Real Estate In A Mixed Asset Portfolio

Real Estate in a Mixed Asset Portfolio

By

Thomas G. Thibodeau

August 12, 2005

Thomas G. Thibodeau is the Global Real Estate Capital Markets Professor at the Leeds School of Business, University of Colorado-Boulder. Financial support for this research was provided by Behringer Harvard Real Estate Investments. The views in this paper are those of the author and do not necessarily represent Behringer Harvard Real Estate Investments.

Preliminary Draft

Do Not Quote
Executive Summary

This paper has three objectives. The first is to compare the returns provided by publicly traded real estate to returns provided by two classes of publicly traded investment alternatives over the past twenty-five years. Using standard mean-variance portfolio theory, the analysis demonstrates that the returns generated by publicly traded real estate dominated the returns provided by the S&P 500 and by the Russell 2000, a small-cap mutual fund. Over the past quarter century, average real estate returns were higher than these two equity alternatives and the real estate returns had less volatility.

The second objective of this research is to address the question: How much in real estate? Just because real estate dominated the equity alternatives on a risk adjusted basis does not mean that investors should invest exclusively in real estate. Real estate returns can be negatively correlated with the returns provided by investment alternatives and portfolios that combine real estate with other investments can provide comparable returns with less volatility (or higher returns with comparable volatility). Using standard mean-variance portfolio theory for three asset classes (real estate, a small cap fund, and the S&P 500), this study concludes that, over the January 1979 through December 2004 period, the optimal portfolio would have held 70% of its assets in real estate. This result is based on the historical investment performance of these three asset classes. It does not mean that investors should allocate 70% of their assets to real estate going forward. It does, however, suggest two things: (1) real estate is indeed a separate asset class that generates returns that are substantially different from other asset classes; and (2) long-term investors should seriously consider allocating a significant portion of their assets to real estate.

The third objective of this paper is to examine the investment performance of privately held real estate during the 1990s. The decade of the 1990s was a period where stocks dominated real estate: the return to investments in the S&P 500 were nearly two times higher than returns provided by equity real estate and the S&P produced those returns at half the volatility. So how did privately held real estate investments perform? Using data on completed partnerships provided by Behringer Harvard Real Estate Investments, private real estate investments outperformed the S&P 500 by a significant margin yielding a 26% annual before tax return during the 1990s. This yield compensates long-term investors for the lack of liquidity associated with private real estate placements. This yield also reinforces the claim that long-term investors should seriously consider allocating a significant portion of their assets to real estate.

1

Introduction

Financial analysts and academics have been analyzing the investment returns provided by stocks, bonds, commodities and other equity and debt investments for decades. Research on real estate investments has been hampered by the lack of comparable data on investment performance. Early research on real estate investment returns typically used data from the National Council of Real Estate Investment Fiduciaries (NCREIF) or returns provided by Commingled Real Estate Funds (CREFs) to evaluate the characteristics of real estate returns. These sources have three disadvantages. First, they are privately held and do not necessarily represent the return characteristics of publicly held real estate investments. Second, the properties in these funds rarely sell. Consequently, the capital appreciation component of the total return for the properties held by these funds is typically measured using an estimate of the market value of the property (e.g. appraisal) in lieu of a transaction price. Returns computed using appraisals are smoothed and have a smaller variance than returns computed from actual transactions. Third, the returns computed from privately held real estate fail to adequately measure the liquidity risk associated with real estate investments. For example, the returns computed using CREFs are based on reported ‘unit prices’. However, fund managers reserve the right to not honor a sell order when property market conditions are unfavorable. In this case, it is difficult to interpret the ‘unit price’ as a cash flow when computing holding period returns.

The first portion of this research compares the returns provided by publicly traded real estate to publicly traded equities over the past quarter century. In that sense, it is an ‘apples to apples’ comparison of historical investment returns. It is important to note that the historical returns are not necessarily representative of future returns. Nevertheless, twenty-five years is a long time with several cycles in the real estate market and several booms and busts in the stock market.

