3Q 2021 GMO Quarterly Letter

Introduction to the 3Q 2021 Quarterly Letter
By Ben Inker

This quarterly is a piece written by my Asset Allocation co-head John Thorndike. In it, he explains the rationale behind our strong preference for non-U.S. stocks despite the stellar performance the U.S. stock market has delivered over the last decade. The research behind the piece is an example of the bread and butter of our historical asset allocation analysis. In this short companion piece I wanted to explain a bit why we are so enamored of this type of analysis, which examines not just historical returns, but the underlying components of those returns.

Historical analysis of markets is crucial for trying to understand future potential returns. The trouble is that a standard way of doing this type of analysis – calculating average returns for an asset over time – can be extremely misleading. It is tempting to assume that over a long enough period of time historical returns are representative of what returns can be expected going forward, but that may not be the case. We believe It is only by taking the further step of delving into the components of that return that it is possible to come up with a reasonable estimate of what returns might be expected going forward. This is probably easiest to see in bonds, so in this piece I will show how a historical view of bond returns can give a grossly misleading impression of what investors might expect going forward. Exhibit 1 shows the rolling 10-year returns to the U.S. 10-Year Treasury Note.


Data from 1961-2021; 10-Year returns start in 1971 | Source: Federal Reserve, GMO

The average decade gave you a return of around 7.6% with a standard deviation of 2.9%. The last decade has seen just about the lowest return over the period, at 3.4%, and judging only from this exhibit one might be tempted to say the odds are the next decade will be better – after all, 99% of all the historical 10-year periods in the last 60 years gave a higher return. Is a 7.6% return a fair expectation for future returns from 10-Year Treasuries? Would 3.4% be a conservative forecast? By analyzing the sources of returns we can get a better idea of why the first forecast would be absurd and even the second is almost certainly unrealistically optimistic. Exhibit 2 shows the drivers of return for 10-Year Treasuries since 1961.


Data from 1961-2021 | Source: Federal Reserve, GMO

Yield change has been a positive piece of returns over this period, but it would be unwise to assume this will continue to be the case simply because it has been true on average historically. The rather large elephant in the room here is the yield component, which has averaged 6.0% over the last 60 years, whereas a 10-Year Treasury Note today yields less than 1.6%. A simple forecast methodology inspired by this component of return breakdown would be the current yield plus average roll-down, or 2%.1 As Exhibit 3 shows, this simple forecast methodology would have explained 71% of the historical variation of bond returns over the subsequent decade, and today’s forecast is lower than any realized 10-year return in this dataset.


Data from 1961-2021; 10-Year returns start in 1971 | Source: Federal Reserve, GMO