characteristics. It is difficult to analyze a market of such goods with the
traditional economic model because it cannot be considered by a single total
price. Consequently, it is necessary to adopt a series of prices (hedonic price) to
express corresponding product characteristics. As a result, the price of the
product is made up of hedonic prices, with each product characteristic having its
own implied price.
On the other hand, American economist Rosen (1976) submitted, within
the context of Lancaster preference theory, the first equilibrium model of
market supply and demand based on product characteristics. Under the
condition of perfect competition market, with maximizing consumer‘s utility
and producer‘s profit as the goal, Rosen (1976) analyzed theoretically the
long-term and short-long-term equilibrium of the heterogeneous goods market.
The model identifies the goods
(Z
)
as the total of their n
characteristics
(
Z
i)
.
i
contains
n
characteristics and indicates the quantity of
each characteristic. In this context, Rosen‘s model can be presented as follows
(Rosen, 1976:37):
)
(
Z
if
Z
(
i
1
,...,
n
)
(1)
The goods are described by numerical values of Z and provide the
consumers with different packages of characteristics. Moreover, existence of
product differentiation enabled by the presence of diverse characteristics
implies that a wide variety of alternative packages are available. Accordingly,
the demand function can be described with respect to price and characteristics
as follows:
)
,...,
,
(
)
(
z
p
z
1z
2z
nP
(2)
This function reveals the hedonic price regression obtained from
comparing prices of brands with different characteristics. In other words, it
gives the minimum price of any combination of characteristics. If two brands
offer the same combination but with different prices, consumers chooses the
cheaper one, and the identity of sellers does not have any affect on their
demand. In this connection, taking the partial derivatives of Equation 2, the
corresponding effect of each characteristic on the price (hedonic price) can be
expressed as follows:
İ ZiZ
P
P
(3)
Lancaster and Rosen‘s approaches try to estimate the combinations of
characteristics –measured objectively and affect the utility– that are comprised
of a number of attributes that the consumer appraises. However, these models
have some basic differences. Lancaster‘s model assumes that the goods are
members of a group, and the goods in a group consist of combinations of
characteristics in accordance with the budget constraint. On the other hand,
Rosen‘s model suggests that goods are in preferential order but consumers are
indifferent for the characteristics while buying a combination of goods.
Moreover, each good is chosen from a bundle of brands and consumed in
certain periods of time. Therefore, Lancaster‘s approach is suitable for all
consumption goods while Rosen‘s model is appropriate for only durable
consumption goods.
Unlike Lancaster, Rosen points out a nonlinear relation between the
price and the inner characteristics of goods. Nonlinearity of the price function in
this model implies that the implicit prices are inconstant.
Rosen‘s model includes two different stages. The first stage determines
the characteristics that affect the price of the good and estimates the marginal
prices for them. This stage develops a price measure but does not directly
provide an inverse
demand function. It only reveals the marginal contributions
of the characteristics to the price. The second stage defines the inverse
demand
function or estimates the marginal demand function through the implicit price
function determined in the first stage.
According to Rosen, the consumer immediately adjusts the budget
constraint for an increase in his/her income, and his/her marginal demand for a
characteristic may change. Rosen assumes that the price the consumer is willing
to pay for a good –or a combination of characteristics– is a function of the
variables that affect consumers‘ pleasure and preferences such as the
consumer‘s utility level, income, age, and education.
Rosen argues that the inverse demand function can be estimated in the
second stage by simultaneous equations using the marginal price as endogenous
variable, and that the inverse demand function is based on the implicit marginal
cost function. However, this identification of the inverse demand function may
be problematic. If the supply of the good has perfect elasticity or the supply of
characteristics is fixed, the marginal price of the characteristics will be
exogenous in the estimation of the inverse demand function. Therefore, Bartik
(1987) opposes Rosen‘s approach to the hedonic price model and argues that
the problem in hedonic estimation is not a result of the interaction between
supply and demand since an individual consumer cannot affect the sellers.
Under a nonlinear budget constraint, the endogeneity of all marginal prices and
the quantity of the characteristics result in hedonic estimation problem. For that
reason, there is no need to include the supply side of the market in the model. In
this regard, the low elasticity of supply of housing in Mugla –as a result of, inter
alia, the scarcity and therefore the high prices of lands for housing– and the
large quantity of the characteristics of housing require that marginal prices of
the characteristics be considered as endogenous. Accordingly, our study focuses
on the first stage of Rosen‘s model and aims to identify the marginal effects of
various characteristics on the housing price in the housing market of Mugla.
3. LITERATURE ON HEDONIC PRICE MODELS IN HOUSING
MARKET
Application of the hedonic price theory to the housing market was, as
mentioned above, first introduced by Ridker and Henning (1967), who analyzed
the effect of air pollution on housing prices. Following this study, a number of
empirical studies appeared in the hedonic price literature regarding the housing
market, a brief list of which may include Kain and Quigley (1970), Straszheim
(1973, 1974), Goodman (1978), Witte, Sumka and Erekson (1979), Palmquist
(1984), Mendelsohn (1984), Blackley, Follain and Lee (1986), Goodman
(1988), Meese and Wallace (1991), Kim (1992), Macedo (1996), Can and
Megbolugbe (1997), Meese and Wallace (1997), Powe, Garrod, Brunsdan and
Willis (1997), Yang (2000), Leishman (2001), Ucdogruk (2001), Bover and
Velilla (2002), Ogwang and Wang (2002), Wilhemsson (2002), Toda and
Nozdrina (2002), Maurer, Pitzer and Sebastian (2004), Wen, Lu and Lin (2004),
Filho and Bin (2005), Cohen and Coughlin (2005), Yankaya and Celik (2005),
Hai-Zhen, Sheng-Hua and Xiao-Yu (2005), Li, Prud‘Homme and Yu (2006).
