Advances in Property Price Modeling

ADVANCES IN PROPERTY PRICE MODELING

Dr.Taher Buyong

Spatial and Numerical Modeling Laboratory

Institute of Advanced Technology

Universiti Putra Malaysia

43400 UPM, Serdang

ABSTRACT

Current trends in mass appraisal of properties are the use of the MRA model. MRA has severe limitations in that it ignores spatial dependence and spatial heterogeneity; two spatial effects inherent in property data. As a result, residuals of MRA display spatial autocorrelation and heteroscedasticity that violated the presumption of OLS estimation used in MRA and may lead to biased estimated coefficients. Consequently, the accuracy of predicted property prices may be questioned. Several advanced models have emerged that try to overcome the limitations of MRA that are grouped as spatial, local and geostatistical models. These advanced models that incorporate spatial effects are proven to be superior to MRA in predictive performance in many studies and one of them should be suitable to be used in the mass appraisal of properties in the country. This paper demonstrates the problem of MRA and the selection of the most appropriate model, using MRA as the benchmark, to deal with a rent and transaction datasets. Problems in regressing residential property data are also highlighted.

Keywords: Mass appraisal, rating valuation, spatial models, local models, geographically weighted regression, spatial prediction.