Bubbles in US regional house prices: evidence from house price–income ratios at the State level

ABSTRACT

We investigate the presence of bubbles in the US house price-income ratio at the State level by applying the recent time series-based econometric test to data from January 1975 to December 2014. We find evidence of bubbles in several States in the 1980s (i.e. California, Hawaii, Massachusetts, New York, etc.), which coincides with some existing studies that investigate housing bubbles or booms and busts using a range of alternative approaches. Our results show the existence of a housing bubble that originates in the early 2000s and collapses in the mid-2000s in more than 20 States and the District of Columbia concluding that the bubbles of the 2000s were more widespread than the 1980s, which is of special interest and importance. Our results seem to be in agreement with the talk given by Alan Greenspan in 2005, who suggest no sign of a nationwide housing bubble but a lot of local bubbles. We also study the importance of the regression model specification with/without an intercept and the regression model with an intercept could lead to false-positive identification of bubbles.

KEYWORDS:

• Bubbles
• generalized sup ADF test
• US regional house prices
• house price–income ratio

JEL CLASSIFICATION:

• C59
• R19
• R39

Acknowledgements

We thank two anonymous reviewers for their helpful comments. We would like to acknowledge helpful comments received from presentation of earlier versions of this article at the University of York and the New Zealand Econometric Study Group (NZESG), Dunedin, New Zealand. Particular thanks go to Professor Peter Phillips for discussions on the role of the intercept in the PSY test.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Intrinsic bubbles are driven solely by fundamentals (Froot and Obstfeld 1991).

² A number of researchers have applied the Chow and Lin’s (1971) GLS procedure that can provide the best linear unbiased interpolations in house prices or property markets (see Kenny (1999), Meese and Wallace (2003), Assenmacher and Gerlach (2008), Goodhart and Hofmann (2008) and Zhou (2010)).

³ A declining price–income ratio implies that the growth of house prices doesn’t outpace the growth in personal income. House prices could drop quite substantially in some States due to unfavourable expectations, low employment and economic recession during the 1980–1990s.

⁴ We assume that a housing bubble should last at least for 6 months. Thus, a potential bubble episode with shorter period is ignored. Figures 6, 8, 10, A1, A3, A5, A7, A9, A11, A13, A15 and A17 compare the backward SADF statistic with the $95\mathrm{\%}$ critical value sequences for the price–income ratio under the regression model with an intercept.

⁵ The critical values under the regression model with an intercept are 2.5343 ($90\mathrm{\%}$), 2.7960 ($95\mathrm{\%}$), 3.4337 ($99\mathrm{\%}$). The critical values under the regression model without an intercept are 3.4989 ($90\mathrm{\%}$), 3.8319 ($95\mathrm{\%}$), 4.5976 ($99\mathrm{\%}$).

⁶ Gabriel, Mattey, and Wascher (1999) explored the housing price patterns in California’s two largest Metropolitan areas (Los Angeles and San Francisco) prior to 2000.

⁷ Riddel (2011) estimated an error correction model that spanned 1978Q2 to 2008Q1 using quarterly housing price data for Las Vegas and Los Angeles with both national and regional economic variables. Riddel (2011) provided support for the contagion hypothesis that income and price in Los Angeles contributed to the run-up house values in Las Vegas from 2002 to 2006.

⁸ Case (1994) reviewed the house prices in the US since the 1950s at national and regional level and discussed the causes of house price behaviours across regions.

⁹ Gupta and Miller (2012) explored the cointegration relationships between house prices in eight Southern California metropolitan statistical areas (MSAs).

¹⁰ Zhou and Sornette (2008) defined a bubble as a price acceleration faster exponential. This definition is the same as Zhou and Sornette (2006).

¹¹ Wheelock (2006) summarized that US States experienced 20 house price booms between 1980 and 1999.

¹² During 1980–1982, Iowa, Wisconsin, West Virginia and Michigan ranked 42nd, 44th, 45th, and 50th, respectively, among all States in real personal income growth, and 45th, 40th, 48th and 50th in employment growth (Wheelock 2006).