The variables and functional forms they used and their findings are
chronologically presented in the appendix.
4. HEDONIC PRICE MODELS FOR THE HOUSING MARKET OF
MUGLA
Data for this study has been gathered for the reference period –May
2007– through a questionnaire of 33 questions on housing prices and 32
characteristics which are considered to have had an effect on them. 178
observations have been obtained through a questionnaire based on face to face
interviews with randomly selected real estate agencies from the urban districts
of Mugla.
These observations consists 100 percent of the population.
This data
has been analyzed through the hedonic price approach. The variables have been
selected with reference to the literature.
Real estate agencies were asked questions regarding the number of
balconies, the number of elevators, the number of houses in the apartment, the
size of the house, the number of rooms, the floor level, the age of the house
(continuous variable), the heating system, the flooring of the living room and
other rooms, the flooring of the bathroom, the material of windows‘ frames,
roof isolation, wall covering, location, structure of the kitchen, satellite TV,
hydrophor pump, parking lot, sunblind, solar water heating, doorman, whether it
is in-garden and in-site, distance to the city centre, direction of the front side,
ground survey, and occupancy (proxy variable).
There are three most frequently used function forms in hedonic price
model: linear, logarithmic, and log-linear. This study utilizes all the three forms
and interprets the common significant variables. The variables have been
analyzed under SPSS 10.0 statistical program, their frequencies have been
determined, and the variables found to be problematic have been excluded from
the analysis.
Table 1 presents the means and standard deviations of the variables.
Average housing price in the City Center of Mugla is 119.240 YTL (New
Turkish Lira). Average number of bathrooms is 1 while that of balconies is 2.
On the other hand, the average number of rooms in the houses questioned is
3.54. The average number of dwellings in an apartment building where a house
in question exists is 11.5 while their average age is approximately 11. In
addition, Table 1 indicates that the houses in the city center of Mugla are on
average 119 meter square and on the 2
ndor 3
rdfloor. 60 percent of the houses
are located on streets while 48 percent of them are located on corners, and 53
percent of them are with front side to south. 53 percent of the houses are as far
as 500-1000 meters to the city center. 39 percent of the houses are in buildings
in which owners of the houses live. Furthermore, 93 percent of the houses are
with clay-tile roofs, 54 percent are with central heating system (furnace) while
34 percent of them are heated with stoves.
Table 1: The Mean and Standard Deviation of Housing Prices and the
Variables that are considered to Have an effect on the Housing Prices
Variable
Mean
Std.
Dev.
Variable
Mean
Std.
Dev.
Housing Price
119.24
33.427
Age of House
10.94
7.457
H
ea
ti
n
g
Stove
.34
.474
D
ist
an
ce
t
o
C
it
y
C
en
tr
e
500–1000m
.53
.501
Central (Floor)
.12
.323
1000-1500m
.23
.422
Central
(Apartment)
.54
.499
1500-2000m
.16
.365
Other
.00
.000
>2000m
.08
.279
Li
v
in
g
R
o
o
m
F
lo
o
r
Stone Tile
.07
.251
O
cc
u
p
an
cy
Vacant
.29
.453
Prefinished
Hardwood
.57
.497
Tenant
.33
.470
Unfinished
Hardwood
.27
.445
Owner
.39
.490
Ceramic
.04
.195
Laminate
.02
.129
F
ro
n
t
si
d
e
to
North
.17
.380
Carpet
.01
.106
South
.53
.501
Other
.03
.181
East
.24
.429
R
o
o
m F
lo
o
r
Stone Tile
.05
.220
West
.22
.415
Prefinished
Hardwood
.54
.500
Th
e
N
u
m
b
er
o
f
Dwellings (in
apt.)
11.51
8.009
Unfinished
Hardwood
.31
.463
Rooms
3.54
.648
Ceramic
.04
.195
Bathrooms
1.07
.251
Laminate
.02
.129
Balconies
1.93
.737
Carpet
.01
.106
Elevators
.28
.451
Other
.04
.195
Lo
ca
ti
o
n
Street
.60
.491
B
at
h
ro
o
m
F
lo
o
r
Stone Tile
.02
.129
Main Street
.39
.490
Glazed tile
.44
.498
Avenue
.03
.181
Ceramic
.54
.499
On the Corner
.48
.501
Other
.01
.075
Ground Survey
.53
.501
W
in
d
o
w
F
ra
mes
Wooden
.11
.317
Sunblind
.08
.270
Aluminum
.22
.419
Solar Water Heating
.28
.451
PVC
.66
.476
Located in-Site
.33
.470
Other
.01
.075
Garden
.61
.490
R
o
o
f
Betony
.06
.241
Ventilation
.70
.459
Clay Tile
.93
.251
Meter Square
119.97
24.772
Profiled Sheeting
.02
.129
Floor Level
2.63
1.339
W
al
l
Plastic Paint
.56
.498
Fire Exit
.20
.399
Oil Paint
.07
.251
Satellite TV
.27
.445
Satin Paint
.37
.483
Doorman
.37
.483
Wallpaper
.00
.000
Hydrophor Pump
.62
.486
This study estimates the hedonic price model through the E-Views
program. To find out the most suitable model, the study utilizes the most
commonly used top-to-down or general-to-specific approaches (Gujarati 1999)
of Hendry (1985) or the so-called London School of Economics (LSE)
Approach. The initial model covered all of the variables, and then the most
insignificant ones were removed from the model in order until a significance
level of
0
.
