Everybody knows that stocks are riskier than bonds. We agree with that, but like to present one chart to cast a little bit of doubt on that picture. Food for thought if you will, for the fact that in finance and personal finance (and most other places in life, for that matter) nothing is completely clear-cut all the time.
Monthly stocks returns are significantly more volatile than bond returns
In the plot below we see monthly (nominal) returns. Clearly, stock returns display more month over month risk evidenced by the wider dispersion of the blue line. Back in the 1930s, you saw the wildest monthly swings but even more recently, we can see all the spikes of October 1987, LTCM situation in 1998, DotCom crash in 2001 and Global Financial Crisis in 2008/9.
Bond returns (expressed as the returns of the 10-year US Treasury Benchmark Bond) had much lower dispersion, though risk has increased quite a bit since the late 1960s to early 1970s. Note that these are default-free US government bonds, so the volatility comes entirely from inflation shocks and interest rate moves (bond price goes down if yield goes up). Even High-Yield (junk) bonds with significant default risk or longer-term US government bonds (20+ year maturity), still have lower fluctuations in monthly returns than stocks. But we just happened to have the returns for the 10-year Treasury handy so we will do our analysis with this bond index.
All of this is true not just visually, but also when comparing the standard deviation of monthly returns (but expressed as an annual value as is common in Finance, i.e., multiplied by the square root of 12). Stock returns have been about three times as volatile than bond returns over the entire sample period (16.7% vs. 5.5%0, and about twice as volatile since 1987 (15.8% vs. 7.4%). What’s the significance of 1987? I like to look at the performance of economic and financial series in the Greenspan/Bernanke/Yellen Federal Reserve era, because the very early period may not be quite as representative for today’s asset returns. Recall, until 1913/14 we didn’t even have a Federal Reserve in this country, and even between 1914 and 1987, Federal Reserve policy was quite a bit different from today’s policy responses to economic shocks.
Bond returns can be very uncertain in the long-run
Let’s look at the chart below. It’s exactly the same data. The only difference is that we plot the cumulative return of $1.00 invested in 1871. To make it more comparable over time due to different inflation regimes we also plot this adjusted for CPI-Inflation. Notice that the equity portfolio would have grown to a staggering $11,000, and yes, this is inflation-adjusted (in nominal terms $1.00 would have grown to a mind-blowing $200,000+). With such massive growth we also have to look at this in a log-scale on the y-axis to be able to compare growth rates over time.
Still the same result, the blue line is more volatile than the orange. Of course, one “risk” of bonds is the lack of significant return over the very long run (2.6% p.a. compared to the 6.6% average real return for stocks). But the chart above reveals another potential risk of bond returns:
Bond returns seem to be very episodic with decade-long super-cycles
You can have many decades in a row of very strong returns, but also multi-decade phases of sideways or even strong negative real returns. There was an 80+ year time window from 1898 to 1981 when bonds had zero real returns. We consider that the mother of all risks! Bonds also had a 40Y return window with -2.08% average real return, compared to +2.62% as the worst equity 40Y return window. The Bond return upside is limited. Only under 5% real return over the best 40-Y window, compared to over 10% for equities.
In contrast to bonds, equities might have wild swings short-term or even over a business cycle, but equity cumulative returns have a tendency to revert back to that strong exponential trend (which shows up as a linear trend due to the log-scale on the y-axis) and that exponential trend has had an impressive growth rate of around 6.2% for the last 145 years. Quite intriguing how equity returns display cycles of greed and fear but always revert back to that trend.
When looking at the cumulative return chart above we see three phases of bond return super-cycles: up (green arrows), sideways or slightly down (orange arrows), down (red arrows). Zooming in some more on what happened during those phases, see the table below:
- During the high return periods (green), we see 6+% annualized real return. That’s because they coincide with low inflation and decreasing interest rates (recall, bonds generate capital gains when yields go down).
- During the low return periods (orange) we usually see an increase in bond yields and inflation.
- During the very low return periods (red) we see 5-10 years of very large increases in inflation and yields, with very bad bond returns outcomes.
- The bond super-cycles (a full cycle of one phase each, i.e., green, orange and red) can be as long as 75 years, e.g. 1941-2016! Equities, in contrast, revert back to their mean growth path much faster, usually within the length of a full business cycle of “only” 8-10 years.
So it seems that stocks and bonds are exposed to different risks to different degrees:
- Stocks have a lot of short-term risks, but in the long-term stock returns are tied to economic growth and thus, in the very long-term, real returns become less risky due to that
- Bonds have relatively little short-term risk around their trend growth rate, but their trend growth path itself has a lot of risk in stark contrast to stocks
It’s that second uncertainty, the one about long-term expected returns, where bonds seem to be riskier than stocks.
