Executive Summary
It’s no great surprise that the sustainability of retirement income from a diversified portfolio is driven by the returns that the portfolio generates over time. And to the extent that stocks are the driver of long-term returns, the valuation of the markets at the beginning of retirement has a significant impact on long-term portfolio returns throughout retirement.
However, the reality is that bonds often comprise a significant portion of the retirement portfolio, and they do play at least some role in long-term retirement income sustainability. As a result, to the extent that interest rates are especially high – or in today’s environment, are unusually low – yields too can play a non-trivial role in long-term expected portfolio returns.
And the historical data really does suggest that at today’s nominal yields, long-term bond returns are likely to be suppressed, even if rates normalize in the coming years. In fact, extremely low starting bond yields, and the potential subsequent impact on prices if/when/as rates rise, are so impactful that it would be reasonable to reduce long-term 30-year real bond returns by as much as about 2%.
Notably, the eventual climb to higher yields does ultimately result in higher nominal yields in the future. Even from an extremely low starting base, long-term real government bond yields have never been worse than 1%/year over a multi-decade time horizon. Nonetheless, given that the historical real return of government bonds has been closer to __%, the potential to be several hundred basis points lower, for decades to come, is a substantial impact that should be reflected in a retiree’s prospective financial plan.
Current Nominal Bond Yields Predict Bond Returns
When it comes to long-term stock returns, valuation has a significant impact on long-term returns. In the short term, markets may be volatile in either direction, as the market-traded price of a stock is based on expectations of future cash flows. In the long term, though, those future cash flows eventually come to bear, and become a material driver on the financial outcome of the stock. Or as Benjamin Graham famously said, “In the short run, the market is a voting machine, but in the long run, it is a weighing machine.”
When it comes to bonds, though, the process of predicting returns is actually much more straightforward. When it comes to stocks, the expectation of future cash flows is only an expectation, subject to significant uncertainty. But with bonds, the committed cash flows are contractually obligated under the bond debenture itself; as a result, short of a default occurring, cash flows, and therefore returns themselves, can be determined with remarkable certainty.
For instance, a 10-year government bond that is bought at issue with a yield of 2.5% and held to maturity will produce a ‘known’ return of exactly 2.5% (presuming no default risk). By contrast, a similar term government bond issued at a yield of 5% will produce a return of 5% when held for 10 years. Knowing the yield at issuance (and having confidence in the default risk, or lack thereof) gives us all the information we need to make a confident forecast of bond returns. The math is what it is.
Accordingly, the chart below shows the correlation between a 10-year bond’s initial yield, and the long-term compounded return over varying time horizons (where the investor annually rolls into a new 10-year bond at the end of each year, thus being impacted by both the yield and the annual price changes that occur due to annual yield shifts). Given that interest rates change slowly – and that the change in interest rates in the following year is already reflected in the price of the prior year’s bond – even when rolling bonds, the 10-year yield almost perfectly predicts the returns for the coming decade.
In fact, the yield of a bond is such a powerful predictor of long-term returns, that even past the maturity of the bond itself, it’s still the primary predictor of cumulative returns, as “most” of the return for the time period is still predicted by the known yield of the bond. It’s only over very long time periods (e.g., 15+ years) that the initial yield loses some of its predictive value; and even then, the yields of the first decade contribute so much to the long-term return, that the 10-year yield up front still has a nearly 0.75 correlation with the 30-year total return of the bond holding!
Assessing Nominal Vs Real Bond Yields
An important caveat to analyzing and predicting returns in the context of a retiree is that ultimately, inflation must be considered as well. For a multi-decade time horizon, even “just” 3% compounding inflation can cut purchasing power by more than half. In the context of bond portfolios in particular, which generally pay out a fixed nominal rate of return, considering the ongoing impact of inflation over time is crucial.
Fortunately, the good news is that in the long run, markets generally price bonds to provide a modest real return over and above (expected) inflation. As a result, higher inflation rates tend to be associated with higher bond interest rates, while lower inflation tends to be associated with lower interest rates. Though notably, the ability of the money supply and interest rate monetary policy to adjust to these dynamics was limited until the US fully abandoned the gold standard in 1971; since then, nominal and real bond returns (when held to maturity) have increasingly moved in lockstep.
As the chart reveals, in recent decades the real return on bonds has actually been slightly more stable and less volatile than nominal yields. Over the long run, though, inflation has still consistently taken a bite out of long-term real bond yields; as a result, nominal 10-year government bond yields have fluctuated between 1.9% and 14.6%, while real 10-year yields have varied from -3.9% to 10.3%, and overall the average nominal yield has been 4.6% while the average real yield was "just" 2.4%.
How Nominal Bond Yields Impact Long-Term Real Bond Returns
Notwithstanding the underlying relative stability of real bond yields, though, the fact that bonds are actually priced and traded in nominal terms is important – because changes in yields, whether due to a shift in real returns, or “just” a shift in inflation expectations, can result in a material change in the price of the bond.
