|
Article Excerpt
In this article we compare public and private real estate equities. In so doing, we control for three of the main differences between these investment alternatives: property-type mix, leverage and appraisal smoothing. With these two restated indices, we then run tests to determine in a statistical sense whether the restated means and volatilities of the two series were different from one another. The clear answer is that they were not. The results of the statistical tests combined with the fact that the average difference between the two (restated) return series has substantially narrowed (to approximately 60 basis points) in the more recent (1993-2001) period jointly suggest a seamless real estate market in which public- and private-market vehicles display a long-run synchronicity. This has important implications for portfolio management. First, public- and private-market vehicles ought to be viewed as offering investors a risk/return continuum of real estate investment opportunities. Second, while the "platform" did not matter in terms of observed return characteristics, the platform may matter with regard to liquidity, governance, transparency, control, executive compensation and so forth; an apparent clientele effect hints at these issues being valued differently by large and small investors.
**********
Returns on publicly traded real estate investment trusts (REITs) have typically exceeded the returns on private real estate equities by approximately 5% per annum. For example, when returns are observed over the 21-year period ended in 2001, REIT returns (as proxied by the National Association of Real Estate Investment Trusts, or NAREIT, Index) averaged 13.5%, while private real estate returns (as proxied by the NCREIF Property Index, or NPI) have averaged just 8.4%. This performance difference of approximately 500 basis points raises the question: Is this differential attributable to some underlying structural reason (e.g., optimal contracting issues, the efficiency of public markets, etc.) or is it just happenstance? In short, does the "platform" (i.e., the format of the legal entity used to hold the assets) matter? This article attempts to examine rigorously this question.
The answer to this question has important portfolio management implications. If one platform clearly and persistently offers superior risk-adjusted returns, then this vehicle ought to, all else being equal, become the dominant real estate investment vehicle over time. If, on the other hand, there is no clearly superior vehicle, then the two platforms can be viewed either as substitutes for one another or as part of a continuum of risk-adjusted return alternatives. While there has been much written about the portfolio enhancement characteristics of private real estate equities, (1) institutional investors have fairly paltry allocations to such investments, and even less so to public real estate equities. (2) Institutional investor reluctance to embrace real estate equities seems at least partly rooted in the apparent mixed signals sent by these two forms of real estate equity.
Circumstantially, the expected persistence of REITs' excess performance is corroborated by their growing popularity. As shown in Table 1, the market capitalization (3) of REITs has exceeded that of the NPI since the mid-1990s. Whereas REITs accounted for only 15-20% of the total market capitalization between these two indices prior to 1993, REITs have averaged an approximate 65% market share in the post-1992 period. The dramatic change in market share is a function of the tremendous growth in REIT market capitalization beginning in the early 1990s. That said, both indices have experienced tremendous gains in total capitalization; on a combined basis, the two indices ended 1981 with a little over $4 billion and ended 2001 with nearly $400 billion (a growth rate exceeding 25% per annum).
A market-efficiency argument (4) might suggest that the increasing popularity of REITs indicates that the observed premium (between public-and private-market real estate equities) is expected to persist.
On the other hand, a different story emerges when the analysis is refined to examining the behavior of (defined-benefit) pension plans. These sophisticated investors (generally supported by sophisticated consultants) clearly have the opportunity to invest in either (the public or private) real estate platform. Yet, the preponderance of large pension funds choose to allocate the bulk of their real estate equity capital to private-market vehicles. See Table 2, which indicates, among other things, that almost 90% of pension fund capital allocated to real estate equities is invested in private-market vehicles (i.e., $80 billion in private vs. $12 billion in public).
