Housing networks in urban China: A panel VAR model with Bayesian stochastic search

Abstract

This paper analyzes housing networks in China established through the inter-city information spillover. Unlike conventional methods that either fail to identify the city-level linkage or impose arbitrary restrictions, we investigate both the direction and magnitude of cross-city housing spillover in a data-driven manner by employing a panel VAR framework with Bayesian stochastic search. Through this, significance of the pair-linkages is examined from both perspectives of return and shock with insignificant ones being detected. Using a monthly dataset of housing prices covering 70 major China's cities over 2005–2021, our results confirm the presence of positive housing spillover that 11.68 % and 12.92 % of the total city-paired linkages are found to exist in terms of return and shock, respectively. With a particular focus on China's key urban agglomerations, housing spillovers exist not only across city-tiers in forms of ‘top-down’ and ‘bottom-up’ but also within the same tier in a ‘peer learning’ pattern, and the spillover magnitude of the latter is shown to be greater than that of the former. Despite the above common pattern, housing networks also demonstrate heterogeneous features across city clusters. Additional analyses reassure robustness of our findings that possess implications for clear comprehension of regional housing price dynamics.