Friday, June 4th, 2021

Adding REITs and TIPS to the mix

In our last note (Is it Time for Commodities? – 20th May 2021) we argued that it was possible to enhance the returns of a systematically-managed portfolio of equities and fixed income, by adding commodities into the mix. We showed that our approach generated higher risk-adjusted and absolute returns, with smaller and shorter drawdowns. This week, we want to expand the range of eligible alternative assets to include REITs and TIPS. There are, of course, other alternatives available to sophisticated institutional investors, such as hedge funds, private equity, private debt, infrastructure, and conventional real estate. We are only going to consider publicly-traded asset classes, with full and continuous price discovery, immediate liquidity, no management fees, low transaction costs, no leverage and no hidden/unreported volatility.

Let us start by recapping the man points of our previous note:

  1. A passive approach with a constant exposure to commodities throughout the cycle does not work. It reduces returns, adds to risk and increases the maximum drawdown.
  • Using a broad, unchanging basket of commodities also does not work, because the behaviour of the portfolio we are trying to enhance will demand different hedges at different stages of the cycle. We can transfer our normal process from equities/bonds to commodities, in order to create a commodity portfolio with higher risk-adjusted returns than the index or any individual commodity.
  • Including commodities within the portfolio is no substitute for actively managing the relationship between equities and fixed income. This is the primary and most liquid way of generating superior risk-adjusted returns.
  • The best approach is to hedge a systematically-managed portfolio of equities and bonds with a systematically-managed portfolio of commodities. This requires that we restrict the number of commodities to just three: crude oil, gold and copper. Any more adds unnecessary complexity and tends to reduce liquidity.
  • Because commodities are so much more volatile than rest of the portfolio, their maximum exposure must be capped at 25% of the portfolio. However, we are very comfortable with the idea of having no exposure to commodities when appropriate, so the average exposure over the 25-year period works out at 8%.
  • We only want to be invested in commodities while they are producing superior risk-adjusted returns. We are likely to have most of our exposure in only one of the three commodities at any one time. The occasions when all three outperform the standard portfolio simultaneously are rare.
  • The relationship between these three commodities and between commodities and the standard portfolio is generally characterized by periods of sustained momentum (up or down), which can be systematically exploited over the medium term, subject to appropriate risk controls.
  • In our experience – based on extensive simulations – any attempt to use a range-trading or mean-reversion strategy, eventually leads to disaster. Adding leverage to this mix (with or without stop-losses) just speeds up the process.
  • We are not trying to harvest any returns from arbitraging different parts of the futures curve against the physical commodity, or different but related commodities. These are essentially mean reversion strategies, no matter how they are structured.
  • We are not interested in the commodities themselves and we don’t need to understand their supply and demand dynamics. We only care about them as financial instruments, which have observable investment characteristics in relation to the standard portfolio.

This sounds like an awful lot of rules, but in practice it saves time, because it stops us worrying about things we can’t control, or collecting information we can’t use. It keeps us focused on the essential task of diversification – only investing in assets when they have a reasonable chance of generating better risk-adjusted returns, than those produced by the standard equity/fixed income portfolio. This is particularly relevant to commodities. Since 2009, our model including commodities has significantly underperformed a basic portfolio comprising a constant 50/50 split between US equities and US Treasuries. There have been two false dawns, when it looked as though it might start working again, but the downtrend so far has unforgiving. Our reaction is simple. If it isn’t working, don’t do it – which is why don’t include commodities in our published models and why we rarely write about them.

The more asset classes we are prepared to consider, the more we will have to leave out of our portfolio at any one time, because they are temporarily not risk efficient.  The rules that apply to commodities will also be applied to REITs and TIPS.

Diversification into REITs

The aggregate index of all US REITs has produced an annualised return of 9.9% since the inception of the model, but this comes with annualised volatility of 23% and a maximum weekly drawdown of 72%, making it less risk efficient than equities or commodities.

However, just like commodities, if we systematically manage the three most liquid and least-correlated sub-indices (and ignore the others) we can generate better absolute and risk-adjusted returns. The three we use are Residential, Retail and Industrial, which offer broad geographical coverage across the USA, unlike Office REITs, which tend to be over-exposed to the major coastal cities. Our systematically-managed portfolio of these three sub-indices has produced an annualised return of 11.5% since inception, with volatility of 25% – clearly superior in absolute terms and marginally more efficient in risk-adjusted terms.

If we substitute REITs for commodities in our systematically-managed portfolio, we find that it outperforms the standard US 50/50 portfolio over the whole 25-year period, but it underperforms the commodity version, which means we should ignore it. However, it has beaten the commodity version since 2009 and there are two recent periods (Q4 2016 to Q2 2018 and Q3 2020 onwards) when it has significantly outperformed the standard model.

Results

Since inception, this model has produced a total return of 1575%, compared with 752% for the systematic equity-bond portfolio and 628% for the benchmark. The annualised figures are 11.7% compared with 8.8% and 8.1%.

The maximum drawdown on a weekly basis is -17.7%, which is worse than the standard portfolio (-12.2%) but much better than the benchmark (-25.0%). There are only two years in which the process produces a negative return: 2008 and 2018. In both cases it is -2.1%.

