February 5, 2024 – Two recent papers in the personal finance area have caused enough of a stir that I’ve gotten numerous requests for comment. I noticed that if I compile all my notes, calculations, simulations, and replies, I already have more than half of a new blog post. So, today I would like to share my results with my other readers who might also wonder what to make of those new research ideas. The first paper claims that the famous “Stocks for the Long Run” mantra is all wrong because stocks don’t outperform bonds as reliably as Jeremy Siegel and many prominent finance pundits claim. The second paper effectively claims the opposite, namely that a 100% equity portfolio, half domestic and half international stocks, handily beats any bond portfolio and all diversified stock/bond portfolios, including life-cycle, i.e., target date funds. Thus, the authors claim they have upended decades of personal finance conventional wisdom on stock/bond allocations, diversification, and target date fund glide paths.
Well, isn’t that ironic; both papers can’t be right! So, which one is right? Or are they both wrong, and conventional wisdom prevails? I started this post and wanted to comment on both papers in one single post but then ran out of space. So, I had to split my material into two posts. Today, I share my thoughts on the first paper and on whether stocks are still a good long-run investment in light of the new data. But stay tuned for the follow-up post, likely later this week or early next week!
Let’s take a look…
How reliably will a 100% Equity Portfolio outperform 100% bonds?
The first paper I mentioned was published in the Financial Analyst Journal (FAJ) and is titled “Stocks for the Long Run? Sometimes Yes, Sometimes No,” written by Edward F. McQuarrie. It’s available to download for free here. The eye-catching result is the following chart indicating that bonds held up remarkably well with stocks during a 220-year time series, except for a few decades between WW2 and the early 1980s:
Thus, the author claims that essentially all of the equity outperformance came from a short sliver of financial market history, and unless that bad bond market scenario repeats, there is no sizable advantage of stocks over bonds.
So does the “stocks for the long run” mantra belong on the trash heap of financial market history? Not so fast. Here are several reasons why I believe stocks remain very attractive for most investors, not just in absolute terms but also relative to bonds…
1: The pre-1871 return data are not that relevant today
I should stress that I’m not saying that the 18th and 19th-century market data are wrong. Professor McQuarrie gives us a better understanding of market history. We should applaud him for carefully cleaning pre-1871 stock market data to better account for equity survivorship bias. I’m merely saying that the pre-1871 record of financial data is not relevant for today’s expected stock and bond returns. The United States was a new country – an emerging market country – in the late 18th and early- to mid-19th century. Thus, corporate and government bonds would have been deemed extremely risky for a recently independent country that had yet to establish itself on the world stage. State bond defaults in the 1840s and the Civil War didn’t help either. So, I am unsurprised that bond yields and returns are roughly aligned with equity returns early on because the U.S. hadn’t reached its status as a global military and financial superpower. U.S. government bonds were far from the financial safe haven asset they are today. But this also implies that most people would be extremely hesitant to use pre-1871 data for calibrating return expectations in 2024 and beyond.
So, to all the people who wonder if I plan to extend my financial history horizon to 1792, my answer would be, “Heck, no!” For example, in my Safe Withdrawal Rate Series, I currently use monthly returns going back to 1871, but I freely admit that any record pre-1914, before we even had a Central Bank, is also a bit of a stretch. But also note that when calculating safe withdrawal rates, most of the time, your failsafe withdrawal rate would come from the 1929 or 1968 cohorts, i.e., the more relevant market history. And, if you are so bothered by my 1871 starting point, you can also calculate safe withdrawal rates focusing only on 1926 and onward, as in some other retirement calculators. The results wouldn’t be too different.
Sidenote: International Safe Withdrawal Rates
A related discussion in economic and financial history is the observation that safe withdrawal rates in the U.S. are an outlier, and they are significantly lower in other countries. See Wade Pfau’s paper on the topic. Many people forwarded this paper to me over the years and asked me to comment. Please see the main result table with the “SAFEMAX,” i.e., failsafe withdrawal rate by country:
Why are some of the safe withdrawal rates so low? Much of Europe and Japan were ravaged in the first half of the 20th century. I was amazed how Norway, Netherlands, and especially Denmark could hold up so well, despite being overrun and occupied by German forces. Denmark has a higher overall failsafe than even the U.S. But in any case, unless we believe a repeat of the WW1 and especially WW2-era destruction of European countries is in store for us again, I find the U.S. safe withdrawal rate calculations a lot more representative. Most importantly, I believe U.S. cities won’t soon look like Dresden in 1945; see the screenshot below. Well, part of Detroit and Baltimore do, but not the entire city, especially not the city center.
Thus, I suspect we can safely ignore the German SWR rate from 1914 and the Japanese SWR from 1940! They are the outliers, not the U.S.! I have to roll my eyes when I hear people say that U.S. SWR calculations are plagued by survivorship bias.
So the lesson here is that just like we should ignore 1914 safe withdrawal rates from Germany, we can also ignore the relative stock/bond performance in 1792!
2: My bond returns look slightly different
McQuarrie uses corporate bonds, while in my safe withdrawal rate toolkit, I’ve always used 10-year U.S. government bonds. Thus, my cumulative Bond real returns are slightly lower; see the chart below. Obviously, I don’t have pre-1871 data, so I started both equity and bond cumulative returns at 100 in 1871. I try to plot the data in the same style as in the McQuarrie paper. Qualitatively, my chart looks the same as the McQuarrie chart zooming in on the 1871-2020 time span. Of course, in the McQuarrie chart, it looks like the Stock and Bond time series was at around 60 or 80 in 1871, while my stock and bond started at 100 in 1871, so that will explain some of the differences. But it’s also true that my average bond returns are very slightly lower, which is expected because I use government bonds only.
