Macroeconomic variables, firm-specific variables and returns to REITs

Su-Jane Chen* Chengho Hsieh** Timothy W Vines*** Shur-Nuaan Chiou****

Abstract. This study investigates the cross-sectional variation in equity real estate investment trusts (EREITs) returns. A pooled cross-sectional, time-series approach is used as an alternative to the two-step Fama-MacBeth regression. With pooling, more powerful tests can be obtained from the limited sample of EREITs available. Beta does not explain return variation. Size is the sole consistent factor explaining prices. None of the variables of Chen, Roll and Ross (1986) is significant when size and book-to-market variables are included in the model. Only the unanticipated change in term structure is significant in versions of the model that exclude firm-specific variables.

Introduction

The Capital Asset Pricing Model (CAPM) of Sharpe (1964) and Lintner (1965) predicts a positive linear relationship between expected security returns and market betas. The CAPM also predicts that market beta is sufficient to describe cross-sectional expected returns. These predictions have been the subject of a great deal of empirical investigation. Much of the evidence does not support the model. Empirical contradictions of the CAPM are documented in Banz (1981), Reinganum (1981), Basu (1983), Rosenberg, Reid and Lanstein (1985) and Bhandari (1988). Average stock returns may be related to firm-specific variables such as size, price/earnings ratio, book-to-market equity ratio, and leverage. More recently, Fama and French (FF) (1992), using cross-sectional tests of stock returns, conclude that the beta fails to describe average stock returns over the past fifty years after introducing two firmspecific variables, size and book-to-market equity, to a traditional single factor pricing model.1

Another development, which casts doubt upon the predictions of CAPM, can be found in the Arbitrage Pricing Theory (APT) literature originally developed by Ross (1976). In fact, the APT has been proposed by researchers (notably Roll and Ross, 1980) as a testable alternative, and perhaps natural successor, to the CAPM. Empirical evidence in favor of the APT can be found in Chen (1983) and Bower, Bower and Logue (1984). Roll and Ross (1980) and Chen (1983) conclude that two to four factors are significant in explaining equilibrium prices. Chen, Roll and Ross (CRR) (1986) show that five macro variables are significant in explaining expected stock returns. They are the unanticipated inflation rate, the change in expected inflation, the unanticipated change in term structure, the unanticipated change in risk premium and the unanticipated change in the growth rate in industrial production.

CRR also show that while market indices are useful in explaining time-series return variation, they cannot explain cross-sectional differences in expected returns when macro variables are included in the model. He and Ng (HN) (1994) combine the five CRR macro variables, the two significant firm-specific variables found in FF and the market index in their pricing model. The cross-sectional variation of average stock returns is not explained by either the market index or the macro variables when the firm-specific variables are included in the model.2 These results raise serious concerns about the usefulness of the CAPM in explaining security returns.

The major objective of this study is to see whether any of the common factors prevailing among "ordinary" common equities is useful in explaining the crosssectional variation in equity real estate investment trust (EREIT) returns. This also will provide useful information in studies that try to relate the returns of real estate assets to the returns of other equities. Financial literature indicates that EREITs may possess distinct risk-return characteristics than ordinary common stocks. For example, real estate is thought to be an inflation hedge. In addition, if the risk premium associated with the market beta is insignificant, any cross-sectional evidence about this relationship that depends on CAPM-based tests may be less meaningful.

This study is the first one to address these issues as they relate to EREITs. Numerous studies have been conducted in evaluating REIT performance (see, for example, Brueggeman, Chen and Thibodeau, 1984, 1992; Kuhle, Walther and Wurtzebach, 1986; Titman and Warga, 1986; Chan, Hendershott and Sanders, 1990; Liu, Hartzell Grissom and Grieg, 1990; Chen, Hsieh and Jordan, 1997; and Peterson and Hsieh, 1997). The majority of studies that investigate the relative performance of REITs and ordinary equities claim that the single index models are not sufficient for studying the risk-return relationship of real estate assets. The consensus is that real estate portfolio returns are the result of a multi-factor return generating function. However, this conclusion should be regarded with reservation, given the fact that all of these studies investigate only time-series real estate returns.

