Signals, Noise & Bond Market Predictions
A hundred years, ain't a damn thing changed.
The bond market can pretty famously predict, or at the very least indicate with some degree of accuracy, that bad things are gonna happen.
Bond market rallies and crashes have anticipated some of the most momentous events in history including, but not limited to the D-Day Invasion and the fall of Nazi Germany2
and the US Civil War:
among other, more mundane things like global recessions.
Recently (past 50 years or so) it’s gotten more complex and we have to think about things like yield curves or the correlation between bonds and equities3, but suffice it to say that for a long time it was simple enough to see if price was going up or down. You just found that out, and you’d know if bonds were predicting Bad Things™.
So, as a financial history hobbyist, I’m going to explain my take on one of financial history’s unanswered questions re: the bond market and predicting things. I imagine it’s probably a good idea to do so now while the concept of geopolitical risk premium is still somewhere in the consciousness of investors.
Eventually, inevitably, it will again be relegated to the back. This is because time is a circle and we are likely living in a simulation with limited data. But, until the matrix resets - let’s see what we can learn about how our trading forefathers messed up, and maybe let it help us think about things more clearly.
Why didn’t the bond market predict World War One?
Well…..maybe a disclaimer first.
Of course, we don’t really have to care about the early 1900s. I mean, why would you? We have computers now. They were basically a different species.
Obviously it was very different, they had to worry about things like high energy prices & producers being unwilling to raise capex despite supply constraints…
Uhh…okay but, back then they had to worry about things like whether inflation had peaked and if unemployment rising would deflate artificially high wages…
No, no…what I’m trying to say here is that it was very different. They were talking about things like the Bank of Japan allowing inflation to run rampant despite wages not rising in a commensurate fashion, as speculation in the yen reached record levels4.
Actually, on second thought, maybe they weren’t so different. Let’s proceed then.
For real, why didn’t bond markets predict World War One?
Simple question. Really.
Yes, of course, it’s easy to laugh. Why should they have?
The fact is, this wasn’t some unforeseen consequence. The idea of the European alliance system causing a global war was not some wacky thing that just came up out of nowhere. The below was written in 1885.
Pretty prophetic, huh? (Although, predicting that Russian creditors will eventually get screwed is kind of like predicting eventually it will rain).
In 1899, the Warsaw financier Ivan Bloch estimated that “the immediate consequence of war would be to send securities all round down from 25 to 50 per cent”. This wasn’t a left field thing. The last one was a prognostication, but by this point - after the Franco-Russian Alliance and later the Entente Cordiale - was basically a consensus view.
“But Citrini, why would you expect the bond market to connect all these dots?”
Because up until around 1880 that was basically all they did. Just prolific geopolitical dot connection. Bond volatility in the global powers pre-1880 was basically 90% war/politics, 10% fundamentals (if that).
“Okay, but it kind of seems like you’re mad at a bunch of people from a hundred and twenty years ago?”
It’s projection. Here’s the thing: a year ago now, we had a very similar occurrence. The ball was wholly and sufficiently dropped by markets before Russia invaded Ukraine. And because we have been living in a world in which the actual. President. of. the. United. States. of. America. can come into the Briefing Room, address the American people in the most straightforward manner imaginable, just:
There’s gonna be a land war in Europe.
And then markets close Lockheed Martin down 7 basis points on the day and the index of the invading country up 1.57%. 5
As the president once said, “trying to make up your mind, huh?”
Because we have lived in that world - the one that bid up LMT with like a 2 day delay after the invasion (efficient, forward looking, etc) - we have a shared commonality with the market participants of July 1914. We are both very bad at predicting geopolitical things.
July 22nd, 1914, just SIX DAYS before war was declared, the Russian chargé d’affaires in Berlin warned a German diplomat that German investors would ‘pay the price with their own securities with the methods of the Austrian politicians’. Still, consols traded relatively tight & global yields didn’t move much more than 5-10bps in anticipation, if at all.
That’d be like if the President of the United States told us Ukraine was going to be invaded and we were all like “ok, boomer” (last time, promise).
So, there’s a lot we can learn about our own follies by discovering why, exactly, our ancestors screwed up like we did.
Perhaps, even, there’s a way we can remind ourselves to not do so. But first, like with any mistake, we must understand why it happened. Let’s establish, firstly…
WE USED TO BE GOOD AT THIS
The solution to this question becomes especially significant to deduce when you consider that until roughly the early 20th century, bond markets (typically benchmarked or represented using UK consols, the prevailing risk free rate of the era) were quite accurate with their predictions. In fact, “quite accurate” does not do it Justice.
