ArticlePDF Available

Residential property value and locational externalities : On the complementarity and substitutability of approaches


Abstract and Figures

It is well-known from the literature that locational externalities influence the price formation of residential property. This effect is usually studied empirically with the hedonic price models, by including various neighbourhood and proximity variables in the model. These regression based techniques have, however, been criticised for a number of reasons. The arguments pertain partly to technical issues such as model flexibility, functional discontinuity and nonlinearity, and data quality, and partly to more fundamental problems regarding the nature of the value formation process. The criticism has attracted experiments with new modelling approaches, each of which adds something substantial to the hedonic approach. The study comprises two parts: it first highlights the rationale behind each broad approach composed of specific modelling techniques currently available, and then demonstrates an improvement of the demand side analysis by applying the analytic hierarchy process. This method enables quantification of qualitative expert judgements, and may lead to conclusions that go beyond the purely economic value framework.
Content may be subject to copyright.
Journal of Property Investment &
Vol. 21 No. 3, 200 3
pp. 250-270
#MCB UP Li mited
DOI 10.1108/14635780310481676
Residential property value and
locational externalities
On the complementarity and
substitutability of approaches
Tom Kauko
OTB Research Institute for Housing, Urban and Mobility Studies,
Delft, The Netherlands
Keywords Location, Property, Pricing, Modelling, Analytical hierarchy process
Abstract It is well-known from the literature that locational externalities influence the price
formation of residential property. This effect is usually studied empirically with the hedonic price
models, by including various neighbourhood and proximity variables in the model. These
regression based techniques have, however, been criticised for a number of reasons. The
arguments pertain partly to technical issues such as model flexibility, functional discontinuity and
nonlinearity, and data quality, and partly to more fundamental problems regarding the nature of
the value formation process. The criticism has attracted experiments with new modelling
approaches, each of which adds something substantial to the hedonic approach. The study
comprises two parts: it first highlights the rationale behind each broad approach composed of
specific modelling techniques currently available, and then demonstrates an improvement of the
demand side analysis by applying the analytic hierarchy process. This method enables
quantification of qualitative expert judgements, and may lead to conclusions that go beyond
the purely economic value framework.
The theory of property value formation examines location as a composite effect
of a set of locational attributes. These are either positive or negative
externalities that contribute to a certain amenity effect, which is internalised in
property values. In the following, locational externalities are defined in a very
broad sense: not only including the very localised negative externalities (social,
physical or visual ones), but also the positive ones (services, greenery, status
etc.), which usually influence the property value within a much broader range
than the effects related to specific disturbances. The detection of a relationship
between property value, often approximated as transaction price, and a
locational externality effect within well-specified locational boundaries is
referred to as economic impact analysis. The practical aspect of impact
analysis may be related to urban planning ± to evaluate in monetary terms the
effect of externalities such as parks, metro, highway and shopping centres for
the community. Alternatively, the purpose may be to determine the grounds
and magnitude of compensation for property owners due to a source of
T h e E m e r a l d R e s e a r c h R e g i s t e r f o r t h i s j o u r n a l i s a v a i l a b l e a t
h t t p : / / w w w . e m e r a l d i n s ig h t . c o m / r e s e a r c h r e g i s t e r
T h e c u r r e n t i s s u e a n d f u l l t e x t a r c h i v e o f t h i s j o u r n a l i s a v a i l a b l e a t
h t t p : / / w w w . e m e r a l d i n s i g h t . c o m / 1 4 6 3 - 5 7 8 X . h t m
This is a revised version of a paper presented at the 8th European Real Estate Society
Conference ± ERES Alicante, Spain, 27-29 June 2001. The author is thankful for comments from
the participants of the conference.
property value
nuisance, for example a hazardous waste site or a high voltage power
transmission line.
The literature dealing with modelling location is abundant. The methods are
often divided into two broad categories: the actual market outcome (prices,
rents etc.) data-based approach (revealed preferences) and the interactive
(interviews, questionnaires etc.) data-based approach (stated preferences or
preferences/perceptions modelling). Within both of these strands of research
there are several particular techniques. Some of them are aimed at estimating
aggregate market outcome (i.e. shadow prices and willingness to pay (WTP)),
whereas others are decision oriented modelling techniques (discrete choice
The most commonly used method is hedonic regression modelling, either in
its simplest form or an extension of it. It is not surprising that hedonic house
price modelling with its spatial extensions may obtain extremely high levels of
explanatory power. As with other micro-economics based methods of the
revealed preferences approach, this method operates within unrealistically
strict assumptions. The contribution outlined in the paper concentrates on one
such assumption, the smooth linear relationship between price and location. Of
all house price characteristics location is particularly discontinuous and
nonlinear ± especially in contexts where the role of regulation is substantial.
Whereas the size of the dwelling exhibits a reasonably linear (or loglinear)
smooth price association, consider for example the price association of CBD
distance location, when the observed radial interval crosses a motorway, or
when the vicinity of the observation includes an environmental amenity such
as a park. In such a case price formation is largely affected by outlier
The paper first outlines the principles of hedonic regression modelling, after
which it comments on flexible and spatial extensions of these techniques, that
still are based on market data analysis. Then it reviews the other broad
approach: interactive data-based modelling. Here one specific technique is dealt
with profoundly ± the analytic hierarchy process (AHP). After this, an exercise
aimed at isolating components of locational quality based on expert interviews
and the AHP follows, and specific findings are reported from metropolitan
Helsinki, Finland. As the resulting empirical models allow insight into the
systematical diversification of locational preferences, the analysis is advocated
as a demand sided improvement of current modelling practice. Lastly, some
conclusions are made about the substitutability and complementarity of the
various methods reviewed, with a particular focus on the results of the expert
interview exercise presented here.
Location in hedonic type of value models
Hedonic price theory explains the price formation (p) of a product (z) consisting
of a set of different qualitative and quantitative characteristics (z1; :::; zn) such
as size, age, amenities, views etcetera. The product as a whole unit is traded in a
competitive market but the individual characteristics do not have a separate
price. The components of zare assumed to be objectively measurable and every
product has a market price which is connected to a certain value of vector z.
The independent variables selected (z1; :::; zn) may be in integer, scale or
dummy variable form, and they are the result of hypothesised relationships
between ziand the dependent variable (Miller, 1982).
The basic idea of the theory is to show how p(z) is determined. The idea is to
compare different goods and assess the value of their differences, ``shadow
prices’’ (i.e. marginal adjustment factors), with respect to all the factors
determining the price. This way, the differences in these factors between the
goods can be adjusted for, and finally the prices of two different property-
baskets can be compared objectively. Hedonic models are very useful, since
they provide an operational theory for dwelling price analysis (e.g. Laakso,
1997; Orford, 1999).
According to the capitalization theory, an environmental improvement or a
public good provided by the local government leads to higher house prices in
the vicinity, unless the good causes significant negative externalities and
become bad. The benefits can be measured as the difference in price per
housing unit, before and after the improvements. Hence, hedonic price models
can with certain conditions be used to evaluate these effects (e.g. Laakso, 1997).