As CRR demonstrate, the fact that the market index is powerful in explaining timeseries asset return variation does not guarantee its significance in explaining crosssectional pricing. The traditional time-series tests provide evidence that can help explain the variation in asset/portfolio returns over time. Cross-sectional tests are designed to explain differences in the returns across various assets/portfolios in a specific time period. Despite the findings that REITs time-series returns were sensitive to the movement of multiple factors in previous studies, it is important for researchers to investigate how well the documented factors explain EREIT returns crosssectionally. This should have important implications for asset selection and relative performance evaluation of portfolio managers.

The next section describes the data and methodology. This is followed by the presentation of the empirical results. The final section contains concluding remarks.

Data and Methodology

The starting point for the data used in this study is all EREITs traded on the NYSE, AMEX and NASDAQ. However, to allow for the estimation of some EREITs characteristics, two additional criteria must be met by the EREITs prior to their inclusion in the final sample. The first criterion is the existence on the Center for Research in Security Prices (CRSP) tapes of forty-eight consecutive monthly returns preceding July of each year in the study period, 1978-94. This allows for the estimation of betas. Second, to allow for estimation of firm-specific variables, the EREIT data must include relevant accounting and market price data. Thus, EREITs without data available in the Compustat database are excluded. Exhibit 1 lists the number of resulting EREITs selected under the criteria for each sample year.

Four pricing models are used to explain the returns to EREITs. The first is the traditional CAPM. The other three are multi-factor models that differ in the number and type of explanatory factors that are included. In the firm-specific variable model (FVM), the factors are attributes unique to individual firms. In the macroeconomic variable model (MVM), economic time-series variables are the presumed pricing factors. In the combined model (CM), all the variables associated with the other three models are combined together as the factors.

The two most well documented firm-specific variables, firm-size (SIZE) and the bookto-market equity ratio (BIM), are used in this study to implement the FVM.3 The two variables are measured in the same way as FF with data provided on the Compustat tapes. The macroeconomic variable set used in this study closely resembles CRR's with one exception. The unanticipated change in the growth rate in industrial production proposed in CRR is excluded from this study, based on the findings of Chan, Hendershott and Sanders (1990). Thus, the macro variables included in this study are the market index, the unanticipated inflation rate, the change in expected inflation, the unanticipated change in term structure and the unanticipated change in risk premium.4 Panel A of Exhibit 2 lists the basic macro variable data, the sources and the notations. Panel B of Exhibit 2 describes how the series are constructed and provides some additional notations.

Since only the innovations or unanticipated changes in the macro variables are of interest, a standard Box-Pierce Q-Statistic is calculated for each of the five series over every four years preceding July of each study year as a check for serial correlation. The inflation (It) and risk premium (PR,) are generally serially correlated. The other series appear to be noisy enough to be treated as pure innovations. To avoid the effects of autocorrelation, proxies for the expected inflation and risk premium are created as the forecasts from autoregressive models.5 The residuals from these fitted processes are treated as the unanticipated inflation (UI) and unanticipated change in risk premium (UPR).

Results

The regression results from each of the four pricing models are reported in Exhibit 3.6 As indicated, the regression coefficient associated with the market beta is not significantly different from zero. Consistent with CRR's (1986) finding using ordinary common stocks and FF's (1992) using nonfinancial stocks, this data does not support the market index as a relevant variable for explaining cross-sectional variation of returns. Thus, the rejection of the CAPM (as far as EREITs are concerned) is suggested.

As with numerous empirical findings on ordinary common stocks, SIZE is found significantly priced among EREITs over time. The significance sustains, although at a lower level, even when all the other factors are present in the same model (CM).7 The negative regression coefficient has similar magnitude between the FVM and CM. The BIM ratio, however, is not significant in either of the two models. There are two plausible explanations. First, this variable may not have the same meaning for the EREITs as for ordinary common stocks. Second, if BE/ME is interpreted as a distress factor, it is possible that this factor behaves in a similar fashion for firms in the same industry and loses its explanatory power. It would be interesting to see whether this phenomenon holds for other industries as well. Future research along this line may be warranted. Based on the test results, the macro variables are generally insignificant in EREIT pricing. The only exception is the unanticipated change in term structure. In the MVM, the risk premium is negatively significant at the 5% level. However, this significance disappears in the CM, possibly because of the impact of SIZE. This inconsistency may be a result of multicollinearity. Nonetheless, the finding here is in line with He and Ng (1994) in that none of the macro variables are significant in explaining the cross-sectional variation of common stock returns when the two firmspecific variables are also included in the model. Although time-series EREIT returns are found to be correlated with some inflation measures (see, for example, Chan, Hendershott and Sanders, 1990), our results show no correlation between EREIT returns and inflation sensitivities in a cross-sectional setting.