20
was achieved, with the following variables: heating by stove;
living room floor–stone tile; bathroom floor–stone tile; wooden window frames;
betony roof; walls–oil paint; house on the corner; house vacant; house with
front side to north; distance to city center-500–1000m. The variables that were
found to be significant and the related models are presented in Table 2. The
existence of heteroskedasticity in the models was tested via
Breusch-Pagan-Godfrey test and rejected at
0
.
05
.
Table 2: Results for the Linear, Logarithmic, and Log-linear Models
Variables Linear Model Log-linear Model Log-Log Model
Coefficient Significance Coefficient Significance Coefficient Significance Central Heating
(Apartment) 12.81457 0.0001 0.117066 0.0000 0.124027 0.0000 Living Room Floor:
unfinished hardwood 5.819712 0.1690 - - - -
Living Room Floor:
Ceramic - - -0.102628 0.0471 -0.093858 0.0638
Bathroom Floor Ceramic 5.101686 0.0498 0.051819 0.0133 0.042196 0.0439 On the Street 4.860699 0.0676 0.035469 0.1235 0.035097 0.1178 Satellite TV 6.782905 0.0538 0.070542 0.0133 0.048301 0.0790 Hidrophor Pump 5.275722 0.0741 0.053024 0.0239 0.054837 0.0184 Parking Lot 3.732982 0.2059 - - 0.029643 0.2226 Modular Kitchen 3.550859 0.1859 0.039558 0.0653 0.029484 0.1641 Sunblind 12.72444 0.0096 0.093856 0.0130 0.098895 0.0074
Solar Water Heating 4.272688 0.1716 0.032140 0.1912 0.039805 0.1019
In-Site -11.02580 0.0003 -0.080453 0.0007 -0.078754 0.0010
Garden - - -0.030564 0.1836 -0.037945 0.1168
Fire Exit - - -0.045414 0.1327 - -
Occupied by Tenant -5.529069 0.0951 - - - -
Occupied by Owner -5.081245 0.1115 - - - -
Front Side to South 5.327686 0.0500 0.044006 0.0422 0.038325 0.0724 Front Side to West -5.490518 0.0830 -0.044016 0.0823 -0.042196 0.0871
On the Corner -3.326854 0.1979 - - - -
Distance to City Centre:
1500-2000m 7.787394 0.0269 0.065401 0.0231 0.066403 0.0175
The Number of Bathrooms 14.86537 0.0054 0.063349 0.1352 0.082357 0.0462
The Number of Rooms -4.772347 0.0615 - - -0.106115 0.1115
Meter Square 0.721088 0.0000 0.005275 0.0000 0.691436 0.0000 The Number of Elevators 13.83910 0.0001 0.104916 0.0002 0.097921 0.0003 The Number of Dwellings
in the Building 0.234487 0.1363 0.002469 0.0482 - -
The Age of the Building - - -0.002554 0.1069 -0.016317 0.1736 F Value 28.86337 0.000000 34.92061 0.000000 36.63798 0.000000 R2/Adjusted R2 0.811703 0.783581 0.816463 0.793082 0.823548 0.801070
Ø 0.424913 6.270495 4.94169
25. INTERPRETATION AND SIGNIFICANCE OF THE COEFFICIENTS
Following are the findings from the hedonic model estimated for Mugla
Center:
Central heating in the apartment, rather than a stove, increases the
hedonic price of the house 12.8 unit in the linear model, 11% in the
log-linear model, and 0.12% in the log-log model. Moreover, the coefficient
of the central heating is positive and significant at 1% in all the three
models.
Ceramic tile floor in the bathroom, rather than stone tile, increases the
hedonic price of the house 5.1 unit in the linear model, 5% in the
log-linear model, and 0.04% in the log-log model. The coefficient of the
ceramic tile floor in the bathroom is positive and significant at 5% in all
the three models.
Location on the street increases the hedonic price of the house 4.8 unit
in the linear model, 3% in the linear model, and 0.03% in the
log-log model. The coefficient of location on the street is positive in all the
three models but significant only in the linear model at 10%. In the
other models, this variable does not have any affect on the hedonic
price.
Satellite TV increases the hedonic price of the house 6.78 unit in the
linear model, 7% in the log-linear model, and 0.04% in the log-log
model. The coefficient of satellite TV is positive in all the three models
and significant at 5% in the linear model and at 10% in the other
models.
Hydrophor pump increases the hedonic price of the house 5.27 unit in
the linear model, 5% in the log-linear model, and 0.05% in the log-log
model. The coefficient of hydrophor pump is positive in all the three
models and significant at 10% in the linear model and at 5% in the
other models.
Sunblind increases the hedonic price of the house 12.72 unit in the
linear model, 9% in the log-linear model, and 0.09% in the log-log
model. The coefficient of sunblind is positive in all the three models
and significant at 5% in the log-log model and at 1% in the other
models.
The coefficient of location in a site is negative and significant at 1% in
all the three models. This variable decreases the hedonic price of the
house 11 unit in the linear model, 8% in the log-linear model, and
0.07% in the log-log model.
Front side to south, rather than to north, increases the hedonic price of
the house 5.32 unit in the linear model, 4% in the log-linear model, and
0.03% in the log-log model. The coefficient of this variable is positive
in all the three models and significant at 10% in the linear and log-log
models and at 5% in the log-linear model.
Front side to west, rather than to north, increases the hedonic price of
the house 5.49 unit in the linear model, 4% in the log-linear model, and
0.042% in the log-log model. The coefficient of this variable is negative
and significant at 10% in all the three models.
A distance of 12000m to the city center, compared to that of
500-1000m, increases the hedonic price of the house 7.78 unit in the linear
model, 6% in the log-linear model, and 0.06% in the log-log model.
The coefficient of this variable is positive and significant at 5% in all
the three models.