Controlling for the locational effect can be done by operationalising suitable
proxies for location and neighbourhood and adding them into the right-hand
side of the model. A standard method is to estimate by means of multiple
regression a function that connects the prices of apartment or property values
with ``shadow prices’’ or marginal adjustment factors for each locational
variable. However, in the empirical hedonic modelling literature locational
proxies may be defined in various ways (cf. Ball, 1973; Miller, 1982; Laakso,
1997; Lentz and Wang, 1998). The following variables are commonly used:
Accessibility factors: general and specific accessibility.
Neighbourhood ± level factors: the physical and social quality of the
environment, or alternatively a composite neighbourhood effect proxy.
Here it is to observe that externality impacts located within or adjacent
to neighbourhoods such as parks are a different story than general
neighbourhood level quality variables (Lentz and Wang, 1998).
Specific negative externalities: damages caused by air pollution, noise,
visual effects etcetera (Miller, 1982). Recent studies confirm that specific
local externalities, such as proximity to industry and refineries affect
property values negatively (see Lentz and Wang, 1998).
Public services and taxes: predominantly included in North American
price studies. The total fiscal package available for each municipality
consists of the benefits of all the public services provided and the costs
or property taxes that are levied on each property. Defined this way, the
public services should have a positive effect on value, whereas the taxes
should have a negative effect on value (Miller, 1982).
property value
Density: we can derive two opposite hypotheses about the relationship
between plot efficiency and house price, assuming all other factors are
equal: the increased land price level caused by higher land use is
switched to the dwelling prices; and more efficient land use implies a
negative house price premium because of a reduced satisfaction for the
resident. This type of effect is analysed in a comparative statics setting
(see e.g. Cho and Linneman, 1993).
If we choose not to split the location into components, we may add location
dummies and estimate a coefficient for each dummy. Alternatively, location
may be treated as a more continuous variable, but in any case through some
sort of standard identification of observations such as X and Y-coordinates,
postal zip codes or census tracts.
Apart from the variable operationalisations, another issue concerns the
evaluation of the model. The inadequateness of a strict model fit criteria for the
evaluation is revealed if the theoretical model behind the equilibrium-based
calculations is opened up. Local externalities are indeed capitalised in land
values and house prices, but what is the spatial and contextual extent of it? The
methodology based on the assumption of a single value model operating on
data from a uniform housing market is not necessary valid, due to multiple
equilibria and various shortcomings. Therefore, marginal adjustment factors
might be more feasible to estimate as separate equations for each area, given
idiosyncrasies pertaining to a certain area, group of people or both.
In principle a hedonic regression cannot detect zonal boundaries, only the
significance of the direction and coefficient of the effect of the value factors as
well as the accuracy and explanatory power within the total sample of
observations. One way of clarifying the issue is to use dummy variables
(Laakso, 1997). Another solution is to split the data into different segments,
which are either a priori predefined or synthesized somehow. Also if
segmentation in a theoretical sense is ignored, the partitioning approach may
be justified (see Needham et al., 1998). One way of managing the segmentation
of data is to chain different statistical techniques to one another (see Bourassa
et al., 1997, 1999).
Market segmentation raises the question, whether a locational externality
capitalization is more appropriately measured with locationally partitioned
data sets and on a submarket specific modelling basis than with total data sets
and with a single hedonic price function (see Michaels and Smith, 1990, for
proximity to hazardous waste sites). GIS-technology may be used as a pre-
processing device for variable construction, and also for data aggregation,
storage and visualisation to improve the analysis of locational value. It is even
more advisable to use an ``export-import approach’’, where the statistical
analysis is undertaken externally to the GIS (see Orford, 1999; Lake et al., 1998;
Rodriguez et al., 1995; Bible and Hsieh, 1996).
All in all, the hedonic approach to house price analysis seems both logical to
grasp and scientifically rigorous. Having said that, there is also reason to be
critical towards the approach. The hedonic models require extensive data sets,
which in most countries are not readily available. Furthermore, problems are
caused by omitted variables, sample selection and choice of functional form
(Hoesli et al., 1997). To be fair to the hedonic theory, it has indeed challenged
these criticisms in an ad hoc way. For the choice of functional form,
traditionally either linear, logarithmic, or multiplicative forms are chosen (see
e.g. Miller, 1982), and more recently a certain Box-Cox transformation
procedure for choosing between alternative functional forms is applied so as to
obtained better specified models (see e.g. So et al., 1997; Laakso, 1997; Orford,
1999). These functions are, however, not truly nonlinear; the nonlinearity only
refers to the argument of respective function.
Flexible regression and advanced spatial modelling approaches
Several estimation methods make fewer assumptions of the data than a fixed
parameter hedonic price model. More generally, Verkooijen (1996) uses the term
``flexible regression’’ rather than ``non- or semi-parametric regression’’ for
estimation techniques such as locally weighted regression (non-parametric
method) and additive models with both parametric and non-parametric
components (semi-parametric method). In multiple regression analysis the
variance is low, whereas the bias is high. In flexible regression it is the other
way around: greater variance is generated but the bias is reduced instead. In
the context of estimating house prices or property values, flexible regression
techniques have been discussed and encouraged by Meese and Wallace (1997);
Pace (1995); Mason and Quigley (1996); Verkooijen (1996); Kyllo
¨nen and Ra
(2000); and Pavlov (2000) to name some recent studies. Coleman and Larsen
(1991) in turn remain more critical towards alternative estimation techniques.
Neurocomputing or the (artificial) neural network comprises an emerging
category of numeric, ``learning’’ or ``intelligent’’ techniques[1] and a sort of
flexible, model-free regression (see e.g. Verkooijen, 1996; Pace, 1995; McCluskey
and Anand, 1999). The nature of the neural network is a ``black box’, which
means that there is no clear functional relationship between the input and
output values. The algorithm learns by training. The basic elements in a neural
network are called neurons or nodes. The connections between them are
determined by weights. Together the neurons process a numerical signal
coming from outside the network in such a manner that a connection, between
input and output information is developed. The connection is referred to as the
``intelligence of the network’’. Throughout the 1990s several studies have been
conducted in order to solve the validity of the neural network as a method of
property value modelling. These studies have arrived at mixed results, while
some of them are for (e.g. Tay and Ho, 1992; Do and Grudnitski, 1992) and
others against (Worzala et al., 1995; McGreal et al., 1998) the approval of
neurocomputing within the appraisal industry. Repeatedly, the same question
is asked: whether the atheoretic nature of the neural network really is a
problem, if the results are good?
property value
The sudden emergence of spatial statistics within the value modelling
paradigm mostly is due to the increased availability of GIS (e.g. Can and
Megbolugbe, 1997). These methods accept the fact that including all
fundamental characteristics of hedonic models is impossible, and allow for the
spatial variation of the data instead. The standard hedonic specifications do not
consider the spatial relations between objects very well, and problems arise
from the spatial dependence of the property values as well as coefficient
dependence regarding the characteristics. Generally, when talking about
spatial house price models, we may separate between two principal lines of
research, spatial lag models that allow for spatial drift and decay effects and
spatial error models that allow for spatial autocorrelation of residuals (e.g.
Meen, 1998). Local spatial errors or spatially lagged variables effectively proxy
for omitted variables correlated with location (Pace et al., 1998)[2]. Interestingly,
Pavlov (2000) combines these two effects into a successful ``semi-parametric’’
approach based on space-varying regression coefficients.