Conclusion

This study employs a single pooled cross-sectional time-series regression approach to investigate EREITs pricing. In doing so, this study also provides further evidence on the cross-sectional comparison between the CAPM-related single index models and the APT-related multi-index models. Only by finding significantly priced REITs factors will the empirical investigation of REITs performance relative to investments in common stocks be meaningful.

The results show that the size factor commands a risk premium in EREIT pricing. There is also some support for the importance of the unanticipated change in the term structure. However, the evidence is not consistent. The insignificance of the market beta across the employed models leads to the rejection of the CAPM, at least for EREITs. Findings in past studies of correlation between EREIT returns and inflation are not confirmed by our cross-sectional results. The other firm-specific variable employed, the book-to-market equity ratio, is not significantly related to EREITs returns. While this study offers some plausible causes for the insignificance, further research is needed. This study does not exclude the possibility that there might be other factors important to EREITs pricing. Thus, exploring the existence of other pricing factors serves as another avenue for future research.

Notes

1 Fama and French (1992) contend that there is a close link between the leverage and book-tomarket for companies in general; that is, when a firm's market leverage (book-value debt/ market-value equity) is higher relative to its book leverage (book-value debt/book-value equity), the firm tends to have a higher book-to-market ratio.

2 The finding on the size factor is in line with the conventional small firm effect found in the literature. Smaller firms earn higher risk-adjusted rate of returns than larger firms do. The returns are risk-adjusted using single-index CAPM. The average firm size is 460 million for all companies as a whole and 70 million for EREITs. Therefore, EREITs are considered small firms. Colwell and Park (1990) and McIntosh, Liang and Tompkins (1991) find the small firm effect within REITs.

3 The commonly used definition of size, total market value of outstanding equity shares, is followed in this study.

4 As noted in Exhibit 2, the yield spread between long-term government bonds and Treasury bills defines the term structure while the yield difference between the low grade and high grade corporate bonds is taken as the risk premium.

5 Chen, Roll and Ross (1986) and Chan, Hendershott and Sanders (1990) use the same approach to extract the unanticipated economic series. As with the two cited studies, the univariate autoregressive procedure under different lags are studied in this research to derive innovations of economic state variables when the original series are determined to be highly autocorrelated.

6 When the four pricing models are examined using the traditional two-step Fama-MacBeth regression, none of the candidate pricing factors is significantly priced. When the size factor is used alone in the two-step regression, the average adjusted R2 is .02.

7 We also estimate the model with size as the sole factor (plus the dummy variables for time). The coefficient estimate is -.0026 with a t-Stat of -2.3. The R2 for the model is .1211 and the f-value is 4.987.

References

Banz, R. W., The Relationship between Return and Market Value of Common Stocks, Journal of Financial Economics, 1981, 9, 103-26.

Basu, S., The Relationship between Earnings Yield, Market Value, and Return for NYSE Common Stocks: Further Evidence, Journal of Financial Economics, 1983, 12, 129-56. Bhandari, L. C., Debt/Equity Ratio and Expected Common Stock Returns: Empirical Evidence, Journal of Finance, 1988, 43, 507-28.

Bower, D. H., R. S. Bower and D. E. Logue, Arbitrage Pricing Theory and Utility Stock Returns, Journal of Finance, 1984, 39, 1041-54.

Brueggeman, W., A. Chen and T. Thibodeau, Real Estate Investment Funds: Performance and Portfolio Considerations, The Journal of the American Real Estate and Urban Economics Association, 1984, 12:38, 333-54.