Return on risk is 1.13 compared with 0.97 for the standard portfolio and 0.93 for the benchmark. The information ratio compared with the pure fixed income portfolio is also 1.13 vs 0.75 for the standard portfolio and 0.64 for the benchmark. In other word, the regime-switching model increases returns by more than the extra risk it incurs.

The regime switching model is also good at managing extreme market conditions. A good risk management system should ensure that the maximum drawdown is never more than three standard deviations below the average annual return.

Multi-Asset Diversification

The score for the regime-switching model is -2.84 compared with -3.81 for the benchmark. The standard portfolio is better (-2.31) but that is because it follows our process and takes less risk. None of the individual asset classes has a score of better than -3.0. The score for our fixed income portfolio is -3.33.

This suggests that we should allow our diversification engine to choose between REITs and commodities in exactly the same way as we choose between different commodities and different REIT sub-indices. We can also incorporate TIPS into this process as well.

We could look at lots of different maturities for TIPS (e.g., 5 and 10 and 30 years). However, we don’t do this when we are looking at our main US Treasuries position, so we have decided to be consistent and only incorporate the rolling 10-year maturity. (There is also the problem that issuance of a full spectrum of TIPS maturities was only completed towards the middle of the last decade.)

We find that this mixed diversification strategy produces returns which are better than the REITs diversification model, but worse than the commodities diversification model. However, it has the lowest overall volatility and the least-worst drawdown of any of these strategies. Risk-efficiency is only marginally below that of the commodities strategy.

A New Regime-Switching Model

It is clearly worth seeing if we could combine all three diversification strategies in a way which produce an even better result. However, if we are going to do this, we must be consistent with first principles and allow the model not be diversified at all – i.e., just a systematically-managed portfolio of equities and bonds. This, in turn, implies that we must allow the model not to be systematically-managed on occasions and revert to the benchmark portfolio of a 50/50 split between US equities and US Treasuries.

This is in danger of becoming very self-referential, so the final part of the process must be really simple and unequivocal. We have five different model portfolios to choose from:

  • Benchmark 50/50 US equities and 7-10 year US Treasuries;
  • Standard systematically-managed portfolio comprising US, Pan European and EM equities plus 7-10 year US Treasuries, US investment grade and US high yield;
  • Systematic equities and bonds plus commodity diversification model (oil, copper & gold);
  • Systematic equities and bonds plus REITs diversification model (residential, retail & industrial);
  • Systematic equities and bonds plus mixed diversification model (commodities, REITs and TIPS).

In all cases, the diversification model is capped at a maximum of 25% of the total model and the through-cycle average exposure is less than 10%.

The test is very simple. At the end of every quarter, we choose the model which has produced the best total return over the previous four quarters and follow that strategy until the end of the next quarter, when we review everything again.

These results are achieved without too many changes of model. Since inception, there have been 102 quarterly reviews but only 37 model changes. On average, each regime lasts just under three quarters.

The most common regime is the standard, undiversified equity/bond model, followed by the benchmark portfolio. The most common diversification regime is commodities, followed by REITs, followed by the mixed model. The model only uses a diversification strategy 45% of the time.

Summary

The starting point for this analysis was to help investors find a way of hedging the portfolios against the impact of a potential surge in inflation. Rather than hedging specifically against reported or expected CPI, we make sure that we have access to the widest possible range of publicly-traded securities, but we only invest in them when there is a clear and obvious contribution to risk-efficient superior returns.

Our conclusions are as follows.

  • There is no single diversification strategy which has consistently offered superior risk-adjusted returns throughout the period since inception. We strongly believe that the next 25 years will be just like the last 25 years, in this respect.
  • There are many occasions when a systematically-managed portfolio comprising just equities and bonds cannot be beaten and any diversification strategy is sub-optimal. There are also occasions when the benchmark performs best. These periods tend to occur in clusters and can be identified simply by monitoring a series of model portfolios.
  • REITs offer a good way of protecting income from the impact of inflation but are very bad at protecting the capital value of the portfolio. They are highly correlated with, and a high-beta version of, equities in the downswing. However, then can provide an effective way of enhancing returns in the early stages of the recovery.
  • TIPS offer complete protection against the impact of reported inflation. However, they provide very little income and tend to underperform when inflation surprises to the downside. The opportunity cost of a significant position over the whole cycle acts as major drag on portfolio returns, but they can be combined with other diversification strategies to reduce portfolio risk on occasions.
  • Commodities offer the best overall portfolio hedge against the impact of expected inflation, but are always significantly more volatile. They tend to underperform when inflation expectations decline. Continuing to invest during these periods is simply counter-productive.
  • In the period since the global financial crisis, the effectiveness of all diversification strategies has been severely reduced. Since the end of 2009, the most common regime is the benchmark 50/50 portfolio, which is used 46% of the time, compared to only 10% of the time before then. For the explanation, we need look no further than the actions of the Federal Reserve. One day this will change, but until then, all attempts at portfolio diversification should be carefully controlled.
  • For what it is worth, the current recommended regime is the REITs diversification model, into which we switched at the beginning of Q2 2021 after four continuous quarters of being in the benchmark regime.

Leave a Reply

Full Blog Archive:


Other Recent Posts