Sustainable development scale of housing estates: An economic assessment using machine learning approach

Abstract

Economic sustainability is often addressed from the perspective of economic growth and at the national level. In contrast, this research attempts to examine the question of economic sustainability of human settlement at a local project level. Urban planners need to strike a balance between dispersal and over-concentration of population in cities. The existing theories suggest that either excessively low or extremely high levels of household concentration is undesirable to a neighborhood. In this study, an economic assessment using machine learning (ML) techniques is used to identify the threshold scale of a housing estate, which comprises many privately owned residential units (like condos) with shared amenities. Using two decades of property transaction data in Hong Kong as our evidence, this study has found that a tipping point exists in the development scale of these housing estates. Housing values initially rise with the number of residential units in a housing estate but gradually fall when it increases beyond a critical limit. This nonlinear economic relationship is attributed to the per household share of common facilities, which does not increase sufficiently to match with the growing population density of the housing estates. The policy implication is that, to optimize housing supply, urban planning should not just focus on increasing the development bulk of housing but should also pay attention to the possible bottlenecks in the provision of shared amenities in the neighborhood.