Understanding real estate price dynamics: The case of housing prices in five major cities of China✰

Abstract

The developing technology of wavelet analysis offers a valuable tool in mitigating many of the limitations of earlier studies of housing price dynamics. This paper applies wavelet analysis to five first-tier cities in China to study housing price changes over time, to decompose housing prices into their trend and cycle components, and to explore co-movement and lead-lag relationships among these cities. We find the average cycle for all five cities to be 3.25 years, which is much shorter than the housing cycles observed in the United States. When we examine the cyclical lead-lag relationships among these cities, we find that during the 2008–2011 period, Shenzhen led Beijing, which led Guangzhou, which led Shanghai, and finally Tianjin followed. However, during the 2011–2014 period, the lead-lag relationships changed to Tianjin leading Shenzhen, then Shanghai, then Beijing, and finally Guangzhou. Although we generally observe a strong co-movement among the city pairs, the co-movement between Tianjin and each of the remaining four cities is weak. The weaker correlation between Tianjin and other cities indicates that real estate investors in these other four cities can improve their risk-return performance by adding Tianjin properties to their portfolios.