A NOVEL USE OF LATENT CLASS MODELLING TO UNDERSTAND THE HETEROGENEITY OF URBAN LAND USE EFFICIENCY IN CHINA
Keywords: Statistical modelling, Urban modelling, Latent class analysis, Data-driven, Urban development, Land use
Abstract. Urban and regional development disparity is a salient issue for both developed and developing countries, and increasing development control brings the efficiency of urban land use to the forefront. Understanding the heterogeneity of urban land-use efficiency is therefore important for balancing growth-oriented land supply and sustainable development. However, two research gaps can be identified from existing studies. Firstly, the disparities of urban land-use efficiency are usually multi-faceted, the measurement of which is thus a major challenge. Secondly, a systematic investigation of land-use efficiency and the associated geographical heterogeneity using multidimensional, longitudinal measurements seems limited. Therefore, this paper presents a novel latent class modelling method for understanding the pattern of such heterogeneity using purposely constructed, repeated cross-sectional measurements, including local government revenue density and employment density. Using data of 272 prefecture-level Chinese cities, our model first identifies the level of urban land-use efficiency can be split into three distinct cities groups based on the performance in 2017. To further investigate the change of the efficiency over the years (2012–2017), two separate models have been applied and a total of five latent groups among Chinese cities are identified with distinct development patterns. This modelling approach represents a viable method for assessing the land-use efficacity, both statically and over time, across a large number of cities of varying development stages. Policy implications are drawn.