تفاوت های عملکردی در نوع اموال متنوع در مقابل سرمایه گذاری املاک واقعی (REITs)

Evidence from the corporate finance literature indicates that diversified firms trade at a discount to otherwise comparable specialized firms. However, very little research has addressed whether a similar diversification discount might exist in equity REITs that diversify across property types relative to those specializing in one property type. Using a sample of 75 equity REITs, the existence of a property-type diversification discount is tested using standard Jensen's Alpha, Treynor Index, and Sharpe Ratio performance ranking methodologies over four commonly employed market proxies. Several variations of these standard tests are also utilized as robustness checks.

Do diversified REITs trade at a discount relative to specialized REITs? The study utilizes the performance ranking methodologies of Jensen (1968), Treynor (1965), and Sharpe (1966) to compare whether the stock market performance of REITs that have chosen to diversify by property type differs from the stock market performance of REITs that have become more specialized in their real estate holdings. Knowing whether benefits exist for property-type diversification within one REIT, versus allowing investors to diversify their own portfolios with REITs from several different property niches, is valuable information for all potential REIT investors, including the growing set of institutional REIT investors. Based on findings from the mainstream corporate finance literature, the expectation is that a REIT maintaining a concentrated focus on one property type will experience better performance than will a REIT choosing to diversify its property investments. At first glance, this expectation of poor performance for property-type diversified REITs relative to specialized counterparts might seem counterintuitive. After all, if a REIT has exposure to a number of property types, then returns are not subject as much to the cash flow variations of any particular property category. When residential property is performing poorly, office property might be performing particularly well, for example. However, the skills needed to effectively maximize profits from apartment complexes are quite different than the skills needed to maximize profits from a high-rise office building. Thus, when a REIT ventures into a new property type, the manager either has to become an expert on the new property type, hire an expert on the new property type, or settle for lower returns from that property type. Since the time commitment to become an expert is quite substantial, but the property is being considered now, the usual route is to hire an expert on the new property type. Clearly, this solution increases salary expenditures for the REIT, which could very well offset any potential diversification benefit. This question of differential performance between property type diversified and specialized REITs has not been addressed as thoroughly as one might expect in the REIT literature, given the historical importance of diversification studies in the finance literature.1 In fact, the application of performance rankings to the REIT industry, in general, has not been as fully explored as has the application of performance rankings to the mutual fund industry. Even with the recent popularity of mainstream corporate finance studies debating the existence of a diversification discount, only one research study was identified that conducted a systematic analysis of whether a similar diversification penalty might exist for equity REITs. That study drew its data from a sample period during which substantial changes were taking place in the operating environment of REITs, making a re-examination of potential differential performance a worthwhile pursuit. In addition to the standard Jensen's Alpha and Treynor Index, which are based on a single-factor market model, this study employs alternative Jensen's Alpha and Treynor Index tests that account for both inflation and size. The multifactor analogs of these performance ranking methodologies have been used in several studies that attempt to rank the performance of some type of real estate asset. These studies have included variables to control for total, expected, and unexpected inflation; shifts in the term structure of interest rates; shifts in market risk premia; changes in industrial production; market capitalization; or total assets, among others. By far the most common additional factor included in these multifactor applications is inflation, due to the traditional view of real estate as an inflation-hedging asset. Ibbotson and Siegel (1984); Hartzell, Hekman, and Miles (1987); and Rubens, Bond, and Webb (1989) all provide excellent examples of studies testing the inflation-hedging properties of real estate assets. Sirmans and Sirmans (1987) provide a detailed review of the inflation-hedging line of literature, and observe that the findings on the effectiveness of real estate as an inflation hedge have been somewhat mixed. The particular size control utilized in this study is the market capitalization, following Litt et al. (1999). The results from the three-factor market model are included in an effort to determine whether any observed differences in performance disappear with the use of an alternative return model. This study also includes among its performance measures an alternative form of the standard Sharpe Ratio, called the “Double” Sharpe Ratio, which adjusts for small sample bias in estimations of the standard Sharpe Ratio that utilize annual return data. Vinod and Morey (1999) proposed the use of bootstrapping to minimize the small sample bias problem. The names and ticker symbols of the REITs that comprise the sample are given in Appendix A. The earliest studies along similar lines often used samples composed of all three types of REITs, equity, mortgage, and hybrid. Some studies even used samples of real estate operating companies (REOCs), which can include builders, property managers, suppliers, and the like. The use of a pure equity REIT sample has become the standard in more recent years, as a way to reduce the potential for heteroskedasticity problems inherent in the use of a mixed sample. The remainder of this study proceeds as follows: Section 2 provides a brief review of the relevant literature. Section 3 details the data and methodology. Results are given in Section 4, and Section 5 presents the conclusions and recommendations for future research.

