Скачать 133.11 Kb.
The Residential Market of Hong Kong: Rational or Irrational?
University of Aberdeen Business School/Property, Edward Wright Building, Dunbar Street, Old Aberdeen, AB24 3QY
Email: firstname.lastname@example.org Tel: (44) (1224) 272361 Fax: (44) (1224) 272181
Division of Economics, School of Humanities & Social Sciences, Nanyang Technological University, Nanyang Avenue, Singapore 639798
AYULIU@ntu.edu.sg , Tel: (65) 67904949, Fax: (65) 67924217
The current study attempts to investigate the proposition that Hong Kong residential market is only driven by a rational speculative bubble, in addition to fundamentals. The fundamentals are chosen according to the present value model, but will account for latent private information. Potential roles of other public information are also explored. The study finds that the influence of the rational bubble on the price growth is highly significant. However, in contrast to Seoul and Singapore housing market (Xiao, 2006; Xiao and Huang, 2007), neither the fundamentals, nor the rational bubble can explain much of the price growth in the market of concern. This finding leaves large room for questioning whether or not this market is more prone to irrationality than its counterparts in Seoul and Singapore.
JEL classification: G12, C13, C52
Keywords: Speculative bubble, rational, irrational, latent information
The property price of Hong Kong is one of the most volatile in the world. Many suspect that such observed price behavior is driven by speculative bubbles. Nevertheless, there have been relatively few research papers devoted to this topic [Xiao, 2005]. In this study, the authors will examine four classes of Hong Kong residential properties: domestic premise class A, B, C and D. These are defined as residential properties with saleable area less than 40 m2, 40-69.9 m2, 70-99.9 m2, and over 100 m2 respectively. The time series are CPI (housing)-deflated monthly price and rent indices from Mar. 1980 to Feb.20062.
The study is first motivated by the observed movement of the price-rent ratio. The present value model suggests that the price and the rent indices should move more or less hand in hand. But instead, the price-rent ratio followed the price movement closely. The second motivation is the observed large discrepancies in price growth across different classes. During the two early expansions in the sample period, occurred in Jan.1991 - Apr. 1994 and Jan. 1996 - Jun. 1997, the luxury end of the market significantly outperformed the lower end. In particular, class D went through the most impressive price growth relative to other classes, especially class A. Between Jan. 1991 and Apr. 1994, class D price rose by 145.68% (in real terms, same for the rest), which is almost three times the price growth in class A. After the Asian Financial Crisis, the market tumbled all the way down. By September 1998, all classes shed more than half of the values. The property market, however, embarked on a remarkable recovery in mid 2003. In this recent expansion, class D again greatly outperformed class A (figure1, table 1). If these price growths are driven by economic fundamentals other than the rent, the impact of these fundamentals on different classes should have been similar. Do these observations imply the existence of a speculative bubble? If so, however much of that is rational?
A bubble exists in a price if the price is other than what is warranted by its fundamentals. The issue of speculative bubble arises because of uncertainties surrounding the assessment of fundamentals. Investors, in assessing the fundamentals, unavoidably have to make use of subjective probabilities of future events, because the relevant actual probabilities are not available by nature. Due to the difficulties associated with the assessment of the fundamental value, economists sometimes define bubbles differently. To them, a speculative bubble does not exist if there is no profit opportunity to be exploited. Skeptics of speculative bubbles argue that investors are rational and markets efficient, hence, speculative bubbles are, theoretically, implausible3. This is because that, when a market is efficient, the current price will fully and correctly reflect all relevant information. There exists no profit opportunity to be exploited, hence no incentive for speculation.
