1、區隔或整合？台灣股票市場與不動產市場之互動Segmented or Integrated? The Interaction between Taiwan Stock Market and Real Estate Market 陳明吉、楊智元Ming-Chi Chen and Chih-Yuan YangAbstractAs the two main components of household portfolios, stocks and real estate are likely to catch peoples attention. Although the number of ext
2、ant studies on the interaction between the stock and real estate markets is large, the views and empirical evidence in those studies show inconsistent results. This paper provides an explanation for the inconsistent results: market imperfection. When the transaction benefit from the disequilibrium b
3、etween the stock and real estate markets can cover the potential cost resulting from market imperfection, the relationship between the stock and real estate markets is integrated. But when there is slight disequilibrium, the price of real estate will not converge since the trading benefit cannot cov
4、er the cost of transaction; as a result, the relationship is segmented. Keywords: market imperfection, stock market, real estate market, threshold vector error correction model (TVECM).11. IntroductionWhat is the relationship between the stock and real estate markets? Because stocks and real estate
5、are two of the main components of household portfolios, this question likely to catch peoples attentionand researchers are no exception. While prior literature focusing on the interaction between the stock and real estate markets is extensive, the views and empirical evidence in those studies show i
6、nconsistent results. One group of researchers has supported the view that there is a connection between these two asset markets (e.g. Liu and Mei (1992), and Li and Wang (1995), while a different group of economists has claimed that the two markets are separate and that such a connection does not ex
7、ist (e.g. Goodman (1978, 1981), Ibbotson and Siegel (1984), Geltner (1990), and Ross and Zisler (1991), and some studies even have shown mixed results (e.g. Liu, Hartzell, Greig, and Grissom (1990), and Ling and Naranjo (1999). These conflicting results cause this paper to wonder whether some key fa
8、ctors have been overlooked.It is unlikely that the stock and real estate markets are uncorrelated because of the wealth effect, investing substitution, and financial liberalization. The large decline in the stock and housing markets that was due to the subprime mortgage crisis of the late 2000s also
9、 reflects the possible connections between the equity and real estate markets. Why, then, have so many studies concluded that the stock and real estate markets are segmented? According to the reports of the Global Property Guide (2007, 2008a, 2008b, and 2009), the transaction costs of buying and sel
10、ling residential property can easily rise above 10% of the property value. For example, in the Philippines, the transaction costs of buying and selling a property are often 35% of the property value because of the 12% value-added tax (VAT). In Russia and Bulgaria, the explicit transaction costs can
11、reach around 25% of the property value for a typical transaction of a resale property. In addition, in the Southeast Asian housing markets, including Taiwan, Thailand, Indonesia, the Philippines, and South Korea, the total transaction costs are all above 10%. In addition, there are implicit costs, i
12、ncluding negotiation cost, search cost, and the cost of information asymmetry in the transaction of real estate. These considerable transaction costs, along with a high degree of market friction, make the real estate market imperfect.Imagine the effect of market imperfections. First, there will be p
13、rice rigidity in the real estate market such that price corrections or adjustments will lag (and be a prolonged process) because of the barrier caused by market friction and transaction 2costs. In addition, only when the trading benefits are enough to cover the potential costs caused by market imper
14、fection will the transaction be triggered, so the high cost from market friction will limit trading behavior, potentially inducing a threshold effect on the markets interactions. For example, when the real estate market slightly deviates from long-run equilibrium, the trader would give up the chance
15、 to invest because the trading profits would no longer cover the transaction costs. On the other hand, when the deviation is greater, the higher arbitrage space would attract investors to trade, and the prices of these markets would go toward their long-run equilibrium.Therefore, when the benefit fr
16、om the disequilibrium between the stock and real estate markets cannot cover the potential cost resulting from the market imperfection, the price of real estate will not adjust (price rigidity), resulting in the relationship between stock and real estate markets appearing to be segmented. However, w
17、hen the disequilibrium is serious enough to provide net profit in the presence of market imperfections, both stock prices and real estate prices will modify at the same time and there will be an appearance of market integration between stock and real estate returns.To verify this explanation and emp
