Abstract
This study tests for the presence of negative bubbles in the REIT markets over the period 1972:01 to 2004:05 using the momentum threshold autoregressive (MTAR) model. There is evidence of asymmetric adjustment towards the long-run equilibrium between REIT prices and dividends indicative of negative bubbles for mortgage and hybrid REITs.
Notes
Researchers (see Case and Shiller (Citation1989, Citation1990), Kim and Suh (Citation1993), Abraham and Hendershott (Citation1996), Bjorklund and Soderberg (Citation1999), and Hendershott (Citation2000) for examples) have investigated whether or not real estate markets have experienced ‘excessive price volatility’ or an asset price bubble, however to the best of knowledge, an examination of periodically collapsing negative bubbles in the REITs markets has not been undertaken.
The following is quoted from Payne and Mohammadi (Citation2004): ‘Equity REITs hold at least 75 percent of their assets in income generating real estate properties. Mortgage REITs hold at least 75 percent of their assets in residential and commercial mortgages as well as short-and-long-term construction loans, acquisition of loans or indirect lending through mortgage-backed securities. Hybrid REITs share the properties of equity and mortgage REITs by both owning properties and lending to owners and operators.’
Taylor and Peel's (Citation1998) residuals augmented least squares Dickey-Fuller tests for periodically collapsing bubbles has been used in the examination of East Asian and Australian stock markets by Sarno and Taylor (Citation1999), asset price bubbles in Latin American emerging financial markets by Sarno and Taylor (Citation2003), and for stock price bubbles in France, Germany, Japan, the USA, and the UK by Capelle-Blancard and Raymond (Citation2004).
McCallum's (Citation1983, Citation1997) minimum state variable criterion could also be invoked to rule out bubbles.
The only other change from Evans's (Citation1991) model, besides the emphasis on the initial value B 0 is the inclusion of absolute values on the conditions in EquationEquations 3a and Equation3b.
Bohl (Citation2003) uses the MTAR model in examining periodically collapsing bubbles in the U stock market.
The Chan (Citation1993) method sorts the estimated residuals in ascending order, eliminating 15% of the largest and smallest values and is typically used in the selection of the threshold value. The idea of eliminating extreme values when testing for bubble behaviour seems counter-intuitive, thus no residuals are eliminated.