Tracking Error: Return Stacking versus Replacement
Many investors look to alternatives as a way to introduce diversification into their portfolio. Traditional portfolio construction, however, requires them to subtract traditional holdings before they add diversifiers, creating a tracking error double-whammy. In this article, we show how return stacking can actually help investors introduce alternatives with less tracking error.
For benchmark-sensitive investors (and, let’s be honest, who isn’t at least a little benchmark sensitive), tracking error is an incredibly important metric.
If you’re unfamiliar with the term, tracking error captures the volatility in the difference of returns between your portfolio and its benchmark. You start by taking the returns of your portfolio and subtracting the returns of the benchmark. What’s left over is your active returns, which captures any return due to active security selection, asset allocation decisions, or off-benchmark bets you’ve made. The volatility of that return stream is your tracking error.
Tracking error is important because it tells you how much risk you’re taking in the pursuit of excess returns. (Of course, not all risk shows up as volatility, but we’re going to handwave that point for now.) In an ideal world, we’d have access to pure, undiluted alpha that shows up with no volatility. In the real world, we must take some risk. Nothing ventured, nothing gained…
While many institutions will measure and track their active risk explicitly, keeping it within a well-defined budget, many other investors will simply manage it by feel, finding their limits when their portfolio and its benchmark diverge too far – especially when that divergence manifests as underperformance.
Ad hoc management of tracking error in stocks and bonds can be as simple as dialing down exposure to active managers in favor of benchmark exposure. For example, a position in a concentrated value equity manager could be pared down in favor of passive large-cap equity exposure. Or you could simply replace them with a lower tracking error value manager.
But things get a bit trickier when investors try to incorporate alternative asset classes or strategies.
Tracking Error with Addition by Subtraction
For most financial advisors we work with, the dollar-weighted average portfolio of their practice looks similar to a 60% stock / 40% bond allocation. The stock and bond allocation is frequently held in a mix of passive and active exposures, and it’s not uncommon to see a 10% allocation to “alternatives,” which is a broad, sweeping term for anything that’s not stocks and bonds.
To make room for the alternatives, the advisor has to reduce exposure in stocks, bonds, or some combination thereof. That means the tracking error of our new portfolio not only comes from the thing we add, but also from the exposure we subtract.
Consider this example: starting with a 60/40 portfolio, we sell 6% stocks and 4% bonds to make room for a 10% position in passive commodities (“54/36/10”).
Our tracking error is going to be from the difference in returns between the 54/36/10 and the 60/40. Note, however, that these portfolios have substantial overlap. In fact, 90% of the portfolio is the same!
The tracking error comes from the difference between the portfolios. It captures not only the returns of the thing we added (commodities), but also the opportunity cost of the benchmark positions we had to sell to make room (stocks and bonds).
This means that when it comes to managing tracking error, we not only have to be thoughtful about what we’re adding, but also what we’re selling. A 50/40/10, a 54/36/10, and a 60/30/10 will all have substantially different tracking errors to a 60/40 portfolio.
In Figure 1, we plot the active returns of these three examples. With the benefit of hindsight, we know that commodities struggled during the 2010s while equities soared, so it is no surprise that the 50/40/10 approach substantially underperformed the 60/40 portfolio.
Figure 1: Active Returns of Different Asset Allocation Schemes
Source: Tiingo. Calculations by Newfound Research. Stocks is the SSgA S&P 500 Trust (“SPY”). Bonds is the iShares Core U.S. Bond ETF (“AGG”). Commodities is the Invesco DB Commodity Index Tracking Fund (“DBC”). Returns are gross of all fees, transaction costs, and taxes with the exception of underlying fund expense rations. Returns assume the reinvestment of all distributions. 60/40 is 60% Stocks / 40% Bonds. 50/40/10 is 50% Stocks / 40% Bonds / 10% Commodities. 54/36/10 is 54% Stocks / 36% Bonds / 10% Commodities. 60/30/10 is 60% Stocks / 30% Bonds / 10% Commodities. Past performance is not indicative of future results.
As we’ve discussed previously, when selling something to make room in the portfolio, we’re implicitly short versus our benchmark exposure. For example, if we sell stocks to buy commodities, we have effectively overlaid our benchmark with a “long commodities / short stocks” trade. Is that actually the long-term trade we want to introduce?
We can also think about this trade as defining the hurdle rate our alternative needs to clear before it adds value to the portfolio. If we sell stocks to make room for commodities, we need those commodities to outperform stocks to add any value. (This isn’t strictly true due to the math of diversification and compounding, but that’s another article for another day.)
Tracking Error with Addition by Stacking
What if we didn’t have to sell anything to make room in our portfolio? What if we could just add our alternative on top? This is, effectively, what return stacking aims to achieve.
What if instead of selling stocks and bonds, we simply borrowed the money to buy commodities exposure? In doing so, we effectively stack the returns of commodities (in excess of the borrowing costs) on top of our 60/40 portfolio!
Figure 2: Active Returns of Different Asset Allocation Schemes
Source: Tiingo. Calculations by Newfound Research. Stocks is the SSgA S&P 500 Trust (“SPY”). Bonds is the iShares Core U.S. Bond ETF (“AGG”). Commodities is the Invesco DB Commodity Index Tracking Fund (“DBC”). Cash is the SSgA 1-3 Month Treasury Bill ETF (“BIL”). Returns are gross of all fees, transaction costs, and taxes with the exception of underlying fund expense rations. Returns assume the reinvestment of all distributions. 60/40 is 60% Stocks / 40% Bonds. 50/40/10 is 50% Stocks / 40% Bonds / 10% Commodities. 54/36/10 is 54% Stocks / 36% Bonds / 10% Commodities. 60/30/10 is 60% Stocks / 30% Bonds / 10% Commodities. 60/40/10 is 60% Stocks / 40% Bonds / 10% Commodities / -10% Cash. Past performance is not indicative of future results.
This means our tracking error is no longer impacted by the opportunity cost of what we might have to sell to make room. Rather, the tracking error is simply due to whatever we’re stacking on top.
And here’s an interesting fact about diversification: if we assume that what we’re selling in the portfolio has the same volatility as whatever we’re adding, then the tracking error created via stacking will always be less than or equal to the tracking error created by replacing holdings.
What’s the trade-off? The stacking approach will generally lead to a higher portfolio volatility than the replacement approach, though it may not be higher than the original portfolio.
For example, selling stocks and bonds from a 60/40 to make room for a strategy like managed futures would have historically lowered the overall portfolio volatility. Stacking managed futures on top of a 60/40, on the other hand, would’ve historically kept the portfolio’s volatility around the same level as the 60/40.
Tracking error is an important concept for every investor to consider. Not only does it help us target the amount of active risk we want to take in our portfolio, but keeping it within prudent levels can help us manage behavioral risks that might emerge if the portfolio drifts too far from its benchmark.
As many investors grapple with introducing potentially diversifying alternatives into their portfolio, we would urge them to consider how it will impact overall portfolio tracking error. The traditional approach to portfolio construction forces us to sell the stocks and bonds we’re comfortable with to make room for diversifiers. In doing so, we create a tracking error double whammy: the addition of something different and the subtraction from our core holdings.
Return stacking can help solve this problem by allowing us to overlay the new, diversifying strategies on top of our core, strategic portfolios, introducing tracking error from only one leg of the trade.