By using all types of specifications the user gets a much better arsenal
against the problems of inefficiency and mis-specification (Pace, 1995).
However, the contextual, spatial and flexible expansions of the hedonic
regression approach to value modelling also are based on the assumption of a
smoothly changing price determination over space (see e.g. Pavlov, 2000).
Arguably, such an assumption is unrealistic, because (especially in the
European contexts) land use regulation interferes almost by definition.
Furthermore, demand side processes involving formation of preferences (i.e.
agency) are traditionally difficult to isolate from market outcome data[3].
Another difficulty is getting information of the locational quality, to later
compare with market data. Because of these limitations a completely different
type of approach is reviewed and applied in the remainder of the paper.
Perceptions/preferences modelling
So far the review of value modelling literature has identified a variety of
difficulties that mainly pertain to the supply side and aggregate level. The
remainder of the paper will address problems with the demand side analysis.
Here the relevant objectives are locational choice, behaviour and preferences,
quality and agency effects. Thus, a substantial share of the plausible
improvements would have to allow for a dis-aggregation of demand.
Recent comments freely encourage the use of interview survey methodology
for residential valuation (see Rodriguez and Sirmans, 1994; see also Strand and
ƒgnes, 2001). The contingent valuation method (CV) is based on a formal
questionnaire about the respondent’s willingness to pay (WTP) or ± in reverse
situations ± his/her willingness to accept (WTA) a given sum of money. It is
fairly commonplace in environmental impact assessment studies (see e.g.
Breffle et al., 1998). Within environmental economics estimates generated by
CV and hedonic modelling have been compared in several contexts (e.g.
Shechter and Kim, 1991; Vainio, 1995a,b; Willis and Garrod, 1993).
On the other hand, when analysing housing prices and preferences,
situations might occur where we need very context sensitive insight into how
various multidimensional values towards housing and environment are being
perceived by the individual. The more recently emerged subset of the stated
preferences approach do not in general aim at an estimate of value or aggregate
demand, but rather at an estimate of choice behaviour in a problem centric
setting of discrete alternative decisions. The idea is to transport the method
down to the level of the individual problem; rather than to calculate an estimate
that can be used for solving several types of problems. The actual problem, that
we do not have past information about, determines the limits of the method.
The prescriptive approaches have been developed, inter alia, as aids to
decision making in complex situations. More specifically, these include
techniques such as the AHP (a demonstration follows in the next section), the
self-explicated utility method and conjoint analysis. The first two are
hierarchical models and thus apply the ``value tree’’ concept, whereas the last
one is based on choice profiles. All three are aimed at making choices according
to preferences in a multiattribute problem setting, in contrast to the purely
economic WTP setting of revealed preferences and CV (e.g. Po
¨nen, 1998a;
Miettinen and Ha
¨inen, 1996). In these methods the weighting of the
preferences becomes a question of elicitation (Ruokolainen and Tempelmans
Plat, 1998; Po
¨nen, 1998a; see also McLean and Mundy, 1998).
The AHP (Saaty, 1990) is based on pair-wise comparison of preferences,
systematically made on each level of a hierarchy of factors presented as a value
tree. At the top of the hierarchy lies the overall objective of the decision,
whereas lower level objectives or attributes lie on the lower levels of the
hierarchy (e.g. Zahedi, 1986). The comparison starts at the lowest level of the
tree, where the elements will usually be elicited with the ordinal scale 1-9,
where the values also correspond to verbal expressions, 1 being equivalent to
``both are of equal importance’’ and 9 being equivalent to ``A has an extreme
importance over B’’[4]. The comparisons are then converted into cardinal
rankings which when summed up equal 1 (e.g. Erkut and Moran, 1991). Finally,
the local weights are transformed into global ones. This means quantifying the
relative contribution of each element in the value tree to the overall goal[5].
The total number of comparisons is (An¡1¤An)/2. For example, a matrix of
four elements generates six comparisons. Each comparison generates a pair-
wise ratio, w1=w2;w2=w1;w1=w3;w3=w1etc. The overall weight is indicated by
the priority vector. Furthermore, the consistency of the comparisons is
measured with the consistency ratio (CR). CR is calculated based on the
expected results of consistent pair-wise comparisons across the matrix. CR is
supposed to be very small and usually a cut-off rule of 10 per cent is applied for
the comparison matrix to be consistent enough, otherwise the comparison is
repeated. This ``eigenvalue’’ method is the most common way to estimate the
relative weights from the matrix of pair-wise comparisons (see e.g. Zahedi,
1986, for a full discussion).
property value
The AHP technique has been applied with success, inter alia, in portfolio
investment and building site selection. One of the classical applications is in
fact a house selection based on assessment of its attributes (e.g. Po
1998a; Saaty, 1990). To name a few other relevant appraisal applications, Ball
and Srinivasan (1994) analysed the role of psychological factors in house
selection, Laakso et al. (1995) and Bender et al. (1997, 1999) investigated
perceptions about the quality of the location and the neighbourhood, and Ong
and Chew (1996) showed how an expert view can be incorporated into
forecasting. Most recently, Strand and Va
ƒgnes (2001) used an expert elicited
utility-based approach to study the relationship between railroad proximity
and property value in Oslo, Norway.
Recently various qualitative AI-techniques have been applied for value
modelling, in order to offer a transparent and ``back to the basics’’ approach to
simulating market behaviour. After all, the issue is not about generating
perfect accuracy, but about simulating human judgement. Rule-based expert
systems emerged as a counter paradigm to neural networks and other
numerical techniques (see Scott and Gronow, 1990; McCluskey and Anand,
1999). Case-based reasoning deals with retrieval of past cases similar to the one
to be assessed (see O’Roarty et al., 1997; Gonzalez and Laureano-Ortiz, 1992).
The theory of fuzzy logic is formally specified as a measure for the degree of
membership for an element’s belonging to a set, and can sometimes be used in
property valuation as well (see Bagnoli and Smith, 1998). Apart from direct
valuation or market modelling applications there is a variety of fairly eclectic
methods of behavioural property research that do not form a coherent group.
The methods reviewed above have one thing in common: to generate
estimates associated with value differences from information collected through
interviews. Within the stated preferences models a distinction was made
between two types of methods: the essentially descriptive tradition of
behavioural research on one hand, and the essentially prescriptive tradition of
decision support research on the other. The majority of the former type of
research relies on mail-back surveys whereas the majority of the latter type
undertake face-to-face interviews. The primal concern of the former tradition is
the demand for attributes, whereas it is preferences and choices in the latter
tradition. Recently also some qualitative methods have gained popularity ±
either computer assisted valuation systems or case studies.
