Equity REIT Beta Estimation

Recent research has provided a summary of procedures for selecting and estimating beta that are assumed to be invariant to the type of company being analyzed. For companies such as real estate investment trusts, however, institutional details may dictate the use of different procedures. Specifically, U.S. REITs qualify for corporate tax exemption only by following rules that require high dividend payouts. Also, they are generally considered to be small-capitalization stocks. We present here a review of how financial services firms calculate REIT betas. The findings indicate that these firms make no special provisions for REITs and that their different procedures for estimating beta generally yield statistically different results among the REIT sample. Then, we present a study of elements that may be important in estimating REIT cost of capital. Through a series of beta estimations, using the same companies over the same period, we tested how treatments of dividends, use of small-cap versus broad market indexes, and use of other specialized procedures influence cost-of-capital results for REITs. The public real estate market has reached a scale and duration of trading that warrant the application of standard valuation models, particularly the capital asset pricing model (CAPM) and multifactor asset-pricing models, to the analysis of real estate investment trusts. Financial analysts have the option of using their own model designs to perform the calculations or obtaining betas from the many commercial financial service providers that now offer betas for REITs. The temptation for analysts and service providers has been to proceed in these modeling efforts as if REITs operate and trade in the same ways industrial and service companies do. Unlike other companies, however, REITs do not pay income taxes but, in return, they do face extremely high dividend payout requirements. Analysts must take special care to accommodate these unique features when applying asset-pricing models to REITs.Our review of how commercial firms calculate REIT betas indicates that they make no attempt to customize their models for REITs. Indeed, in at least one case, we found that the firm uses the standard corporate tax rate to unlever REIT betas. Our statistical analysis of commercial service estimates for the period January 1993 through November 1997 revealed no difference among the providers in the betas they assign to REITs collectively but significant differences among the betas they assign to individual REITs.Our examination of various specifications in asset-pricing models for REITs considers the effects of dividends, the use of a small-capitalization market index instead of a broad index, and the introduction of additional variables frequently recommended to improve the results of single-factor models. Our findings are as follows: First, because of the unusual dividend pattern in REIT returns, we found statistically different estimates of the cost of equity capital from various service providers depending on whether the provider firm included or excluded dividends in the return calculation. Thus, differences among the REIT betas produced by commercial services may be rooted in how the services treat dividends in the model.Second, neither the substitution of a small-cap stock index for a broad market index nor taking cross auto-correlation into account had a statistically significant influence on REIT cost-of-capital estimates.Third, the only asset-pricing model innovation that generated statistically significant improvement in cost-of-capital estimates was the addition to the CAPM of a factor representing the premium return on small-cap versus large-cap stocks and a factor representing the difference in return on stocks of high market-to-book companies and stocks of low market-to-book companies. The introduction of these additional factors increased the average R2 from about 3 percent to 11 percent and doubled the number of significant betas in the sample studied.No real surprises emerged from straightforward estimations of REIT betas. The betas were low (i.e., less than 0.4 in most cases), which is consistent with the long-term nature of the contractual obligations supporting the income streams from the REITs' underlying assets. The low average R2 for the sample regression (0.03) when the S&P 500 Index was used as the market is not new evidence, but it does confirm that stock market return indexes do a poor job of explaining variations in REIT returns—which may be related to the traditional wisdom that real estate is a good diversifier for common stock portfolios in the United States. The REIT company is a unique entity, not only because of its statutory tradition but also because REITs are the only companies for which the underlying assets trade in their own active secondary markets.