Modelling of commercial property market segmentation to improve price prediction accuracy in Malaysia

• Main Author: Usman, Hamza
• Format: Thesis
• Language: English; English; English
• Published: 2021
• Subjects: HF Commerce; T Technology (General)
• Online Access: http://eprints.uthm.edu.my/4799/1/24p%20HAMZA%20USMAN.pdf; http://eprints.uthm.edu.my/4799/2/HAMZA%20USMAN%20COPYRIGHT%20DECLARATION.pdf; http://eprints.uthm.edu.my/4799/3/HAMZA%20USMAN%20WATERMARK.pdf; http://eprints.uthm.edu.my/4799/

The commercial property market is strategic to the global economy. Significant attention is therefore given to its pricing by various stakeholders. The most common price modelling technique is the traditional hedonic price model. The commercial property market is too complex to be modelled by the traditional single equilibrium model. Property market segmentation models are used to improve the accuracy of price modelling, mostly reported in the housing market. This research, therefore, aims to propose a commercial property market segmentation model to improve price prediction accuracy in Malaysia. 14,043 commercial property transaction records obtained from Malaysia’s National Property Information Centre (NAPIC) was used. The submarkets were delineated using conventional hedonic, data-driven and spatial econometrics approaches. The evidence of submarket existence was determined using the Chow test and weighted RMSE, MAE and MAPE. The research found a significantly high level of spatial dependence in Malaysia’s commercial property market. Submarkets were efficiently delineated using all the methods except using submarket dummies. The research proposed the spatial error model using adaptive kernel maximum KNN spatial weight matrix as the optimal model for commercial property market segmentation in Malaysia. The proposed model improved the model fit by 19.76 per cent, reduced the RMSE, MAE and MAPE by 20.82 per cent, 24.63 per cent, and 25.92 per cent, respectively. The research shows that accounting for spatial dependence in the commercial property market reduces error, improves model fit and increases the accuracy of price modelling. The research has contributed to the existing body of knowledge by extending the commercial property market segmentation from a priori methods to the empirical data-driven and spatial econometrics approach in Malaysia. The implication to policymakers, financial institutions, the economy, property valuers, and property investors is that the findings will guide them in making informed decisions regarding the differentiated commercial property market.