Mobile Edition
Print
Get the Mag
Weekly UpdatesSocio-economic characteristics are composed of average income of the household, household size, the length of time the inhabitants have lived in Istanbul and the length of time the inhabitants have lived in Istanbul. In order to capture the neighbourhood quality characteristics, the satisfaction from schools, health services, cultural facilities, playground facilities, neighbour satisfaction, and neighbourhood quality are examined in this study. The neighbourhood quality characteristics (satisfaction levels) are measured on a 1 to 7 Likert scale, 1 being "appalling" response, and 7 being an "excellent" response. The locational factors gauge the urban structure based on the built and natural environment elements. The travel time to jobs, schools and shopping areas (or centres) are examined with the intention of measuring the transportation infrastructure. The earthquake risk percentage measurement has been taken into account and was derived from predictions by the JICA (Japanese Agency for International Cooperation) (IBB, 2007).
The dependent variable is based on the data collected from the real estate agencies, as explained in the data section. The following hedonic price function is employed to estimate the factors affecting housing prices:
P = [[beta].sub.0] + [[beta].sub.1][X.sub.1] + [[beta].sub.2][X.sub.2] + [[beta].sub.3][X.sub.3] + [[beta].sub.4][X.sub.4] + [epsilon] (1)
where: P is the vector of logarithm of trans action prices; [X.sub.1] is the vector of variables for property characteristics; [X.sub.2] is the vector of variables for socio-economic characteristics; [X.sub.3] is the vector of variables for neighbourhood quality characteristics, and [X.sub.4] is the vector of variables for locational factors. [[beta].sub.i] (i = 1, 2, 3, 4) is the vector of coefficients and [epsilon] is the error term. A log-linear functional form was employed because of the econometric problem arising from the occurrence of heteroscedacity in regression. Because the data from 348 submarkets with different characteristics are combined in the analysis, the errors are heteroscedastic. In order to reduce the error variance, a log-linear functional form was selected to improve the efficiency of parameter estimation (Rephann, 1998).
4. RESULTS FROM THE ISTANBUL HOUSING MARKET
The results of the hedonic price model are presented in Table 4. The overall [R.sup.2] is 0.609 which compares well with others reported in the literature (Malpezzi, 2003; Rothenberg et al., 1991).
A logarithmic functional form is employed in this study. Overall the model produces implicit prices that are reasonably consistent with a priori expectations of the likely signs and magnitude.
In terms of the property characteristics, living area in the housing unit has the largest impact on the housing price. A 1% increase in the living area of the housing unit will change the logarithm of the housing price by 0.0000645. The second most important variable among the property characteristics is site. This variable has been crucial since the 1999 Marmara Earthquake. High income level households have moved towards peripheral areas that have less earthquake damage risk with a solid ground formation and this tendency formed a phenomenon of gated communities with their own security, social and recreational facilities. The movements of high income group have been followed by the middle income group. Filtering has been assumed as shifts of households across dwelling qualities and changes in dwelling qualities (Rothenberg et. al., 1991). The middle income household group has preferred to live in the sites similar to the gated communities where there is a perceived high standard life quality.
The tendency to live in gated communities, or in sites, is not only because of the high life quality, existence of social and recreational facilities, but also because of the earthquake risk. The regulation system for construction of new buildings did not involve the high load bearing capacity construction rules before the 1999 Marmara Earthquake. This new regulation system and changing preferences of home purchasers mean that the supply side began to construct structurally higher load-bearing capacity buildings, and on more solid ground formations. A 1% increase in the earthquake risk percentage in a neighbourhood will have a significant impact on house prices. Since the Marmara Earthquake in 1999, inhabitants also prefer to live in low storey buildings as it is perceived that they will cause less damage. As a result most of the gated communities have detached houses.
In comparison to most studies on housing prices, age has an unusual sign. A 1% increase in the age of the housing unit will lower housing price. Similar results for Istanbul were found by Ozus et. al. (2007), and Onder et. al (2004). It is argued that as the average age of housing units in a neighbourhood increases, it is expected that there will be more social and recreational facilities, and public investments such as schools. This result is also related with the variable "Living Period in Istanbul (the length of time the inhabitants have lived in the city)" in the socio-economic characteristics group because as the length of time the inhabitants have lived in Istanbul raise, the housing values also increase. Not only public facilities but also class concerns of the home buyer's causes such a result. The original inhabitants in Istanbul seek to avoid the ghetto areas where new migrants locate. As the income increases the housing values rise too.
Neighbourhood quality is also important and neighbour satisfaction is a significant variable. Previous studies have showed that that individuals prefer to live near others like themselves and decisions about whether or not to move and where to locate are influenced by a perception of the behaviour and characteristics of the current and potential neighbours (Ioannides, 2002). Interestingly, despite of the insights of access-space theory, the travel time to work does not affect values significantly. The reason for that unexpected result may be because of the polycentric structure of Istanbul. This finding is similar to others where there has been a rise in the spatial pull of several of the subcenters in the region of Los Angeles County (Richardson et. al., 1990) that has a polycentric urban pattern like Istanbul.
5. CONCLUSIONS AND FURTHER RESEARCH
This paper reports on the first stage of a larger research project. This research project seeks to build on the existing studies of the Istanbul market. Specifically the research aims to develop a model of house prices that captures neighbourhood-level price differences. The research employs a multi-level modelling framework as the main analytical tool. The results of the multi-level model are examined in several ways. First, the results are compared to those generated by two different forms of the standard hedonic model. The first hedonic model estimates house prices within Istanbul, but largely ignores neighbourhood differences. The second model includes neighbourhood dummy variables as a proxy for submarkets within the model. This analysis compares the estimated coefficients, significance and explanatory power of the models. Secondly, the spatial pattern of the residuals will be explored. This analysis will use GIS techniques to systematically examine the weaknesses of the different modelling approaches. However, at present, much of this research is still at the development stage. The present paper reports the results of the basic hedonic model, which were then compared with other published studies of Istanbul.
In this research, it is aimed to find out the determinants of the housing prices in Istanbul. A market-wide hedonic price model is employed by taking into property characteristics, socio-economic characteristics, neighbourhood quality characteristics and locational characteristics. The dataset used for this hedonic model is composed of two dataset. The data of property characteristics is provided from two major real estate agent's websites and this data set contains 2,175 transactions of single-family homes sold in Istanbul in November 2006 and in April. The second dataset provides information about the socio-economic and the neighbourhood quality characteristics. This dataset is derived from a survey that was undertaken by Istanbul Greater Municipality. The data of the locational characteristics such as travel time to jobs, schools and shopping areas (or centres) are taken from the second data set. The earthquake risk percentage measurement which is one of the most important locational characteristics is taken into account from predictions by the JICA (Japanese Agency for International Cooperation) (IBB, 2007).
The results of the hedonic model suggest that the housing price is determined by four types of characteristics: property, socio-economic, neighbourhood quality and locational characteristics. Among the property characteristics, living area being in a low storey building, being in a secured site (with swimming pool and garage), are found to have a positive impact on housing value. On the contrary to most studies on housing prices, age has a counterintuitive sign. Such similar results for Istanbul were found by Ozus et al. (2007) and Onder et al. (2004). Among the socio-economic characteristics, the length of time the inhabitants have lived in Istanbul, average income of the household and neighbour satisfaction, as a variable in the behaviour characteristics, have positive impacts on housing value. As expected, earthquake risk as a locational variable with a negative impact.
The results of this study also display the demand side preferences so that these can be used as a guide to improve the understanding within the supply side and investors. In addition to supply side and investors, policy makers and urban planners can use the results in order to analyze housing market behaviour.
FEED