18、irically examine the interaction between Taiwan stock and real estate markets, this paper employs the threshold vector error correction model. The sample period covers from 1973Q2 to 2009Q4. The empirical results show that, regardless of the size of disequilibrium, stock prices adjust toward equilib
19、rium between the stock and real estate markets. Only when the deviations of long-run equilibrium are large enough to exceed a threshold value do housing prices start to adjust. Therefore, when the deviations are larger, the stock and real estate markets are integrated, but when the deviations are sm
20、aller, the markets are segmented. The empirical results are very robust in that similar conclusions result when different proxies of housing price are used. This study contributes to the literature in several ways. First, the question of the relationship between the returns from two major assets hel
21、d by the private sector stock and real estateremains unanswered despite a large and growing literature on the interaction of financial and real estate economics. This study provides a reasonable explanation to reconcile the inconsistent results, and the empirical results provide additional evidence
22、to fill this gap in the literature. Second, considering the nonlinear interaction between stock and housing price movements, this study demonstrates the trade-off between trading profits and total costs, including explicit transaction cost and potential cost, caused by market 3imperfections. The con
23、vergence process triggers when deviations from the equilibrium of the housing and credit markets are large enough to cover the costs, indicating that the government can maintain a reasonable range of housing prices by decreasing the degree of market friction. Finally, our empirical results are of im
24、portance not only so investors and real estate companies can anticipate the variation of the financial environment, but also so government policy-makers can adjust monetary and fiscal policies correctly in response to the variations and linkages between markets. Otherwise, the asymmetric adjustment
25、and nonlinear interaction between the stock and real estate markets is likely to reinforce the boom-bust cycles in the economy and to increase the fragility of the financial sector. The remainder of this study is organized as follows. Section 2 reviews the literature concerning the relationship betw
26、een the stock and real estate markets. Section 3 describes the empirical model and econometric methodology. Section 4 introduces the Taiwan stock and real estate markets. Section 5 presents the data sources and reports the empirical results. Concluding remarks are in the final section. 2. The Relati
27、onship between the Stock and Real Estate MarketsBecause stocks and real estate are the two major assets held by the private sector (Chen, 2001), the beginning of the 1980s saw a number of attempts to validate the relationship between the prices of these two assets. Obviously, there are significant d
28、ifferences in the market characteristics, transaction processes, and regulatory limitations of the stock and real estate markets. Especially because of factors such as differences in location, architectural, and neighborhoods, real estate products have always had an important characteristic of heter
29、oskedasticity. Therefore, Liu, Hartzell, Greig, and Grissom (1990) contended that there is no direct flow of information between the real estate and stock markets because of legal restrictions and indirect barriers such as information costs, amounts, and quality, so the two markets are segmented, th
30、at is, unrelated.However, Gyourko and Keim (1992) presented a compelling argument that the real estate and stock markets would be expected to be co-related, since a large part of a companys value is tied up in real estate. For instance, Zeckhauser and Silverman (1983) found that about twenty-five pe
31、rcent of corporate value is associated with real estate. Under these circumstances it would appear that part of the risk in stock market returns is related to changes in the value of company-owned land and structures even though a portion of this risk may be uncorrelated with the firms core business
32、 risk. 4Broadly, then, the part of the real estate market risk associated with the general health of the economy should result in a positive correlation between property returns and returns on the stock market. Thus, there are common economic factors that may be expected to affect on the returns in
33、both markets (Okunev, Wilson, and Zurbruegg, 2000). Brueggeman, Chen, and Thibodeau (1984) indicated that, if real estate is considered an investment vehicle, real estate and other financial assets such as stocks and bonds would have a economically substitute relationship. Kim (1993) and Chen (1998)
34、 also contended that real estate and stocks have characteristics of investment substitutions and that a rising stock market drives a rise in investment substitution markets, resulting in a directly positive relationship between the prices of real estate and stocks. Moreover, according to the life cy
35、cle hypothesis proposed by Ando and Modigliani (1963), when stock prices rise beyond expectations (or fall less than expectations), investors wealth increases and, as a result, their consumption levels also rise. Green (2002) inferred from this hypothesis that the prices of stocks and real estate ar
36、e correlated since homes are a kind of consumer goods, at least as they relate to the wealthy. These two viewpointsinvestment substitution and wealth effectssuggest that there is a positive relationship between stock and real estate markets, at least to some degree.However, from the viewpoint of an
37、individuals portfolio allocation, given a fixed amount invested in funds, investing more money in real estate market must reduce the investment in the stock market. Therefore, an oncoming rise in the stock market would drive the demand for stocks and attract more cash from other markets, resulting i
38、n less money invested in real estate market. Such an “elbow-giving effect”1 would result in a negative relationship between the prices of real estate and stocks.Therefore, based on previous literature, it is unclear whether the real estate and stock markets are segmented or integrated and to what ex
39、tent, if any, they are related. Clearly, depending upon the methodologies, data and sample periods used, researchers have derived different conclusions to the question of the relationship between the real estate and stock markets. 1 Some studies, e.g., Yu and Tzeng (2003), have used the term “crowdi
40、ng-out effect” to describe this effect. However, in economics, the crowding out effect has been used to indicate the reduction in private consumption or investment that occurs because of an increase in government spending. To avoid confusion, this study uses the term “elbow-giving effect” to describ
41、e the trade-off of investment asset allocation.53. Empirical Model and Econometric Methodology3.1 Stochastic Processes of Stock and Real Estate ReturnsTo set up a model that describes the relationship between stock and real estate markets, we start with the stochastic process of stock returns. Becau
42、se of the uncertainty, but mean reversions, of stock prices, a Geometric Brownian Motion (GBM) is usually used as a reasonable approximation of stock price dynamics, as in equation (3.1):, (3.1)tdwttSdSwhere is the stock returns; and is a Wiener Process with mean of tdwzero and unit variance. The dr
43、ift parameter represents the expected rate of stock Sreturn, which relates to the risk of stock, the interest rate level, and the degree of risk-averse of investors. The diffusion parameter represents the volatility of stock returns. This process can also be used in the real estate market, so the ho
44、using return is:, (3.2) tdwttREdREwhere is the return of real estate, while and are the expected t REREreturn and volatility of the real estate market, respectively.3.2 Model of the Interactive Relationship between Stock and Real Estate ReturnsA model used to examine whether the real estate market i
45、s affected by the movements in the stock market may be described by the models used by Amihud and Mendelson (1987) and Chiang et al. (1995), as follows:, (3.3)( )( )( tdwtREtStdREwhere is the instantaneous change in the real estate index, and t )(tS)(tREare the values of the stock and real estate in
46、dexes, is the speed of adjustment coefficient, is the standard deviation of per unit time, and is a )(tdREtdwWiener process with mean of zero and unit variance.6The appealing intuition behind Eq. (3.3) is that movements in the real estate and stock markets occur together in time but, for two reasons
47、, the process does not satisfactorily describe movements in the real estate index. First, the process permits the value of the real estate index to be negative. Second, the volatility of changes in the real estate index is assumed to be constant, while empirical evidence suggests that it is not. (Se
48、e, for example, Abraham and Hendershott (1996). Therefore, Okunev and Wilson (1997) proposed the following model to overcome these deficiencies while retaining the property of mean reversion between the two markets.Assume that the relationship between the real estate and stock markets is, (3.4)kteRE
49、tS)()(where is some parameter and k is a constant. Taking the logarithms of Eq. (3.4) gives. (3.5)(log)(logtREktSThis type of formulation has been commonly used in testing for a linear relationship between two variables. If is a stationary process, then Eq. (3.4) )(trepresents the long-term cointegration relationship. Okunev and Wilson (1997) assume follows an Ornstein Uhlenbeck (O-U) process, which is written in terms )(tof and given by:,