One-unit increase in the number of bathrooms increases the hedonic
price of the house 14.86 unit in the linear model, and 6% in the
log-linear model. On the other hand, 1% increase in the number of
bathrooms increases the hedonic price 0.08% in the log-log model. The
coefficient of this variable is positive in all the three models and
significant at 1% in the linear model and 5% in the log-log model,
while it is insignificant in the log-linear model.
One-unit increase in the meter square of the house increases its hedonic
price 0.72 unit in the linear model, and 0.5% in the log-linear model.
On the other hand, 1% increase in the meter square of the house
increases its hedonic price 0.06% in the log-log model. The coefficient
of this variable is positive and significant at 1% in all the three models.
One-unit increase in the number of elevators in the building increases
the hedonic price of the house 13.83 unit in the linear model, and 10%
in the log-linear model. On the other hand, 1% increase in the meter
square of the house increases its hedonic price 0.09% in the log-log
model. The coefficient of this variable is positive and significant at 1%
in all the three models.
F values regarding the findings above are significant at
0
.
01
level.
In addition,
R
2values indicate that these variables explain approximately 81%
of the change in the hedonic price of the house.
6. CONCLUSION
The findings revealed by the econometric model in this study on
estimating hedonic price parameters in the real estate market in Mugla province
have met the theoretical and economic expectations. In other words, considering
the structure and features of Mugla, the estimated coefficients are found to be
significant in terms of both the characteristics of housing and its location (its
position, whether it is in-site or not, etc.).
As expected, the variables that positively affect the housing price have
been found in all linear, logarithmic, and log-linear models to be central
heating, ceramic bathroom floor, location on the street, satellite TV, hydrophor
pump, modular kitchen, sunblind, solar water heating, front side facing south,
1500-2000 meters to the city center, the number of bathrooms, square meter of
housing, and elevator.
The affect of particularly two variables on housing prices needs to be
interpreted taking into account the structure and characteristics of Mugla.
In-Site Location of Housing: The analysis has revealed by all the three
models that in-site location of housing negatively affect the price of housing,
implying that in-site location decreases the price of housing. The reason
underlying this argument may be that housing in Mugla is mostly in cooperative
type, and their prices are relatively lower, which is attributable to the common
opinion that the material and workmanship used in this kind of housing is of
low quality.
Distance to the City Center (1500–2000m): Our findings suggest that
the price of houses is higher if distance to the city center is 1500–2000 meters.
This is mainly because the closer areas, up to an approximate distance of 2 km,
are not allowed for construction in Mugla. That is why such a distance is
considered close to the city center, and in this manner housing tends to become
more expensive as it gets closer to such a distance.
This study has the merit of identifying the factors that may affect the
housing prices in Mugla at present and in the future by means of hedonic
models, providing a data set on the real estate market for both buyers and
sellers.
7. REFERENCES
Bartik, T. J. (1987). ―The Estimation of Demand Parameters in Hedonic Price
Models‖, The Journal Of Political Economy, 95(1): 81-88.
Blackley, D.M., J.R. Follain and H.Lee. (1986). ―An Evaluation of Hedonic
Price Indexes for Thirty Four Large SMSAs‖, AREUEA Journal, 14(2):
179-205.
Bover, O. and P.Velilla. (2002). ―Hedonic House Prices Without
Characteristics: The Case of New Multiunit Housing‖, European
Central Bank Working Paper Series, Working Paper No.117.
Can, A. and I. Megbolugbe. (1997). ―Spatial Dependence and House Price
Index Construction‖, Journal of Real Estate Finance and Economics,
14: 203–222.
Cohen, J. P. and C. C. Coughlin. (2006). ―Airport-Related Noise, Proximity,
and Housing Prices in Atlanta‖, Federal Reserve Bank of St. Louis
Colwell, P.F. and G. Dilmore. (1999). ―Who Was First? An Examination of an
Early Hedonic Study‖, Land Economics, 75(4): 620-626.
Court, A. T. (1939). Hedonic Price Indexes with Automotive Examples in
Dynamic of Automobile Demand, General Motors, New York.
Cowling, K. and J. Cubbin. (1972). ―Hedonic Price Indexes for United Kingdom
Cars‖, Economic Journal, 82 (September): 963-978.
Filho, C. M. and O. Bin. (2005). ―Estimation of Hedonic Price Functions via
Additive Nonparametric Regression‖, Empirical Economics, 30: 93–
114.
Goodman, A. C. (1978). ―Hedonic Prices, Price Indices, and Housing Markets‖,
Journal of Urban Economics, 5(4): 471-84.
Goodman, A. C. (1988). ―An Econometric Model of Housing Price, Permanent
Income, Tenure Choice, and Housing Demand‖, Journal of Urban
Economics, 23: 327-353.
Goodman, A. C. (1998). ―Andrew Court and the Invention of Hedonic Price
Analyzis‖, Journal of Urban Economics, 44(2): 291-298.
Gujarati, D.N. (1999). Temel Ekonometri, Çev:Ü.Şenesen ve G:G:Şenesen,
Literatür Yayıncılık, İstanbul.
Harsman, B. and J.M. Quigley. (1991). Housing Markets and Housing
Institutions: An International Comparison,. Kluwer Academic Press,
Boston.
Kain, J. F. and J. M. Quigley. (1970). ―Measuring the Value of Housing
Quality‖, Journal Of The American Statistical Association, 65 (330):
532-548.
Kaul, S. (2006). Hedonism and Culture: Impact on Shopper Behaviour, Indian
Institute of Management Publication.
Kim, S. (1992). ―Search, Hedonic Prices and Housing Demand‖,
The Review of
Economics and Statistics, 74(3) : 503-508.