The top 25 pension (including endowment and union) funds with an allocation to private real estate equities represent, in the aggregate, an approximately $80 billion commitment. (5) Each of these funds has, on average, $54 billion of assets committed to various investment vehicles, most of which is allocated to stocks and bonds. However, the top 25 pension funds with an allocation to REITs only represent, in the aggregate, an approximately $12 billion commitment. (6) Each of these funds has, on average, only $39 billion of assets committed to various investment vehicles. Alternatively stated, the funds with a commitment to private real estate are generally 40% larger in terms of total asset size than those funds investing in public real estate. Perhaps as telling are the 10 funds that are common to both lists (i.e., they have among the 25 largest commitments to private real estate as well as among the 25 largest commitments to REITs); these pension funds have private-market allocations that dominate their public-market allocations by a ratio of approximately 6:1. This substantive discrepancy suggests an argument in favor of a clientele effect (7) by which large institutional investors tend to favor private real estate equities, while individual and small institutional investors (8) favor public real estate equities. If this effect were to persist, then this might suggest that there is something intrinsically preferable about private-market vehicles since large investors are capable of investing either privately or publicly (9) (while individuals and small institutions rarely have and/or devote enough wealth to attract the attention of conventional real estate advisory firms).
Thus, the circumstantial evidence does not conclusively suggest that one market might outperform the other. Alternatively, an empirical examination of the returns from the two markets might resolve the central question--at least from a historical point of view--of whether REITs have, in a statistically reliable sense, outperformed private real estate. However, before such an approach can be satisfactorily implemented, there must be an appreciation of the substantive differences between the two data sets. Without controlling for these differences, the statistical procedures might be jeopardized. Given our data constraints, we restated the NAREIT index (for the period 1981-2001) to exclude noncore property types. We then delevered the REIT series. We then restated the NCREIF index so that it had the same weightings of core property types as the reconstituted NAREIT index. Finally, we restated the NCREIF index for appraisal smoothing. We then ran tests to determine in a statistical sense whether the restated means and volatilities of the two series were different from one another. They were clearly not.
The next section reviews the relevant literature. The third section describes our data and methodology. The fourth section presents our results. And the final section presents our concluding thoughts.
Literature Review and Theoretical Arguments
Earlier research (10) has also attempted to examine public- and private-market returns; however, the examinations have not been as rigorous. Barkham and Geltner (1995) and Geltner and Rodriguez (1997) use book, rather than market, values of equity, use long-term government bonds to proxy for the costs of indebtedness (rather than firms' reported interest expense) and do not control for differences in property-type allocations. Giliberto and Mengden (1996) examine cash flows, market values and total returns. They support the view that the wrapper does not matter when they examine the cash flow patterns generated by the different real estate vehicles. However, they do not adjust for differences in property-type mix, leverage or appraisal smoothing. Geltner and Kluger (1998) use NAREIT-reported characteristics to construct "pure-play" (unlevered) REIT portfolios, which they compare in turn to the "unsmoothed" NCREIF returns by property type. They generally found, even after making such adjustments, that pure-play REITs exhibited higher (quarterly) mean returns with higher volatility. While Seiler, Webb and Myer (1999) use a battery of statistical and financial measures to conclude that significant differences exist between the two series, they control for neither leverage nor appraisal smoothing. Despite a number of methodological differences, the study which most closely resembles ours is that of Moriarty, Riddiough and Yeatman (2004). However, they do not adjust the NCREIF returns for appraisal smoothing--though they do attempt to quantify the effects of property development as well as differential investment management fees--and they conclude their study period with the 1998 data. Moreover, none of these studies frames the empirical questions in terms of statistically testable hypotheses as presented herein. In this respect, the Capozza and Lee (1995) study, which examines performance by REIT characteristics, is more in keeping with the spirit of this study.
Yet other studies have looked to the stock market in order to investigate a link with publicly traded investment vehicles, which are then extended to an analysis of private real estate investment vehicles. For example, Gyourko and Keim (1992), looking at a variety of securitized real estate firms (not just REITs), find that the stock market reflects information about the real estate market, which is later imbedded in the appraisal-based private real estate return series. This type of investigation has been extended to include the bond market. See Clayton and MacKinnon (2003), who find that most of REIT volatility can be attributed to factors representing large- and small-cap stocks and to a lesser degree to the bond market; however, private real estate factors represent a negligible portion of REIT volatility.