They consistently nailed it, to be honest.
You could use yield changes as a ranking system of significance in a modern history book.
Note the date on that last one - July 31st. Franz Ferdinand was assassinated a full month prior. UK consols really dropped the ball (not as bad as their continental cousins but still badly in comparison to their signal value prior).
Not just that, but they continued dropping the ball. To understand how unusual this was, let’s go back to 1848.
Cannon fire is bad for money
In the mid-1820's, conventional wisdom among the financially literate was that “'cannon fire is bad for money” - a phrase coined by the Comte de Villele in the mid-1820s.6:
And this was basically the only reason bonds, or as they were referred to then “funds” (or the french rentes, if you wanna be a drama queen about it), ever moved more than a percent or two in a quick fashion.
In 1848, with images of guillotines still fresh in the mind of the older generation, French peasants begin talking about things like “better working conditions” and “liberté, egalité, fraternité” - I know, petrifying.
This is what global bond yields did the second they got wind of that nonsense:
(This is spread to consols, basically g-spread but for olde tymes.)
Yes, Russia - a country very definitively not France - saw its 5 per cent bonds go from yielding ~2% to ~8%. Because there was a revolution. In France.
Russian aristocrats do not play that shit, homie.
And the crazy part is, they were kind of right to do so. The 1848 revolutions were basically the Arab Spring of Europe (other way around, but as we have established - time is a circle). They spread like wildfire, and during the industrial era workers not working was like…idk, the New York Stock Exchange forgetting to do the opening auction. It just didn’t happen.
Of course it did - sometimes - and when it did it greatly affected a country’s ability to repay its debt. So that’s score 1 for pre-1880 bond market.
Between 1845 and 1880, not just war, but the mere threat of war would send bond markets spiraling.
And they kind of kept the streak going, generally, something would happen and chances are if it was able to be predicted bonds had sold off in a way that they really only did when global political risk was an issue. It would not be a mistake to say,
If you lined up all the outliers of bond volatility before 1880, the dates listed would mean much more to a political historian than a financial one.
What happened in 1880?
Within the span of a decade, the behavior of the bond market in response to political tensions changed. Drastically. See, around 1880, it seems that the idea that cannon fire is bad for money fell out of vogue. Markets became very complacent with geopolitical risk. And when I say around 1880, I mean +/- 2.5 years.
Here’s an example of how drastic this change was. In 1876, close to the end of the Russo-Turkish war, Russian yields rose 68bps in the span of three days and stocks collapsed.
By the time the carnage was over, a column in The Economist asked why it happened at all - since Russia had basically already won,
“The decline in Russian stocks, with an army in possession of the Turkish capital, is remarkable. Peace might enable Russia to acquire a large war indemnity, to disband its troops, and to cease extra expenditure.”
Then we pass some time, and in 1885 - only 9 years later, Russia has an armed conflict in Afghanistan. Called the Panjdeh incident, the Russians captured an Afghan border fort, threatening British interests in the area. Seeing this as a threat to India, Britain prepared for war. That’s a setup to a World War right there. Much more risk than Russia-Turkey when Russia has already won. But yields didn’t even flinch this time. Consols went down a touch, but mostly nothing happened. Russian stocks actually went up.
Juxtapose the earlier Economist writer’s confusion as to the decline in Russian assets with this one’s confusion as to the lack of a decline in Russian assets, with some great quotes that are still relevant today:
Contrast that to 1905. Russia is deeply involved in the Russo-Japanese war and then, similar to the revolution that sent yields up 4x in 1848, a revolution happened in Russia.
So now that we know how bond yields a thousand miles away in Russia reacted to a revolution in France, we probably have a good idea of how - 57 years later - Russian bond yields reacted when there was a revolution in Russia?
If you guessed “go up more than they did when the revolution was in France”, you’d be quite wrong.
And that brings us to the market’s attitude when Franz Ferdinand is assassinated by Gavrilo Princip. Let me set the scene.
EXT. WALL ST - MORNING
It’s 1914, you’ve just gotten your paper (The Economist, before it was a contra7) from the local newsie and you read something like this:
This is one of those headlines you see but don’t actually read and definitely do not process. It’s hot in Berlin & two countries you have never interacted with dislike each other, who cares!
This article was published after Franz Ferdinand was killed, like two full weeks after.
Then they get a bit more concerned. The first time global markets allude to financial instability is July 22nd, 1914 - 6 days before WWI is declared and about 3 weeks after Gavrilo Princip decided to have a sandwich that lead to him setting into motion millions of lives lost - in the Times.