-, Some Additional Evidence on the Performance of Commingled Real Estate Investment Funds: 1972-1991, Journal of Real Estate Research, 1992, 7, 433-48. Chan, K. C., P. H. Hendershott and A. B. Sanders, Risk and Return on Real Estate: Evidence from Equity REITs, The Journal of the American Real Estate and Urban Economics Association, 1990, 18, 431-52.

Chen, N. F., Some Empirical Tests of the Theory of Arbitrage Pricing, Journal of Finance, December 1983, 38, 1393-414.

Chen, N. F., R. Roll and S. A. Ross, Economic Forces and the Stock Market, Journal of Business, 1986, 59, 383-403.

Chen, S., C. Hsieh and B. D. Jordan, Real Estate and the Arbitrage Pricing Theory: Macrovariables vs. Derived Factors, Real Estate Economics, 1997, 25, 505-23. Colwell, P. F. and H. Y. Park, Seasonality and Size Effects: The Case of Real-Estate-Related Investment, The Journal of Real Estate Finance and Economics, 1990, 3, 251-60. Fama, E. F. and K. R. French, The Cross-Section of Expected Stock Returns, Journal of Finance, 1992, 47, 427-65.

Fama, E. F. and J. D. MacBeth, Risk, Return, and Equilibrium: Empirical Tests, Journal of Political Economy, 1973, 71, 607-36.

He, J. and L. K. Ng, Economic Forces, Fundamental Variables, and Equity Returns, Journal of Business, 1994, 67, 599-609.

Kuhle, J., C. Walther and C. Wurtzebach, The Financial Performance of Real Estate Investment Trusts, Journal of Real Estate Research, 1986, 1, 67-75. Lintner, J., The Valuation of Risk Assets and the Selection of Risky Investments in Stock Portfolios and Capital Budgets, Reviews of Economics and Statistics, 1965, 13-37. Liu, C. H., D. Hartzell, T. Grissom and W. Greig, The Composition of the Market Portfolio and Real Estate Investment Performance, The Journal of the American Real Estate and Urban Economics Association, 1990, 18, 49-75.

McIntosh, W., Y. Liang and D. Tompkins, An Examination of the Small-Firm Effect within the REIT Industry, Journal of Real Estate Research, 1991, 6, 9-18. Peterson D. J. and C. Hsieh, Do Common Risk Factors in the Returns on Stocks and Bonds Explain Returns on REITs? Real Estate Economics, 1997, 25, 321-45. Reinganum, M. R., Misspecification in Capital Asset Pricing: Empirical Anomalies Based on Earnings Yields and Market Values, Journal of Financial Economics, 1981, 9, 19-46. Roll, R. and S. A. Ross, An Empirical Investigation of the Arbitrage Pricing Theory, Journal of Finance, 1980, 35, 1073-103.

Rosenberg, B., K. Reid and R. Lanstein, Persuasive Evidence of Market Inefficiency, Journal of Portfolio Management, 1985, Il, 9-17.

Ross, S. A., The Arbitrage Theory of Capital Asset Pricing, Journal of Economic Theory, 1976, 341-60.

Sharpe, W. F., Capital Asset Prices: A Theory of Market Equilibrium under Conditions of Risk, Journal of Finance, 1964, 19, 425-42.

Titman, S. and A. Warga, Risk and the Performance of Real Estate Investment Trusts: A Multiple Index Approach, The Journal of the American Real Estate and Urban Economics Association, 1986, 14, 414-31.

*Department of Accounting and Finance, University of Wisconsin-Eau Claire, Eau Claire, WI 54702.

**Department of Economics and Finance, Louisiana State University-Shreveport, Shreveport, LA 71115 or chsieh@pilot.lsus.edu.

***Department of Economics and Finance, Louisiana State University-Shreveport, Shreveport, LA 71115 or Tvines@pilot.lsus.edu.

****Department of Finance, National Chung Cheng University, Min-Shiung, Chia-Yi 621, Taiwan or finsn@ ccunix.edu.tu.

The authors thank two anonymous referees and Bradford D. Jordan for their constructive comments. Su-Jane Chen acknowledges the financial support from the College of Business at the University of Wisconsin-Eau Claire.

Copyright American Real Estate Society 1998
Provided by ProQuest Information and Learning Company. All rights Reserved