What explains this phenomenon?
Equities are tied to the real macroeconomy. Corporate profits drop during recessions, often many times more than GDP, which explains the overreaction in equity returns, but equity total returns will eventually catch up to their growth path, tied to the economy’s growth path.
Bonds on the other hand, while having low month-to-month volatility, can languish for many decades, usually during periods when inflation and interest rates rise. Being completely nominal financial instruments, if inflation and/or rate hikes wipe out your returns, there is no snapping back of your returns back to any real time trend. Your losses are baked in and permanent! It’s true that eventually you can get a good phase again, but that might be after many decades of low or negative returns. The total length of the orange and red phases was 19 years (1901-1920) and a whopping 40 years (1941-1981), almost an entire retirement.
Implications of these bond super-cycles are far-reaching
First, if the pattern we observed over the past 145 years continues, that’s a very bad sign for bonds right now. The current “green phase” of strong bond returns has already lasted longer than the other two and is due to turn into a lower return regime. It’s almost impossible to continue the trend growth rate of 6+% real returns observed over the past 35 years because nominal yields for 10Y government bonds are at 1.80% (nominal) while writing this. At 1.75%-2.00% expected inflation, it’s hard to see how you get even any positive real rate of return from government bonds. Of course, you could go for corporate bonds with higher yields, but you then introduce equity correlation through default risk. There’s no free lunch.
The other unpleasant implication from our results is that over the past 145 years, bond returns went through 3 very good sub-periods, but only 2 of the bad periods each. That means the average bond return over the cFIREsim simulation period (2.6%) is likely unsustainable in the long-term. Why is that? If you look at bond returns over one complete super-cycle (one green, one orange and one red period) you get a much lower average returns. For example for the past full that started in 1941 and has so far lasted 75 years, the average bond return was only 1.63% real, a full percentage point lower than the full sample average return. Ouch!
What risk matters most for retirees?
Of course, both types of risk matter. Short-term fluctuations in equity returns (especially negative ones) are poison in retirement. But if the drawdown is only temporary and equities are likely to snap back fast it may not sink your retirement. But for early retirees with a 50-60 year retirement horizon uncertainty about long-term expected returns is a big issue. When planning with a 3%, even 4% withdrawal rate, adding an asset with a potentially low or even negative decade-long real return trend, may be riskier than equities, at least in the long-term.
Do we have anything nice to say about bonds?
There’s an upside: Bonds have had a negative correlation with stocks for quite some time. The one way bonds can add a real boost to the portfolio is to buy them on margin. Say, you own a 100% equity portfolio and want to add bonds to the mix. Instead of selling equities to buy bonds, thereby incurring the steep opportunity cost of low yielding bonds, one could buy Treasury Futures and get the bond exposure without giving up the equity returns. That’s not for everybody and some people might find the use of leverage too scary. But in this case, the leverage is used to reduce equity volatility. Currently, we do a version of this, but we will defer the details to a later post.
50 thoughts on “When bonds are riskier than stocks”
That’s bad news for bond investors. Everybody tells me that bonds have less risk and you can’t have 100% stocks because that’s too risky. Very intriguing, the concept of having two measures of risk, a short-term where bonds seem secure and long-term where bonds can be such a hit or miss, depending on whether we’re in a green orange or red phase!
Thanks! Yes, for early retirees the short-term risk is a concern, but low returns over 40 years is a far bigger problem for your retirement.
Hi thanks for stopping by. Don’t be myopic and look only at the month over month risk. Long-term is equally or even more important!
Wow, this post is a real gem!
I just visit your blog by chance, and I’m happy I did. I got attracted to this article because recently I wrote about bonds; “Bond; A Defensive Strategy To Grow Your Wealth.” I realized there is much more to write about it and to think to do a comprehensive research on corporate bonds as most people only invest in treasury bonds (me too for now).
I think corporate bonds are much better and less risky. More of this later.
I love your analytic approach on the long term cycles for bonds, I never see something like this and make a lot of sense for me. I wrote a similar post on stock market cycles, so we share the belief that cycles exist and are very useful to take better investment decisions.
Your bond cycle example makes sense because bonds yield are linked to central bank policies which have cycles on their own.
Now the FED is into a tightening cycle which last usually 1-3 years, So, I agree with your forecast for a negative cycle on the long-term and medium bonds.