Lancaster, K. J. (1966). ―A New Approach to Consumer Theory‖, Journal of
Political Economy, 74: 132-57.
Leishman, C. (2001). ―House Building and Product Differentiation: An Hedonic
Price Approach‖, Journal of Housing and the Built Environment, 16:
131–152.
Li, W., Prud‘Homme M. and Yu K. (2006). Studies in Hedonic Resale Housing
Price Indexe, OECD-IMF Workshop.
Macedo, R. (1996). ―Hedonic Price Models with Spatial Effects: An
Application to the Housing Market of Belo Horizonte, Brazil‖, Revista
Maurer, R., M. Pitzer and S. Sebastian. (2004). ―Hedonic Price Indices for the
Paris Housing Market‖, Allgemeines Statistisches Archiv, 88: 303-326.
Meese, R. and N. Wallace. (1991). ―Nonparametric Estimation of Dynamic
Hedonic Price Models and the Constructions of Residential Housing
Price Indices‖, AREUEA Journal, 19(3):308-332.
Meese, R. and N. Wallace. (1997). "The Construction of Residential Housing
Price Indices: A Comparison of Repeat-Sales, Hedonic-Regression, and
Hybrid Approaches", Journal of Real Estate Finance and Economics,
14: 51–73.
Mendelsohn, R. (1984). ―Estimating the Structural Equations of Implicit
Markets and Household Production Functions‖, The Review of
Economics and Statistics, 66(4): 673-677.
Ogwang, T. and B. Wang. (2002). ―A Hedonic Price Function for a Northern
BC Community‖, Social Indicators Research, 61: 285-296.
Palmquist, R. B. (1984). ―Estimating the Demand for the Characteristics of
Housing‖, Review of Economics and Statistics, 66(3): 394-404.
Powe, N.A., Garrod G.D., Brunsdon C.F. and K.G. Willis. (1997). ―Using a
Geographic Information System to Estimate a Hedonic Price Model of
the Benefits of Woodland Access‖, Forestry, 70(21): 39-149.
Ramanathan, R. (1998). Introductory Econometrics with Applications, The
Dryden Press, Fourth Edition.
Ridker, R.G. and J.A. Henning. (1967). ―The Determinants of Residential
Property Values with Special Reference to Air Pollution‖ The Review of
Economics and Statistics, 49: 246–257.
Rosen, S. (1974). ―Hedonic Prices and Implicit Markets: Product
Differentiation in Pure Competition‖, Journal of Political Economy, 82:
34-55.
Straszheim, M.R. (1973). ―Estimation Of The Demand For Urban Housing
Services From Household Interview Data‖, Review of Economics and
Statistics, 55: 1-8.
Straszheim, M.R. (1974). ―Hedonic Estimation of Housing Market Prices: A
Further Comment‖, The Review of Economics and Statistics, 56(3):
404-406.
Toda, Y. and N. N. Nozdrina. (2004). ―The Spatial Distribution of the
Apartment Prices In Moscow in 2002: Hedonic Estimation from Micro
Data‖ ENHR Conference July 2nd-6th 2004, Cambridge.
Üçdoğruk, Ş. (2001). ―İzmir İlinde Emlak Fiyatlarına Etki Eden Faktörler:
Hedonik Yaklaşım‖, D.E.Ü. İ.İ.B.F. Dergisi, 16(2): 149-161.
Wallace, H.A. (1926). ―Comparative Farmland Values in Iowa‖, Journal of
Wen H., J. Sheng-Hua and G. Xiao-Yu. (2005). ―Hedonic Price Analysis of
Urban Housing: An Empirical Research on Hangzhou, China‖, Journal
of Zhejiang University Science, 6A (8):907-914.
Wen, H.Z., J.F. Lu, and L. Lin. (2004). An Improved Method of Real Estate
Evaluation Based on Hedonic Price Model, International Management
Conference.
Wilhelmsson, M. (2002). ―Household Expenditure Patterns for Housing
Attributes: A Linear Expenditure System with Hedonic Prices‖, Journal
of Housing Economics, 11: 75–93.
Witte, A. D., H. Sumka and J. Erekson. (1979). ―An Estimate of a Structural
Hedonic Price Model of the Housing Market: An Application of
Rosen's Theory of Implicit Markets‖, Econometrica, 47: 1151-72.
Yang, Z. (2000). ―An Application of the Hedonic Price Model with Uncertain
Attribute: The Case of the People's Republic of China‖, Property
Management, 19(1): 50-63.
Yankaya, U. ve M. Çelik. (20059. ―İzmir Metrosunun Konut Fiyatları
Üzerindeki Etkilerinin Hedonik Fiyat Yöntemi İle Modellenmesi‖,
Appendix: Examples of Hedonic Pricing Models in House Markets
Study,
Data and
Functional
Form
Variables
Conclusions and Evaluation
Ridker & Henning (1967) , 167 Observation About House Selling, Linear Functional Form
Dependent Variable: Median value
of owner-occupied single family housing units
Independent Variables:
(1) An index of annual geometric mean sulfation levels, (2) Median number of rooms per housing unit, (3) Percentage recently built, (4) Total houses per square mile of tracts, (5) Time zone for central business district, (6) Percentage non-white housing units, (7) School quality, (8) Occupation ratio, (9) Highway accessibility, (10) İllionis/ Missouri dummy variable, (11) Persons per unit, (12) Median family income, (13) Index of annual geometric mean concentrations of suspended particulates gathered by high-volume air samplers, (14) Percentage substandart, (15) Crime rate, (16) Shopping area accessibility, (17) Industrial area accessibility, (18) Social area analysis indexes.
—This study was one of the earliest to apply hedonic price theory for analyze the housing market and calculated the impact of improving environmental quality on housing price.