From their inception in 1960 to the present day, REITs have offered a way for investors to participate in real estate markets without necessarily committing a large sum of money or developing an expertise in property ownership. As interest in REIT shares grows among individual and institutional investors alike, questions surrounding REIT performance take on more importance. For example, investors should be concerned with whether benefits exist to owning shares of one REIT that is property-type diversified versus owning shares in REITs from several different property niches. This study utilizes the standard performance ranking methodologies of Jensen (1968), Treynor (1965), and Sharpe (1966), as well as their multifactor analogs, to compare whether the stock market performance of REITs that have chosen to diversify by property type differs from the stock market performance of REITs that have remained more specialized in their property holdings. Both parametric and nonparametric tests are employed to determine whether property-type diversified REITs outperform property-type specialized REITs. Using a single-factor model of market returns, t-tests indicate that three of the four market proxies used provide Jensen's Alpha rankings that are significantly better for property-type diversified REITs at the 10% level for the period 1995–2001. In addition, one market proxy (out of four) gives superior Treynor's Index rankings for property-type diversified REITs that is significant at the 10% level for the same sub-period. Finally, the Sharpe Ratio indicates superior performance for property-type diversified REITs that is significant at the 10% level, also for the earlier sample sub-period. Using the three-factor market model, the t-tests indicate that only the SNL Equity REIT Index market proxy provides Treynor's Index rankings that are significantly better for property-type diversified REITs at the 10% level during the earlier sample sub-period. Neither the Sharpe Ratio nor the “Double” Sharpe Ratio lend themselves to testing in a multifactor setting. Turning to the nonparametric tests, using a single-factor model of market returns, Mood's Median Test indicates superior performance significant at the 10% level for property-type diversified REITs when performance is measured using the value-weighted Jensen's Alpha, the small capitalization Jensen's Alpha, or the “Double” Sharpe Ratio over the period 1995–2001. Conversely, Mood's Median Test indicates superior performance for property-type specialized REITs with Jensen's Alpha results from the value-weighted index and with the “Double” Sharpe Ratio over the entire sample period. Similarly, Mood's Median Test over the period 2002–2006 indicates superior performance by property-type specialized REITs measured by the S&P 500 Jensen's Alpha, the SNL Jensen's Alpha, or the standard Sharpe Ratio. The two-sample Wilcoxon Rank-Sum Test using the single-factor market model rankings shows superior performance for property-type diversified REITs, significant at the 5% level, for the “Double” Sharpe Ratio over the earlier sample sub-period. Using the multiple factor market model, Mood's Median Test indicates superior performance that is significant at the 10% level over the first part of the sample period for property-type diversified REITs when performance is measured using the S&P 500 Treynor Index. Similar to the single-factor results, Mood's Median Test indicates superior performance for property-type specialized REITs with Jensen's Alpha results from the S&P 500 index and Treynor Index results from the value-weighted index over the latter part of the sample period. The two-sample Wilcoxon Rank-Sum Test fails to indicate significant differences in performance rankings of property-type diversified versus property-type specialized REITs using the multifactor market model rankings. Clearly, the multifactor market model results are not quite as convincing as the results from the single-factor market model tests. However, it is intriguing that the multiple factor model results uncover superior performance for property-type diversified REITs using the REIT index as the market proxy, given that this particular difference is not evident in the single-factor model results. Taken together, the results from these tests strongly indicate a significant difference in performance between property-type diversified REITs and property-type specialized REITs. However, the difference seems to depend on overall market conditions during the sample period. Somewhat surprisingly, the diversified REITs are actually the better performers when overall markets are performing well. The results provide limited evidence that property-type specialized REITs perform better when overall market conditions are not as favorable. However, as the focus of this study is on whether property-type diversified REITs perform better than more specialized competitors, further investigation of this possibility is left to future research. The result that property-type diversified REITs perform better over any time period is contrary to findings in the mainstream corporate finance literature that diversified firms often trade at a discount to their more specialized counterparts. However, the existence of a diversification discount has also been challenged in the corporate finance literature by several recent works. Surveying Table 1, the table of summary statistics, the only categories that exhibit significant differences between the property-type diversified and property-type specialized subsamples are the percent of the portfolio invested in apartment properties, office properties, and, for the period 2002–2006, retail properties. The failure of any financial and accounting variables to register as significantly different lends credence to the argument presented here that identifiable differences in performance may well be due to property-type diversification. These differences lend themselves to two possible (and non-mutually exclusive) explanations for the performance differences noted. First, property-type diversified REITs seem to have a significantly higher percentage of their portfolios devoted to office property, on average, versus all property-type specialized REITs. If it were the case that office property outperformed other property types during the first part of the sample period, this could explain the somewhat counterintuitive results for 1995–2001. Second, property-type diversified REITs seem to devote significantly less of their investment to apartments and retail properties, on average, versus all property-type specialized REITs. If it were the case that one or both of those property types underperformed other property types during the first sample sub-period, that could also provide an explanation for the results from that period. The reverse of these arguments would also explain the lack of significant findings during the latter part of the sample period. There are several potential avenues for future research related to performance differences between property-type diversified and specialized REITs. First, the results provided from the multifactor models were not as convincing as the results from the single-factor models, presumably due to the inclusion of the two additional return-generating factors. This observation is curious, and certainly worthy of future investigation. Second, the use of multiple market proxies makes reporting of results somewhat cumbersome. Roll, 1977 and Roll, 1978 argument that the choice of market proxy matters for performance ranking studies has been retested numerous times for mutual funds; however, very few studies exist that test whether choice of market proxy matters in performance rankings of publicly traded real estate securities. If it were the case that the choice of market proxy did not matter for real estate securities, then the presentation of results in similar future works would be greatly simplified. Third, the differential performance of property-type diversified and property-type specialized REITs in overall rising markets versus overall flat or declining markets hinted at in some of the results warrants further research. If it were the case that property-type diversified REITs performed better in rising markets and fared no worse than their specialized counterparts in flat or declining markets, this would certainly be valuable information for individual and institutional investors. Finally, alternative specifications of property-type concentration are possible, and perhaps even desirable. For example, if property-level cash flows could be obtained, then concentration could be based on the amount of cash flow generated by each type of property in the portfolio, rather than on the amount invested in each property type. Obviously, this is not intended to be an exhaustive list of future possibilities. Instead, this list provides some idea of the potential future directions that might be taken and, hopefully, serves as an invitation to fellow researchers to explore questions surrounding property-type diversification in more depth.