Expert elicited locational quality models
To continue the updating of the methodology, the next task at hand is to
explain how the behavioural modification propagated so far could improve the
current modelling practice. While the broad hedonic literature here serves as a
guideline for the overall goal of the exercise, the review has recognised
problems that require a more pragmatical and qualitative treatment. Basically,
demand side improvement has been called for: dis-aggregation of preference
profiles to capture the systematic variation in preferences among actors
(agency) and measurement of the rather intangible quality factor. Below some
empirical material is presented as a support to this argument. The key to the
analysis is the intensive element instead of the extensive one; to look at context
specific, fuzzy and qualitative effects, and forget about requirements of
statistical sample selectivity. However, it needs to be borne in mind that, within
the argument of a nonlinear and discontinuous locational influence on house
prices, plenty of other directions of improvement, some of which address
supply related issues (land use regulations, for example), can be feasible as
To clarify the associations between locational value formation and agents
operating in the residential property market, an experimental study was
conducted about the locational value factors for two separate submarket
settings: multi-storey flats and single-family houses in suburban Helsinki
metropolitan area. To cover all aspects of the problem, the respondents had to
meet two criteria: first, a pursuit as stakeholder based on professional
responsibility, and second, a deep local knowledge gained by professional
experience. Hence the participants were real estate agents and assessors; rental
agents (relatively rare in Helsinki); building companies in the owner occupied
sector; few large owners of rental houses; other owners (especially the
municipality); managers of private housing corporations; planning officers; and
``other experts’’.
Because of the simultaneous commodity and asset nature of housing in
economic analysis, the house has to both satisfy the needs of the dwellers and
generate the yield and/or profit compatible with alternative investments (e.g.
Meen, 1998). Now the question arises, whose agents’ values are at stake and
with respect to which factor is the relative importance to be determined? The
problem was solved by making explicit to the respondents that when choosing
the preferred combination of characteristics, the relevant preferences always
have to be checked against preferences of the consumer, the would-be end user
in question.
The variables are presented in the hierarchy illustrated in Figure 1, which is
based on several different studies (e.g. Hoesli et al., 1997; Laakso, 1997; Laakso
et al., 1995; Miller, 1982).
The idea is to split the problem into parts and compare fairly subjective
values against each other. The comparisons are begun on the lower levels of the
value tree (attributes), from which one proceeds to the upper levels (sub-criteria,
criteria and goal). Altogether 18 comparisons are performed for each segment.
Later some actual alternatives (i.e. either multi-storey or single-family homes)
can be linked to each model for rating and comparison.
The relative strengths between the attributes for each aggregate model are
seen in Figures 2 and 3. Not surprisingly, the external accessibility is the single
most important attribute. This obviously has a mathematical explanation,
when these attributes where positioned on an upper level in the hierarchy, but
still comparisons across different levels can be made if one uses global weights
¨nen, 1998b). However, the status and the commercial services are very
important as well, and a bit surprisingly also the municipality matters, which
property value
Figure 1.
The hierarchical
structure of the
value tree
would suggest some similarities to a Tiebout-type of trade off between different
local government packages. In that case, prospective investors and residents
would make the choice of location based on the net benefit of municipality
characteristics including taxation.
The responses were also classified into three to five groups for each
segment, based on similar patterns of preferences. The dis-aggregated profiles
for flats are shown in Figure 4. Figure 5 shows the model, where the 10 per cent
consistency rule (cut-off rule) was applied. For the flats, the elicitation does not
differ substantially between the total model (see Figure 2) and this model: the
attributes with the highest (external distances) and lowest (density)
magnitudes are the same, which indicates a certain consistency in the results.
Figure 2.
The locational attributes
in the aggregate model
for flats (22 respondents)
Figure 3.
The locational attributes
in the aggregate model
for single-family houses
(22 respondents)
property value
The same differentiation of preferences was then done for the single-family
segment. Using the same notions as above, the three resulting dis-aggregated
models are shown in Figure 6. When the cut-off rule was applied on the single-
family segment the group of respondents shrank to four, the result was a
slightly different profile than the aggregate model of all respondents (see
Figure 4.
The locational attributes
in the dis-aggregated
models for flats
Figure 7). However, the two strongest (ext. dist. and status) and the two
weakest preferences (taxation, density) remained the same (cf. Figures 3 and 7).
The exercise showed that it is possible to analyse the rather abstract and
fuzzy concept of preference-based locational and housing quality by a three
stage process: first, splitting it into different elements; second, assessing the
relative importance of each element; and third synthesising the calculations
Figure 6.
The locational attributes
in the dis-aggregated
models for single-family
Figure 5.
The locational attributes
in the aggregate model
for flats with a 10 per
cent cut-off rule in
inconsistency (three
property value
into mathematical models ± an aggregate model as well as three or four
diversified models for each supply side segment.
Agency relationships
In the models discussed above (see Figures 2-7) substantial differences in
elicitation were found equally within professions or interest groupings, as well
as across these categories. This preliminary finding does not support the use of
a priori grouping of respondents into models. However, a more formal testing
of whether one group of participants elicits systematically higher or lower
preferences than another was undertaken. This was based on a significance
measure, where a difference of 0.045 between any two groups was considered
enough for an agency effect to be verified, for a given attribute. This measure is
derived from the formula (Max ± Min)/10, previously used by Kaufman and
Escuin (2000) in a somewhat similar problem setting (Max = 0.450, Min =
The respondents were partitioned into three categories, depending on their
relation to their function as an urban housing or land market participant. The
three categories were ``investors’’, ``assessors’’ and ``occupants’’. The
composition of each interest group was as follows:
(1) The investor’s view: builders, investors and municipality as a land-
owner (asset value).
(2) The assessor’s view: estate agent and assessor (exchange value).
(3) The occupant’s view: planners, housing managers and rental agents (use
For some of the attributes agency effects were detected. The investors may put
an extra emphasis on social factors (a somewhat surprising finding) and
taxation (direct profitability calculations), the assessors may emphasise
accessibility within the area (due to good access to exceptional location specific
information) and the occupants on CBD accessibility and general satisfaction.
The extended analysis confirmed that preferences towards location specific
attributes may depend on whether the respondent belongs to a predefined
category that recognises the asset value, exchange value or the use value of the
Figure 7.
The locational attributes
in the aggregate model
for single-family houses
with a 10 per cent cut-off
rule in inconsistency
(four respondents)
Alternative locations
The expert elicited models may subsequently be used in problems involving
assessment of alternative locations, the fourth stage in the process. The
alternative locations were comparable housing locations from both segments,
where investment was considered in the near future. The idea was to assess
seven micro-locations from each segment. The comparisons were performed
based on various sources of information (statistical yearbook, earlier research,
cartographical displays and own judgement) for each of the 12 locational
variables for each model (see Figures 8 and 9). The suitable locations
represented suburban Helsinki, Espoo and Vantaa, and they were given by
some of the respondents during the interviews.
The ranking order between the locations does not change too much in either
segment. However, for flats (see Figure 8), one location changes place from best
location to sixth best depending on the model used for elicitation. This location
(Kontulankaari 4) has fairly good accessibility and services but is bad in a
social sense. Consequently, the model Ia ranks it as the best location whereas
the model II ranks it as the second worst location.
Similar effects were noted in the models for single-family houses (see Figure
9): the order between the seven cases was fairly stable across the models.
Nevertheless, some differences between the ranking order existed dependent on
models, but it was only among the order of the four best locations: ``Marttila’’,
``Kuunarinkuja’’, ``Nuottalahdentie’’ and ``Soukan rantatie’ changed order
reciprocally in the models. To be specific, ``Marttila’’, a location with good
accessibility and service attributes, received the highest ranking in all the other
models, except in the model I and in the ``cut-off’’ model (albeit narrowly in the
latter model). In these two models the social factors had a dominating role, and
therefore ``Kuunarinkuja’’obtained the highest ranking.
