Abstract. Identifying explosive bubbles that are characterized by periodically collapsing phenomenon over time has been a major concern among researchers in finance and marketing. The paper advances a new date-stamping strategy for identifying the existence, as well as the origination and termination points of multiple bubbles, by using the generalized version of the sup ADF test (GSADF, hereafter) backwardly, which was originally introduced by Phillips, Shi and Yu (PSY, hereafter, 2012). This new method is more effective in identifying the existence of bubbles than the date-stamping strategy (SADF) when multiple bubbles occur. The backward GSADF test is as a rolling window of the ADF test with a double-sup window selection criterion that significantly improves its discriminatory power in simulations. This paper, thus, employs the backward GSADF test to identify housing bubbles in China by using the quarterly data of the real estate price index of 35 large and medium cities from 1998Q1 to 2013Q4. After running the test, we find statistical evidence of real estate bubbles that exist in the price indexes of Beijing, Tianjin, Shanghai, Yinchuan, Shenzhen, Urumqi, and Shenyang. These bubbles are either short-term bubbles or extremely short-term bubbles. We also discover that the ratio of city with housing bubbles is higher in the eastern region than in other regions in China.

Keywords: Date-stamping strategy, Generalized sup ADF test, Multiple bubbles, Real Estate Price Index.