There is, however, no consensus on whether or not markets are efficient. An efficient market requires expectations to be formed rationally. If expectations are “backward looking” instead, it is possible to predict prices based on past trends, and excess profits could be earned. There are considerable evidences of backward looking expectations in the formation of asset prices, including real estate prices (Ito and Iwaisako, 1995, Cho, 1996; Malpezzi and Wachter, 2002; Schnusendberg, 1999; Lee and Ward, 2000). But evidences of profit opportunity did not emerge until recently. Daniel and Titman (2000) show that a meta-strategy, that mechanically picks strategies based on the performance of stocks in the previous 10 years, has continued to earn consistent profits. Lee and Ward (2000) also found unexploited profit opportunities in the UK real estate market.
To verify the existence of a speculative bubble empirically is, nevertheless, a thorny issue. This is because that, in an empirical study, a bubble is typically inferred, rather than directly observed, from a fundamental model4. Such a fundamental model itself is often a subject of dispute. Sometimes, the disputes arise about what the fundamental is; sometimes, the disputes center on the specification of a model (Flood et al., 1986). Even if economists can agree that a speculative bubble exists, they are often divided about the characteristics of the bubble. To Robert Shiller, “The idea that there has been a speculative bubble … is inherently a statement about some less-than-rational aspect of investor behavior” (Shiller, 2001). Many economists are, however, uncomfortable with the notion of irrationality. Perhaps because of that, there are far more theoretical and empirical studies devoted to the rational speculative bubbles than to the irrational ones (Flood and Garber, 1980; Brock, 1982; Tirole, 1982, 1985; Blanchard and Watson, 1982; Obstfeld and Rogoff, 1983; Mussa, 1984; O’Connell and Zeldes, 1988; Diba and Grossman, 1988a; Taylor and Peel, 1998; Basile, and Joyce, 2001).
This study will adopt the present value approach which is well accepted in the financial economics literature, but will allow for information which is available to the investors but latent to the researchers. The unobserved information is inferred using Kalman filter. This filter assumes normality of data. Unlike daily stock returns which has high kurtosis, monthly returns on real estate are far close to normal, hence the use of this filter is justified in the current context (see Xiao, 2007). The role of other public information on expectation formation is also explored.
The present value model was initially developed for the stock market. The property market has a number of features which differ from those of the stock market. For example, it involves high transaction cost, it suffers from restrictions in supply in cosmopolitan cities, and its institutional arrangements differ from those for the stock market (Xiao, 2005). However, both stocks and properties are assets which store the wealth of the economy. As such, the decisions to hold one asset or the other must have a common ground: the expected return to the asset. This is the rationale behind the used of the present value model in this study.
In addition, a rational bubble proxy is also explicitly incorporated into the present value model. The use of that proxy is based on Shiller’s (1990) positive feedback theory. The study finds that the influence of the rational bubble on price growth is highly significant. However, in contrast to Seoul and Singapore housing market (Xiao, 2006; Xiao and Huang, 2007), neither the fundamentals, nor the rational bubble can explain much of the price growth in the market of concern. This finding leaves large room to question whether or not the residential market of Hong Kong is more prone to irrationality than its counterparts in Seoul and Singapore.
The real-estate market has the longest and the most reliable history of boom and bust. That history stretches back to the early 1800s (Carrigan, 2004). Because of the great uncertainties surrounding its fundamental value, real estate has always been a major object of speculation.
The Western economists have long been fascinated by the speculative bubble in the real estate market. Abraham and Hendershott (1994) investigated 30 cities in the United states. In their model, Abraham and Hendershott incorporate a proxy for the tendency of a bubble to burst and a proxy for the tendency of a bubble to swell5. They found that the proxy does indeed work and is especially useful in explaining the large cyclical swings in real housing prices in the West. The lagged appreciation term, which represents speculative pressures in the market, performs admirably in soaking up volatility. Scott (1990) and Brooks et al. (2001) employ variance-bound tests to test the rationality of real-estate share prices. Scott analyzes price indices of 13 REITs. His sample stretches from the late 1960s and early 1970s to 1985. Brooks et al. examine the prices of U.K. property stocks. Both studies indicate the existence of irrational, speculative bubbles. Bjorklund and Soderberg (1999) examine the 1985–1994 cycle in the Swedish property market and contend that the ratio of property value to rent increased too much, indicating that a bubble may have existed.
