RENTAL VALUATION OF RESIDENTIAL PROPERTIES AROUND FEDERAL UNIVERSITY OF TECHNOLOGY, AKURE, NIGERIA: A GIS APPROACH

Abstract

There have been continuing efforts at ensuring more objectivity in property valuation. This has brought about different statistical techniques in property valuation by researchers. Therefore, this study considered the use of Multiple Regression Analysis (MRA) and Geographically Weighted Regression (GWR) in rental valuation of residential properties. Consideration was given to the spatially related nature of property in the analysis by using the GIS for the analysis, to ensure spatial referencing. Self-contained residential properties in the immediate neighborhood of the Southgate axis of the Federal University of Technology, Akure were considered in order to examine the impact of the proximity to the University Southgate axis on the residential properties. One hundred and ninety five self contained residential properties, which represent the total sample of the properties in the study area, were considered for the analysis of this study. A downloaded image from the google earth software ®, which was projected to the appropriate coordinate system, georeferenced and digitized was created as the shape file for the study area on the ArcGIS 10. 2 software. The self-contained residential properties were administered questionnaire to elicit data on the characteristics of each of the properties, which influence their rental values. A hand-held Global Positioning System (GPS) was also taken along to obtain the coordinates of each of the properties considered. The data were then coded on the ArcGIS 10.2 software and the analyses were carried out. The results of Multiple Regression Analysis show that all the property characteristics significantly predict the rental values of the properties at p < .050. However, due to the spatial nature of the variables, an analysis on the Local Indicator of Spatial Association (LISA) was carried out to examine the possible local spatial clustering or outlying among the independent variables. The results of LISA analysis show that there were statistically significant local associations among these pairs of the independent variables: ‘bathroom quality and toilet quality’; ‘floor finishing and wall finishing’; ‘window quality and door quality’, all at p < .005. Therefore, in order to avoid inflation of the results of the Multiple Regression Analysis, one variable from each of the pairs was selected to explain each spatially associated pair in recalibrating the Multiple Regression Analysis model. The results of the recalibrated Multiple Regression Analysis show that all the independent variable significantly predict the rental value of self contasined residential properties in the study area at p < 0.050. The Variance Inflation Factor (VIF) was < 5 for each of the independent variables. The Wald statistics result was e = 683.9395, p < 0.050, showing that there was statistically significant prediction. The Jarque-Bera statistics was not statistically significant, indicating the normality of the distribution of the residuals of the explanatory variables, following a Gaussian pattern. However, the Koenker (BP) test result was significant at p < 0.050, showing that the data were not spatially stationary. The significant result of Koenker (BP) test therefore qualifies the data for a spatially varying analysis and the Geographically Weighted Regression was used for the spatially varying analysis on GIS. The results from the GWR analysis show that there are differences in the prediction coefficient of the property characteristics per location. The predictive strength of each of the property characteristics at different locations on the study area were then produced on maps to show the variation. The predicted rental values of these properties were also presented on map to show the predictions made from the results of the Geographically Weighted Regression visually. The Geographically Weighted Regression results were also calibrated on the AcrGIS 10.2 for prediction of other self-contained residential properties in the study area. The results of the Multiple Regression Analysis and the Geographically Weighted Regression were then compared using the R2, AIC and Moran’s I value. The R2 of Multiple Regression Analysis was 69.1% while the R2 of Geographically Weighted Regression was 77.4%. The AIC value of the Multiple Regression was 3943.26 while that of Geographically Weighted Regression was 3880.44. Also, the Moran’s I value of the Multiple Regression Analysis was 0.3459 while that of Geographically Weighted Regression was 0.0253. These three tests show that the Geographically Weighted Regression results were better than that of Multiple Regression Analysis. However, the results of both Multiple Regression Analysis and Geographically Weighted Regression show that the distance from the University Southgate axis has negative effect on the rental values of the residential properties. The use of GIS for the analyses in this study gave opportunity to accommodate the spatial nature of property values and the property characteristics. It allowed the analysis and presentation of results on maps and also aided the prediction of the rental values of properties with known coordinates, especially with the Geographically Weighted Regression which produced better results than the Multiple Regression Analysis. It is therefore recommended that, considering the spatial nature of property, property valuers and researchers should adopt the use of Geographically Weighted Regression as a statistical technique for property valuation and also consider the use of GIS in property valuation, which enables the presentation of the analyses and results on map and aid visualization for better understanding.