Foreign political news, ugh. You probably throw this limey rag in the trash. Then a few days go by and this article finds its way onto your desk (unbeknownst to you, WWI will begin in 2 days).
old timey note: this article is using the actual meaning of the word “incredible”, that is to say, not credible.
Ok, now comes decision time. You think “well, I remember when there was all that hubbub in the Balkans. That could have lead to war, but all it lead to was a bad month and then everything bounced back”. Because you are living in the same kind of conditions that we were in February 2022, the kind where geopolitics doesn’t really matter and is just a mild annoyance at worst and a dip buying opportunity at best. Sure, big move in rates. But it’s not that bad. And stuff like this was happening in eastern Europe a lot.
It’s kind of like once there’s a terrorist attack, you sell everything. But then you just start ignoring it. Al Qaeda was scary, but ISIS is a distraction. After all, this group was called the Black Hand. Super cringey name. You’re not gonna dump your holdings for some Serbian edgelords. Just three weeks ago this was “dullness” because of the “Austro-Serb complications”8. How bad could it really get?9
(Spoiler Alert: Bad)
And, of course, that was what everyone thought. And that’s why the bond market didn’t predict WWI. And you’d be forgiven for thinking you were 100% right in that, at least for a few months.
The reaction was not pronounced at the onset, although liquidity fled the market at some points (mainly because of dealers lacking the requisite pluckiness, according to the Times).
It was a very gradual, very painful lesson that it’s probably better to react like they did in pre-1880s and price in all the risk at once so you can get down to pricing whether or not it’s still there (which typically entails going up, or, if you’re a European natural gas investor, buying a bullet and renting a gun).
From peak to trough, consol prices declined 44 per cent between 1914 and 1920, an increase in yields of 251 basis points off historically low yields (~1 7/8% pre-Archduke Assassin).
The figures for French rentes were similar (a 40 per cent price drop and a 222 basis point yield increase).
Britain and France were the two great powers that emerged on the winning side of the war. The other three all suffered defeat and revolution. The Bolshevik government defaulted outright on the Russian debt, while the post-revolutionary governments in Germany and Austria reduced their real debt burdens drastically through hyperinflation. The fact is this: sometimes you should actually be worried about geopolitics. It’s very rare, but there’s about a 50% chance in your lifetime that some individual geopolitical event will send otherwise safe, high quality, liquid assets down 50-100%.
Generally Accepted (aka wrong) Reasons Markets F*cked Up:
Whenever an academic field of study has more than one commonly accepted reason for something to have occurred, you can be pretty sure they’re all wrong. Academia does not stomach “it might be this, or that” unless it truly has no clue.
Explanation 1: Buy and Hold Investors
Does anything happen in finance without Buy and Hold investors getting blamed?
The idea goes like this: The growth of private and public savings banks, designed to attract large volumes of relatively small deposits, created a new market for government bonds, not least because such institutions were often obliged (for prudential as well as political reasons) to hold such securities as their principal assets. In Britain the number of individual holders of consols remained small at around 200,000 (a fact lamented by The Economist), but the number of small savers with accounts at trustee savings banks and post office savings banks soared from 1.5 million in 1860 to 10.2 million in 1910. This phenomenon occured in Russia as well as France, where rentiers (French is so over the top) or holders of smaller denominated rentes became more commonplace. This investor base doesn’t care about what’s going on over on the Continent.
This might have played a role, but still - an investor base that only buys and holds and nevers sells onto the market does not reduce volatility in times of crisis. One need only ask Kuroda about this thesis, he will tell you.
Explanation 2: The Gold Standard
International monetary standardization may well have been a factor in the long-term convergence of great power yields. Britain had been on gold since the 1820s, while France and Germany adopted gold in the 1870s; Austria-Hungary and Russia followed suit in the 1890s, though the Austrian currency was never freely convertible. As The Economist noted:
But this would only affect the spread on these bonds. It shouldn’t have made them more sanguine about the risks of a global war. Gold convertibility can and will be abandoned whenever it is convenient, and there were precedents for this.
There is a simpler explanation: information.
Signal vs. Noise
The monetary, international, and fiscal changes discussed above cannot explain why the London bond market failed to anticipate a great power war. After all, historical experience strongly suggested that such a war would cause at least some combatant countries to suspend gold convertibility, might well lead to the closure of stock markets, and would certainly cause a sharp deterioration in the fiscal positions of all the powers involved. If the political crises in the decade before 1914 increased the risk of such a war even momentarily, bond yields should have ticked upwards in response. It is apparent that the concept of the geopolitical risk premium became less and less important gradually leading up to the late 1800s and then virtually disappeared. Yet the evidence from the financial press is unambiguous. The crises over Morocco and the Balkans had negligible impacts on investor sentiment. Even the assassination of the Archduke Franz Ferdinand on June 28th 1914 had no significant effect on bond yields.