—The variables which causes multicollinearity problem also featured in the study and introduced the results if including these varibles or not. Thus, adjustments for multicollinearity choosing four different estimating method.
—The most important results are statistically significant and all are fairly reasonable within the context of the area. —Sulfation levels to which any single-family dwelling unit is exposed were to drop by 0.25 mg./100cm2/day, the value of that property could be expected to rise by at least $83 and more likely closer to $245.23.
—Characteristics specific to the property [variable (2),(3) and (4)] all turned out to be important explanatory variables. The sign and magnitudes of their coefficients are as expected.
— Both variables (5) and (9) are statistically significant. The coefficients attached to variable (5) , however, are not quite as expected.
—Variable (8) proved to be best estimated among neighbourhood characteristics. The coefficients of variable (7) are positive. Kain & Quigley (1970) , 1184 Observation In The Entire Model And 854 Observation For The Restricted Model About House Selling,,Semi-Logarithmic and Linear Functional Forms
Dependent Variable: Dwelling unit
price
Independent Variables:
(1) Basic residential quality, (2) Dwelling unit quality, (3) Quality of proximate properties, (4) Nonresidential usage, (5) Avarage structure quality, (6) Proportion white in census tract, (7) Median schooling of adults in census tract, (8) Public School achievement, (9) Number of major crimes, (10) Age of structure, (11) Number of rooms (natural log.), (12) Number of bathrooms, (13) Parcel area (hundreds of sq. ft.), (14) First flor area (hundreds of sq. ft.), (15) Single detached, (16) Duplex, (17) Row, (18) Apartment, (19) Rooming house, (20) Flat, (21) No heat included in rent, (22) No water included in rent, (23) No major appliances included in rent, (24) No furniture included in rent, (25) Hot water, (26) Central heat, (27) Duration of occupancy (years), (28) Owner in building
—This article estimates the market value, or the implicit prices of specific aspects of the bundles of residential services consumed by urban households. Quantitative estimates were obtained by regressing market price of owner-and renter-occupied dwelling units on measures of the qualitative and quantitative dimensions of the housing bundles.
—The measures of residential quality obtained by using factor analysis to aggregate some 39 indexes of the quality. —For renters equations used 25 variable and for owners equations used 15 variable in the study.
—The analysis indicates that the quality of the bundles of residential services has about as much effect on the price of housing as such objective aspects as the number of rooms, number of bathrooms, and lot size.
—For renters, among the first 5 quality variables, variable (1), (2) and (5) are statistically significiant in the model which have restristed observation. For owners, among the first 5 quality variables, variable (3) and (5) are not statistically significiant in the model which have restristed observation. For renters equations only 16 variable and for owners equations only 5 variable are statistically significiant at %5 significance level.
—The most striking difference when the model is reestimated for the entire obvervations is the increase in the significance of the coefficients.
Study,
Data and
Functional Form
Variables
Conclusions and Evaluation
Straszheim (1973), Household Interview Data For 1965 (100-200 Observation About House Selling), Linear Functional Form
Dependent Variable: Price of
standartized dwelling unit
Independent Variables:
(1) Probability of ownership, (2) Number of rooms in dwelling units, (3) Structure age dummies, (4) Lot size dummies, (5) Structure condition dummies, (6) Unsound condition dummies, (7) Sample size
—Separate equations were estimated for owner and rental units, and for each geographic submarket.
—It was found strong relationship between house price and variables (3), (4) and (7).
— Variables (3) and (4) were always statistically significiant.
—Analysis of covariance tests reveal satatistically significiant differences in the equations across zones.
—There is substantial spatial variation in the price of most attributes of housing services.
Straszheim (1974), Pooled Data Of The 3 Different District About House Characteristics, Linear Functional Form
Dependent Variable: House
selling price
Independent Variables:
(1) Number of rooms, (2) Built in pre1940, (3) Built in 1940–1945, (4) Built in 1950–1959, (5) Lot size less than 2 acre, (6) Lot size between 3–5 acre, (7) Lot size greater than 5 acre, (8) Unsound condition dummy , (9) Sample size
—F-tests reveal that the geographic stratification reduces the residual sum of squared errors. A few of the individual submarket equations are presented to illustrate the range of estimates obtained.
—Though suburban submarkets exhibit more price homogeneity, there are also limits to how wide a geographic area can be employed.
—The discussion of hedonic price estimation might more usefully be directed to the criteria which should be employed to define homogenous submarkets within urban areas. Goodman (1978), 1835 Observation About House Selling, Box-Cox Functional Form
Dependent Variable: House
selling price
Independent Variables:
(1) Lot size in sq. ft., (2) 1 if house is all brick; 0 otherwise, (3) 1 if hardwood floors; o otherwise, (4) Number of covered garage spaces, (5) Age of house in years, (6) Number of rooms excluding bathrooms, lavatories, (7) Number of full bathrooms, (8) Number of Lavatories, (9) Indoor living space in sq. Ft., (10) Number of fireplaces, (11) Percentage balack population, (12) Percentage families with income less than 5000 $, (13) Percentage of population over age 25 with 13 or more years of education, (14) 1 if black is greater than %5 and less than %15, (15) Principle components measure of neighbourhood attitudes
— This study appears to clarify several aspects of housing analysis using hedonic prices, with respect to market segmentation, functional form and behavior of prices within submarkets. In positing various spatially and temporally separate submarkets, covariance analysis indicates heterogeneity of coefficients.
-—Model results showed that, Variables affect the house prices differently in urban and suburb areas and for both structure and neighbourhood characteristics the price are up to 20% higher than the suburbs.
— Intrametropolitan examination of structural and neighborhood quality reveals that the relative valuation of physical improvements in housing is smaller in the central city than in the suburbs, while the relative valuation of improved neighborhoods is relatively constant.