The resulting scores are readily applicable for land value determination in
various different ways. Perhaps the most apparent application is to construct a
measure of the locational quality to include into hedonic models (see Bender et
al., 1997). Alternatively, as shown by Ball and Srinivasan (1994), by comparing
the association between actual price levels and the AHP generated ``ranks’’ a
price/quality relationship may be obtained. These ratios may indicate two
(1) The performance of the property portfolio regarding its components.
(2) The existence of other than purely monetary values in the vicinity of the
subject property.
Final remarks
No single method can capture all the aspects of locational value formation.
Each method covered has its cons and pros with regard to the total task at hand
and therefore the evaluation criteria to apply cannot be uniform across the
range of methods. The crucial question to ask is an explorative one: exactly
what do we want to achieve by using a particular method? While spatial and
property value
Figure 8.
The relative importance
of the alternative
locations in the models
for flats
Figure 9.
The relative importance
of the alternative
locations in the models
for single-family houses
property value
flexible hedonic regression models solve one problem: how to deal adequately
with nonlinearity and discontinuity in the price formation of land and housing,
other problems, related to the behavioural processes involved, remain.
Therefore, methods based on hypothetical data might bring added value to the
The main achievements of the proposed modelling strategy have been on the
demand side. In a currently popular social science jargon: in agency more than
structure. On one hand, real life decisions to achieve economic profitability and
more social goals alike are also based on qualitative judgement of the situation
and ``soft factors’’ ± not solely on heavy model calculations. To encompass the
explicitly behavioural aspects of location within the modelling application
therefore seems promising, as the end user may welcome a pragmatical way of
thinking. On the other hand, the prospects are also good for those who have
their preference in a more orthodox modelling approach. The formally derived
part worth utilities of an overall value may conveniently be understood within
the context of a strictly rational hedonic model. In this case adding an
interactive element provides a means to handle the problematic locational
quality factor more efficiently than a method that uses only secondary data
A word of warning is also necessary. Based on the study it appears as if the
use of expert elicited multiattribute modelling methods should be restricted to
property investment related applications, where an arguably ``subjective’’ value
is not a problem for the commissioner. The method in itself is less promising in
a realm where the target is different, an ``objective’’ value (tax assessment, for
example); unless the expert elicitation is used as a support to hedonic models, to
put a more qualitative spin on the analysis. Here we see the core idea of the
argument: from the perspective of progress in value modelling related research,
methodological triangulation combines market data-based and interactive
approaches to an extended approach, where each method covers the blind spots
of the other method (see Strand and Va
ƒgnes, 2001). Alas, while incorporating
new elements into the value model brings a more differentiated analysis, it also
brings a less clear result.
1. The genetic algorithm is another AI-technique, which has proved successful in
applications involving the efficient investigation of large search spaces. The genetic
algorithm performs an artificial ``breeding’’ of a replacement population from a randomly
generated population of previous encoding (Cooley et al., 1994).
2. The grid adjustment method (referred to as ``real estate’s homegrown spatial statistical
estimator’’ (Pace et al., 1998)) uses the estimated coefficients of regression models for
adjustment of comparable sales and assigns more weight to comparables similar to the
subject property, thus avoiding the high standard error problem associated with the
regression method.
3. It is, however, plausible that the hedonic WTP models (i.e. specific demand models, see e.g.
Laakso, 1997) can capture the variation in preferences for different groups, in other words
4. The scale can be ordinary like here or balanced (Po
¨nen, 1998b). A balanced or
multiplicative scale 1:8, based on the formula en¤…1=8†¤ln8 , is sometimes used instead of the
standard linear scale 1:9. Property valuation is one example of such a problem, where this
scale is suggested to be more accurate than the standard linear one (see Laakso et al., 1995;
Bender et al., 1997, 1999).
5. Expert choice ± software is used for constructing the hierarchy, performing the
comparisons and calculating the ordinal and cardinal weights for each element, either
variable or alternative.
Bagnoli, C. and Smith, H.C. (1998), ``The theory of fuzzy logic and its application to real estate
valuation’’, Journal of Re al Estate Research, Vol. 1 6 No. 2, pp. 169-99.
Ball, J. and Srinivasan, V.C. (1994), ``Using the analytic hierarchy process in house selection’’,
Journal of Re al Estate Finance and E conomics, Vol. 9, pp. 69-85.
Ball, M.J. (1973), ``Recent empirical work on the determinants of relative house prices’’, Urban
Studies, Vol. 10 , pp. 213-33.
Bender, A., Din, A., Hoesli, M. and Laakso, J. (1999), ``Environmental quality perceptions of urban
commercial real estate’’, Journal of Property Investment and Finance, Vol. 17 No. 3,
pp. 280-96.
Bender, A., Din, A., Favarger, P., Hoesli, M. and Laakso, J. (1997), ``An analysis of perceptions
concerning the environmental quality of housing in Geneva’, Urban Studies, Vol. 34 No. 3,
pp. 503-13.
Bible, D.S. and Hsieh, C.-H. (1996), ``Applications of geographic information systems for the
analysis of a partment rents’’, Journal of Real Estate Research, Vol. 12 No. 1, p p. 79-88.
Bourassa, S.C., Hamelink, F., Hoesli, M. and MacGregor, B. (1997), ``Defining residential
submarkets: evidence from Sydney and Melbourne’’, Real Estate Research Unit Working
Paper Series 3, Department of Property, University of Auckland, Auckland.
Bourassa, S.C., Hamelink, F., Hoesli, M. and MacGregor, B. (1999), ``Defining housing
submarkets’, Journal of Housing Economics, Vol. 8, p p. 160-83.
Breffle, W.S., Morey, E.R. and Lodder, T.S. (1998), ``Using contingent valuation to estimate a
neighbourh ood’s willingness to pay to preserv e undeveloped urban land ’’, Urban Studies,
Vol. 35 No. 4, pp. 715-27.
Can, A. and Megbolugbe, I. (1997), ``Spatial dependence and house price index construction’’,
Journal of Re al Estate Finance and E conomics, Vol. 14 Nos 1/2, January/March, pp. 203-22.
Cho, M. and Linneman, P. (1993), ``Interjurisdictional spillover effects of land use regulations’’,
Journal of Housing Research, Vol. 4 No. 1, pp. 131-63.
Coleman, J.W. and Larsen, J.E. (1991), ``Alternative estimation techniques for linear appraisal
models’’, The Appraisal Journal, Octob er, Vol. LIX No . 4, pp. 526-3 2.
Cooley, R.E., Pack, A .D., Ho bbs, M. and Clew er, A.D.E. (1994), ``A gen etic algorithm for m odelling
locational effects on residential property prices’’, The Cutting Edge ± 94 Conference
Proceedings, pp. 1 79-93.
Do, A.Q. and Grudnitski, G. (1992), ``A neural network approach to residential property
appraisal’’, The Real Es tate Appraiser, Decembe r, pp. 38 -45.
