Here’s the next installment of the inflation series, joint with my blogging buddy Actuary on FIRE. Check out the other parts here:
Today’s post is about one issue I raised in the post last month: What asset classes – if any – are useful in hedging against inflation? Simple question, not an easy answer. It all depends on the horizon!
Keeping up with inflation over the long haul. And I mean really long haul!
To set the stage, let’s look at how different asset classes have helped to keep up with inflation over the decades. I have long-term return data on four asset classes: Stocks, Bonds, Cash (i.e. money market, short-term instruments), and Gold going back to January 1871:
- Stocks: I use the S&P500 index (backfilled prior to the official index start), with dividends reinvested (i.e., Total Return Index)
- Bonds: 10-year U.S. Treasury Bonds.
- Cash: I have 3-month Treasury Bill returns since 1934 and backfill the returns prior to that with the 1-year T-bill returns.
- Gold: The spot price in dollars, London Fixing.
Here are the cumulative real returns since 1871, so returns adjusted by the U.S. CPI inflation rates, see chart below. Equities leave all other asset classes in the dust! One single dollar invested in 1871 would have grown to over $14,000 by the end of 2017 if invested in stocks. But only about $45 if invested in bonds, $19 if invested in Cash (CDs, short-term instruments) and only a measly $2.90 if invested in gold (e.g., 1 ounce of gold).
The large differences in the final value are, of course, due to compounding return differences over a very large horizon, almost 150 years. Equities returned about 6.7% real on average, more than four percentage points above the bond and cash return and 6 percentage points above gold return:
In other words, all four asset classes kept up with inflation over the very long-term. But that’s a low bar! And quite intriguingly, the one that’s normally seen as an inflation hedge, gold, has the least potential of outpacing inflation over very long horizons. Especially if we knock off another 0.25% from the high ETF expense ratios in, say, the iShares IAU ETF or 0.40% in the SPDR GLD ETF we are basically left with about zero real returns over the long haul. Someone, please page William Devane!
In fact, the only asset class that had any chance at all to not just keep up with inflation but also supports a withdrawal rate in the neighborhood of 3.5-4% would be equities. Every percentage point we move to a 10-year bond or shorter-term fixed income investment will erode the prospect of making it through a 50 to 60-year retirement. It’s one of the reasons why I’m still pretty optimistic about stocks and why I dislike bonds so much. With bonds, you trade lower short-term volatility for a higher probability of running out of money in the long-term!
Inflation hedging over an intermediate (~10-year) horizon
In the long-run, we’re all dead, as some old (and already dead) economist once said. So, protecting our portfolio from the forces of macroeconomic shocks over shorter horizons should still be a high priority! Think sequence of return risk, i.e., the fact that low returns during first 5-10 years in retirement can sink a retirement plan even though the longer-term average return looked really good. And returns over 5 to10-year windows will look wildly different from the long-term averages. In other words, what the table with the average real returns above ignores is that returns were not a straight line up for those asset classes. In the cumulative return chart, I already included a “trend line” for each return series. For the math geeks, I construct this via a Hodrick-Prescott Filter to generate a trend series that “hugs” the raw data series over the duration of roughly a business cycle, i.e., about 6-10 years. And two fun facts: Ed Prescott taught me macroeconomics when I studied in Minnesota. And he won the Econ Nobel Prize in 2004! In any case, the trend lines for the cumulative returns aren’t straight lines, see chart below:
Let’s look at the annualized growth rates of those trend series and also include the annualized inflation trend (constructed the same way; HP-Filter), see chart below:
- Long-lasting inflation shocks, those that last for a roughly a whole business cycle, are rare and normally the result of a large external shock: The two World Wars and the 1970s-80s oil shocks.
- Nominal assets (bonds and cash) always perform poorly during those times. That’s expected: inflation erodes the value of nominal assets and the inflation shock causes a rise in interest rates. That implies a drop in the bond price even in nominal terms!
- During WW1, all four asset classes performed very poorly.
- During WW2, equities actually performed really well due to the bounce back from the Great Depression, while the other three performed poorly. Including gold!
- During the oil shocks, only Gold performed well, while bonds, cash and equities had some their worst trend growth rates.
In other words, hedging against these kinds of multi-year inflation shocks will be hard in practice. Simply because we don’t know what’s the exact nature of that external shock! The only certainty we have is that nominal bonds will get hammered by inflation. It’s between hard and impossible to pin down what exact asset class will do well in that next big inflation shock.
