Publication: Spatial hedonic pricing model for real property valuation in jordan using geographic weighted regression
Date
2020-08-01
Authors
Adwan, Zubeida Ali Al
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Property valuation or assessment is a compulsory property tax activity to be imposed
on all properties in Jordan. Currently, the lengthy, time-consuming and costly
processes have been performed manually. Furthermore, the assessment needs to be
updated from time to time in order to keep up with the current market value. As such,
there is an increasing need to develop alternative valuation models that can estimate
large-scale property values in a short time with little workforce and low cost.
Hedonic property price models are increasingly used in economic methods that are
used to assess properties automatically. Real property is a group of characteristics
and utilities where the property parameters are related to the estimated total value of
the transaction. By collecting value-added data on different buildings, a regression
analysis can be used to determine the relationship or correlation between each of the
characteristics of the valued price transaction. This study applies a spatial hedonic
pricing model to the real estate market in Amman Jordan. The Geographic Weighted
Regression (GWR) was used within the framework of GIS to correlate the adopted
thirteen properties namely structural, location and neighborhood characteristics with
their corresponding price and to obtain results in the form of reports and maps. As a
first step in the GWR procedure, the exploratory ordinary least square (OLS)
regression was carried out and the redundant property parameters were excluded.
After the implementation of the GWR, coefficients showing the effects of property
variables were identified, their values and standard residuals were mapped and
analyzed. The random spatial distribution of the standard residuals shows that there
is no spatial clustering in these residuals and therefore the model estimation is out of
multicollinearity. Noticeable spatial variations in the values of both coefficients and
their standard residuals were investigated. The predictivity test also shows spatial
variation of the model predictivity goodness from location to location. The results
verified the need for local models to measure spatial variations, as it was evident that
spatial heterogeneity was revealed in output coefficient raster surfaces. The study has
also suggested that more property variables be included in order to enhance the
specified price model reality.