Erkut, E. and Moran, S.R. (1991), ``Locating obnoxious facilities in the public sector: an
application of the analytic hierarchy process to municipal landfill siting decisions’’, Socio-
Economic Planning Sciences, Vol. 25 No . 2, pp. 89-102.
Gonzalez, A.J. and Laureano-Ortiz, R. (1992), ``A case-based reasoning approach to real estate
property app raisal’’, Expert Systems With A pplication, Vol. 4, pp. 229-46 .
property value
Hoesli, M., Giacotto, C. and Favarger, P. (1997), ``Three new real estate price indices for Geneva,
Switzerland’’, Journal of Rea l Estate Finance and E conomics, Vol. 15 No. 1, pp. 93-109.
Kaufman, J.L. and Escuin, M. (2000), ``Thinking alike: similarities in attitude of Dutch, Spanish,
and American planners’’, APA Journal, Vol. 66 No. 1, pp. 1-12.
¨nen, L. and Ra
¨ty, T. (2000), ``Asuntojen hinta-laatusuhde Joensuussa, semiparametrinen
estimointi’’ (in Finnish), VATT-tutkimuksia 67.
Laakso, J., Bender, A., Din, A., Favarger, P. and Hoesli, M. (1995), ``An analysis of perceptions
concerning the e nvironmental quality o f residential rea l estate in Gen eva’’, paper pr esented
at the 1 st IRES Co nference, Stoc kholm 28 June -1 July.
Laakso, S. (1997), Urban Housing Prices and the Demand for Housing Characteristics, The
Research Ins titute of the Fin nish Econom y (ETLA) A 27, Helsinki.
Lake, I.R., Lovett, A.A., Bateman, I.J. and Langford, I.H. (1998), ``Modelling environmental
influences on property prices in an urban environment’’, Computers, Environment and
Urban Systems, Vol. 2 2 No. 2, pp. 1 21-36.
Lentz, G.H. and Wang, K. (1998), ``Residential appraisal and the lending process: a survey of
issues’’, Journal of Real Esta te Research, Vol. 15 Nos 1/2, pp. 11-39.
Mason, C. and Quigley, J.M. (1996) ``Non-parametric hedonic housing prices’’, Housing Studies,
Vol. 11 No. 3, pp. 373-85.
McCluskey, W.J. and Anand, S. (1999), ``The application of intelligent hybrid techniques for the
mass appraisal of residential properties’’, Journal of Property Investment and Finance,
Vol. 17 No. 3, pp. 218-38.
McGreal, S., Adair , A., McBu rney, D. and Pa tterson, D. (1998), ``Neural networ ks: the pre diction of
residential values’, Journal o f Property Valua tion and Investment, Vo l. 16 No. 1, pp. 57-70.
McLean, D .G. and Mundy, B. (1998 ), ``The addition of c ontingent valuation and conjoint analysis
to the required body of knowledge for the estimation of environmental damages to real
property’’, Journal of R eal Estate Practice a nd Ed ucation, Vol. 1 No. 1, p p. 1-19.
Meen, G. (1998), ``25 years of house price modelling in the UK. What have we learnt and where do
we go from he re?’’, pape r presented at the E NHR Conferen ce, Cardiff, Septem ber, available
Meese, R.A. and Wallace, N.E. (1997), ``The construction of residential housing price indices: a
comparison of repeat-sales, hedonic-regression, and hybrid approaches’’, The Journal of
Real Estate Finance and Economics, Vol. 14 No s 1/2, Januar y/March, p p. 51-74.
Michaels, R.G. and Smith, V.K. (19 90), ``Market segmentation and valuing amenities with hedonic
models: the case of hazardous waste sites’’, Journal of Urban Economics, Vol. 28 No. 2,
pp. 223-42.
Miettinen, P. and Ha
¨inen, R.P. (1996), Y mpaÈ ristoÈ n arvottaminen ± Ta loustieteelliset ja
monitavoittei set me netelmaÈ t (in Finnish), Helsinki University of Technology, Systems
Analysis L aboratory B19, Es poo.
Miller, N.G. (1982), ``Residential property hedonic pricing models: a review’’, in Sirmans, C.F.
(Ed.), Urban Housing Markets and Property Valuation. Research in Real Estate, Vol. 2,
JAI Pr ess, Greenwich, C T, pp. 3 1-56.
Needham, B., Franke, M. and Bosma, P. (1998), ``How the city of Amsterdam is using econometric
modelling to value real estate’, Journal of Property Tax Assessment and Administration,
Vol. 3 No. 2, pp. 25-46.
Ong, S.E. and Chew, T.I. (1996), ``Singapore residential market: an expert judgemental forecast
incorporating the analytical hierarchy process’’, Journal of Property Valuation &
Investment, Vol. 14 No. 1, pp. 50-66.
Orford, S. (1999), Valuing the Built Environment ± GIS and House Price Analysis, Ashgate,
O’Roarty, B., McGreal, S., Adair, A. and Patterson, D. (1997), ``Case-based reasoning and retail
rent determination’’, Journal of Property Research, Vol. 14 No . 4, pp. 309-28.
Pace, R.K. (1995), ``Parametric, semiparametric, and nonparametric estimation of characteristic
values within mass assessment and hedonic pricing models’’, Journal of Real Estate
Finance and Economics, Vol. 11, pp. 195-217.
Pace, R.K., Barry, R. and Sirmans, C.F. (1998), ``Spatial statistics and real estate’’, Journal of Real
Estate Finance and Economics, Vol. 17 No . 1, pp. 5-13.
Pavlov, A.D. (2000), ``Space-varying regression coefficients: a semi-parametric approach applied
to real e state markets’’, Real Estat e Economics, Vol. 28 No. 2 , pp. 249-83.
¨nen, M. (1998a), On Attribute Weighting in Value Trees, Systems Analysis Laboratory
A73, H elsinki Univers ity of Te chnology, Espoo.
¨nen, M. (1998b), discussion via Internet, Summer, European American Centre for Policy
Analysis (EAC), Delft.
Rodriguez , M. and Sir mans, C.F. (19 94), ``Quantifying the v alue of a vie w in s ingle-family hou sing
markets’’, The Appraisal Journal , Vol. LXII No. 4, pp. 600-3.
Rodriguez, M., Sirmans, C.F. and Marks, A. (1995), ``Using geographic information systems to
improve real es tate analysis’’, Journal of Real Estate Res earch, Vol. 10 No. 2, pp. 163-73.
Ruokolainen, A. and Tempelmans Plat, H. (1998), ``A system for optimizing private house
owner’s spendings and benefits’’, paper presented at the 4th Design and Decision Support
Systems in Arc hitecture and Urb an Plann ing Conferen ce, Maastric ht, 26-29 July.
Saaty, T.L. (1990), ``How to make a decision: the analytic hie rarchy process’, European Journal of
Operational Research, Vol. 4 8, pp. 9-26.
Scott, I. and Gronow, S. (1990), ``Valuation expertise: its nature and application’’, Journal of
Valuation, Vol. 8 No. 4, pp. 362-75.
Shechter, M. and Kim, M. (1991), ``Valuation of pollution abatement benefits: direct and indirect
measurement’’, Journal of Urban Economics, Vol. 30 No. 2, pp. 133-51.
