Dynamic Relationships among Composite Property Prices of Major Chinese Cities: Contemporaneous Causality through Vector Error Corrections and Directed Acyclic Graphs

Abstract

The present study is the first one that investigates dynamic relations among composite real estate price indices of ten different cities in China during the years from 2005 to 2021. Utilizing the data recorded on a monthly basis, we apply VECM (vector error-correction modeling) and DAGs (directed acyclic graphs) in order to characterize contemporaneous causal relations among the ten real estate price indices. We use the PC algorithm to identify a pattern with non-directed edges and the LiNGAM algorithm to determine the causal ordering, based on which we calculate the results of innovation accounting. The LiNGAM algorithm adopted here effectively utilizes non-normality for facilitating the arrival of complete causal orderings. Our results show that price dynamics revealed through processes of price adjustments due to shocks to prices are rather sophisticated and such dynamics are, in general, dominated by price indices of Shanghai and Shenzhen, which are two top-tier cities among the four top-tier cities in China. This indicates that policy design on composite property prices should be focusing on price indices of Shanghai and Shenzhen.