Going back to the corporate bonds; “Of course, you could go for corporate bonds with higher yields, but you then introduce equity correlation through default risk. There’s no free lunch.”
Here there is a miss-concept in the bond market created by most developed countries for their benefit.
Treasury bonds are the safest place to park money so as result corporate bonds are riskier.
In my opinion, this is an excellent sale pitch.
In reality and logically, I approach the risks in this way.
I invest using mutual funds or ETFs for diversification risks. In the case of Treasury bonds, any ETF has only one issuer, the US Treasury. So, if the US government default, the Treasury bonds are worthless losing 100% of the investment.
People can argument that never would happen, but looking at how the US government is spending their budget and the results they bring, I would rather borrow money to a drunk hooker.
What about Investing in a corporate bond ETFs which holds several hundred companie’s debts?
First, companies produce value and second, even if few companies default, the ETF portfolio would lose only a fraction of the percentage.
In conclusion, in my humble opinion, corporate bonds are less riskier than Treasury bonds if bought through an ETF.
This article is definitely a must-read, so I’m going to add it to the “Best Of The WeB” series next month.
Thanks to share
Hi, thanks for that nice compliment. Yes, we agree that the stock market goes through cycles and like your recent post on that. Gladly, the stock market cycle has much faster mean reversion than the bond market.
So, um, how is it possible to have S&P 500 returns dating back further than the S&P 500?
Great question: Some very smart people have reconstructed several economic and financial time series all the way back to the 1800s, even late 1700s. For equity data that would involve constructing the index of large-cap US stocks from historic stock quotes in newspapers.
Just be aware that it really was an S&P 90 until 1957, a bit of a misnomer… There is an S&P 400 series of sorts that goes further back in time, but it isn’t very well known (a research article from Prof. Wilson and Jones). See this thread on Bogleheads: https://www.bogleheads.org/forum/viewtopic.php?t=205081
That is definitely one of my homework items: Figure out where my data source fits in!
Cheers and thanks for the reference!!!
Excellent article. I have been making similar points for a while now on the Bogleheads forum, but your article is much better articulated than I could ever be, very nicely done. I especially liked the chart with the arrows, cleanly illustrating the ‘episodic’ property of (past) bonds returns.
One suggestion though: the ‘noisy’ chart at the beginning should really be shown in both nominal and real terms (i.e. two charts). When doing so, you start to appreciate much better that bonds aren’t as safe as common wisdom (distorted by recency bias) says… This is still not quite the right representation, as this still only capture short-term volatility (which is pretty much irrelevant in truth) as opposed to cumulative effects (which is where bonds can become so damaging), but still, this should make many people stop and think while seeing bonds returns in real terms.
The noisy chart in the beginning with the M/M returns will look very similar. Most of the M/M noise is due to asset performance and not inflation volatility. Inflation vol was an issue pre-1913, though. I will look into that and see how much it changes the results.
Give it a try… My point wasn’t about volatility, but about the false sense of comfort given by nominal bonds returns.
Nominal Returns, for comparison:
That was fast! Thank you. Hm, you’re right, besides the first 50 years, one has to squint a bit hard to see the difference, as the monthly returns introduce quite some noise. I made charts like that with annual returns, and we see the key points (the post WW-II and oil crisis bond crisis) much more clearly. As a side note, this was nothing compared to what happened in the UK or in France, and I am not even speaking of Germany. Oh well, thanks for following on my suggestion. And for an excellent article.
Yes, you’re right, big difference between MoM and YoY returns! Because the asset returns are essentially uncorrelated month to month but inflation has severe serial correlation you’ll see a big difference in YoY real vs. nominal returns. Especially in the 1970s. 🙂
ERN, a quick technical question. How did you proceed to get monthly returns for bonds in the early days (e.g. pre SBBI/Ibbotson, pre-1926)?
I once found a snapshot of 10-year benchmark bond returns that someone had created, apparently by looking at the 10y Treasuty yields, and then deducing monthly returns that are a function of the
a) yield level *1/12
b) yield changes times (-1)*duration estimate
c) an estimate of the roll yield using the yield curve slope.
Pre-1926 that’s more art with little bit of science, but we have to trust the economic historians who put this together.
Hi ERN–I just ran across this study that says that the best results come from going to around 30% stocks at the beginning of retirement and then increasing that to about 70% over 30 years.
It looks like 100% -> 100% is not as successful as most of the other options. (At least for Table 6 : )
I didn’t spot any glaring problems with the set up (they they just did Monte Carlo simulation and did not also backtest, but I’m not sure that would have changed the results significantly.)