Traders in 1914 were oblivious to global political risk. Just like we were in early 2022. 1880-1910 was relatively peaceful, at least from a global markets impact point of view. And peace leads to complacency. But there were other peaceful times, and the bond market still leapt at the chance to send yields up 2 or 3x in response to political risks.
Here’s a question to keep in mind: what is the average reaction of someone who is not accurately deciphering between signal and noise?
By the early 1900s, rumours of war could cross Europe in a matter of hours rather than days. It’s that increased access to information between 1880 and 1914 that changed the dynamic, in my opinion that's the missing link.
But wouldn’t more information work to enhance the significance of political risk, making the market more sensitive to such rumors? . Moreover, increased international integration should have made the market more sensitive to such rumours - no?
It’s easy to think of all information as being an edge, a positive thing, when you are operating as a market participant. This could not be further from the truth. Information is great, sure. But it’s better to have none of it than all of it. Having too much information leads you to conflate knowing things with understanding them. Information is like a safety blanket. It doesn’t really protect you, except maybe from your own emotions.
There exists a sweet spot for information that pre-1880 constantly existed inside of.
Pre-1880, most of the time, all information really was an edge. You just had to do some really crazy things to get it, like deal with pigeons and boats:
Getting a faster boat across the Atlantic was even more lucrative for traders. In the middle of the 19th century, it took eight days to cross the Atlantic, and any person who could get important news across the ocean more quickly could profit from it. In one extreme example, when it became known in New York City in 1865 that the South had lost the war, financier Jim Fisk chartered several ships that were faster than the mail ships then in use and sailed them to London with orders to his brokers to sell short Confederate bonds10. When the news became known in London that the South had lost, the price of the bonds went to zero, and Fisk of course made a fortune.
If you were plugged in, you were getting a telegraph weekly. Maybe daily in the later years if you were a big shot. Newspaper in the morning. And then people would tell you this that or the other thing and you’d judge it. That was it. That was the information.
Before 1880, information - news - likely could be ascribed the virtue of importance solely by the fact that it had traversed the arduous multi-step journey to reach the haven of your eardrums. The guy writing the Economist has 80 pages or so to fit all the world’s news for that day. If you hear about something more than a few times, it’s probably a good chance you’re gonna hear about it a lot more. Before about 1860, the fact that you even knew about something happening on a different continent meant it was probably a good enough reason to sell.
And then ticker tape came out. And then the telephone started becoming widely used. Everyone has a telegraph and gets messages constantly.
On March 3, 1846, the New York Herald reported that “certain parties in New York and Philadelphia were employing the telegraph for speculating in stocks.” Traders, however, had plenty of help, for a price. One news entrepreneur, Daniel C. Craig, made a fortune selling information on European news to traders. Using a combination of pigeons and express couriers, he got news from European steamers which had just sailed into Halifax harbor in Nova Scotia to Boston before anyone.
The information was then sent to New York via telegraph. Craig was so effective in gathering information ahead of others he was later hired by the newly-formed New York Associated Press. The press was apoplectic about traders and the agents who supplied them with information. There was particular anger directed against telegraph operators who were believed to have close relationships with traders; editorials railed against the telegraph’s potential for falling “in the hands of bad men.” Receiving information first was so valuable that some speculators, at times in cahoots with telegraph operators, were willing to cut the telegraph wire after the news had been received! Because telegraph operators were among the first to receive tradable news, some operators cut out the speculators altogether and went into business for themselves. This continued for decades: when the Civil War broke out, several Western Union employees made a fortune in the gold markets by leveraging their advance knowledge of war news. Forty years later, the telegraph was still a stock speculators’ realm. In 1887 the president of Western Union said 87% of the company’s revenue came from stock and commodity speculators, and from racetrack gamblers.
Most information is noise. Even in the above example, the edge had nothing to do with the information and everything to do with the latency. The news you needed to pigeon to your buddy was probably very obvious to the pigeon guy, because he was filtering.
