Housing bubble detection using GSADF : exuberance in non-fundamentals

• Main Authors: Ong, Daphne Jia Lin, Heng, Yan Si, Wen, Yuanyuan
• Other Authors: Wang Wei Siang
• Format: Final Year Project
• Language: English
• Published: 2019
• Subjects: DRNTU::Social sciences::Economic theory::Macroeconomics
• Online Access: http://hdl.handle.net/10356/77191
• Institution: Nanyang Technological University

The SADF and GSADF tests have been widely used in empirical studies to identify bubbles. These tests provide a more consistent result and additionally provide date-stamping for bubbles as compared to traditional methods such as the cointegration test. Price-to-rent ratio was commonly used in the GSADF test to identify bubbles. However, most studies did not place importance on how housing prices are affected by the changes in macroeconomic variables. Therefore, this paper attempts to analyze the presence of housing bubbles while taking into account the changes in aggregate economic conditions. The Dividend-Ratio Model was used to decompose the log price-to-rent ratio into fundamental and non-fundamental components to reduce the false positive detection of bubbles. The fundamental component accounts for macroeconomic variables influencing housing prices, while the non-fundamental component captures irrational or speculative behaviours. After accounting for the market conditions, the GSADF test can then be used to identify irrational bubbles in the non-fundamental component. Carrying out the analysis, it was found that the GSADF test on log price-to-rent ratio had detected two bubble periods while the GSADF test on non-fundamental component did not detect any bubble period over the whole sample period in the Singapore’s housing market. Hence, the decomposition had greatly reduced the chance of over-identification of bubbles. Furthermore, the GSADF test was found to be more superior than the SADF test as it allows for the detection of multiple exuberances. Therefore, it was shown that the GSADF test was more accurate and superior than the SADF test in presence of multiple exuberances. In addition, the GSADF test carried out on the non-fundamental component of the log price-rent-ratio provides lesser room for over-identification of a bubble.