So which is better? 100% steady, or 100% -> 30% -> 100% ?
Thanks for the link!
I am skeptical about Monte Carlo simulations. They tend to penalize equities because you can’t properly replicate the mean-reversion feature of equity valuations.
I am also skeptical about 30Y horizons.
So, qualitatively I would agree with the conclusion, i.e., you want to use a rising equity glidepath. But I wouldn’t trust the specific hard numbers. Starting with an equity share of only 10-30% is asking for trouble when we have a 50-60-year horizon. I am still thinking about how to model the time-varying asset allocation, so stay tuned for future installments of the safe withdrawal rate series.
Looking forward to it!
The only advantage I could think of is that it would dampen the sequence of return risk for stocks in the first 10 years pretty dramatically. That, and the past is not rife with 2.x% yielding bonds like now.
Nearing the end of 2017 with a high equity mkt and high corp bond risk, if one is 62 retired and owns a high percentage of corp bond funds at this time, what may be the lesser of evils….trading some corp bond funds for equity funds, cash, or riding the corp bonds?
What percentage of your portfolio is in bonds? If you have to bridge a few years until you max out your SocSec benefits at age 70 to hedge against longevity risk, you might be good to have 8 years worth of expenses in corporate bonds.
50% of our portfolio in multisector bond funds can supply enough income for us, worse case. The rest is currently in high yield bond funds because I’ve been waiting for an equity dip to trade these for equities for better portfolio diversification, which hasn’t happened. So now with high yields facing head winds and equities scarey high, I’m truly wrestling to find some logic in either continuing to ride the high yields for later trading for equities when equities someday collapse, or trading some now for equities?? Trying to time this global mkt has never worked well for me. Thanks for any insight you can offer.
Thanks, Keith! Bonds could be under pressure soon if the Fed keeps raising rates. Since spreads between corporate bonds and Treasurys are very tight, one would expect that all bonds will be in for some pain. Even HY bonds. That’s why I never even try to time the market. In the worst possible case for you, bonds might get hammered and stocks keep rising.
But best of luck!
ERN – huge fan here. You’re really revolutionized my thinking on bonds, asset allocation and withdrawal rates.
Quick question for you – how do you feel about levered bond funds? You mentioned that you were running a leverage strategy in this article with futures, but could a UBT or UST be similarly implemented to reduce allocation while keeping most equity returns?
Cheers and thanks for all the great information.
Thanks! Good question: I don’t like those funds because of their expense ratios. If you want leveraged bond exposure you can do this yourself for a fraction of the cost with futures. Treasury futures are some of the most liquid and easy to trade instruments out there.
And I also like the idea of keeping equities and buying bonds on margin, see here: https://earlyretirementnow.com/2016/07/20/lower-risk-through-leverage/
Has your opinion on the near term future outlook changed for bonds now that interest rates are going down and inflation is low? What’s confuses me is my understanding is that this should normally be a very favorable time for bond investments, at least in the short term, no? Yet the reality doesn’t not appear to support this. In other words, why aren’t we seeing bond yields rise in this aforementioned economic cycle?
And on a separate but related note, where would you suggest us early retirees park our safe/conservative money now in order to at least beat inflation and or to relatively safely eke out 3-4% returns over the next 3-5-10 years? What do you think about spreading money/risk among REITS, corporate bonds, treasuries, dividend stocks, money market funds and perhaps CDs? I was thinking of splitting my 40% of conservative funds in roughly equal parts between these vehicles.
Would love to see a post dedicated to the rather wonky and unusual interest rate investing environment we currently find ourselves in.
Thanks in advance as always!
With the 30y yield under 2% it WAS a good time to be in bonds and ride down the bond yields. But the future looks pretty grim. Unless there’s major recession where yields drop even more.
In the past, expansions saw rising yields. Correct. But central banks didn’t meddle as much.
Where to park your cash? Unless you do the sign-up benefit dance it’s hard to get 3-4% SAFE returns.
REITs can be even more volatile during a recesssion (see 2008/9).
Corporate bonds have higher yields for a reason: they also have equity beta. The higher the yield the more beta.
See my SWR Series posts 29-31 on the dangers of yield-chasing.
At least we’re not Germany (yet). Enjoy your 1.5-2.0% interest while it lasts! 🙂
Thanks again my friend! But I wish the outlook wasn’t so pessimistic. And I wish it was easier to earn a safe lil ole 4% 🙂 Was looking through some of my very old investment files – from late 80s-90s and noticed my money market cash at Schwab was sometimes earning 8%+!!! Also noticed that I cashed in about $25,000 in Vanguard S&P 500 mutual fund around 1998 for down payment on my house. I don’t even want to know what that would be worth now. Actually I do and searched a bit to see but couldn’t really discern what the 21 year compounded result would be. Also had $3-6k in each of Matthews China Fund, T Rowe Price International mutual fund, and Icon Technology Fund. Sold those long ago too 🙁 Even had a few hundred shares of Webvan around 2001. Imagine if that was Amazon instead. Hindsight is not only 20/20 it can also be frustrating as hell 😬
Very true! Sometimes it’s best not to look back too carefully! 🙂
Great site. Lots of reading ahead of me. Starting with March 2016 and working my way up. And let me further preface this by disclosing my limited understand of all the moving parts. On bonds, I’m concerned about 1) the recent positive correlation between stocks and bonds (perhaps something that is temporary and due to current market dynamics, including Fed intervention) as one would otherwise need to revisit asset allocation accordingly, and 2) where do we go from here with the Fed rate at 0 (with nowhere to go – unless one believes negative rates help) and the risk of high inflation assuming an eventual depreciation of the USD given the MMT currently being applied. Any thoughts? My initial reaction is to stay away from bonds for now, but not inclined to go heavy in stocks.
1: There is no positive correlation between bonds and stocks. If you look at longer daily time windows, the correlation is actually very negative now: -0.5 SPX vs. 10Y T-bonds
2:Bonds right now are still a good hedge: If the economy tanks the yield will go to <0.
If this is all a big nothing-burger, bonds will go down but equities rally.
I’d like to offer a few observations.
1) The high real bond returns during the green periods for the first two supercycles had mild (on average) deflationary tailwinds. From your numbers for real bond returns and CPI, I back-out the annualized nominal total returns as 4.61% and 5.02%, respectively for these two periods; lower than the corresponding real returns because of deflation.
2) The high real bond returns during the current green period has been in the face of a moderate (on average) inflationary headwind. I estimate nominal total bond return as 9.13% for the 1981-2016 period, again using your numbers.
3) The mild inflation (2.75%) during the 1981 to present green period is pretty close to the average inflation during the orange periods from the first two supercycles.
4) The negative real returns during the orange periods of the first two supercycles look disturbingly close to current real (negative) yields.
It seems likely that we might already be in an orange period, with a red period right around the corner. Without naming names, those in charge of such things might see high inflation as an easy way out of a number of problems – burgeoning government debt, massive student load burdens, underfunded pension plans and social security (deliberately understate CPI in calculating COLAs), tax revenues (again, raising taxes by understating CPI in adjusting tax brackets)
As if I don’t have enough things keeping me awake at night.
btw, I love your blog. While I missed my opportunity to FIRE, as a hands-on recent retiree, I’ve found a far richer lode of insight in your blog than almost anything else I’ve run across.
All good points. A few comments:
2: 1981-current had a disinflationary trend. You started with double-digit inflation, then mid-single digits and then slowly down to 2% and now even slightly below that.
3+4: That’s exactly my concern! 🙂
Another couple of observations and an aside:
5) The length of the current green period (now approaching 40 years) is a good bit longer than the green periods of the prior two super-cycles. Another indication we’re approaching a new orange/red period?
6) The bond return for 1940 to 1950 is also sliqhtly negative, although not marked as a red period. This coincides with an average inflation of 5.71% from 1/1/1940 to 1/1/1950. (calculated from a CPI table I found online – neglected to make of a note of where I retrieved it)
7) The red arrows would also fit the real S&P returns pretty nicely if shifted up. Nominal returns for the S&P 500 were positive from 1970 to 1980, but the real returns were negative, in line with bond returns.
For those of us doing long-term planning, it would be dangerous to count on the negative correlation between stocks and bonds. The paper PIMCO_Quantitative_Research_Stock_Bond_Correlation_Oct2013.pdf addresses this in some detail; the authors point out that correlations between the S&P 500 and long-term Treasuries have ranged from -93% to +86% from 1927 to 2012. In particular, the correlation was positive from the mid-60’s to the late 90’s (see PIMCO Fig. 2).
5: Definitely. It looksmlike a new orange period is around the corner!
7: Very good point! The S&P returns don’t look so bad during the 1970s/80s. Only when you compute returns either ex-CPI or net of a risk-free return, would the S&P look really bad.
Yes, that’s a good point! I’d like to hedge against both a Great Depression (negative correl.) and the 1970s (positive correlation) when running risk models for SWRs. 🙂