Real estate returns are measured using the National Association of Real Estate Investment Trusts (NAREIT) equity index over the January 1979 through December 2004 period. In 2004, the total equity market capitalization of Real Estate Investment Trusts (REITs) included in the NAREIT index was about $275billion. The first part of this research compares monthly and annual equity REIT returns to returns provided by large corporations (as measured by the S&P 500) and to returns provided by small cap companies (as measured by the Russell 2000).

This research compares the nominal holding period returns, inflation adjusted returns, and risk-adjusted returns for these three investment alternatives. The inflation adjustment is made using the All Urban Consumers’ Consumer Price Index (CPI). Risk, or uncertainty in the expected return, is measured using the standard deviation in returns. Risk-adjusted performance is evaluated using standard mean-variance portfolio theory.

The second portion of this research asks the question: ‘How much in real estate?’ That is, using the historic properties of the returns for these three funds (Equity REITs, Russell 2000 and S&P 500) for the past 25 years, what allocation of funds to each asset class would have produced the highest return for the least risk?

Finally, this paper examines the return characteristics of privately held real estate during the 1990s. Research has demonstrated that the return characteristics for privately held real estate are indeed different than the return characteristics of publicly held real estate. Until 1997, REIT returns resembled returns in the equities market more than they resembled returns to the underlying properties. This relationship changed when the speculative bubble in technology stocks emerged in 1997—stock market returns increased significantly while REIT returns decreased as equity capital chased returns in technology stocks. The burst in the speculative bubble in 2000 changed that relationship again. Since 2000, equity returns in the stock market have declined while REIT returns have increased significantly—by 26.4% in 2000, 37.1% in 2003 and 31.6% in 2004! Furthermore, privately held real estate is fundamentally different from publicly traded real estate because privately held real estate is illiquid and investors must be compensated to take the additional liquidity risk. This compensation for taking additional liquidity risk makes direct real estate investments an ideal vehicle for long-term investors.

The third part of this research examines the returns provided by Behringer Harvard Real Estate Investments during the 1990s. Behringer Harvard is a real estate investment firm located in Dallas, Texas. During the 1990’s the firm completed 14 partnerships: 7 apartment properties, 2 industrial properties, one land acquisition and four office properties. The capital market for privately held real estate is very different than the market for publicly held real estate. The supply of private equity is much smaller than the supply of public equity. In addition, the liquidity risk in private real estate investments is substantially larger than the liquidity risk in public markets. Consequently, the return characteristics for private real estate investments can be very different than the return characteristics for publicly traded assets. To compensate investors for the additional liquidity risk, private real estate investments typically offer investors significantly higher expected returns. During the 1990s, the return to publicly held real estate (as measured by equity REITs) was 9.14%. The return to the S&P 500 was 18.2%. After eliminating flips, the average return for completed partnerships in Behringer-Harvard’s portfolio was 26.1%!

A Quarter Century of Investment Returns

The investment returns provided by three alternative asset classes (real estate, small caps and the S&P 500) are evaluated using three different measures of return: (1) nominal, (2) inflation adjusted (or real), and (3) risk-adjusted. Returns are measured using holding period return. The holding period return is the sum of the percentage change in asset price (i.e. the capital appreciation component of the total return) plus the dividend yield during the period.[1] The average return is computed two ways: (1) the arithmetic mean return, and (2) the geometric mean. The arithmetic mean return is the simple average of the holding period returns over the entire holding period. The geometric mean return compounds the periodic (either monthly or annually) return and is a better way to measure the investment’s total return.

Risk, or uncertainty in investment returns, is measured using the square root of the variance (or standard deviation) in returns. The variance is simply the average of the squared deviations from the arithmetic mean.

Finally, when evaluating investment alternatives it is important to control for differences in risk. Investments with more risk must offer investors higher expected returns to compensate the investors for the additional risk. One way to evaluate differences in risk is to compute the ratio of the standard deviation in returns to the mean return. This ratio is called the coefficient of variation. The coefficient of variation, or CV, is interpreted as the amount of total risk per unit of return. The lower the coefficient of variation, the lower the risk (per unit of return), and the more desirable is the investment.