For instance, imagine two bonds that both have a real yield of 2%: the first has a 12% nominal yield while inflation is 10%, and the second has a 2% yield while inflation is 0%. As long as the interest rate environment remains the same, the net real return is identical for both (at least on a pre-tax basis). However, if inflation shifts to 5% (mid-way between the two scenarios), the results are dramatically different. The 12% bond yield will fall to “just” 7% to maintain the 2% real return, producing a significant increase in the price of the bond that will coincide with the 5 percentage point shift in yield. Conversely, the 2% bond yield rising to 7% would trigger a significant bond price decline, dramatically impairing the total return of the bond (and likely making the return negative for a period of time).
Given this dynamic, the starting nominal yield of bonds can have a significant impact on their long-term real return, even if real returns themselves are somewhat uncertain, because of the price appreciation or price losses that may occur in the interim (which matters to a retirement investor who is rolling the bonds to maintain duration, and/or taking ongoing withdrawals for spending).
<<<INSERT CHART TO SHOW THIS. IT WILL BE A COMBINATION OF THE PREVIOUSLY-CREATED GRAPHS #2 AND #3. SO TWO LINES, ONE FOR CORRELATION BETWEEN STARTING NOMINAL YIELD AND REAL RETURNS, THE OTHER BETWEEN STARTING REAL YIELD AND REAL RETURNS.>>>
As shown in the chart above, the starting nominal yield of bonds is a better predictor of long-term real returns on bonds than just looking at the real yield of the bonds, because of the implied appreciation/depreciation potential that is captured in the nominal bond yield! In fact, despite not knowing what real returns will be in a decade or two, the longer the time horizon the better nominal yields predict future real returns, simply given the inevitable shifts in yield that occur over time from their initial nominal levels (and the associated bond price changes that occur).
In fact, even (and especially) over 30-year time periods, there is a material difference in the long-term real returns on (rolling) 10-year government bonds, based solely on the nominal yield of the bonds at the beginning of the time period. Thus, while the long-term average real return on bonds is 2.4%, as the chart below reveals, when yields begin in the lower 1/3rd of historical returns (nominal yields less than 3.4%), the average 30-year real return is only 0.1%. By contrast, when yields start in the top 1/3rd (starting yields greater than 4.2%), the average 30-year real return is a whopping 3.7%.
Interest Rate Cycles And Long-Term Real Bond Returns
The chart above suggests that financial advisors doing long-term retirement projections should consider cutting long-term real bond returns by about 2%, given that today’s 10-year Treasury yield below 2.0% is squarely in the bottom-third range of historical yields.
At the same time, though, the question still arises of whether interest rates may soon rise by enough to offset this long-term real return impact, especially given that interest rates have moved in historical cycles from highs to lows and back again. And we already know that real returns on equities tend to move in 15-year cycles as well, driven primarily by the expansion and contraction of market valuation.
And in point of fact, real bond returns actually exhibit a similar 15-year cycle, where the 15-year real return on a (rolling) 10-year government bond portfolio is followed by a subsequent 15-year real return that moves in the opposite direction
However, the caveat is that even as interest rates move in 15-year ‘mini-cycles’, which in turn drives headwinds or tailwinds to bond real returns (in the form of price changes that occur due to interest rate changes), yields can still spend multi-decade periods of time above or below the long-term historical average.
For instance, in most of the ‘modern era’ of interest rates since the Great Depression, interest rates that went below the long-term average stayed there for nearly 40 years, and once rates swung above the long-term average, they stayed there for the subsequent several decades.
In fact, similar to today’s environment, when nominal interest rates troughed in the 1940s and 1950s, the subsequent 30-year real returns were not only below-average, but actually negative for several starting years, a combination of the low starting yields and the bond price losses of the subsequent interest rate increases when they occurred.
<<<INSERT CHART TO SHOW THIS. AGAIN THERE’S A MOCKUP ON THE ‘LAGGED RETURNS’ TAB.>>>
Ultimately, what all of this suggests that is from today’s low-yield starting point, it really is reasonable to curtail long-term real bond returns for retirees to a material degree – something on the order of 2% below the long-term historical average, or a forward-looking real bond return of just 0.1%, as even if yields do normalize in the coming years, the impact on bond prices will still drag down long-term real returns to a material degree.
Notably, reducing long-term real bond returns for today’s low-yield environment is different than reducing long-term stock returns given high valuations, as stock returns tend to exhibit stronger but “shorter term” cycles, such that it’s better to project periods (or “regimes”) of below-average returns followed by subsequent periods of higher returns. In the case of bonds, however, it appears that the cycles are so slow to play out, especially when including the impact of bond price changes, that from a high or yield low extreme adjusting projected real bond returns for 30+ years is still an effective approach when modeling retirement!
So what do you think? What bond returns do you use in retirement projections? How much of a haircut are you applying, and over what time horizon, given today's low-yield environment? Please share your thoughts in the comments below!
Leave a Reply