Do REITs Behave Like Property or Equity?

By Watson Wyatt Investment Consulting

ONE WAY of investing in property is through Real Estate Investment Trusts (REITs), which are listed vehicles that own property. Investors often wonder whether these REITs behave more like equity than property. If they did, then they would not diversify equity returns as well as other property investments.

Summary

REITs tend to behave more like equity in the short term and direct property in the long term, once lagged effects are taken into account. In that sense, it is as if the renowned investor Benjamin Graham was talking about REITs when he said, "In the short run, the market is a voting machine. In the long run, it's a weighing machine." The results from our analysis and various academic studies support this view. We consider each in more detail below, after summarizing the main features of REITs.

Our analysis of 20 years of US data finds that REIT returns were leading and amplified indicators of direct property returns. Very poor performance of REITs therefore typically preceded poor performance of direct property.

These historical return patterns have implications for REIT investors. Consider an investor that wants to use REITs to gain a proxy exposure of six percent to direct property. Given their historical return patterns, perhaps resulting from the leverage used by most REITs, the investor might want to allocate around four percent to REITs, retaining the rest in cash. Doing so would provide a proxy allocation to direct property of around 6 percent.

If one assumes that these historical patterns will continue, then REITs pass one test for inclusion in an investor's portfolio. That is, REIT returns do:

• behave like those of direct property over the longer term (that is, for periods of, or longer than, two years that include lagged data); and
• do not have as strong a positive relationship with equity over the long term.

Taken on their own, these findings suggest that investors might wish to access the returns of global property by investing in REITs and perhaps other listed property securities¹. Before doing so, however, investors should compare the attributes of REITs alongside those of other property options.

We address these attributes, including leverage, liquidity, fees and return volatility, in a future paper. We devote the rest of this paper to a more technical discussion of the points above.

Background

Investors that buy a REIT gain exposure to the future income stream of a property portfolio. For its dividends to be exempt from corporation tax, a REIT must pay no less than 90 percent of its income to investors. REITs must also have at least 75 percent of their assets in, and income from, specified property activities.

US-based REITs have three sub-categories, which denote the specified property activity:

• Equity REITs invest in equity investments in the private property market; and
• Mortgage REITs invest in property-related debt, such as Mortgage Backed Securities.
• Hybrid REITs are all remaining REITs.

Most REITs have a high degree of financial leverage or debt. Equity REITs in the US, for instance, typically had leverage of 66 percent on 31 March 2008².

Given the size and history of the US REIT market, we chose it as the basis for our description of REITs. Investors should be aware, however, that REIT structures vary by jurisdiction.

Our analysis of REIT returns

To gauge whether REITs behave more like property than equity, we reviewed relevant academic literature and analyzed historical REIT returns. We now consider this analysis.

We studied the total returns for US equity, REITs and direct property from 1979 to 2008, using local currency data for the S&P 500 Index, the NAREIT Index and the NCREIF Index, respectively³. We focused our analysis on the US market, as its data on REITs is relatively reliable and extensive.

First, we calculated the correlations that REITs exhibited with both equity and direct property. As commentators have suggested that REITs behave differently in the short and medium term, we also considered different periodicities of data. That is, we analyzed the correlations over:

• 120 separate quarters;
• 60 separate semi-annual periods;
• 30 separate years; and
• 15 separate two-year periods (or 'biennial')⁴.

 

Figure 1 contains the results of this analysis.

So, did REITs behave like property or equity over this time? According to these positive correlations, they behaved like both⁵. This suggests that positive returns from REITs usually accompanied good returns on equity and direct property from the same period⁶. (As they use returns from the same periods - be they quarterly or otherwise - academics call these correlations 'contemporaneous'.) More specifically:

• REITs typically resembled equity more than direct property. For example, the correlation between the annual returns of REITs and equity exceeded that for REITs and direct property (0.46 vs. 0.23).
• REITs resembled equity slightly less, and direct property slightly more, as the holding period lengthened and lagged data was introduced. For example, the correlation between the quarterly and biennial returns of REITs and direct property were 0.22 and 0.32, respectively.

 

That said, commentators often suggest that the annual returns of direct property are correlated with the annual returns of REITs from prior years. We therefore tested the same data as before, but checked for lagged (and leading) relationships of this form.

For the quarterly returns of REITs and direct property, for instance, that left us with five correlations to calculate:

• the contemporaneous correlations (as in Figure 1);
• the correlation between the quarterly return of REITs and the quarterly return of direct property from one year (and two years) earlier; and
• the correlation between the quarterly return of REITs and the quarterly return of direct property from one year (and two years) later.