Table 1 reports return characteristics for the three investment alternatives: (1) monthly for the entire1979-2004 period; (2) annually for the same quarter century; and (3) annually for the 1980s and for the1990s. The annual returns in Table 1 are computed by compounding monthly

Table 1
Nominal Returns
Equity REITs / Russell 2000 / SP 500 / Inflation
Monthly: 1979-2004
Mean
Arithmetic / 1.24% / 1.24% / 1.17% / 0.33%
Geometric / 1.17% / 1.08% / 1.07% / 0.33%
Standard Deviation / 3.77% / 5.64% / 4.38% / 0.33%
Coef. Of Variation / 3.04 / 4.55 / 3.74 / 0.99
Annual: 1979-2004
Mean
Arithmetic / 15.98% / 15.26% / 14.79% / 4.10%
Geometric / 15.61% / 14.28% / 14.26% / 4.22%
Standard Deviation / 14.85% / 18.56% / 15.45% / 2.98%
Coef. Of Variation / 0.93 / 1.22 / 1.05 / 0.73
Annual: 1980-1989
Mean
Arithmetic / 16.05% / 15.65% / 18.19% / 5.14%
Geometric / 15.64% / 14.52% / 17.55% / 5.10%
Standard Deviation / 9.50% / 15.89% / 12.04% / 3.05%
Coef. Of Variation / 0.59 / 1.02 / 0.66 / 0.59
Annual: 1990-1999
Mean
Arithmetic / 10.64% / 14.80% / 18.98% / 2.94%
Geometric / 9.14% / 13.40% / 18.20% / 2.93%
Standard Deviation / 17.89% / 17.46% / 13.44% / 1.18%
Coef. Of Variation / 1.68 / 1.18 / 0.71 / 0.40

1

returns. The results in Table 1 indicate that the average monthly return (as measured by the geometric mean) for equity REITs, exceeded the returns provided by the Russell 2000 and by the S&P 500. The geometric monthly return was 1.17% for REITs, 1.08% for small caps and 1.07% for the S&P 500. Real estate also had the least risk over this twenty-five year period: the standard deviation in monthly returns was 3.77% for REITs, 5.64% for the Russell 2000 and 4.38% for the S&P 500. Real estate’s higher mean return and lower standard deviation produced the smallest coefficient of variation (total risk per unit of return) for these alternatives.

Table 1 also provides summary statistics for the rate of inflation (as measured by the All Urban Consumers’ Consumer Price Index, or CPI) over the 1979-2004 period. The average rate of inflation was 0.33% per month (or about 4% per year) over this period. However, the average rate was heavily influenced by very high rates of inflation early in the period: 13.3% annually in 1979; 12.5% in 1980 and 8.9% in 1981.

The annual returns for the entire 25-year period yield similar results: public real estate investments yielded higher returns with less risk compared to small caps and the S&P 500. Real estate provided a mean annual return of 15.61% compared to 14.28% for the Russell 2000 and 14.26% for the S&P 500. The standard deviation in nominal returns was 14.85% for real estate, 18.56% for small caps and 15.45% for the S&P 500.

Table 1 also reports return characteristics for two sub-periods: the 1980s and the 1990s. In each of these two sub-periods, average nominal returns were highest for the S&P 500: 17.55% during the 1980s (compared to 15.64% for REITs and 14.52% for Russell 2000) and 18.2% for the 1980s (compared to 9.14% for REITs and 13.4% for the Russell 2000). Real estate risk was substantially lower than the risk of both small caps and the S&P 500 during the 1980s, but real estate risk increased dramatically during the 1990s relative to the equities market.

Graph 1 provides cumulative returns for the three alternative assets and for the CPI for the 1979-2004 period. The graph shows that returns for all three investment alternatives were highly correlated from 1979 through 1997. In 1997, speculative interest in technology stocks propelled the stock market to record highs. At the same time, the increased demand for technology stocks reduced demand for real estate investments in public capital markets, driving REIT prices and returns (in public markets) down. The speculative bubble in the stock market burst in 1999, reversing the trends in returns for both stocks and real estate: returns in the S&P 500 declined significantly through 2002 while REIT returns have been high since 1999. At the end of 2004, the cumulative return to the equity REIT index rivaled the S&P 500 at its 1999 peak.