A comparison of DEA and statistical regression analysis in exploring property value effects of ferry terminals in Brisbane

Abstract

Property value effects of urban river ferries that run linear services to multiple stops are not well understood. Brisbane's CityCat, CityHopper and CityFerries together service 24 terminals in a connected system. This paper employs the Data Envelopment Analysis (DEA) approach to measure the property value effects of that system. The prediction performance of DEA is then compared with statistical regression methods. The research findings suggest that DEA has the capability of portfolio analysis to understand which types of properties benefit from the ferry service more. By contrast, the statistical regression models have the better flexibility in selecting the explanatory variables and greater power in identifying the magnitude of impact on property price from each of explanatory variable. This research sheds light on how DEA can be applied as a tool to evaluate the relationship between property value and price determinants, which has been conventionally undertaken by statistical regression methods.