-— Aggregation of hedonic price coefficients into standardized units yields significantly higher housing prices in the central city than in its suburbs, as well as differential effects of structural and neighborhood improvements among submarkets.
Study,
Data and
Functional
Form
Variables
Conclusions and
Evaluation
Witte & Sumka & Erekson (1979), 500 Observation About House Rents In Four City, Linear Functional FormDependent Variable: House rent
Independent Variable (For Consumers): (1) Total annual current
income of household, (2) Age in years of the head of household, (3) 1 If the head of household is employed in blue collar job; 0 otherwise, (4) 1 If the head of household is employed in white collar job; 0 otherwise, (5) 1 if the highest level of education attained by head of household is high school; 0 otherwise, (6) 1 if the head of household has education beyond the high school level; 0 otherwise, (7) The number of persons in the household, (8) 1 if the household head is female and 0 if male.
Independent Variable (For Supplier): (1) The number of years the landlord has owned the dwelling, (2) 1 if there is a lease; 0 otherwise, (3) 1 if the family has lived in the dwelling for five years or more; 0 otherwise, (4) 1 if the landlord is resident in the dwelling; 0 otherwise, (5) 1 if rent is collected more often than once a month; 0 otherwise, (6) 1 if a professional manager is employed; 0 otherwise, (7) The number of rental units owned, (8) 1 if ownership is of the corporate form; 0 otherwise, (9) 1 if the rental housing owned was acquired by inheritance; 0 otherwise, (10) 1 if the head of household is black; 0 otherwise.
Other Independent Variables: (1) Dwelling quality, (2) Dwelling
size, (3) Lot size, (4) Neighbourhood quality, (5) A measure of accessibility
—In this study, estimated a simultaneous system of hedonic price equations suggested by the work of Rosen. This system consisted of bid and offer curves for each of housing bundle attributes, dwelling quality, dwelling size, and lot size.
—Empirical results confirm the theoretically expected negative coefficient for each attribute in its own bid price equation and the expected positive or zero coefficient for each attribute in its own offer function.
—An examination of cross price relationships among the attributes revealed an intriguing and generally logical pattern of interactions both on the demand and the supply sides.
Palmquist (1984), .20297 Observation About House Selling, Linear, Semi-Logarithmic, Log-Linear and Inverse Semi-Logarithmic Functional Forms
Dependent Variable: House selling price
Independent Variables:(1) Area of lot in sq. ft., (2) Finished interior
area in sq. ft., (3) Finished interior area squared, (4) Number of bathrooms, (5) Year of construction, (6) Number of stalls in garage, (7) Number of stalls in carport, (8) 1 if garage is detached from house, (9) 1 if there is underground wiring, (10) 1 if there is a dishwasher, (11) 1 if there is a garbage disposal, (12) 1 if there is central air conditioning, (13) 1 if there is wall air conditioning units, (14) 1 if there is a ceiling fan, (15) 1 if the date of sale was 1976, (16) Excellent condition, (17) Fair condition, (18) Poor condition, (19) Brick or stone exterior finish, (20) 1 if there is a full basement, (21) 1 if there is a partial besement, (22) 1 if there are one or more fireplaces, (23) 1 if there is a swimming pool, (24) The annual arithmetic mean of the particulate air pollution level, (25) The median age of the residents of the census tract, (26) The median family income of residents of the census tract, (27) The percentage of workers in the census tract that has a blue collar job, (28) The percentage of houses in the census tract that has changed ownership within the last five years, (29) The percentage of the population of the census tract that is classified as non-white, (30) The percentage of the population of the census tract over 24 years old that has graduated from high school, (31) The percentage of the structures in the census tract with 1.00 or less persons per room, (32) The number of work destinations within the census tract divided by the area of the census tract, (33) Adjusted monthly housing expenditure, (34) Hedonic price of sq. ft. of living space, (35) Hedonic price of bathrooms, (36) Hedonic price of the percentage of the census tract with high school degrees, (37) Hedonic price of racial homogeneity, (38) Hedonic price of lot area, (39) Hedonic price of reduction in age of house, (40) Age of the purchaser, (41) 1 if the purchaser is single, (42) Number of dependents in the family making the purchase, (43) 1 if the purchaser is black.
—First, linear hedonic regression equations were constructed and in the second srtage estimated logarithmic linear estimates for house characteristics in the study. To reduce the costs of estimation, the search was restricted to the four functional forms most frequently used: linear, semi-logarithmic, log-linear and ınverse semi-logarithmic.
—With approximately 200 coefficients estimated, there are only 17 with incorrect signs and none of these are for the most important variables. Hedonic regression results showed that variables (3), (8), (18), (24), (28) and (29) affects house prices negatively. First 32 variables except variables (3), (8), (18), (24), (28) and (29) affects house prices positively and the variables which were positively affects house prices have expected signs and magnitudes, also they were statistically significiant. In the second stage, variables (33), (34) (42) and (43) were more effective on the house prices and these variables which were statistically significiant have positive coefficients.