So, H.M., Tse, R.Y.C. and Ganesan, S. (1997), ``Estimating the influence of transport of house
prices: evidence from Hong Kong’’, Journal of Property Valuation and Investment, Vol. 15
No. 1, pp. 40 -7.
Strand, J. and Va
ƒgnes, M. (2001), ``The relationship between property values and railroad
proximity: a study based on hedonic prices and real estate brokers’ appraisals’’,
Transportation, Vol. 28, pp. 137-56 .
Tay, D.P.H. and Ho, D.K.K. (1992), ``Artificial intelligence and the mass appraisal of residential
apartments’’, Journal of P roperty Valuation and Investment, Vol. 10 N o. 2, pp. 525-40.
Vainio, M. (1995a), Traffic Noise and Air Pollution, Helsinki School of Economics and Business
Administration A-102, Helsinki.
Vainio, M. (1995b), ``Liikenteen melu ja ilmansaasteet: Ulkoisvaikutusten arvo hedonisten
hintojen ja contingent valuation -menetelmilla
¨’’ (in Finnish), Kansantaloudellinen
aikakauskirja, 91 , 2/95, pp. 240-4.
Verkooijen, W.J.H. (1996), ``Neural netwo rks in economic modelling’’, doctoral dissertation, Ce nter
for Econ omic Researc h, Tilburg University, Tilbur g, 205 pp.
Willis, K.G. and Garr od, G.D. (1993), ``Not from e xperience: a co mparison of ex perts’ opinions an d
hedonic price estimates of the incremental value of property attributable to an
environmen tal feature’’, Journal of Property Research, Vol. 10, pp. 193-216.
Worzala, E., Lenk, M. and Silva, A. (1995), ``An exploration of neural networks and its application
to real e state valuation’’, Journal of Real Estate Research, Vol. 10 No. 2, p p. 185-201.
Zahedi, F. (1986), ``The an alytic hierarchy p rocess ± a survey of the me thod a nd its applica tions’’,
Interfaces, Vol. 16 No. 4, pp . 96-108.
... Location as a property attribute has been widely examined in real estate research. Property location has been proven to significantly influence property value in different real estate markets around the world as reported by McCluskey et al. (2000), Kauko (2003), Ge and Du (2007), amongst others. This variable was ranked as most highly significant in the Lagos metropolis, which conforms to existing literature, suggesting that homebuyers will consider the location a property when making residential decisions. ...
... Which suggest that homebuyers in the metropolis consider the characteristics of a neighbourhood when making real estate decisions. This was established in the studies of Han et al. (2002), Kauko (2003) that homebuyers are willing to pay a premium for a property situated in a neighbourhood characterised with pleasant features. This conforms to the study of Iroham et al. (2014) that found that neighbourhoods that are characterised with modern buildings designs, good roads, ocean view, building constructed with modern building materials commands higher property value. ...
... Location simply means the ease of access to public facilities such as workplace, school, public transport, shopping centers, recreation centers, health facilities, open space (Kauko, 2003(Kauko, , 2007. Most homebuyers are willing to pay more for a house in a good location and consider it as an important factor. ...
... Most homebuyers are willing to pay more for a house in a good location and consider it as an important factor. Location has a strong association with the property value (Kauko, 2003(Kauko, , 2018. This is because a majority of the residents prefer that if places such as cinema, university, shopping complex and workplace are located within their living area, they are not interested in spending time for travelling purposes. ...
Full-text available
ABSTRACT Objective: To analyze the performance of the first trimester Down syndrome screening in a single medical center in Bitola Macedonia. Materials and methods: From January 2016 to January 2022, a total of 2977 pregnant women at gestational age of 10 weeks to 13 weeks 6 days received first trimester “combined test” for Down syndrome screening. The test combines the ultrasound scan of nuchal translucency thickness and maternal biochemical serum levels of pregnancy-associated plasma protein A (PAPP-A) and free beta-human chorionic gonadotropin (bhCG). A positive screen was defined as an estimated Down syndrome risk 1/270, and either chorionic villous sampling or amniocentesis was performed for fetal chromosomal analyses. Results: 7% of pregnancies were proven to have fetal chromosome anomalies. The detection rates for trisomy 21and trisomy 18 were 82% (176), 18% (39). Conclusion: The first trimester combined test is an effective screening tool for Down syndrome detection with an acceptable low false positive rate. The best timing of screening will be between 11 and 12 weeks’ gestation.
... Location as a property attribute has been widely examined in real estate research. Property location has been proven to significantly influence property value in different real estate markets around the world as reported by McCluskey et al. (2000), Kauko (2003), Ge and Du (2007), amongst others. This variable was ranked as most highly significant in the Lagos metropolis, which conforms to existing literature, suggesting that homebuyers will consider the location a property when making residential decisions. ...
... Which suggest that homebuyers in the metropolis consider the characteristics of a neighbourhood when making real estate decisions. This was established in the studies of Han et al. (2002), Kauko (2003) that homebuyers are willing to pay a premium for a property situated in a neighbourhood characterised with pleasant features. This conforms to the study of Iroham et al. (2014) that found that neighbourhoods that are characterised with modern buildings designs, good roads, ocean view, building constructed with modern building materials commands higher property value. ...
... The impact of these attributes on property values is perceived differently by the different stakeholders because of the heterogeneous nature of real estate properties. The characteristics of the property market (imperfect, heterogeneous, complex legal interest, land laws and regulations among others) make the services of a real estate professional inevitable to a rational real estate investor/stakeholder ( [55,26,36,17] have reported that property value is determined by some sets of attributes which have been categorised into groups. The classifications of these attributes as posited by Chin & Chau (2002) [13] are locational, structural and neighbourhood factors. ...
... Some scholars see crime as a subset of neighbourhood characteristics (Sirman et al., 2006;Adegoke, 2014;Chin & Chan, 2002), Some scholars overly omitted to mention crime as an attribute of the residential property value (Kauko, 2003;Teck-Hong, 2011). ...
Full-text available
The need to curb the soaring trend of residential neighbourhood crime, considering its negative impact on the neighbourhood in general and property value in particular is not to be given the expected attention in academia and government settings. However, the consequences of residential neighbourhood crime are found to be devastating. Hence, this study proposes a Socio-Environmental Design Factors (SEDeF) model for residential neighbourhood crime toward improving property value in Nigeria. Purposive and systematic sampling techniques were adopted, while logistic regression and Structural Equation Modeling (SEM) was used to analyse the data and achieve the set objectives. The findings showed that social risk factors (poverty, unemployment, juvenile delinquencies, illiteracy, and homelessness) and the environmental design factors (natural access control, surveillance, efficient maintenance, territorial functioning, and target hardening) are capable of influencing residential neighbourhood crime in Nigeria. The results of the analysis find the set hypotheses to be significant. This is shown through the regression weights and p-values of the influence of the social risk factors and environmental design factors on residential neighbourhood crime to be 0.69 (0.000) and 0.14 (0.000), respectively. Also, the impact of residential neighbourhood crime on property value gives regression weight and p-value at 0.47 and 0.000, respectively. The model fitness is further guaranteed by the R2, which stands at 52%. The interpretation of these results is that applying social development programmes to tackle the social risk factors and purposeful manipulation of the residential neighbourhood through design could go a long way to decrease neighbourhood crime and boost property values. This research serves as an awakening call to the Nigerian government, policymakers, and researchers to tackle property crime to ensure housing sustainability and property value appreciation, among others.