After making these calculations, we selected the strongest correlation of the five for each asset and periodicity analyzed. By strongest, we mean the correlation with the highest value, irrespective of whether it is positive or negative. We present these strongest correlations in Figure 2, along with the lead or lag with which they are associated.

The results from Figure 2 show similar behaviour to those in Figure 1, but also reveal that:

• In shorter time periods, REITs resembled equities more than direct property, irrespective of lagged effects. As with many traded assets, the prices of REITs and equities also often fall dramatically in times of market turmoil.
• When lags were allowed, REITs resembled direct property more than equities
  over longer time periods.
• REIT and equity returns most resembled each other when no lagged or leading data was used. This suggests that information quickly finds its way into the prices of REITs and equities.
• REIT returns most resembled the direct property returns that were lagged by one or two years. This behavior could reflect the use of appraisal pricing in property markets, where information takes a considerable time to feed into prices.

The strongest correlation that we found was between the biennial returns of direct property and lagged REITs, at +0.65. This correlation of +0.65 was so high that we decided to study the data underlying it in more detail. We do so in Figure 3, (overleaf) by plotting the 14 separate data points in question. (The point in the bottom-left quadrant, for example, plots the -18.8 percent return to REITs over 1989 and 1990 against the -9.6 percent return to direct property over 1991 and 1992.)

REIT returns as a leading and amplified indicator of direct property returns
Considering the 14 data points in Figure 3 provides us with more details than we could have gleaned from merely one correlation statistic. By examining the line that provides the best fit of these 14 data points, we learn that REIT returns typically lead those of direct property in a super-cyclical way⁷. That is, over the thirty years analysed, every 1.5 percent of extra return for REITs typically preceded a 1 percent increase in the returns to direct property.

Assuming that this pattern continues, it has major implications for investors that use REITs as a proxy exposure for direct property. History suggests that investors should allocate around 4 percent to REITs in order to gain an exposure of 6 percent to direct property (6 percent being the product of the 1.5 multiple noted above and 4 percent). The remaining 2 percent exposure should be invested in cash. This relationship might well reflect the different gearing levels of REITs and direct property.

Some investors might go further and use this historical relationship to proxy their near-term expected returns of direct property. While we would be most cautious about extrapolating this relationship too far (given its limited data), we note that REITs returned -48.9 percent over 2007 and 2008. If the pattern in Figure 3 continues then one should expect a return from the NCREIF Index of -35.5 percent over 2009 and 2010. This compares with the cumulative return implied by US property derivatives over the corresponding time of -33 percent.

That said, the limited liquidity of direct property curbs an investor's ability to profit in this way.

 

Three main academic papers

Using a range of statistical techniques, academics and practitioners have long considered the relationships between REITs and direct property. We studied three papers that tackle these matters, all of which reach similar conclusions to ours.

Whilst the study by Myer and Webb (1993) is old, it introduces a more technical statistical approach to determine the long-term structural relationships of REITs. It finds that REIT returns have previously had strong long-term links with direct property⁸. These findings are reinforced by the more recent and elaborate paper by Morawski et al (2008).

Consistent results are also found by Lee and Stevenson (2005) in a less technical study of the effects of including REITs in a portfolio.

Notes
1. While we have not studied listed property securities in this analysis, it seems reasonable to assume that their return behavior would not be unlike those of REITs. That said, listed property companies that are not REITs tend to have higher leverage, pay lower dividends and, because of their ability to retain more cash, are more likely to be involved in development. For further details, see McDonald (2005).
2. See NAREIT (2009). We define leverage by taking the REIT's total debt and dividing it by the total market capitalization (which includes debt).
3. NAREIT, the National Association of Real Estate Investment Trusts, is the organization that represents the real estate investment trust industry in the US, and produces the index of the same name. The National Council of Real Estate Investment Fiduciaries (NCREIF) is a similar body for US real estate professionals, and produces the NCREIF Property Index. Both of these indices used in our analysis only represent properties from the US, although their sector exposures differ to some extent.
4. In doing so, we ensured that we calculated the correlations using data that did not overlap (for instance, by using separate calendar years). Otherwise, we would have been forced to place less importance on the earliest and latest return data available - a key problem with analyzing rolling multi-period returns.
5. We also decided against analysing longer periods - of, say, three years. We did so, as this would only provide us with ten observations and so not enough information to calculate a correlation with confidence.
6. In fact, according to Lee and Stevenson (2007), Equity REITs have tended to behave more like mid-or small-capitalization stocks with a value orientation.
7. As the box in Figure 3 shows, despite the limited data, other statistical techniques confirm that this best-fit line provides a reasonable guide to the historical return patterns of REITs and direct property.
8. In statistical parlance, the authors found that equity REIT returns typically 'Granger caused' the returns on direct property.