Circling back to conducting the types of analyses considered in this article, another complicating factor is the pricing of REITs relative to their underlying net asset value (NAV). Historically, the aggregate premium or discount fluctuates substantially over time. Yet, the mean premium/discount is approximately zero. (11) This analysis is conceptually similar to the analyses used to examine closed-end bond funds, which also display fluctuations in their premium/discount to NAV. Shleifer (2000) argues that, in addition to more traditional reasons, (12) investor sentiment also plays a large role in these fluctuations. As a result, entrepreneurs can profit by assembling assets and packaging them to "noise" traders when such traders are particularly optimistic. And while such analyses related to REITs differ in two important ways: (1) the REITs are open-end (not closed-end) funds (13) and (2) the underlying assets are illiquid and often difficult to value, it seems plausible to argue that the large premiums (to NAV) observed in the REIT marketplace of the mid-1990s coincided with the real estate market's recovery from the real estate recession of the late 1980s/early 1990s. Consequently, any analysis of REIT returns (relative to private real estate returns) includes this imbedded investor sentiment consideration.
At a more fundamental level, the question is: Should the platform (or "wrapper") matter? Linneman (1997, 2002) asserts that, since real estate is inherently a capital-intensive industry, large publicly traded REITs ought to experience economies of scale with regard to operating performance as well as procuring both capital and managerial talent, thereby driving out smaller (publicly and privately traded) real estate firms. The notion that the increase in marginal costs declines with firm size is a notion well grounded in economic theory and supported empirically in the context of the REIT market; for example, see Capozza and Seguin (1998) and Bers and Springer (1998). While Linneman's assertion has not been met with universal acceptance (e.g., see Sagalyn 1997, 2002, and Vogel 1997 for moderately to strongly, respectively, dissenting points of view), it certainly seems plausible to argue that large publicly traded REITs (which, by construction, dominate a market-weighted return series such as the NAREIT Index) might enjoy certain scale-related benefits that private advisory firms (which, by agreement, represent the bulk of firms contributing to the NPI) do not. If so, then these benefits would contribute to the observed and persistent excess performance of REITs vis-a-vis private real estate equity instruments.
Conversely, large size is not limited to the REIT market. The assets under management by the largest real estate advisory firms compare favorably to their REIT counterparts. (14) So, the economies-of-scale argument should not be seen as de facto in favor of REITs. Moreover, classical financial theory would seem to suggest that the legal entity should have little to do with the long-run performance of an asset class. To put the assertion more bluntly: It's what's underneath the wrapper that matters in the long run. Of course, this argument presupposes that, on a comparative basis, the competing legal entities neither preclude management from acting optimally nor create incentives for management to act suboptimally. Modern financial theory takes a more nuanced view of optimal contracting and, accordingly, the subtleties of the legal environment in which REITs operate (essentially, the 1960 legislation that granted REITs pass-through tax status, along with subsequent "modernization" acts) and in which real estate advisory firms operate (essentially, the 1974 passage of Employee Retirement Income Security Act [ERISA] combined with Investment Company Act of 1940) matters.
Finally, the corporate finance literature is replete with claims that publicly traded corporations (15) suffer from a series of agency conflicts (i.e., management vis-a-vis shareholders); see, for example, Jensen and Meckling (1976). Of course, the frailties of the public equities market must be balanced against similar concerns voiced with regard to the investment advisory profession's relationship to its customers (16) (generally, pension funds); see, for example, O'Barr and Conley (1992). Ultimately, it is a matter of which format (publicly traded REITs vs. privately placed advisory relationships) offers fewer infirmities.
Data and Methodology
To examine this question empirically, there must be an appreciation of the substantive differences between the two data sets. Without controlling for these differences, the statistical testing procedures might be jeopardized. These differences include the following.
Data Availability
First, it is necessary to utilize several data sources. The starting point for the publicly traded REITs was NAREIT, which provided the names (and ticker symbols) of each firm in their equity REIT index for every year from 1980 through 2001. They also provided, beginning in...
|