I can also report the return stats for the entire time series and the same sub-periods as in the McQuarrie chart. Stocks outperformed bonds by 421 bps (basis points = 0.01% points) annually over the entire time span. 215 bps during the early period, a whopping 857 bps during the four-decade bond drawdown, and then 325 bps during the last 40 years.
Sidenote: how much extra bond return was there in the McQuarrie data?
Maybe the McQuarrie results make more sense when using a different bond return series. By how much would the McQuarrie bond index outperform my Treasury benchmark bond index? If I rescale the starting points to 60 for equities and 80 for bonds in 1871 and add an additional 0.75% per year return to the bond index, I’d match the rough shape of the McQuarrie Bond index, i.e., reach a 1941 peak of around 2000, drop by 50% during the 1941-1982 bond bear market and we get the 10x real return over the 1981-2019 bond bull market (1,000 to 10,000). See the chart below.
Thus, even with an additional 75bps per year, calling the third subperiod “Fluctuating Advantage,” as McQuarrie does, feels highly inappropriate. Even during one of the longest and strongest bond bull markets on record, bonds had trouble keeping up with stocks, which actually went through four(!) bear markets over those four decades (1980-1982, 1987, 2000-2002, and 2007-2009), plus several close calls (e.g., 1991, 1994/95, 1997/98, 2018).
And we all know what happened in 2020, right? Once the U.S. fixed-income market entered its bear market in 2020, bonds fell far behind stocks again, which brings me to the next point…
3: What about that 2020-2023 bond bear market?
Even though the paper came out in January of 2024, it’s curious that the returns ended before the pandemic. The multi-decade bond bull market ended in mid-2020, and your 10-year benchmark bond took a serious beating. Thus, with an additional four years of return data, 1981-2023 doesn’t look so hot anymore for the bond portfolio. Here’s the cumulative return chart in that same McQuarrie style, again using my 10-year Treasury bond index. Notice the sharp downdraft of bond returns over the last four years!
And here’s the return stats table. The stock outperformance during the last period is back to 428 bps, almost indistinguishable from the overall time series average of 447bps:
So, I admit that the 857 bps outperformance from 1942 to 1981 may have been something of an aberration, but over the last 42+ years, the stock return engine has delivered a sizable outperformance over safe government bonds; by more than 4.25 percentage points!
The “stocks for the long run” theme definitely looks legitimate, as the last 80+ return years indicate. Clearly, you can improve the bond performance if you venture into riskier bonds with higher yields and higher expected returns. But that means you add equity-style risk through the backdoor. The best you can hope for is an additional 75-100 bps for corporate investment-grade bonds (92bps for LQD over IEF over the 1/2003 to 1/2024 time span, according to Portfolio Visualizer). Even with high-yield bonds, you’ll have trouble making up a 400 bps annual return advantage.
4: What about 30-year rolling windows?
Just because stocks outperform bonds over the very long run, 150 years in my sample, it’s no guarantee that the same is true for individual investors over much shorter horizons. So, let’s check how stocks performed vs. bonds over 30-year rolling windows. Over my sample, 1901-2023, there was only one single 30-year window where bonds outperformed stocks by a mere 0.06 percentage points: June 1902 to June 1932, i.e., the sample that ended at the bottom of the Great Depression bear market and comprised not just that equity drawdown but also five additional bear markets, including the terrible 1907 banking panic and the 1920-1921 depression. But for the rest of the sample, equities outperformed bonds very reliably.
Conclusion
In my view, the McQuarrie paper is a big “nothing burger.” The first main result, looking at the 1792-1871 newly adjusted stock and bond return data, is irrelevant from today’s perspective because the U.S. is a much more stable and developed economy today with a bond market that acts as an international safe haven. Realized returns have been lower over the past 80+ years now, and the same is true for expected bond returns going forward.
The second main result, looking at 1981-2019 data, seems highly deceptive because the author focuses on one single bond bull market, ignoring both bond bear markets before and after that long stretch of strong bond returns. Recall that the 10-year yield dropped from 15.84% in September 1981 to 0.52% in August 2020 (intra-month peak and trough). The massive duration effect from this drop in the yield created above-average bond returns lasting four decades. Even back in 2020, you’d have to be insane to extrapolate that forward with a then-yield of under 1%. But even when deceptively focusing on that one bond bull market with that once-in-a-lifetime duration effect, bonds had trouble keeping up with stocks going through four bull and bear markets. So, stocks are still a great long-term investment, both in absolute terms, with a 6-7% annualized real return, and also in relative terms, with a 400+ bps annualized excess return over bonds.
Does that mean we should forego bonds and invest 100% in equities? Not necessarily. Bonds can still add value because they tend to diversify risky stock portfolios, especially during demand shock recessions, i.e., when yields drop in response to accommodative monetary policy (1929-1932, 2000-2002, 2007-2009, and 2020). Especially when you’re in retirement and facing Sequence of Return Risk. So the consistent stock outperformance doesn’t necessarily point toward a 100% equity portfolio. But that will be the topic of the next post when I discuss my views on the second paper and the pros and cons of a 100% equity portfolio. Stay tuned for that one.
Thanks for stopping by today! Please leave your comments and suggestions below.
Title picture credit: pixabay.com