Study,
Data and
Functional
Form
Variables
Conclusions and Evaluation
Goodman (1988), 2857 Observation About House Selling, Box-Cox Functional Form
Dependent Variable: House selling price Independent Variables: (1) 1 if central air, o
otherwise, (2) 1 if heating breakdowns in past 90 days, 0 otherwise, (3) Number of full bathrooms, (4) Number of bedrooms, (5) Number of utility breakdowns in past 90 days, (6) Age of house in year, (7) 1 if full cellar, 0 otherwise, (8) 1 if electricity used for cooking, 0 otherwise, (9) 1 if open holes or cracks, 0 otherwise, (10) 1 if additional heating equipment used, 0 otherwise , (11) 1 if Steam heat, 0 otherwise, (12) 1 if gas heat, 0 otherwise, (13) Rating neighborhood, 1 (best)…,4 (worst), (14) 1 if fuses blown in past 90 days, 0 otherwise, (15) Number of lavatories, (16) Years residing in dwelling unit, (17) Number of rooms without hot air ducts, (18) 1 if plaster broken over 1 foot 2 , 0 otherwise, (19) 1
if Access to other rooms through bedroom, 0 otherwise, (20) 1 if sign of rats ın past 90 days, 0 otherwise, (21) Number of rooms, (22) Logarithm of property tax rate, (23) Garage/carport available for use, (24) Location dummies, (25) City dummies, (26) 1 if passenger elevator in building, 0 otherwise, (27) Number of extra features included in rent, (27) Number of stories in building, (28) 1 if heat included in rent, 0 otherwise, (29) 1 if Single family Structure, 0 otherwise.
—Hedonic price methods define price indices for owner and renter housing and define value-rent ratios for the investment components of the housing purchase. Permanent income is estimated for both owners and renters. Tenure choice is estimated using the price, value-rent ratios, permanent and transitory incomes, and sociodemographic variables. Housing demand is estimated for both owners and renters.
—The adjusted R2 is 0.6025 for the value
regressions, as opposed to 0.4585 for the rent regressions. Abathroom adds 26 % to the house value and 28.5 % to the apartment rent. An additional room adds 7.3 % to the value and 6.0 % to the rent. An owner (renter) unit loses About 0.53 % (0.28 %) of value (rent) per year.
— Neighborhood effects are considerably weaker for renter units. A unit improvement in the quality of neighborhood Structure leads to a 3.8 % rent increase; for owner housing the percentage increase is 7.5 %.
—There is significant regional variation in owner housing prices, there is less variation in quality-adjusted rents.
—Variables (2), (5), (13), (16), (17), (19) and (22) were affects house value and rents negatively.
Meese & Wallace (1991), Time Series Data For 2 Different City For The Years Between 1970–1988, Trans-Log and Log-Linear Functional Form
Dependent Variable: House selling price
Independent Variables:
(1) Number of bathrooms, (2) Sq. Ft. Of floor space (m2), (3) Number of total rooms, (4) Index
of house condition, (5) Federal mortgage, (6) Multiple sales dummy variable, (7) Mortgage assumability dummy, (8) Residential zoning dummy, (9) Swimming pool dummy, (10) Fireplace dummy, (11) Age of dwelling (years).
—In this paper advocating the use of nonparametric regression techniques to construct housing price indices.
—The analysis includes an examination of the variation in the implicit price of house attributes over time, diagnostic checks of the adequacy of the fitted hedonics, and simulated confidence intervals for the Fisher Ideal Price index.
—For Diedmont city variables (6) and (11) have positive and negative coefficients respectively. Variables (1), (2), (3), (4), (7), (9) and (10) have positive signs and have positive effects on the house selling prices. Moreover, variable (5) states less expensive houses and have negative signs. —For San Francisco city only variable (5), (7) and (8) have negative signs.
Study,
Functional
Form, Data
Variables
Conclusions and Evaluation
Can & Megbolugbe (1997), 944 Observation About House Selling, Linear Functional Form
Dependent Variable: House selling
price
(1) Living area in sq. ft., (2) Land area in sq. ft., (3) Age of the structure, (4) Composite neighborhood quality score
Neighborhood quality variables:
—Owner-occupancy rate —Median household income —Percentage of residents with college education
—Percentage of households paying at least 30 % of income on monthly housing costs —Median value of owner-occupied housing
—Vacancy rate
—Median age of housing stock —Percentage of detached signle-family units
—Percentage of white-headed, balack-headed and hispanic-balack-headed
—Study illustrates the importance of spatial dependence in both the specification and estimation of hedonic price models. In this article, presenting the importance of spatial dependence on the specification of a house price function due to the presence of spatial spillover effects in the operation of local housing markets. With the spatial models which constructed in the article, it would be possible to adjust the confidence intervals of the metropolitan –level indices to reflact the localized dependencies in the house price determination process. —Models also achieve very reasonable estimates of marginal prices for selected attributes. Spatial dependence plays an important role in the house price determination process.
— Variable (2) is not statistically significant in both 6 regression equations of study. Except variable (2), other variables have a hidh significance levels and have different effects on the house selling prices.
—This study rpresents an attempt to derive useful house price indices from large data sets containing only alimited number of variables.
- —The R2 value in the spatial hedonic expansion models
which have strong consequences was more than spatial hedonic and traditional hedonic models.
Meese & Wallace (1997), 27606 Observation About House Selling In Two District Over An 18 Years Period, Translog and Log-Linear Functional Forms
Dependent Variable: House selling price
Independent Variables:
(1) Number of bathrooms, (2) Number of bedrooms, (3) Sq. ft. of lot size, (4) Number of rooms, (5) House quality index, (6) Age of structure.
—This article examines a number of hypotheses that underpin the repeat-sales and hedonic approaches to the construction of housing price indices, as well as the practical problems associated with the implementation od either approach.
—Study examines a hybrid procedure that combines elements of both the repeat-sales and hedonic regression techniques.
—In this article, documenting the shortcomings of repeat-sales price indices when they are constructed on municipality-level data sets. The indices suffer from sample selection bias and nonconstancy of implicit housing characteristic prices, and the yare quite sensitive to small sample problems.
—The standart variance specification of repeat-sales approaches appears to be inappropriate for data at the municipality level.
—Repeat sales methods reject the assumption that changing attribute prices over time.