One little side note on the chart above: One redeeming feature of the gold return pattern is that the trend returns are almost an exact mirror image of equity trend returns, at least after WW1. So, forget about how well gold hedges inflation. It tends to hedge equity performance, in that it performs really well when equities suffer (Great Depression, 1970s/80s, Global Financial Crisis). Something one would miss by looking at only the monthly correlations. There is no guarantee that this pattern will prevail, of course. But in the past, a little bit of gold exposure (maybe around 5%) would have done well in my Safe Withdrawal Rate calculations!
Hedging inflation over shorter horizons
Great news! I don’t have to reinvent the wheel because some smart folks have already done all the heavy lifting on this one. I found this pretty cool research on Unexpected Inflation Hedging. Heck, one of the authors actually shares my first name, so this better be really good!
The authors first make the point that what matters for asset returns is not so much inflation but rather the surprise or what they call “unexpected inflation.” For example, 5% realized inflation might be really bad for your bond portfolio if 2% were expected. But 5% inflation might be really good for bonds if 8% were expected (and priced into bond yield beforehand) and then inflation surprised significantly on the downside. It’s a subtle difference, see chart below, but it’s the correct and proper way of doing this exercise.
In any case, the authors now correlate different asset returns over 12-month windows with the unexpected inflation (UI) readings over the same window. Specifically, they estimate UI betas, i.e., a slope parameter to gauge how much over/underperformance is coming from a 1% inflation surprise, all else equal. See results below. The broad bond and equity indexes have negative UI betas, meaning that if inflation surprises on the upside then, all else equal, stocks and bonds underperform. Actually, the S&P500 UI beta (not reported in the chart below) would have been around -2, which is even a bit more negative than the MSCI World index. So, even though equities hedge against inflation in the very long term, there is likely a negative short-term impact due to uncertainty about how aggressive the central bank will be in addressing the inflation shock.
So, in other words, a portfolio with just the broad equity and bond indexes, say 60/40 or 80/20 will still have a substantially negative UI beta (regression slopes aggregate linearly when mixing and matching asset classes, so a 60/40portfolio would have 0.6x the equity and 0.4x the bond UI beta). The way the UI betas were constructed, one would need a portfolio UI beta of +1.0 to be immune in real terms. Thus, if our portfolio has a UI beta of -2, then for every 1% of an inflation surprise, we would lose 3% of purchasing power!
Commodities (energy, precious metals, industrial metals) would have generated quite impressive positive UI betas. But in order to make a difference in the overall portfolio UI beta, we’d need a pretty substantial portfolio weight. But that comes at a price! Commodities don’t create a meaningful average return over the long run. Gold, oil, copper and aluminum and the like don’t multiply, so they don’t create any dividends and only measly capital gains. So, in order to hedge against inflation, we’d have to sacrifice some expected return! How about some of the commodity-linked equity sectors? They do have equity-like returns but also tend to have much higher volatility than the broad market index.
So, to illustrate some of the tradeoffs I outlined above, the authors of the study come up with a pretty ingenious methodology: simply draw an efficient frontier in 3D (!) instead of the usual two dimensions, i.e., add an inflation risk dimension, where inflation risk is measured as (-1)-times the portfolio UI-beta, see chart below:
Of course, we can always bring this 3D efficient frontier back to 2 dimensions. For example, we could specify a risk target of, say, 10% and then slice through the surface, see below:
And the line along that cut now becomes a new efficient frontier to illustrate the tradeoff between expected returns (y-axis) and inflation risk (x-axis), see below. And again, this looks like a real efficient frontier, i.e., an investor has to sacrifice some of the expected return to reduce inflation risk (=increase the UI beta):
Hedging against inflation is no easy task. It all depends on the horizon! Over the very long-term, equities are an extremely powerful inflation hedge. Makes sense because over long horizons, corporations have the power to raise prices and raise their corporate profits pretty much exactly in line with inflation. Over shorter horizons, commodities or commodity-linked equity seem like a great inflation hedge. But they both have drawbacks. Commodities have low expected returns and commodity-linked equities tend to have higher volatilities than the broad index.
What would I do personally? I am happy with the fact that equities keep up with inflation over the very long-term. The opportunity costs of low long-term returns make commodity investments very unappealing, not to mention the high expense ratios in some of the ETFs. But I’m also concerned about a potential future recession that may look less like 2008/9 and more like the 1970s; equities and bonds go down, i.e., bonds don’t provide any diversification. I am probably going to increase our real estate investments going forward. That would include buying a primary residence again (yes, we do consider a home an investment – it hedges housing inflation). But we also plan to move more money into our multi-family private equity funds. Similar to the REITs, rental properties tend to have positive UI betas. And they have decent expected returns and relatively low volatility. But that’s a topic for a future post in this series! Stay tuned!