Are REITs a Distinct Asset Class in a RP portfolio?

I hinted at this in my VNQ post - while I do invest in VNQ and DFREX, I am not altogether sure I should. Should I have an overweight allocation to REITs beyond what I’d get anyway in broad index funds? Are REITs a distinct asset class? Do they belong in an RP portfolio?

This is the second in a series on Real Estate investing in an RP Portfolio. The first previewed the discussion about REITs with my look at VNQ. More posts coming on: expert opinion, backtesting, alternatives to REITs and more.

To go back to the basics of RP, I’ll return to Ray Dalio’s original formulation of the “Holy Grail” of investing, and relay the three fundamental criteria for thinking about asset allocation:

  1. Positive Expected Return - keeps the savvy investor out of pure speculation.
  2. Uncorrelated with other Asset Classes - ideally, we’d like negative correlation with other asset classes, but zero correlation is fine, too. We’ll evaluate this quantitatively, looking at correlation coefficients over time.
  3. Exposure to Different Risks - related to #2, but for this, I’m looking for a plausible story of why this asset class will respond differently than others. This is a more qualitative judgment; I want to see if the non- or negative correlation makes sense or is just a data surprise like Bangladesh’s butter production numbers.

With these criteria in mind, let’s see how REITs stack up:

Positive Expected Return

REITs have performed quite well over pretty much any time frame you can think of that end with the current day. Using Portfolio Visualizer’s Asset Class Backtest tool, I looked at the performance of REITs compared to: 1) The Total Stock Market, 2) The Total Bond Market, and 3) the Classic 60/40. Since the data goes back to 1994, I used that 28 year timeframe, a 16 year timeframe back to 2006 (so I could capture the global financial crisis), a 5 year and the past 18 months. Here are the compounded annual growth rates over those periods:

Clearly, we can expect a positive return going forward, well, at least as much as we can expect it with any asset class. Returns over 28 years, 16 years, and 5 years are all good, and then in the recent market downturn, REITs have been impressive in the face of broad market declines. As for particular ticker symbols, REIT index funds have had a rough go of things year to date, but when considering short-, intermediate-, and long-term returns, have been positive.

So, to judge the first criterion: yes, REITs have a positive expected return.

But wait…there’s a bit more to consider. The returns for REITs have typically not beaten the US stock market as a whole. They’ve been close, for sure, and it’s not like every asset class has to beat the US stock market to be in an RP portfolio. But the data present a question: if the pattern of price ups and downs with REITs hasn’t on the whole been different than the US stock market, then would it be worth OVER-weighting them in a portfolio? Based on return data, there really would be no reason to move resources from the stock market as a whole to invest more in REITs unless you derived a significant diversification benefit from them. To get at that, we have to turn to the second and third criteria.

Correlation with other Asset Classes

To answer this, I looked at Portfolio Visualizer’s Asset Class Correlation tool which features a giant correlation matrix of fifteen asset classes and the correlations between all pairs. Since the data for this go back to 2008, I use that 14 year timeframe, plus 5 year and 18 month timeframes. Since the matrix is unwieldy, I’ve simplified the table to just show the correlations between VNQ and IVV (S%P 500 ETF) and then VNQ with AGG (US Aggregate Bond Market ETF).

Correlations between .74 on the low side and .9 on the high side show that VNQ functions like a slightly different collection of stocks rather than as a distinct asset class. For the record, the correlations between VNQ and established alternatives like gold and commodities were generally very low, ranging from 0 up to about .3.

Another look at correlations over time show that REITs vary slightly from both US and international stocks, but not to a great degree. The following chart from Daniel Sotiroff in Morningstar compares VGSIX (Vanguard’s REIT Index mutual fund) with VTSMX ( Total US Stock Market), VGTSX ( Total International Stock Market) and VBMFX (Total Bond Market).

Again, correlations ranging typically in the .3 to .7 range are something, but they do not scream “uniqueness.” Additionally, if considering the eleven sectors that make up the stock market (h/t again to Sotiroff), there is little to say that this is all that notable. Of the eleven, the REIT sector has the third lowest correlation with the total market at .59 in the period from 2000 to 2019, trailing utilities (.4) and consumer staples (.57) and just ahead of telecommunications (.62) and energy (.64). We’d hardly consider consumer staples a separate asset class, and its claim might be stronger than that of REITs (meanwhile, utilities are intriguing; more on them to come).

To judge the second criterion: no, correlation figures don’t show that they really are that distinct.

Exposure to Different Risks

I’ll go into this one more in my next blog post that gathers expert opinion on risk factors and REITs, but for now: is there a plausible story I can tell about REITs that suggest they will behave differently than the general stock market, which, let’s recall, has a higher expected return?

When you think about them, the reason we cluster REITs together is not based on what they invest in, but rather the specific elements of the tax code which allow them to avoid one layer of taxation at the corporate level as long as they pass along 90% of the income to shareholders. It is not just connection to real estate which defines them, and indeed, if we really did want to collect every equity asset related to real estate, we’d have a lot of head scratching omissions; home builders like Toll Brothers (TOL) and NVR, and building material suppliers like Builders FirstSource (BLDR) are not REITs, even though they are sol clearly tied to the sector. These are all just regular equities, available even in industry-specific ETFs like XHB.

Do companies like Prologis (PLD), a global logistics company that owns warehouses, Welltower (WELL), which owns and operates housing for seniors, or Public Storage (PSA), where people keep all the extra stuff they’ve bought but can’t/won’t throw away, have any more claim to represent real estate than those above? To my mind, no, but PLD, WELL, and PSA all have elected to organize themselves as REITs, whereas the others haven’t. When people talk of REITs, they often mistakenly think they are a true mirror to reflect real estate in its entirety. Real estate is a large and complex part of the economy that may have its own risks, but REITs are just defined by how they are taxed.

As for this third criterion, there is more to the argument and I’ll be reviewing more of the literature to answer this aspect of the question. But just based on my first-pass assessment, I’m going to lean slightly to the “no” side on this.

Conclusion and Next Steps

Judging REITs by Dalio's “Holy Grail” criteria , it’s looking more and more that REITs aren’t really a distinct asset class. This does not, of course, mean that REITs are a bad investment. With a positive expected return and an always uncertain future, it’s impossible to say that investment in them is a losing proposition. There will be some diversifying effect on your portfolio if you include them, as well, though this would happen if you had a carve out for consumer staples or utilities, too.

What is at issue is whether they deserve an overweight allocation. REITs are 4 to 5% of the market already, and if you own VTI, you’re getting all the REITs in VNQ, plus all the other real estate-related businesses above which happen to not be organized as REITs.

In the next post, I’ll look at how some academics and pros have answered this question, as we move towards a more definitive judgment.