Maybe I have been approaching REITs wrong - maybe it’s the packaging that is the problem, not the REITs themselves. In this test, I took an unusual approach (for me) - chose ten individual REITs to see how they might compare. Good news: better than VNQ.
The backtests I have done so far have pointed to some problematic, though still very tentative, points about REITs and whether they have a place in a Risk Parity portfolio:
- You may not even need them at all
- You can probably do better by taking whatever allocation you were going to devote to REITs and splitting it up into two-thirds Small Cap Value and one-third intermediate-term Treasuries.
- Utilities could very well do everything you want REITs to do and do them better: higher returns, less correlation, and paying out qualified instead of ordinary dividends.
But maybe I have been approaching REITs wrong by focusing on REIT Index funds such as VNQ. Broad-based index funds are generally my preference over narrow funds or especially over individual assets, but there are times where simply going with the index ETF that claims to represent an entire asset class can steer you wrong. This is certainly the case with investing in BND, or Vanguard’s Total Bond Fund, for example. The problem with BND is that it purports to be a one-stop shop for bonds, but winds up being neither fish nor fowl. By stuffing together some bonds for diversification with others for stability, and then with an overlay of bonds for income, you wind up really getting none of these characteristics. People are then surprised when it doesn’t do what they want them to do, but that is a bit like expecting the Pushme Pullyu from Dr. Doolittle to take you where you want to go.
So, maybe REITs do have a proper role in an RP portfolio if we move away from a Total REIT Fund and instead choose individual REITs. To be clear, I don’t advocate picking individual securities (or rather, not anymore. I tried and failed that approach - repeatedly! - in my younger days), but I’ll play along and try to create a realistic slate of REITs for investment. I’ll focus on REITs that will provide more diversification in my portfolio.
Selecting Individual REITs
I’ll look for ten of them, and invest equal amounts in each. To find these ten, I’ll need a process:
- To find REITS that are less correlated with the market, I’ll focus on those with a beta (or correlation to the stock market) less than that of VNQ, which is .83. Since I already analyzed VPU as a superior alternative to VNQ, and its beta is .48, I’ll try to keep my choices close to that, so I’ll go with a maximum beta of .58.
- I might get some fly-by-night ones with that, so I’m going to take the ten largest by market cap with a market beta of .58 or lower.
This is a pretty simple screen, for sure. If anyone just points out that I didn’t choose the ten “best” REITs, then point taken. Ideally, I’d like to just pick the ones that are going to go up in the future (the Will Rogers approach to investing), but that is a game with no end, so my two simple criteria will have to do.
Here are the ten to make my list (from Barchart):
An interesting bunch of companies here: cell phone towers, digital infrastructure, storage, and refrigerated warehouses. In case you were wondering, COLD was the 41st REIT by market cap, so my screen picked up about a quarter of available REITs. until it got to ten. It was also skewed toward the largest REITs: six of the top nine by market cap met the correlation criterion. As for dividends, since they aren’t a big concern of mine, I didn’t prioritize them, but I did the math and found that the yield for these ten would come in at 2.61%, just a bit lower than VNQ’s rate of 2.95%.
The Test Portfolios and the Comparisons
To explore how they might do, I’ll use a modified version of a Risk Parity portfolio suitable for someone really into REITs. It has a large REIT allocation so that we can really see their impact:
- 40% Total Stock Market (VT)
- 20% Long-term Treasuries (TLT)
- 10% Commodities (PDBC)
- 30% REITs, so either all as VNQ or as 3% each to the ten companies above.
I also wanted to have some good comparisons, so in addition to the VNQ-dominated portfolio and the one with the individual REITs, I’ll run a third portfolio with Vanguard’s Utilities ETF (VPU) instead of VNQ, and then use a standard 60/40 portfolio as a benchmark, as well.
If interested, you may want to check out the full correlation matrix for the REITs along with annualized returns since February 2018, when COLD was launched. One quick takeaway: nine of the individual REITs had a higher annualized return than VNQ at 8.83%, while the tenth (DLR) was still a respectable 7.45%. Meanwhile, the highest came in at 24% (EXR). Even though I didn’t have any type of screen for REITs with higher returns, those are what came up, so we’ll likely have a backtest that paints a rosy picture for individual REITs.
Backtests
In a nutshell, holding individual REITs is certainly an improvement over VNQ, and an improvement over holding a utilities index fund, as well. One very important caveat, though: the backtests don’t go back very far, just to February 2018 (you can see the full backtest here), so keep that in perspective. For the record, for all portfolios, dividends are reinvested and they are rebalanced annually. The results:
Key things to notice:
- The portfolio with the individual REITs is the best of the four, with a CAGR 1.78% higher than the Risk Parity with Utilities portfolio. The volatility of the utilities portfolio was slightly better that the individual REIT portfolio, but due to its lower returns, its Sharpe ratio was inferior to the individual REITs portfolio: .82 compared to .94.
- Having the ten individual REITs was much better than the portfolio with just VNQ: 2.44% higher CAGR than the VNQ version, with a lower standard deviation and a higher Sharpe ratio.
- Having Commodities and REITs in the portfolio, whether in index fund form or with individual REITs, was better than the standard 60/40, so score one for Risk Parity in general with this.
After seeing the results, I wondered if the short timeframe was too limited. COLD is the youngest stock, so I ran another backtest where I eliminated COLD and allocated its 3% to the top 3 (an additional one percent to AMT, CCI, and EQIX) instead. I also switched out DBC for PDBC to gain another few months for the backtest. These moves stretched the backtest three years back to 2014. By the way, I tried extending the backtest even farther, but this meant to dropping so many REITs that I was getting away from my original test.
The results for this nine individual REIT portfolio were consistent with the first: the RP portfolio with the individual REITs was the best of the bunch, and by a considerable margin. The CAGR was 1.77% higher than RP with utilities and almost 3% higher than with VNQ. Once again, it was a smidge more volatile than the utilities version, but its higher CAGR meant a higher Sharpe ration anyway.
Meanwhile, the portfolio containing VNQ was the worst of the four, even lower than the 60/40. In a way, this is further strength for the argument that individual REITs have performed well - we can't just say the last eight years were a great time for REITs in general and we just cherry picked off the top of those. The fact that the general REIT strategy finished fourth but the selective strategy finished first is indeed important.
Conclusion
Call this a victory for individual REITs, especially compared to the broad VNQ but also compared to indexed utilities. By focusing on individual REITs with lower correlation to the stock market than the Total REIT fund has as a whole, you get better backtest results in terms of compound annual growth rate and Sharpe Ratio. These backtests were not that long, however, so I only offer that judgment very tentatively. One test went back to 2018 and the other to 2014, so this isn't ground as solid as in the VNQ vs. VPU test, which went back to 2005 and 2010. Still, its enough to suggest that if you aren't tied to index funds in your investing strategy you can benefit from holding REITs individually.
The original version of this post:
