2022: Worst. Bond. Market. Ever?

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
NiceUnparticularMan
Posts: 3822
Joined: Sat Mar 11, 2017 7:51 am

Re: 2022: Worst. Bond. Market. Ever?

Post by NiceUnparticularMan »

jeffyscott wrote: Fri May 13, 2022 8:15 am And then there's Mrs. Real, who says: "You are both being foolish, the real value of your money is what matters. The Treasury is issuing these newfangled things they call TIPS and I-bonds. With these you will currently be guaranteed about 3-4% over and above inflation. I can put up to $30K in I-bonds and as much as I want in TIPS and be guaranteed to get 3-4% above inflation for anywhere from 5 to 30 years." :)
Yeah, I often wonder if financial history will write of this period something like:
Once IP bonds appeared, it was obvious to most academics that long-term personal investors should be shifting from nominal bonds to IP bonds. However, several decades of inflation tracking lower than expectations, combined with declining real yields, led to favorable conditions for rolling nominal bonds, and persuaded many personal investors to continue investing in rolling nominal bonds rather than comparable IP bonds.

This all changed in 20XX, which marked the beginning of a long period of inflation tracking higher than expected, with increasing real yields . . . .
Part of me thinks this story is more or less inevitable, it is more just a question of when 20XX will occur.

And while it is far too early to tell, of course, certainly 2022 has so far done a good job throwing its hat in the ring.
Booglie
Posts: 133
Joined: Sun Aug 15, 2021 2:17 am

Re: 2022: Worst. Bond. Market. Ever?

Post by Booglie »

McQ wrote: Thu May 12, 2022 10:37 pm
vineviz wrote: Wed May 11, 2022 7:05 pm
rockstar wrote: Wed May 11, 2022 6:25 pm So the take away is: stay short to maturity.
Boy I hope that's not what people are taking away from this year's adventure.
The perils of moving to a short duration

Here are some concrete numbers in support of the cautionary note I believe vineviz is trying to convey.

It is December 2001. Mr. Short is still shivering from 911, and dares not think how much worse things might yet get. Even though he is seeking income over a 20-year horizon, for safety he puts his $100,000 fixed income allocation into a 5-year Treasury note, yielding 4.42%, per the SBBI. He’ll roll it over into a new one when it matures.

Mr. Long remains unperturbed. Since he has a 20-year horizon, he puts the $100,000 into a 20-year bond yielding 5.75% (again, SBBI Appendix A)

Mr. Short is appalled by this choice: “you’d take on all that volatility risk for an extra 133 bp in yield?!?”

[...]

And here in 2022 you *know* which is the better analog--right? Don't you?
It's not so simple.

Imagine the following scenario: Mr. Long buys a 20-year bond yielding 5.75%. Inflation and yields skyrocket to 15%, and Mr. Long's bond drop 50% on the short term. Because this is a nominal bond (and not post-fixed after CPI), Mr. Long not only would take a long time to recover, but now he's losing -10% in real terms per year.

But there is more: just after Mr. Long's invested his money and their market-to-market price fell 50%, he found out his wife had cancer. Her chances are good, but it's a pretty aggressive disease that will need his savings. Had his savings not fell 50%, he'd be able to survive, but now he's forced to liquidate all of it.

In other words, long-term duration bonds carry a higher duration / uncertainty risk. Do you know how well will be the economy in 5 years? 10 years? Next year? No one could forecast COVID how would play out in December 2019. And yet, look at the impacts it has up to this day.

Good planning can limit the risks of duration risks, but it can never eliminate them!
User avatar
vineviz
Posts: 11576
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: 2022: Worst. Bond. Market. Ever?

Post by vineviz »

NiceUnparticularMan wrote: Fri May 13, 2022 8:49 am
Yeah, I often wonder if financial history will write of this period something like:
Once IP bonds appeared, it was obvious to most academics that long-term personal investors should be shifting from nominal bonds to IP bonds.
I hope that's not how the history gets written: it's been obvious to "academics" that inflation-protected bonds were a good idea for individual investors at least as far back as the early 1940s. In fact Barron's published a book titled "Investing in Purchasing Power" back in 1925.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
User avatar
vineviz
Posts: 11576
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: 2022: Worst. Bond. Market. Ever?

Post by vineviz »

Booglie wrote: Fri May 13, 2022 8:58 am Imagine the following scenario: Mr. Long buys a 20-year bond yielding 5.75%. Inflation and yields skyrocket to 15%, and Mr. Long's bond drop 50% on the short term. Because this is a nominal bond (and not post-fixed after CPI), Mr. Long not only would take a long time to recover, but now he's losing -10% in real terms per year.

But there is more: just after Mr. Long's invested his money and their market-to-market price fell 50%, he found out his wife had cancer. Her chances are good, but it's a pretty aggressive disease that will need his savings. Had his savings not fell 50%, he'd be able to survive, but now he's forced to liquidate all of it.
Changing the terms of discussion from nominal returns to real returns means we are now talking about a different risk from the one we were just discussing. Inflation risk is important, but it's not the same as interest rate risk.

You're framing this a contest to see who makes the most money, which overlooks the key lesson: Mr. Long has much less uncertainty about his nominal cash flows than Mr. Short has. If real yields change, Mr. Short's future cash flows change. This isn't true for Mr. Long.

Plus, some of the embedded assumptions in this scenario are questionable. There's an implicit assumption that Mr. Short gets off free from the price drop, but if US history is any indication he's facing a severe capital loss as well. The yield on the 5 year would likely increase by more than the increase on the 20 year,
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
Booglie
Posts: 133
Joined: Sun Aug 15, 2021 2:17 am

Re: 2022: Worst. Bond. Market. Ever?

Post by Booglie »

vineviz wrote: Fri May 13, 2022 10:33 am Plus, some of the embedded assumptions in this scenario are questionable. There's an implicit assumption that Mr. Short gets off free from the price drop, but if US history is any indication he's facing a severe capital loss as well. The yield on the 5 year would likely increase by more than the increase on the 20 year,
No matter how you slice, it Mr. Short is only tied to his bonds / bond ETFs for 5 years at most. Mr. Long is tied for 20 years.
User avatar
vineviz
Posts: 11576
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: 2022: Worst. Bond. Market. Ever?

Post by vineviz »

Booglie wrote: Fri May 13, 2022 10:42 am
vineviz wrote: Fri May 13, 2022 10:33 am Plus, some of the embedded assumptions in this scenario are questionable. There's an implicit assumption that Mr. Short gets off free from the price drop, but if US history is any indication he's facing a severe capital loss as well. The yield on the 5 year would likely increase by more than the increase on the 20 year,
No matter how you slice, it Mr. Short is only tied to his bonds / bond ETFs for 5 years at most. Mr. Long is tied for 20 years.
You say that like it's a good thing.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
NiceUnparticularMan
Posts: 3822
Joined: Sat Mar 11, 2017 7:51 am

Re: 2022: Worst. Bond. Market. Ever?

Post by NiceUnparticularMan »

vineviz wrote: Fri May 13, 2022 9:58 am
NiceUnparticularMan wrote: Fri May 13, 2022 8:49 am
Yeah, I often wonder if financial history will write of this period something like:
Once IP bonds appeared, it was obvious to most academics that long-term personal investors should be shifting from nominal bonds to IP bonds.
I hope that's not how the history gets written: it's been obvious to "academics" that inflation-protected bonds were a good idea for individual investors at least as far back as the early 1940s. In fact Barron's published a book titled "Investing in Purchasing Power" back in 1925.
I didn't put a specific date on when academics first thought that IP bonds would be a good idea.

But personal investors can only shift into things that actually exist.
User avatar
SimpleGift
Posts: 4165
Joined: Tue Feb 08, 2011 3:45 pm

Re: 2022: Worst. Bond. Market. Ever?

Post by SimpleGift »

NiceUnparticularMan wrote: Fri May 13, 2022 11:06 am I didn't put a specific date on when academics first thought that IP bonds would be a good idea.

But personal investors can only shift into things that actually exist.
A little known (and totally extraneous to this thread) historical fact, but inflation-indexed bonds first appeared in America in the late 1700s. Robert Shiller wrote a paper on these bonds:
Robert Shiller wrote:The world’s first known inflation-indexed bonds were issued by the Commonwealth of Massachusetts in 1780 during the Revolutionary War. These bonds were invented to deal with severe wartime inflation and with angry discontent among soldiers in the U.S. Army with the decline in purchasing power of their pay. Although the bonds were successful, the concept of indexed bonds was abandoned after the immediate extreme inflationary environment passed, and largely forgotten until the twentieth century.
And we had a Forum thread discussing Shiller's paper: First Inflation-Indexed Bonds in America, 1777-1780
User avatar
Kevin M
Posts: 12168
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: 2022: Worst. Bond. Market. Ever?

Post by Kevin M »

McQ wrote: Thu May 12, 2022 10:37 pm PS: of course, the 2001 starting point was chosen for rhetorical effect. Results would look different with, say, a 1965 starting point.

And here in 2022 you *know* which is the better analog--right? Don't you?
Image

In Dec 2001, the ACM 10-year term premium was 2.438%. In April 2022, the term premium was -0.193%.

Your thoughts?

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)
User avatar
vineviz
Posts: 11576
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: 2022: Worst. Bond. Market. Ever?

Post by vineviz »

NiceUnparticularMan wrote: Fri May 13, 2022 11:06 am I didn't put a specific date on when academics first thought that IP bonds would be a good idea.

But personal investors can only shift into things that actually exist.
TIPS and inflation-linked savings bonds exist BECAUSE academics spent decades explaining why they were a good idea. This wasn't a realization that came to them AFTER "IP bonds appeared", as you wrote.

If we're going write a story, let's make sure we properly identify the hero.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
User avatar
vineviz
Posts: 11576
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: 2022: Worst. Bond. Market. Ever?

Post by vineviz »

SimpleGift wrote: Fri May 13, 2022 11:18 am A little known (and totally extraneous to this thread) historical fact, but inflation-indexed bonds first appeared in America in the late 1700s. Robert Shiller wrote a paper on these bonds:
Robert Shiller wrote:The world’s first known inflation-indexed bonds were issued by the Commonwealth of Massachusetts in 1780 during the Revolutionary War. These bonds were invented to deal with severe wartime inflation and with angry discontent among soldiers in the U.S. Army with the decline in purchasing power of their pay. Although the bonds were successful, the concept of indexed bonds was abandoned after the immediate extreme inflationary environment passed, and largely forgotten until the twentieth century.
And the Rand Kardex Company issued an inflation-linked corporate bond in July 1925.

http://museumofmoney.org/exhibitions/bond/index.htm

The bond was the brainchild of Irving Fisher who, in addition to being a notable economist (he's the father of monetarism, price indexes, and more), invented the filing system that was a key product of Rand Kardex. Fisher is, unfortunately, possibly most famous for declaring on Oct. 16, 1929, that stocks had reached “what looks like a permanently high plateau".
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
incognito_man
Posts: 106
Joined: Fri Sep 08, 2017 11:00 pm
Location: Madison, WI

Re: 2022: Worst. Bond. Market. Ever?

Post by incognito_man »

FCM wrote: Sat May 07, 2022 6:51 am Thank goodness for my 401K stable value fund, which is where I have the bulk (about 60%) of my income generating funds!
I did the same thing. I'm guilty of trying to market time - but right now am feeling pleased with my decision to from 100% equities to 65/35 over the last 6 months or so - with the 35% in BNYM: https://nb.fidelity.com/public/workplac ... mmary/OXCT
Booglie
Posts: 133
Joined: Sun Aug 15, 2021 2:17 am

Re: 2022: Worst. Bond. Market. Ever?

Post by Booglie »

vineviz wrote: Fri May 13, 2022 11:03 am
Booglie wrote: Fri May 13, 2022 10:42 am
vineviz wrote: Fri May 13, 2022 10:33 am Plus, some of the embedded assumptions in this scenario are questionable. There's an implicit assumption that Mr. Short gets off free from the price drop, but if US history is any indication he's facing a severe capital loss as well. The yield on the 5 year would likely increase by more than the increase on the 20 year,
No matter how you slice, it Mr. Short is only tied to his bonds / bond ETFs for 5 years at most. Mr. Long is tied for 20 years.
You say that like it's a good thing.
Being stuck with 20-year bonds is NOT a good thing if the nominal rates are worse than inflation and bond yields are rising. Imagine having being stuck with -5% real yield for 20 years?

If the bond is adjusted for inflation and there is a real yield, it's not so bad. But the only bonds in the American economy that are adjusted for inflation have zero real yield currently.
User avatar
vineviz
Posts: 11576
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: 2022: Worst. Bond. Market. Ever?

Post by vineviz »

Booglie wrote: Fri May 13, 2022 1:25 pm Being stuck with 20-year bonds is NOT a good thing if the nominal rates are worse than inflation and bond yields are rising. Imagine having being stuck with -5% real yield for 20 years?

If the bond is adjusted for inflation and there is a real yield, it's not so bad. But the only bonds in the American economy that are adjusted for inflation have zero real yield currently.
I think you're still missing the important point.

When you buy a bond that matches your investment horizon, your return over that horizon is certain unless that bond defaults. Sure, some other bond might have done better or might have done worse. But that return uncertainty has a name: risk.

For an investor with a 20 year investment horizon, only a default-free bond with a duration of 20 years can provide a certain return over that horizon.

The real/nominal thing is a distraction from this essential lesson. If you want a certain REAL return then buy a 20-year REAL bond. If you want a certain NOMINAL return then buy a 20-year NOMINAL bond. Inflation doesn't change the implication.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
User avatar
whodidntante
Posts: 10695
Joined: Thu Jan 21, 2016 11:11 pm
Location: outside the echo chamber

Re: 2022: Worst. Bond. Market. Ever?

Post by whodidntante »

Real returns are what we should all think about, or at least it's what I am going to think about. Bonds have absolutely stunk up the room lately. The best thing is that maybe we can have a brief reprieve from the "bonds are for safety" mantra now that the risk showed up.
User avatar
jeffyscott
Posts: 10832
Joined: Tue Feb 27, 2007 9:12 am

Re: 2022: Worst. Bond. Market. Ever?

Post by jeffyscott »

Booglie wrote: Fri May 13, 2022 1:25 pm If the bond is adjusted for inflation and there is a real yield, it's not so bad. But the only bonds in the American economy that are adjusted for inflation have zero real yield currently.
Not quite 0 for all. There's a positive real yield based on quotes or the treasury real yield curve for TIPS with about 7 years or more to maturity. The treasury real yield curve shows 20 year at about +0.5%.
The two greatest enemies of the equity fund investor are expenses and emotions. ― John C. Bogle
User avatar
Kevin M
Posts: 12168
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: 2022: Worst. Bond. Market. Ever?

Post by Kevin M »

jeffyscott wrote: Fri May 13, 2022 2:01 pm
Booglie wrote: Fri May 13, 2022 1:25 pm If the bond is adjusted for inflation and there is a real yield, it's not so bad. But the only bonds in the American economy that are adjusted for inflation have zero real yield currently.
Not quite 0 for all. There's a positive real yield based on quotes or the treasury real yield curve for TIPS with about 7 years or more to maturity. The treasury real yield curve shows 20 year at about +0.5%.
Based on actual TIPS quotes today from Fidelity at about 12:30pm, the 4/15/2028 (5.9 years) has a quoted ask yield of +0.04%. The highest ask yield was for the 2/15/2045 (22.8 years) at +0.74%. Yields decline from there to 0.56% for the 2/15/2052 (28.8 years).

Closest thing to a 20-year is the 2/15/2042 (19.8 years) at 0.67%.

Image

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)
Topic Author
McQ
Posts: 379
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: 2022: Worst. Bond. Market. Ever?

Post by McQ »

vineviz wrote: Thu May 12, 2022 4:24 pm
SimpleGift wrote: Thu May 12, 2022 3:56 pm But history shows that once inflation gets into the 4%-5% range, the correlation has tended to shift positive, with bond and stock returns much more closely aligned (chart below). Higher inflation = higher bond/stock correlations.
I think this is an inappropriate inference.

If you replicate the chart but instead of using inflation on the x-axis you simply use the calendar year of the observation, you get an equivalent result: once the calendar year gets into the 1990 to 2022 range, the correlation has tended to shift negative.

Basically all that PIMCO demonstrated is that stock/bond correlations were higher in earlier time periods (when inflation was also higher) than in later time periods. In fact, they had to add a dummy variable in their inflation model to account for this.
We also add a breakpoint dummy variable (D1997) to evaluate whether inflation and real rates correctly capture the regime shit from positive to negative correlation, which the data in Figure 1 suggest occurred around 1997.
The stock – bond correlation

I’m not sure I can resolve the dispute between SimpleGift and vineviz about whether the correlation between stocks and bonds is inflation-linked or time-linked; but I can add a historical perspective on how that correlation has varied over time.

Preface

IMHO, many investors suffer from what I’ll call “Markowitz hangover,” or maybe, “the Aristotelian misinterpretation of Markowitz.” Let me explain.

Everyone knows that diversification is good. Everyone knows that the lower the correlation between two assets, the greater the diversification benefit, aka, “the free lunch.”

And anyone who has ever runs some numbers through Markowitz’ formula learns to dream of the negatively correlated asset, offering the ultimate in diversification benefit.

Now here is the Aristotelian misinterpretation: that standard deviation, correlation, etc. are intrinsic properties of the assets in question. In the present context, that would be the belief that stocks and bonds have a characteristic or customary correlation, a level from which there may be brief departures, but to which the correlation will sooner or later revert.

The Markowitz hangover is the elision between what we would like to be true—that ideally stocks and bonds will have a negative correlation—to what is actually true. In the spirit of behavioral finance, just a few instances of realized negative correlations will encourage the belief that stocks and bonds, by nature, have a correlation whose values mostly cluster below zero. After all, and again in the Aristotelian spirit, these two are fundamentally different kinds of assets, making a sustained positive correlation per se unlikely.

The more perceptive investor—I’ve read some wise comments from Nisiprius to this effect—will recognize that the profound differences between the stock asset and the bond asset argue only for independence, i.e., correlations that cluster either side of zero. The expectation of a negative correlation is just Markowitz hangover.

Now let’s look at the history: what has been the realized correlation between stocks and bonds? And which data series should we choose to represent “bonds” and “stocks?”

Correlations 1973 – 2022

Although it is conventional to chart such values back to 1926, I consider the Great Depression, and the extreme financial repression during WW II, to add noise rather than signal. Plus, if we start in 1973, we can compare Total Bond, the golden-haired child here at Bogleheads, with an intermediate duration, as a bond proxy, to long Treasuries, the safe but all too volatile long duration asset (the role of variance in the Markowitz formula is under-appreciated, in my view).

I’ll use CRSP and later VTI as my stock proxy (i.e., Total Stock Market index).

In the SBBI, rolling 60-month correlations are charted. Let’s start with that interval. Here is the chart.

Image

A first impression: for the 25 years through the late 1990s, the correlation was quite high, often over 0.50. Something changed after June 1998: the correlation began to fall rapidly, going negative around 2002 during the 2000-2003 bear market in stocks, and staying there through the beginning of the 2008-09 crisis. As that crisis developed, the correlation became positive again for a few years, until 2012, when an abrupt reversal drove the correlation negative again, where it stayed. Even as of April 2022, the Treasury correlation had not yet turned positive over the trailing 60 months.

A second impression: total bond and long Treasuries behaved the same up until the Great Financial Crisis, after which Treasuries had a persistently lower correlation with stocks (in fact, BND correlation with stocks went positive in December 2021).

A third impression: I see no characteristic level to which correlations revert. Rather, I see shifts in regimes, periods when the correlation runs positive for years or decades, and periods where it runs negative.

A more jaundiced observer might only see random variation.

Shorter rolls

A problem with 60-month rolls is the mismatch between the roll length and how long periods of crisis last. Folklore has it that at the nadir of a crisis, all correlations approach 1.0. Conversely, if the time span is too short, then the correlation will have a large error of estimate, i.e., be noisy. As a compromise (sticking with monthly data; daily data would offer a different solution, at the cost of truncating the historical record), here is the same chart but now with 24-month rolls.

Image

Unsurprisingly, given the shorter time frame, the impression of volatility is enhanced. The chart has to scale between +/- 0.85 to handle the range of correlation values. Especially before 2000, the oscillation between high and low values is head-snapping.

In other respects, the 24-month rolls show a similar pattern over time. Before 1998, correlations were mostly positive and often quite high. The exception is the rolls that include October 1987. After 1998, correlations are again mostly negative, except for the great financial crisis, and also in recent months. On this shorter time scale, the correlation with stocks for both total bonds and long Treasuries has abruptly moved into positive territory in 2022.

So much for historical context. I’ve tried to stay at the descriptive level in this post. I am hoping others will offer interpretations; I’ll be happy to address.

The one interpretation I will assert: the past five decades give no support to the idea that stocks and bonds have a characteristic or customary correlation. In particular, there’s no evidence that stock and bond correlations tend to mean-revert to even a value of zero, much less to a negative value.
They that read the footnotes, they shall be saved; but they that pass over the appendices, they shall wander forever.
garlandwhizzer
Posts: 3389
Joined: Fri Aug 06, 2010 3:42 pm

Re: 2022: Worst. Bond. Market. Ever?

Post by garlandwhizzer »

vineviz wrote:

For an investor with a 20 year investment horizon, only a default-free bond with a duration of 20 years can provide a certain return over that horizon.

The real/nominal thing is a distraction from this essential lesson. If you want a certain REAL return then buy a 20-year REAL bond. If you want a certain NOMINAL return then buy a 20-year NOMINAL bond. Inflation doesn't change the implication.
Personally, I disagree strongly with this point of view. Real returns in the future are all I care about. In 20 years you must purchase things with inflation adjusted real dollars of 2042, not nominal dollars from 2022 whose purchasing power has been reduced by 20 years of cumulative inflation. I have no interest whatsoever in a promised nominal fixed income return particularly in the distant future. Too much inflation risk and not sufficient yield inducement to take it. Not to mention the principal value volatility of long duration bonds along the way which becomes critical if the holder unexpectedly has to liquidate them before maturity for emergencies or life setbacks.

Fixed income academic models do not take that uncertainty into account. What if you will unexpectedly have to liquidate fixed income earlier than expected of what the value in real inflation dollars at that point? Down by a third like current LTT? Or if you wait for the LTT to mature, both its yield and its principal value will be reduced by 20 years of cumulative inflation the degree of which is completely unpredictable up front. If you want real protection to purchasing power in 20 years, you need TIPS and equity exposure, not today's nominal LT bonds which have low yields by historical standards. Those whose fixed portfolios have been dominated by LTT have been devastated by principal losses in the recent bond bear market. They are fully aware now of the risk of long duration bonds posed by inflation. That's the kind of hard lesson, I suspect, that they won't soon forget.

Garland Whizzer
User avatar
vineviz
Posts: 11576
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: 2022: Worst. Bond. Market. Ever?

Post by vineviz »

McQ wrote: Fri May 13, 2022 3:50 pm
A first impression: for the 25 years through the late 1990s, the correlation was quite high, often over 0.50. Something changed after June 1998: the correlation began to fall rapidly, going negative around 2002 during the 2000-2003 bear market in stocks, and staying there through the beginning of the 2008-09 crisis. As that crisis developed, the correlation became positive again for a few years, until 2012, when an abrupt reversal drove the correlation negative again, where it stayed. Even as of April 2022, the Treasury correlation had not yet turned positive over the trailing 60 months.

A second impression: total bond and long Treasuries behaved the same up until the Great Financial Crisis, after which Treasuries had a persistently lower correlation with stocks (in fact, BND correlation with stocks went positive in December 2021).
I can't pretend to know how the next 50-100 years play out, but I think it's worth trying to understand whether "something changed" between the earlier time periods and later time periods.

McQ is the expert on the total bond time series data, but a couple of features of US Treasuries come to mind.

For one thing, before the 1980s the FOMC mostly operated without any regard to the concept of expected inflation. Not only did they lack the tools to easily decompose the nominal bond yields into real rates and expected inflation, for the most part they lacked an economic understanding that the distinction even mattered. The data suggest that "something happened" around 1998 to create an inflection point in the way that stocks and bonds traded. One candidate "something" is 30-year TIPS, which were first auctioned by the Treasury in April 1998.

Other possibly impactful changes that occurred in the years between 1985 and 1998 include the issuance of non-callable 30-year Treasuries (1985) and the removal of the 4-1/4% interest rate ceiling on bonds with maturities over 7 years (1988).

Is the cumulative impact of these structural changes in the Treasury market enough to explain the 1998 inflection point? Time will tell.
McQ wrote: Fri May 13, 2022 3:50 pm A problem with 60-month rolls is the mismatch between the roll length and how long periods of crisis last.
On the other hand, the ideal rolling period is one that matches the investment horizon of the investor. Even 60-month rolls are much shorter than market-implied investment horizon for the representative investor, which is closer to 100 months than to 24 months.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
User avatar
vineviz
Posts: 11576
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: 2022: Worst. Bond. Market. Ever?

Post by vineviz »

garlandwhizzer wrote: Fri May 13, 2022 4:19 pm
vineviz wrote:

For an investor with a 20 year investment horizon, only a default-free bond with a duration of 20 years can provide a certain return over that horizon.

The real/nominal thing is a distraction from this essential lesson. If you want a certain REAL return then buy a 20-year REAL bond. If you want a certain NOMINAL return then buy a 20-year NOMINAL bond. Inflation doesn't change the implication.
Personally, I disagree strongly with this point of view. Real returns in the future are all I care about.
You're not disagreeing, though. You're agreeing.

In the passage you quoted, I specifically said that if you want certainty around real returns (which apparently you do) then you should buy real bonds.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
User avatar
SimpleGift
Posts: 4165
Joined: Tue Feb 08, 2011 3:45 pm

Re: 2022: Worst. Bond. Market. Ever?

Post by SimpleGift »

McQ wrote: Fri May 13, 2022 3:50 pm A first impression: for the 25 years through the late 1990s, the correlation was quite high, often over 0.50. Something changed after June 1998: the correlation began to fall rapidly, going negative around 2002 during the 2000-2003 bear market in stocks...(snip)

So much for historical context. I’ve tried to stay at the descriptive level in this post. I am hoping others will offer interpretations.
Thank you for the history, Professor McQ. I hope we haven't hopelessly sidetracked your original thread topic!

As for what caused the pronounced shift in stock-bond correlations around 1998-2002, one interpretation is that investors' inflation expectations underwent a pronounced shift around this time (chart below). All through the latter 1980s and 1990s, there was a bit of a hangover from the runaway inflation of the 1970s, with the fear that inflation was perhaps not really dead. By 2002, inflation expectations had become well-anchored at a lower level.
Today, with one-year inflation expectations approaching 4%, the stock-bond correlation is positive again. The theory is that a significant shift to a higher inflation regime (or higher inflation expectations) causes common discount rate changes in both stocks and bonds, so their returns become more highly correlated.
User avatar
vineviz
Posts: 11576
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: 2022: Worst. Bond. Market. Ever?

Post by vineviz »

SimpleGift wrote: Fri May 13, 2022 4:39 pm As for what caused the pronounced shift in stock-bond correlations around 1998-2002, one interpretation is that investors' inflation expectations underwent a pronounced shift around this time (chart below). All through the latter 1980s and 1990s, there was a bit of a hangover from the runaway inflation of the 1970s, with the fear that inflation was perhaps not really dead. By 2002, inflation expectations had become well-anchored at a lower level.
I suggest that rather than trying to find a bright line between a "high inflation regime" and a "low inflation regime", the more relevant variable would be related to the amount of uncertainty around expected inflation.

It was roughly in the late 1990s that the FOMC started thinking about even having an inflation target, and not until the 2000s were participants starting to openly discuss such a target.
For instance, Atlanta Fed President Robert Forrestal, during the first meeting of 1995 said, “I would be against an inflation target and I would associate myself entirely with the views of Governor [Janet] Yellen,” who had noted potential risks while others expressed favoring a target. The question remained unsettled into the late 1990s, as exemplified by President Melzer’s statement at the November 1997 meeting: “What are the FOMC’s intentions? Do we like seeing inflation below 2%? Does the public know it? I think, as I have said before, that we ought to be more explicit about our longer-term objective. In that event, it would be much less likely that our actions would be misinterpreted as being anti-jobs or anti-growth.”
https://www.frbsf.org/economic-research ... on-target/

There's also a fascinating article at Quartz about the inflation target.
At least since 1996, the US Federal Reserve has used monetary policy with the aim of keeping inflation at 2%—a number that Ben Bernanke, the former Fed chair, made an explicit policy target in 2012. And it isn’t the only central bank in the developed world to shoot for 2%.

The Bank of Canada has the same target, as do Sweden’s Riksbank, the Bank of Japan, and the European Central Bank. The Bank of England is so devoted to its 2% target that its governor must write a letter to the chancellor of the Exchequer if inflation moves more than a percentage point in either direction. Andrew Bailey, the current governor, sent such a letter in May, pointing out that weakened economic activity during the pandemic had caused prices to tumble in the 12-month period ending February.

But why did these banks uniformly gravitate to the 2% figure? And where did that number come from?

From New Zealand, it turns out: specifically, from a finance minister who was put on the spot during a TV interview in 1988.
https://qz.com/2022696/where-did-the-fe ... come-from/
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
User avatar
SimpleGift
Posts: 4165
Joined: Tue Feb 08, 2011 3:45 pm

Re: 2022: Worst. Bond. Market. Ever?

Post by SimpleGift »

vineviz wrote: Fri May 13, 2022 4:52 pm It was roughly in the late 1990s that the FOMC started thinking about even having an inflation target, and not until the 2000s were participants starting to openly discuss such a target.
Agree that the emergence of explicit inflation targeting, by the Fed and central banks around the world, made a significant contribution to lowering inflation expectations and uncertainty worldwide from 1990 onward.

However, it's difficult to judge how much inflation targeting actually contributed to the shift in the inflation expectations regime around 2000. In 1990, only one country had an explicit inflation target (New Zealand), and by 2000 there were just 10 countries — compared with over 40 countries today, both developed and emerging.
Topic Author
McQ
Posts: 379
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: 2022: Worst. Bond. Market. Ever?

Post by McQ »

Kevin M wrote: Fri May 13, 2022 11:47 am
McQ wrote: Thu May 12, 2022 10:37 pm PS: of course, the 2001 starting point was chosen for rhetorical effect. Results would look different with, say, a 1965 starting point.

And here in 2022 you *know* which is the better analog--right? Don't you?
Image

In Dec 2001, the ACM 10-year term premium was 2.438%. In April 2022, the term premium was -0.193%.

Your thoughts?

Kevin
Hello Kevin: I haven’t worked with term premia much, or studied them historically. Term structure isn’t very meaningful in the deeper history where I spend most of my time (before WW II, there were mostly just long bonds and commercial paper/ short government notes; short and intermediate bonds weren’t typically issued, and only came into being through the aging of a long bond). In fact, before 1933, the SBBI had to manufacture the 5-year Treasury series from a yield curve interpolation because there were none to be found on the market (and I double-checked that absence by going back to the Bank and Quotation record for those years).

You are probably familiar with Durand’s 1942 paper, which mostly found flat yield curves from 1900 to about 1930. I’ve long wondered whether the Treasury’s forcing of a positive yield curve during and after WW II didn’t create what we now think of as the “normal yield curve.”

Long story short, I have no thoughts to share on the chart. But I’d be happy to respond to a more specific assertion or challenge.
They that read the footnotes, they shall be saved; but they that pass over the appendices, they shall wander forever.
Topic Author
McQ
Posts: 379
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: 2022: Worst. Bond. Market. Ever?

Post by McQ »

Thanks to SimpleGift and vineviz for insightful comments on the correlation charts (insightful, as in, gave me new insights!)

And not to worry, SimpleGift, I'm approaching this thread broadly as "let's situate the bond market debacle of 2022 within a larger historical context," making correlations just as relevant as returns.
They that read the footnotes, they shall be saved; but they that pass over the appendices, they shall wander forever.
User avatar
SimpleGift
Posts: 4165
Joined: Tue Feb 08, 2011 3:45 pm

Re: 2022: Worst. Bond. Market. Ever?

Post by SimpleGift »

McQ wrote: Sat May 14, 2022 1:28 pmAnd not to worry, SimpleGift, I'm approaching this thread broadly as "let's situate the bond market debacle of 2022 within a larger historical context," making correlations just as relevant as returns.
These financial history threads may seem a bit arcane and largely dispensable to many investors (as in, Who cares what happened in 1947?) — but I believe they can help prepare us do-it-yourself amateurs for the range of possible financial market outcomes that might be experienced over our investing careers. For example, if only having experienced the markets of the last twenty years, since 2000:
  • a) Who would expect intermediate-term bonds to lose 15% of their inflation-adjusted value in a year? Turns out, according to the historical record since 1800, this has happened 5 or 6 times per century.

    b) Who would expect stock and bond returns to be positively correlated, and to both lose (and gain) value at the same time? Turns out, since 1800, there were several decades-long periods when this was the case.
Not only can historical insights help investors with portfolio design, to anticipate what sort of market extremes might be ahead, but they can also help with emotional stability when these anticipated extremes are actually happening!

Just my two cents.
User avatar
Kevin M
Posts: 12168
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: 2022: Worst. Bond. Market. Ever?

Post by Kevin M »

McQ wrote: Sat May 14, 2022 1:24 pm
Kevin M wrote: Fri May 13, 2022 11:47 am
McQ wrote: Thu May 12, 2022 10:37 pm PS: of course, the 2001 starting point was chosen for rhetorical effect. Results would look different with, say, a 1965 starting point.

And here in 2022 you *know* which is the better analog--right? Don't you?
Image

In Dec 2001, the ACM 10-year term premium was 2.438%. In April 2022, the term premium was -0.193%.

Your thoughts?

Kevin
Hello Kevin: I haven’t worked with term premia much, or studied them historically. Term structure isn’t very meaningful in the deeper history where I spend most of my time (before WW II, there were mostly just long bonds and commercial paper/ short government notes; short and intermediate bonds weren’t typically issued, and only came into being through the aging of a long bond). In fact, before 1933, the SBBI had to manufacture the 5-year Treasury series from a yield curve interpolation because there were none to be found on the market (and I double-checked that absence by going back to the Bank and Quotation record for those years).

You are probably familiar with Durand’s 1942 paper, which mostly found flat yield curves from 1900 to about 1930. I’ve long wondered whether the Treasury’s forcing of a positive yield curve during and after WW II didn’t create what we now think of as the “normal yield curve.”

Long story short, I have no thoughts to share on the chart. But I’d be happy to respond to a more specific assertion or challenge.
Thanks for replying. Let me try a more specific challenge.

I don't think there is any serious challenge to the assertion that the lowest risk asset is one that matches your holding period and unit of account (e.g., US dollar or real consumption). I think I first learned this framework from this 2012 post by forum member bobcat2 Risk Basics: What is the risk-free asset?. So in your example, sure, the 20-year nominal Treasury has less risk for a 20-year holding period and the US dollar as the unit of account. Whether the outcome from rolling is higher or lower, it still demonstrates the variance of rolling vs. holding; i.e., holding for the term is the lower risk choice.

But, you used a specific time period to compare outcomes between rolling a shorter-term security and holding the longer-term security until maturity You acknowledged that you picked the starting point to demonstrate the point. You acknowledged that picking a different starting point would result in a different. Your challenge was whether or not we know which is a better analog.

The challenge is that if we want to make a rational decision about the best expected outcome in comparing holding vs. rolling, the term premium is a relevant input to consider. My understanding about the term premium is that when it's positive, holding the longer-term security has a higher expected return, and if it is negative, rolling the shorter-term security has a higher expected return.

In other words, if we're not looking for the risk-free asset, but the one with the higher expected return, shouldn't we be taking the term premium into account? Let's assume an investor with 100% fixed income, so we don't get into the whole portfolio discussion.

You chose a starting point when the term premium was positive and reasonably high. What if we look at some starting points where the term premium was very low or even negative? I see some negative values in the early 1960s (especially if looking at daily data)?

Of course one could challenge the ACM term premium model itself.

Thanks,

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)
sandan
Posts: 378
Joined: Wed Apr 03, 2013 12:48 pm

Re: 2022: Worst. Bond. Market. Ever?

Post by sandan »

A discussion of the worst bond market ever without including the history of the republic and Alexander Hamilton is misleading.

A 90% drop is excluded. The events leading to the drop and correction are much more relevant than any chart.

https://allthingsliberty.com/2016/08/bo ... -republic/
User avatar
SimpleGift
Posts: 4165
Joined: Tue Feb 08, 2011 3:45 pm

Re: 2022: Worst. Bond. Market. Ever?

Post by SimpleGift »

McQ wrote: Wed May 11, 2022 5:00 pm Yes, you can lose 20% real, in one year, on the safest bond you can buy. It’s happened before, more than once.
After a double-digit real bond loss, how long has it taken historically for bonds to recover their real value? Have they recovered within 7 years (the duration of an intermediate-term bond fund)? Recovered within 10 years?

From the McQuarrie database of real bond returns, the table below lists each instance of a double-digit real bond loss since 1794, and then calculates the compound annual growth rate over the subsequent 7 years. If the initial bond loss had not recovered in 7 years, the CAGR after 10 years was calculated.
  • Image
    Data Source: McQuarrie database, 1794-2019
In most historical instances of double-digit real losses, bonds more than recovered their value within 7 years. The exceptions were 1918 (when bonds recovered within 10 years), 1947 (when interest rates were capped at 2.5% until 1951, while inflation soared), and 1968 (which saw two subsequent oil supply shocks and runaway inflation).

Barring monetary controls or runaway inflation, bond returns have been fairly resilient after steep real losses.
Marseille07
Posts: 9238
Joined: Fri Nov 06, 2020 1:41 pm

Re: 2022: Worst. Bond. Market. Ever?

Post by Marseille07 »

SimpleGift wrote: Sun May 15, 2022 12:10 pm In most historical instances of double-digit real losses, bonds more than recovered their value within 7 years. The exceptions were 1918 (when bonds recovered within 10 years), 1947 (when interest rates were capped at 2.5% until 1951, while inflation soared), and 1968 (which saw two subsequent oil supply shocks and runaway inflation).

Barring monetary controls or runaway inflation, bond returns have been fairly resilient after steep real losses.
The road to recovery won't be pretty because the yield curve got extremely low. In 2001, the Ten was around 5% and went down during the downturn. We don't have much room here in 2022.
85% US + FM + RSU | 15% Cash
User avatar
SimpleGift
Posts: 4165
Joined: Tue Feb 08, 2011 3:45 pm

Re: 2022: Worst. Bond. Market. Ever?

Post by SimpleGift »

Marseille07 wrote: Sun May 15, 2022 1:31 pm The road to recovery won't be pretty because the yield curve got extremely low. In 2001, the Ten was around 5% and went down during the downturn. We don't have much room here in 2022.
Good point. Prevailing interest rates were much higher during previous episodes of real bond losses and recoveries. Plus, deflation was much more common in the years before 1940. So past circumstances for bond recoveries, in real terms, may look more optimistic than current circumstances.
Booglie
Posts: 133
Joined: Sun Aug 15, 2021 2:17 am

Re: 2022: Worst. Bond. Market. Ever?

Post by Booglie »

Marseille07 wrote: Sun May 15, 2022 1:31 pm
SimpleGift wrote: Sun May 15, 2022 12:10 pm In most historical instances of double-digit real losses, bonds more than recovered their value within 7 years. The exceptions were 1918 (when bonds recovered within 10 years), 1947 (when interest rates were capped at 2.5% until 1951, while inflation soared), and 1968 (which saw two subsequent oil supply shocks and runaway inflation).

Barring monetary controls or runaway inflation, bond returns have been fairly resilient after steep real losses.
The road to recovery won't be pretty because the yield curve got extremely low. In 2001, the Ten was around 5% and went down during the downturn. We don't have much room here in 2022.
You do, if the US government is willing to sacrifice job positions.
It's a matter of picking your poison.
Which is worse: unemployment or inflation?
Marseille07
Posts: 9238
Joined: Fri Nov 06, 2020 1:41 pm

Re: 2022: Worst. Bond. Market. Ever?

Post by Marseille07 »

Booglie wrote: Sun May 15, 2022 1:59 pm You do, if the US government is willing to sacrifice job positions.
It's a matter of picking your poison.
Which is worse: unemployment or inflation?
We really don't. We don't operate like Turkey, CPI is 8.2% and the FF rate will only be 3% by EOY and that's not even a guarantee.
85% US + FM + RSU | 15% Cash
User avatar
Kevin M
Posts: 12168
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: 2022: Worst. Bond. Market. Ever?

Post by Kevin M »

Marseille07 wrote: Sun May 15, 2022 2:07 pm
Booglie wrote: Sun May 15, 2022 1:59 pm You do, if the US government is willing to sacrifice job positions.
It's a matter of picking your poison.
Which is worse: unemployment or inflation?
We really don't. We don't operate like Turkey, CPI is 8.2% and the FF rate will only be 3% by EOY and that's not even a guarantee.
CPI is not 8.2%. That is year over year (YoY) change in seasonally adjusted CPIAUC for April. The monthly change was 0.33%, which is 4.06% annualized. Monthly change for March was 1.24%, which is 15.95% annualized. Inflation appears to be coming down.

CPIAUCSL for April was 288.663. For March it was 287.708, and for April 2022 it was 266.727. From these numbers you can compute monthly and YoY change in CPI.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)
Topic Author
McQ
Posts: 379
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: 2022: Worst. Bond. Market. Ever?

Post by McQ »

Kevin M wrote: Sat May 14, 2022 4:58 pm
McQ wrote: Sat May 14, 2022 1:24 pm
Kevin M wrote: Fri May 13, 2022 11:47 am
McQ wrote: Thu May 12, 2022 10:37 pm PS: of course, the 2001 starting point was chosen for rhetorical effect. Results would look different with, say, a 1965 starting point.

And here in 2022 you *know* which is the better analog--right? Don't you?
Image

In Dec 2001, the ACM 10-year term premium was 2.438%. In April 2022, the term premium was -0.193%.

Your thoughts?

Kevin
Hello Kevin: I haven’t worked with term premia much, or studied them historically. Term structure isn’t very meaningful in the deeper history where I spend most of my time (before WW II, there were mostly just long bonds and commercial paper/ short government notes; short and intermediate bonds weren’t typically issued, and only came into being through the aging of a long bond). In fact, before 1933, the SBBI had to manufacture the 5-year Treasury series from a yield curve interpolation because there were none to be found on the market (and I double-checked that absence by going back to the Bank and Quotation record for those years).

You are probably familiar with Durand’s 1942 paper, which mostly found flat yield curves from 1900 to about 1930. I’ve long wondered whether the Treasury’s forcing of a positive yield curve during and after WW II didn’t create what we now think of as the “normal yield curve.”

Long story short, I have no thoughts to share on the chart. But I’d be happy to respond to a more specific assertion or challenge.
Thanks for replying. Let me try a more specific challenge.

I don't think there is any serious challenge to the assertion that the lowest risk asset is one that matches your holding period and unit of account (e.g., US dollar or real consumption). I think I first learned this framework from this 2012 post by forum member bobcat2 Risk Basics: What is the risk-free asset?. So in your example, sure, the 20-year nominal Treasury has less risk for a 20-year holding period and the US dollar as the unit of account. Whether the outcome from rolling is higher or lower, it still demonstrates the variance of rolling vs. holding; i.e., holding for the term is the lower risk choice.

But, you used a specific time period to compare outcomes between rolling a shorter-term security and holding the longer-term security until maturity You acknowledged that you picked the starting point to demonstrate the point. You acknowledged that picking a different starting point would result in a different. Your challenge was whether or not we know which is a better analog.

The challenge is that if we want to make a rational decision about the best expected outcome in comparing holding vs. rolling, the term premium is a relevant input to consider. My understanding about the term premium is that when it's positive, holding the longer-term security has a higher expected return, and if it is negative, rolling the shorter-term security has a higher expected return.

In other words, if we're not looking for the risk-free asset, but the one with the higher expected return, shouldn't we be taking the term premium into account? Let's assume an investor with 100% fixed income, so we don't get into the whole portfolio discussion.

You chose a starting point when the term premium was positive and reasonably high. What if we look at some starting points where the term premium was very low or even negative? I see some negative values in the early 1960s (especially if looking at daily data)?

Of course one could challenge the ACM term premium model itself.

Thanks,

Kevin
Hi Kevin: to your point, here is Mr. Long and Mr. Short, but now purchasing their bonds at the end of 1965 (the bête noir start year for retirement since at least when William Bengen wrote in 1994). The 20-year Treasury yielded 4.50%, the 5-year 4.90%, per Appendix A-9 and A-13 of the SBBI.

Image

Results are almost the inverse: now Mr. Short does much, much better, as he rolls into higher and higher yielding intermediate bonds. But is that because the term premium was negative in the start year, or because 1965 marked the beginning of one of the steepest rises in yields ever seen? Can't get very far with two datapoints, of course.

Not sure of your ultimate goal. Would it be to build a regression model, with term premium (magnitude) and term premium (present/absent) the key predictor variables? Would the dependent variable be surplus/deficit income for choosing to own the longer bond?

Complexities abound (express the term premium as 20-year minus 5-year, or use a more complicated estimate of the slope from 1- to 30-year?) In any case, you probably know that the Treasury Constant Maturity series can be downloaded from the Federal Reserve site (and others). That will take you back to 1954 or so, and with monthly data, call it 800 possible start points more or less. A big enough sample, I'd say, to do some decent modelling.

If you do construct such a regression model, I hope you will present the results in a thread.
They that read the footnotes, they shall be saved; but they that pass over the appendices, they shall wander forever.
Topic Author
McQ
Posts: 379
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: 2022: Worst. Bond. Market. Ever?

Post by McQ »

sandan wrote: Sun May 15, 2022 11:26 am A discussion of the worst bond market ever without including the history of the republic and Alexander Hamilton is misleading.

A 90% drop is excluded. The events leading to the drop and correction are much more relevant than any chart.

https://allthingsliberty.com/2016/08/bo ... -republic/
Just a reminder: all my data presentations in this thread are limited to declines experienced by “the safest bond you can buy.” Debt from the Revolutionary war or incurred under Articles of Confederation, as discussed in the linked article, hardly counts; that would open the door to Weimar Germany bonds, Argentine bonds, etc. etc.

Even post-Hamilton (the 3s and 6s were floated in 1790), I dispute that US Government bonds meet the “safest” criterion during the 19th century—that’s why I switched to British Consols before 1918. Contrary to the thrust of the article to which you linked, absent one or two years during the Napoleonic Wars, Consols yielded less than US Treasuries and were arguably more free of default risk.

In that connection, the linked article is incorrect: the Treasury did default on notes issued in 1814. The mess wasn’t cleaned up until shares of the 2nd Bank of the US were swapped for the unpaid notes. (You can find the annual reports of the Secretary of the Treasury for those years on Hathitrust.org and I believe FRASER.) That default kept rates high in the US for quite a while.

One last point: I benefited greatly from the Sylla et al. database in constructing my critique of Jeremy Siegel’s thesis. In the course of that work I spent a lot of time in the individual city spreadsheet tabs for the 1790s (downloadable from EH.net). I decided I couldn’t construct a defensible stock index any earlier than January 1793, although there are scattered observations in the Sylla data back to 1791 and before.

The Treasury data is better; I believe the earliest trade in the Sylla data was November 1790. But these omitted years don’t affect the analysis of “how bad has it been,” because …

A great rally occurred in these bonds once the market decided that the Hamiltonian program was going to work out. Fortunes were made by intrepid speculators who travelled out to the countryside with a list in their pocket of the older bonds which were eligible to be swapped for the Hamiltonian bonds (technically, the Hamilton bonds were a refunding). They’d find some old soldier or merchant, not up on the latest news from the Federal government, long disgusted by the worthless paper with which they’d been saddled, and offer to buy it for a nickel or a dime on the dollar. Back at the Treasury, with the bonds swapped par for par, that got them a US 3s initially trading at about 50% of par, and soon at 70.

In short: I must reject the charge that important events have been left out, making the charts presented thus far “misleading.”
They that read the footnotes, they shall be saved; but they that pass over the appendices, they shall wander forever.
User avatar
Kevin M
Posts: 12168
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: 2022: Worst. Bond. Market. Ever?

Post by Kevin M »

McQ wrote: Sun May 15, 2022 5:25 pm
Kevin M wrote: Sat May 14, 2022 4:58 pm
McQ wrote: Sat May 14, 2022 1:24 pm
Kevin M wrote: Fri May 13, 2022 11:47 am
McQ wrote: Thu May 12, 2022 10:37 pm PS: of course, the 2001 starting point was chosen for rhetorical effect. Results would look different with, say, a 1965 starting point.

And here in 2022 you *know* which is the better analog--right? Don't you?
Image

In Dec 2001, the ACM 10-year term premium was 2.438%. In April 2022, the term premium was -0.193%.

Your thoughts?

Kevin
Hello Kevin: I haven’t worked with term premia much, or studied them historically. Term structure isn’t very meaningful in the deeper history where I spend most of my time (before WW II, there were mostly just long bonds and commercial paper/ short government notes; short and intermediate bonds weren’t typically issued, and only came into being through the aging of a long bond). In fact, before 1933, the SBBI had to manufacture the 5-year Treasury series from a yield curve interpolation because there were none to be found on the market (and I double-checked that absence by going back to the Bank and Quotation record for those years).

You are probably familiar with Durand’s 1942 paper, which mostly found flat yield curves from 1900 to about 1930. I’ve long wondered whether the Treasury’s forcing of a positive yield curve during and after WW II didn’t create what we now think of as the “normal yield curve.”

Long story short, I have no thoughts to share on the chart. But I’d be happy to respond to a more specific assertion or challenge.
Thanks for replying. Let me try a more specific challenge.

I don't think there is any serious challenge to the assertion that the lowest risk asset is one that matches your holding period and unit of account (e.g., US dollar or real consumption). I think I first learned this framework from this 2012 post by forum member bobcat2 Risk Basics: What is the risk-free asset?. So in your example, sure, the 20-year nominal Treasury has less risk for a 20-year holding period and the US dollar as the unit of account. Whether the outcome from rolling is higher or lower, it still demonstrates the variance of rolling vs. holding; i.e., holding for the term is the lower risk choice.

But, you used a specific time period to compare outcomes between rolling a shorter-term security and holding the longer-term security until maturity You acknowledged that you picked the starting point to demonstrate the point. You acknowledged that picking a different starting point would result in a different. Your challenge was whether or not we know which is a better analog.

The challenge is that if we want to make a rational decision about the best expected outcome in comparing holding vs. rolling, the term premium is a relevant input to consider. My understanding about the term premium is that when it's positive, holding the longer-term security has a higher expected return, and if it is negative, rolling the shorter-term security has a higher expected return.

In other words, if we're not looking for the risk-free asset, but the one with the higher expected return, shouldn't we be taking the term premium into account? Let's assume an investor with 100% fixed income, so we don't get into the whole portfolio discussion.

You chose a starting point when the term premium was positive and reasonably high. What if we look at some starting points where the term premium was very low or even negative? I see some negative values in the early 1960s (especially if looking at daily data)?

Of course one could challenge the ACM term premium model itself.

Thanks,

Kevin
Hi Kevin: to your point, here is Mr. Long and Mr. Short, but now purchasing their bonds at the end of 1965 (the bête noir start year for retirement since at least when William Bengen wrote in 1994). The 20-year Treasury yielded 4.50%, the 5-year 4.90%, per Appendix A-9 and A-13 of the SBBI.

Image

Results are almost the inverse: now Mr. Short does much, much better, as he rolls into higher and higher yielding intermediate bonds. But is that because the term premium was negative in the start year, or because 1965 marked the beginning of one of the steepest rises in yields ever seen? Can't get very far with two datapoints, of course.

Not sure of your ultimate goal. Would it be to build a regression model, with term premium (magnitude) and term premium (present/absent) the key predictor variables? Would the dependent variable be surplus/deficit income for choosing to own the longer bond?

Complexities abound (express the term premium as 20-year minus 5-year, or use a more complicated estimate of the slope from 1- to 30-year?) In any case, you probably know that the Treasury Constant Maturity series can be downloaded from the Federal Reserve site (and others). That will take you back to 1954 or so, and with monthly data, call it 800 possible start points more or less. A big enough sample, I'd say, to do some decent modelling.

If you do construct such a regression model, I hope you will present the results in a thread.
Thanks. Before reading this, I did my own simple model.

The lowest 5-year ACM term premium in the 1960s was -0.21% on 1/4/1962. Using 5-year and 20-year CMT data from FRED, and just using a simple total return model assuming either zero-coupon bonds or that coupons are reinvested at the original YTM:

Image

Here it is graphically, just connecting the endpoints for the 20-year, since all that matters is the 20-year return:

Image

My ultimate goal is simply to show a counter-example to the one you originally shared, where initial conditions are more similar to today. We now have two, although at the end of 1965, the 5-year ACM term premium was +0.12%, so my example is closer to today's conditions. However, 5-year yields at the beginning of 1962 and the end of 1965 were somewhat higher than now.

Thanks again,

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)
Topic Author
McQ
Posts: 379
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: 2022: Worst. Bond. Market. Ever?

Post by McQ »

SimpleGift wrote: Sun May 15, 2022 12:10 pm
McQ wrote: Wed May 11, 2022 5:00 pm Yes, you can lose 20% real, in one year, on the safest bond you can buy. It’s happened before, more than once.
After a double-digit real bond loss, how long has it taken historically for bonds to recover their real value? Have they recovered within 7 years (the duration of an intermediate-term bond fund)? Recovered within 10 years?

From the McQuarrie database of real bond returns, the table below lists each instance of a double-digit real bond loss since 1794, and then calculates the compound annual growth rate over the subsequent 7 years. If the initial bond loss had not recovered in 7 years, the CAGR after 10 years was calculated.
  • Image
    Data Source: McQuarrie database, 1794-2019
In most historical instances of double-digit real losses, bonds more than recovered their value within 7 years. The exceptions were 1918 (when bonds recovered within 10 years), 1947 (when interest rates were capped at 2.5% until 1951, while inflation soared), and 1968 (which saw two subsequent oil supply shocks and runaway inflation).

Barring monetary controls or runaway inflation, bond returns have been fairly resilient after steep real losses.
I like this analysis, SimpleGift--it gives me a fresh perspective on my bond data!
They that read the footnotes, they shall be saved; but they that pass over the appendices, they shall wander forever.
Topic Author
McQ
Posts: 379
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: 2022: Worst. Bond. Market. Ever?

Post by McQ »

Bonds in the Long Run: A Different Apercu

However bad 2022 has been, it’s worth remembering that for a long period prior to 2022, bond holders had enjoyed strong returns year after year. A follower of John Bogle might well urge, “Keep the 2022 bond market in perspective. Take the long view. Forget 4-month rolls. How has your bond allocation performed over the long term?”

Everyone has a different idea of how “long-term” might most usefully be defined from the standpoint of the individual investor. Suffice to say, “twelve-month holding period” is not on the list. Ten years would meet the standard for some; twenty years for rather more. And since it has been possible to buy a Treasury with a 30-year maturity most of the time since the 1970s, that holding period might top the list. Then again, a 45-year-old who begins to throttle back on stocks and to add bonds, in the spirit of Target Date funds, probably should use a 50-year planning horizon.

Next, once bond performance is to be measured over multi-decade intervals, it becomes important to focus on real returns. Inflation has been too variable, especially since the lapse of the gold standard, for 30-year nominal returns in one century to be comparable to 30-year returns in another.

Last, it is useful to work with annualized returns (geometric roots). It helps in keeping the chart scale under control.

Here is a version of a chart that appeared in my market history paper (https://papers.ssrn.com/sol3/papers.cfm ... id=3805927). It stops in January 2019, just before the pandemic and the Fed’s response sent safe bonds soaring (reminder, I use January to January returns to map onto 19th century compilers). Later in this post I’ll update it through 2022; this version helps to set the scene. For simplicity, it shows 20-, 30- and 50-year rolls to January of the year named, i.e., the annualized real return received by holders of long bonds over those intervals.

I found that 10-year rolls cluttered the chart (but see below); FWIW, and as will be seen later, 10-year rolls typically showed the same minima (e.g., 1982). In any case, almost all bonds in the index had maturities longer than ten years, arguing again for the longer rolls.

Image

The first takeaway: 1982 marked a kind of world-historical low in real returns on long bonds. Rolls ending that year show the lowest 20-, 30-, and 50-year rolling returns across the 200-year history, negative over each interval.

And that’s important: most investors active today (always excluding Taylor) have spent their entire investing career thinking that a rally from off a world-historical low point reveals the usual, customary and reasonable returns to be expected from investment grade bonds. Nonsense: it shows only what can happen during one of the greatest and longest bull markets ever seen in bonds.

The second takeaway: the decades following World War II show most of the worst rolls. With one exception, all of the negative annualized returns occur in this time frame. The exception, unsurprisingly, came just after WW I, courtesy of another great inflationary spike. In general, the 19th century was a much better time to be a US bond holder than the 20th century—until after 1982.

The period after World War II is unusual in another respect. Here is a chart showing ten-year rolls for bonds and also for stocks (real total return in both cases).

Image

The years after WW II are the only period where stocks walloped bonds, 10-year roll after 10-year roll, decade after decade. At all other points in this history, over 10-year rolls either stocks and bonds soared together (1820s, 1850) or tanked together (1842, 1918), or moved fitfully and moderately in the same or different directions.

That unusual post-war period dominated early editions of the SBBI, setting up the narrative that stocks inevitably beat bonds over longer intervals, a case made most famously by Jeremy Siegel, as in the chart reproduced in the post upthread by Jo Money. (With better 19th century data, I was able to show that had Siegel written in 1942 rather than 1992, he could not have fielded that narrative from the historical data to that point; the long-term performance of stocks and bonds had been about the same from 1793 - 1942).

In the US, stocks beat bonds with a stick from 1949 to 1969; and thus the myth of a necessary, inevitable equity premium arose.

Stepping back: looking on the bright side, even over 20-year rolls bonds can sometimes achieve stock-like returns (as in 1835, 1886, 1940, 2002). But bonds can also clock negative real returns over 20-, 30-, and even 50-year intervals. It’s happened before.

Chilling thought to conclude the section: those no good, terrible, horrible real returns on bonds are clustered in the years following the last time long Treasuries yielded less than 2.0% ...

Updating through April 2022

For this chart I added dashed lines which extend the 20-, 30-, and 50- year lines to January 2020, 2021, 2022, and “2023” (treating returns through April of this year as if they were from a full year). The Vanguard Long ETF is used for the post-2019 returns (BLV). Again, these are real total returns.
I also added and extended 10-year rolls. An interpretation follows the chart.

Image

The added data causes the rolls to rise on the chart through 2020, as the pandemic flight to safety pushed long bond yields to what may be generational lows. Then the 2022 decline caused all lines to turn lower. The reversal is moderate for the 20-, 30-, and 50-year rolls, where the last roll charted covers the periods from January 2003, January 1993 and January 1973, respectively. These lines remain far above the lows seen in the decades prior to 1982.

Results for the 10-year are a little different (with the last roll charted running from 2013 to “2023”). The annualized real return has dropped below 1.0%, after having peaked near 7.0% just a few years ago.

Conclusion: if your planning horizon spans 20 years or more, 2022 thus far might be considered just a part of the normal ebb and flow of fluctuating asset returns. You take the bad with the good.

But if your planning horizon, or holding period for long bonds, uses a 10-year metric, then you’ve seen a pretty dramatic reversal over the past few months. Through the start of this year your bonds had racked up a nice ten-year record; now, not so much.

I hope the charts help BH to see how very volatile real returns on long bonds can be over periods as short as ten years; and how very bad these returns can be, and for how long.

Caveat investor.
They that read the footnotes, they shall be saved; but they that pass over the appendices, they shall wander forever.
NiceUnparticularMan
Posts: 3822
Joined: Sat Mar 11, 2017 7:51 am

Re: 2022: Worst. Bond. Market. Ever?

Post by NiceUnparticularMan »

vineviz wrote: Fri May 13, 2022 12:49 pm
NiceUnparticularMan wrote: Fri May 13, 2022 11:06 am I didn't put a specific date on when academics first thought that IP bonds would be a good idea.

But personal investors can only shift into things that actually exist.
TIPS and inflation-linked savings bonds exist BECAUSE academics spent decades explaining why they were a good idea. This wasn't a realization that came to them AFTER "IP bonds appeared", as you wrote.
That's not what I wrote.
sandan
Posts: 378
Joined: Wed Apr 03, 2013 12:48 pm

Re: 2022: Worst. Bond. Market. Ever?

Post by sandan »

McQ wrote: Sun May 15, 2022 5:45 pm
In short: I must reject the charge that important events have been left out, making the charts presented thus far “misleading.”
Thanks. I missed the caveat about the safest possible bonds that could be bought.
Topic Author
McQ
Posts: 379
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: 2022: Worst. Bond. Market. Ever?

Post by McQ »

sandan wrote: Mon May 16, 2022 9:10 am
McQ wrote: Sun May 15, 2022 5:45 pm
In short: I must reject the charge that important events have been left out, making the charts presented thus far “misleading.”
Thanks. I missed the caveat about the safest possible bonds that could be bought.
Not to worry. The "safest bonds" constraint is key to the exercise. In its absence, the answer to the question, "How bad can it get for bonds?" is always, "minus 99.xx%," as in government bonds of Russia in 1917, Weimar Germany in 1923, Japan in 1945, and Hungary, Zimbabwe, Argentina and a host of others at other time points.

If it isn't a "safest bond," i.e., the full faith and credit obligation of a world hegemon, then it is just a risk asset, and by definition, the spectrum of outcomes for a risk asset includes "minus 99.xx%"
They that read the footnotes, they shall be saved; but they that pass over the appendices, they shall wander forever.
Post Reply