South Korean economists have been very active on the topic of speculative bubbles in the country’s real estate market. Kim and Suh (1993) show that a bubble existed in the nominal and the relative land price between 1974 and 1989. Lee (1997) conducts a test of bubble in the land price between 1964 and 1994. Using a structural model with GNP, interest rate and money supply as fundamentals, he found the hypothesis that only market fundamental drove the land price in Korea can be rejected.
Kim and Lee (2000) adapt the idea that the existence of an equilibrium relationship excludes the possibility of a price bubble. They conclude that, in the long run, South Korea’s nominal and real land prices are cointegrated with market fundamentals (approximated by nominal and real GDP respectively). However, in the short run, such cointegration relationship does not exist. This is consistent with the notion that speculative bubbles are periodically collapsing (Blanchard and Watson, 1982). In the short run, speculative forces could drive prices away from market fundamentals. In the long run, fundamental forces will eventual reassert themselves. However, in the light of Taylor and Peel (1998), it is not clear how robust are their procedures.
Lim (2003) conducted two bubble tests based on the present value relation on the housing price of Korea, one is a modified volatility test (MRS test) suggested by Mankiew et al (1985), another combine unit-root test suggested by Diba and Grossman (1988b), and cointegration test by Campbell and Shiller (1987). Their MRS test show that the null hypothesis of market efficiency is rejected, indicating the existence of irrational bubble. Their unit-root test and cointegration test however suggest that bubbles do not exist! This is in contrast with the findings of Xiao (2005) in her PhD thesis, which employ a Markov-switching ADF approach. However, the data series employed in her thesis are not identical to those by Lim (2003).
The large swings of property prices in Japan in the late 1980s and early 1990s have intrigued many researchers. Ito and Iwaisako (1995) attempt to measure how much of the asset price variation observed in Japan in the late 1980s and the early 1990s can be attributed to changes in the “fundamentals”. The fundamental model they use is the standard present value model. Their conclusion is that “it seems impossible to offer a rational explanation of the asset price inflation in the second half of the 1980s by changes in fundamentals (page 10).” This point is reemphasized by the authors in a variance decomposition study (page 20).
To the best knowledge of the authors, there have been relatively few studies devoted to the speculative bubbles in Hong Kong property market, even the experiences of this city suggest that it is one of the most interesting for this topic. Chan, et al’s (2000) is one of those. Their study uses a standard present value model with constant discount rate. Their model allows for a misspecification error, extracted using Durlauf and Hall (1989a,b). A rational bubble is then inferred from the model residuals. The data they use are monthly averaged rentals and quarterly averaged prices of the private domestic properties within the class A (Q1 1985 – Q3 1997). Their results show that there exists misspecification error in the model noises as well as bubbles. However, as the bubble is inferred from model residuals, it could well contain information not captured by the model. And they call that bubble rational, which could indeed be a mixture of rational and irrational ones.
If economic agents are risk-neutral, the real price of one property, Pt, will be equal to the expected discounted present value of the real rent flows during the ownership period, Dt, plus the price at which the property can be sold at the end of the ownership period, Pt.+1, Mathematically,
Where Rt = the time-varying real discount rate. In a static world, it can be shown that
Where c is a constant, a discount factor for time preference, Et[.] a rational expectation operator, and pt and dt are the logs of Pt and Dt. If the transversality condition,, is satisfied, we would have the fundamental solution:
which is simply a function of the expected future rent flows. However, the transversality condition may fail to hold. In such a case, we would expect the price to contain a rational speculative bubble,
Suppose the log of the property price and the log of the rent are integrated of order one; and suppose the growth of the rent follows an AR(p) process. Then the following model may be estimated instead: