ارتباط متغیر با زمان بین بازده بازار سهام و بازده دارایی واقعی

Direct investment in commercial or residential real estate is found to provide valuable diversification benefits for Australian investors though this is not so evident for indirect real estate investment vehicles like listed Australian real estate investment trusts (A-REIT). Further, multivariate analysis of Australian real estate and share market quarterly returns, spanning the period from the 3rd quarter 1986 to the 3rd quarter 2009, suggest that the correlation between real estate returns and share market returns is time-varying. Finally, while all of the asset class correlation coefficients increased with the Global Financial Crisis period this broad movement in asset class correlation is not evident in during the Wall Street Crash of 1987.

Real estate forms an important asset class for mutual funds both locally and globally (Higgins, 2007), accounting for around 10% of UK portfolio investment, around 5% of US portfolio investment (Blake et al., 1999) and close to 10% for Australian mutual funds.2 Yet, it has been argued that the large institutions, particularly pension and superannuation funds, could benefit from even greater exposure to real estate investment (Blake et al., 1999, Brounen and Eichholtz, 2003, Hudson-Wilson et al., 2003 and Lee and Stevenson, 2005). This is very much an empirical question determined by the correlation between available real estate investments and existing asset classes that make up pension and superannuation fund portfolios. There are two key questions that we address in this paper. The first question is whether direct investment in commercial or residential real estate improves upon diversification benefits available in the share market. Unconditional correlation between returns on direct real estate investment and returns on shares suggests considerable diversification benefits as this is quite low but the critical factor for investors is whether this correlation will change over time: in particular, whether the correlation increases around crisis periods like the Wall Street Crash and the more recent global financial crisis. The destruction of diversification benefits achieved by share portfolios around financial crises is well known (Brooks and Del Negro, 2004, Campa and Fernandes, 2006 and Caporale et al., 2005) but little is known about the impact of other asset classes on portfolio diversification during these turbulent periods, particularly real estate. The second question concerns whether listed real estate investment trusts provide the same level of diversification potential as direct property investment. We use Australian share price returns and three classes of Australian real estate returns in analysis of time variation in correlation between these asset classes in an effort to better understand the quite low levels of institutional real estate investment noted in the literature. To date, Australian real estate research has focused on describing the market (Higgins, 2007), modelling the determinants of real house prices (Abelson et al., 2005 and Bodman and Crosby, 2003), surveying Australian fund manager attitude to real estate investment funds (Keng, 2004) and assessing the performance of Australian listed A-REITs (Higgins and Ng, 2009). We extend this literature by analysing the time-varying correlation that exists between returns on different classes of real estate investment and share price returns. Real estate, as an asset class, is perhaps more complex than might initially be thought, with at least two distinct sub-classes: commercial real estate and residential real estate. Further, investment can take place either through direct investment or indirect investment. Direct real estate investment is achieved through purchase of property while indirect investment is normally achieved through acquisition of shares or units in listed or unlisted entities that hold property (e.g. Australian real estate investment trusts or A-REITs). Both liquidity costs and transaction costs are an important consideration in direct investment in either residential real estate or commercial real estate. Indeed, it is possible that these costs could negate the benefits of diversification in some markets. There are also problems with the valuation of direct real estate investment because real estate tends to be held for long periods of time and is rather heterogeneous in nature. Thus, because the market value of real estate is infrequently observed, appraisal based valuation is commonly relied upon for real estate portfolios. Nevertheless, there is a range of corrections available for these estimation issues (Byrne and Lee, 1995, Cho et al., 2003, Geltner and Goetzmann, 2000, Georgiev et al., 2003, MacGregor and Nanthakumaran, 1992 and Newell and MacFarlane, 1996). Indirect investment offers a more liquid real estate investment but there is also some scepticism in the literature about the performance of this asset class (Brounen and Eichholtz, 2003, Byrne and Lee, 1995, Clayton and MacKinnon, 2001, Feldman, 2003 and Georgiev et al., 2003). Estimation issues, arising from share market microstructure effects, could also affect returns calculated for this class of real estate investment (Brounen and Eichholtz, 2003 and Georgiev et al., 2003). While we do not attempt to remove equity market microstructure effects, due to the arbitrary nature of these corrections, we do correct for serial correlation induced by appraisal based valuation and stale prices in our final model specification, using a simple time series adjustment. A major contribution of this paper is in the analysis of time-varying correlation between returns generated by the three real estate investment sub-classes and Australian share market returns. It is well known that share returns tend to move together both over time and across markets and there is some evidence of increasing correlation during crisis periods. We extend the present literature through multivariate estimation of the correlations between Australian real estate returns and the Australian share market returns to assess whether a similar relation exists across these two asset classes. While the correlation between share returns and returns on direct real estate investment is quite low, the correlation is considerably higher for A-REITs. It appears that investment in indirect forms of real estate such as A-REITs offers more limited diversification benefit relative to direct real estate investment. Finally, our study provides further incentive for the use of multivariate GARCH (MGARCH) models in analysis changes in correlation over time. This approach can provide considerable insight into the variation that can take place in conditional correlation estimates over time (McClain et al., 1996 and Moschini and Myers, 2002).

It has been argued in the literature that combining direct investment in commercial or residential real estate with investment in shares could provide considerable diversification benefits. The results reported here support this argument while also suggesting the need for some caution, particularly with respect to event like the global financial crisis. The Global Financial Crisis period is of particular importance because correlations increased across all of the series included in this study. Thus diversification benefits from direct investment in commercial real estate were reduced as the value of commercial property dropped along with the falls in the value of A-REITS and the share market more generally. This was quite unlike the 1987 Wall Street Crash period where direct property investments were not as sensitive to the events that resulted in a stock market fall in value of around 25% on one day. At present, DCC models do not allow us to test the impact of these crisis periods within the conditional correlation equation. We leave this extension to the DCC model to future research. Consistent with criticism found in the literature, conditional correlations between A-REITs and share market returns are quite high and increased further during both the Wall Street Crash and the global financial crisis. These results suggest that indirect property investments like A-REITs, at least to some extent, behave more like shares than the underlying assets that they purport to mimic. It should be noted, though, that diversification benefits are still feasible even where correlation coefficients push through to values a high as 0.578. In this paper we explore the diversification benefits that can be achieved from adding real estate investment to a share portfolio and find fairly clear support for diversification into residential and commercial real estate, though the support for inclusion of A-REITs is less convincing. Regardless, we are unable to model the important costs that arise from direct investment in real estate markets, which tend to be less liquid than A-REITs. We expect that, in certain states of the world, these liquidity costs will far outweigh the additional benefits arising from diversification through direct property investment. Yet, while the question of liquidity is critical to an investor with a short investment horizon, the impact of liquidity may not be so critical for a large pension fund with a 20-to-30 year investment horizon. Investors with a longer investment horizon have more time to plan for the purchase and sale of real estate. Nevertheless, the decision to invest in real estate requires careful analysis of the costs and benefits of investment. Liquidity is not such an issue with listed securities like A-REITs but the evidence of diversification benefits is less clear cut when these securities are combined with a well-diversified portfolio of shares. The question of why A-REIT returns are more highly correlated with the share market returns is an important one and does require further analysis. It is possible that the differences in management structure, the differences in underlying legal structure and the differences in the ultimate investor group that trades in these securities all have an impact on the relative performance of the different classes of real estate investment, though further analysis of this question is beyond the scope of the paper due to a lack of appropriate data.