... Obtiene que el precio y los ingresos tienen efectos importantes en la elección del régimen de tenencia, mientras que otras variables socioeconómicas como la edad tienen un efecto más complejo. Kauko (2003) analiza el efecto sobre el precio de la vivienda de muchos atributos relativos a la localización, entre ellos el estatus de la vecindad, medido por el nivel de ingresos y educación. Obtiene que es un atributo importante en las decisiones de localización. ...
Conference Paper
Full-text available
rea Temática: 9. Ciudades, áreas metropolitanas, mega-regiones y redes. Resumen: Los modelos de estimación del precio de la vivienda se han orientado, generalmente, a analizar el impacto de las características intrínsecas a ésta. En el presente trabajo nos centramos en aspectos extrínsecos, concretamente, aquellos relativos a las características socioeconómicas de la vecindad. El objetivo es identificar la incidencia de características del entorno socioeconómico en el precio de la vivienda. Se trata de determinar qué variables nos serán de utilidad para la formación de un indicador del nivel socioeconómico. Los datos han sido recogidos de la Oficina de Estadística de Barcelona. En la investigación se parte de un número elevado de variables relativas a las características socioeconómicas, como el nivel de estudios, la tasa de ocupación, renta de la población, tipo de actividades económicas y precio de los locales comerciales, entre otras. A partir de la técnica de análisis factorial se seleccionarán, de entre todas las variables consideradas, las más relevantes, con el objetivo de definir un indicador operativo del nivel socio-económico. El propósito último del trabajo es la incorporación de dicho indicador, junto a otros determinantes del valor de la vivienda, en una estimación del precio basada en un modelo de Red Neuronal Artificial. Palabras Clave: Mercado inmobiliario, precios de la vivienda, entorno socioeconómico, análisis factorial, Barcelona Clasificación JEL: R31
... No matter which type of housing property investor chooses, location may be one of the most important considerations (Hassan, et al., 2021a). Kauko (2003) found that location has a strong correlation with the increase in housing property value and property investment behavior. If location becomes more sought-after, then well-located properties will show good returns (Sean et al., 2014). ...
There is a vast literature that seeks to define and identify spatial submarkets in metropolitan housing systems. These tend to use one of three methods to delineate submarkets: a priori geographies, ad-hoc sub-division and data-driven approaches to grouping units. Recently, analysts have increasingly used multilevel modelling strategies to analyse spatial segmentation in the housing market. Despite the increasing prevalence of multilevel approaches, there is no existing systematic analysis of which of these three main approaches to submarket definition has the greatest effectiveness when employed in a multilevel modelling framework. This paper addresses the gap in the literature by comparing the utility of these main approaches to submarket definition. It develops and evaluates three separate, distinct multilevel models of submarkets to a data set comprising of 2175 transactions in the Istanbul housing market, an emergent market context. The results show that multilevel models with a priori submarket dummy variable can predict price more accurately than the models with ad-hoc sub-division or data-driven stratified submarkets. Similarly, test results indicate that multilevel models with neighbourhood submarket dummy variables (a priori) perform better than other models. These test results show that granular definition of submarkets tend to perform better in terms of predictive accuracy than less spatially granular models. The paper also suggests that real estate agents’ views of submarket structures might be particularly useful as inputs into micro modelling processes in contexts where datasets are thin.
The paramount ingredient for the socio-economic development of any country is the land and buildings. An accurate appraisal of these land and buildings’ value has a colossal effect on the state’s economy. Still, the value of these buildings has been outrageous with realtors’ emergence. Various computational techniques have been employed to resolve the issue of the appraisal of the value. The factor analysis, correlation analysis and linear regression analysis are employed in this paper in order to model the residential house prices. This study is to be carried out in the Chengalpattu neighbourhood, where the modelling of the residential house prices is considered. All the factors which account for house price were determined. The modelling is performed by taking the market prices and the various factors that account for the valuation using various software.KeywordsHouse pricingRegressionUrban density
Full-text available
Recent research in theoretical statistics indicates that even if the data best fit a linear specification, a model is not necessarily correctly specified if estimated by ordinary least squares (OLS) regression analysis. Alternative estimation techniques have been developed that account for nonnormal residuals and result in more efficient estimators. This article reports the results of an empirical test to determine whether these alternatives provide better predictors of real property selling price compared to OLS. Based on several different criteria, the results indicate that for small samples the alternative formulation techniques do not produce statistically better predictions compared with those derived using OLS.
Executive Summary: A problem often exists when using traditional approaches to valuation in assessing damages to contaminated real property. Cost, income, and sales comparison approaches may understate the impact on value when historical and recent sales are limited or nonexistent, or when potential buyers are unaware of or not fully knowledgeable regarding the nature and extent of the contamination. When one or more of these conditions is present, reliance on traditional methods is limited in developing an opinion of damages for litigation. Three complementary survey-based techniques are presented as supplementary tools in estimating damages in impaired property situations: contingent valuation, conjoint analysis and perceived diminution. Due to the fact these approaches may be employed when conditions limit the use of other methods, the approaches presented are an important addition to the required body of knowledge for any analyst addressing impaired property value. This study briefly explores the theoretical foundations that underlie the use of these measures. It then addresses contingent valuation, conjoint analysis and perceived diminution methods applied to damage assessment of real property. Disguised examples of actual study materials and findings generated are employed.
This paper serves as an introduction to the Analytic Hierarchy Process - A multicriteria decision making approach in which factors are arranged in a hierarchic structure. The principles and the philosophy of the theory are summarized giving general background information of the type of measurement utilized, its properties and applications.
This paper develops a statistical method for defining housing submarkets. The method is applied using household survey data for Sydney and Melbourne, Australia. First, principal component analysis is used to extract a set of factors from the original variables for both local government area (LGA) data and a combined set of LGA and individual dwelling data. Second, factor scores are calculated and cluster analysis is used to determine the composition of housing submarkets. Third, hedonic price equations are estimated for each city as a whole, fora prioriclassifications of submarkets, and for submarkets defined by the cluster analysis. The weighted mean squared errors from the hedonic equations are used to compare alternative classifications of submarkets. In Melbourne, the classification derived from aKmeans clustering procedure on individual dwelling data is significantly better than classifications derived from all other methods of constructing housing submarkets. In some other cases, the statistical analysis produces submarkets that are better than thea prioriclassification, but the improvement is not significant.
This dissertation addresses the statistical aspects of neural networks and their usability for solving problems in economics and finance. Neural networks are discussed in a framework of modelling which is generally accepted in econometrics. Within this framework a neural network is regarded as a statistical technique that implements a model-free regression strategy. Model-free regression seems particularly useful in situations where economic theory cannot provide sensible model specifications. Neural networks are applied in three case studies: modelling house prices; predicting the production of new mortgage loans; predicting the foreign exchange rates. From these case studies is concluded that neural networks are a valuable addition to the econometrician's toolbox, but that they are no panacea.