This was invented in 1869 and became commonplace in the late 1800s, the Edison Universal Ticker Machine:
Pre-Civil War, there existed The Economist, The Times, some Investor’s Almanacs and that’s about it. Here are some completely coincidental founding dates of various publications you may be familiar with:
London Financial Guide (later, Financial Times): 1888
Customers' Afternoon Letter (later, The Wall Street Journal): 1889
Reuter's Telegram Company (later, Thomsen Reuters): 1865
New York Associated Press (later, Associated Press): 1846, in 1899 used Marconi’s wireless telegraph to cover a yacht race, the first news test of the new technology.
And you’ve got subscriptions to all of them! And the broker has a telephone now. He’s getting updates, and he’s relaying them to you, and you’re dealing with unreliable news and bad news and biased news from sources that are dealing with the same exact thing. And so you either filter effectively or you become inundated.
You try to separate the signal from the noise.
And guess what, since things (on balance, at least in the course of global events and geopolitics) are more often “not a catastrophe” than they are “a catastrophe” you begin to think you’re pretty good at this. You don’t really need to worry about it because you can tell when things are a big deal. The idea that “people are always overreacting” becomes a truism for you. Because, hey, it’s kind of true. You know what’s real. Stuff comes in that is similar to past stuff and you coolly and calmly say “no”.
But what you’re really doing is handily ignoring data that you don’t like or that doesn’t fit with your worldview, and worse you’re mistaking it for accurately filtering signal from noise and not jumping to act emotionally11. After all, emotion is the enemy of the investor. Right? 12
The average reaction of a group of people not accurately deciphering between signal and noise? It’s not so much of a reaction as a lack of one. Indecisiveness, dismissal of counter narratives, downplaying the downside. These are all far more harmful over the long term to an investor than any single decision, but not necessarily more harmful in their immediacy or volatility. Noise drives you to make no decision and justify it as making one. So you can stay with your indecisiveness for as long as markets allow you. And in WW1 they allowed you to for a while.
We can learn from that. We’re constantly barraged by mostly meaningless information. In my own experience, I have found out two things to occur: information that typically has high signal value will either be actively sought out or will find me itself. I try to stay abreast of events, even subconsciously, but I don’t try to understand them all. I let them coalesce and form some collection that can be drawn upon later on, so that the dots can be connected. The headline is enough.
The Ukrainian invasion is a good example. I was on the fence about it. Then the signal found me (the President said it was going to happen on live television, doesn’t get much more signal value than that). Could it have been wrong? Yeah, sure, it could have been some approval rating BS or whatever. But the risk/reward of acting like there’s not going to be an invasion after something like that is not in your favor.
The Chinese reopening is another one. You want high signal value? It was pretty obvious. The change in tone, the drastic changes in stance, the lack of any rhetoric after the party congress. I didn’t need to be constantly reading SCMP’s takes on whether or not it was going to happen - that’s noise.
Overall, I really just wanted to write about why bond markets didn’t predict WW1 (but did predict Hitler’s rise and fall, so, you know, maybe give the yield curve inversion a little longer on the whole recession thing, idk). But this has always been what I’ve thought happened. It wasn’t the fiscal or monetary or market integration variables, it was the market psychology. An entire generation that was getting way more noise than signal because of technological advances, with little in the way of filtering as an aggregate market participant, that also had become complacent with geopolitical risk.
Sounds familiar, right?
or, arguably, cause
classic distressed case
just adding this because it’s fascinating, but the stock-bond correlation was basically screaming for a decade that the US was headed for some bad (read: stagflationary) juj
Someone tell Kuroda that SFSOs and YCC and all this monetary stuff is the wrong way to go about it, he just needs to tell Kishida to put a tax on geishas.
Stop me if you’ve heard this one before: “They’re not really going to do that, it’s just the media being dramatic.
Cannon fire was not necessarily bad for money this time. It was bad for some money, like Russian sovereign debt holder’s money and anyone who bought the dip on RSX (I mean..) but if you bought broad European indices when the first shot was fired you’d be positive in terms of total return right now. Marko Papic’s rebuttal to this is:
Kidding, of course. The Economist has perpetually been and will always be a contra. Here’s the front page a week after Franz Ferdinand was shot:
“Russia invaded Ukraine already! They took Crimea. Nothing happened. It’ll be fine”
Fun Fact: The Confederacy was exchanging confederate paper for Union paper up until the very end of the war. This probably would have been a good signal that wouldn’t even have required a boat. If you think you’re going to win the war, why would you exchange your notes (which, presumably, will be worth money) for theirs (which, presumably, would not).
The market can’t bottom until the VIX spikes, bro (also note they upgraded from “Austro-Serb complication” to “Austro-Serb crisis”)
This is, of course, from a more primitive time when they were not aware you can simply blame everything on market makers. They actually thought market makers were people and not vague demonic entities