Housing demand prediction in Penang, Malaysia

National Property Information Centre (NAPIC), Malaysia recorded that for first half of 2018 there was a decline about 37.8% on the residential stock especially for new planned supply and an increment of 18.2% on the overhang units. However, in Penang, it was estimated the demand to reach a total of 46,740 units due to the growing number of populations, formation of new households and the replacement of existing houses. Crucial housing challenge in Malaysia, especially in Penang is majorly related to a mismatch in demand and supply for affordable housing. The goal of this research is to predict a housing demand in Penang based on 4 categories of housing which are consist of low-cost, low-medium cost, medium cost, and high-cost houses based on household formation in Penang. The research focused on Penang due to its high and rapidly urbanized area compared to other states in Malaysia. The research used Census Data 2010 from Department of Statistic Malaysia to determine the headship rate. The Headship Rate Method applied to obtain the household formation. An extensive literature review was done on studies on housing demand and prediction method. A questionnaire was designed and approved by the expert before distributed to 400 households. The questionnaire was distributed to five districts in Penang and the respondents or household were divided according to 14 levels of age group. Multinomial Logit analysis was carried out to determine the choice of probability on house type selection and to produce a Choice Probabilities (CP) Model. The result showed that medium cost housing was the most preferable type of house with the highest CP value of 0.4875. Then, using data collected from Department of Statistics Malaysia, population prediction for 14 level age groups from years 1970 until years 2020 were determined using Double Exponential Smoothing method. Total housing demand in Penang was obtained by multiplying the population with headship. The actual and projected data for the prediction model was validated using MAPE and result show indirectly confidence level for 33.50% which categorized as acceptable. This prediction can assist local authorities, developers, consultants, contractors to plan which type of housing to construct based on demand in the future