The Butterfly Test: What If Tokyo Has Nothing to Do With Lagos?

By The Lokoja Contrarian · Published 28 September 2026 · Seven Gates Research · Data cut-off 28 September 2026; latest market readings 24 September

Here is the story, told the way it gets told at a good bar near a trading floor.

The Bank of Japan raises rates. Japanese government bond yields climb. A Japanese life insurer that spent twenty years buying American bonds, because Japan paid nothing, discovers that Japan now pays something, and buys a little less American. The yen carry trade, in which the world borrowed yen at close to zero and put the money somewhere more exciting, gets uncomfortable. Treasuries lose a buyer at the margin. US long-term yields rise. Credit spreads widen. Emerging-market borrowing costs follow. Nigerian Eurobonds reprice, capital turns cautious, and eventually the naira feels it.

A butterfly flaps its wings in Tokyo. Somewhere in Lagos, a basket of tomatoes costs more.

It is a lovely story. It starts in a central bank and ends in a kitchen, and it lets a Nigerian investor feel clever about the Japanese bond market. It is also exactly the kind of story that parts people from their money, because every link sounds plausible and almost nobody checks the chain.

So before we believe it, we are going to try to kill it.

How to kill a good story

Start in Tokyo. Suppose the BOJ keeps raising rates and Japanese investors keep buying Treasuries anyway. The first link has failed.

Suppose the yen strengthens sharply and nobody who borrowed it panics. Leveraged funds hold their positions, volatility stays asleep and global assets shrug. Another link gone.

Suppose the US 10-year sits above 5% for six months while high-yield spreads stay tight, refinancing windows stay open and defaults stay boring. Then 5% may not be a prelude to anything. It may simply be the new price of American money: expensive, and survivable.

And suppose that through all of it Nigeria keeps adding to its reserves, the naira stays orderly and Nigerian spreads over Treasuries do not widen. Then the butterfly never reached Lagos.

At that point the honest move is to drop the thesis. Renaming it, extending the timetable or explaining that the butterfly was flying into a headwind would all be easier, and all dishonest. Markets supply stories without limit and facts on a tighter ration. The story has to answer to the facts, and Figure 1 is where we make it answer.

A research house named after seven gates was always likely to find seven of them in this problem. In this case the count is honest, and the rule in the corner of Figure 1 is the whole method. A chain is only as strong as the link you have actually checked.

A seagull, a coffee break and a missing title

In the winter of 1961 Edward Lorenz, a meteorologist at MIT, wanted another look at a weather simulation he had been running on a Royal McBee LGP-30, a desk-sized machine full of vacuum tubes. To save time he restarted the run halfway through, typing in numbers from an earlier printout. The printout showed three decimal places, 0.506, where the machine had been carrying six, 0.506127. Lorenz went down the hall for a coffee. When he came back, the new weather had drifted so far from the old that the two runs might have belonged to different planets. A rounding difference of about three hundredths of one per cent had rewritten the forecast.

His first image for the idea was not a butterfly. In 1963 he repeated a colleague's quip that, if the theory were right, one flap of a seagull's wings could change the course of the weather for ever. The butterfly arrived in December 1972, when Lorenz was due to speak at a meeting of the American Association for the Advancement of Science and had not supplied a title. The session's convenor, Philip Merilees, wrote one for him: Predictability: Does the Flap of a Butterfly's Wings in Brazil Set off a Tornado in Texas?

One of the most famous metaphors in science came out of an administrator chasing a speaker. Even chaos theory had a filing deadline.

Now the awkward part. We used the butterfly above to say that small things can cause big things a long way off. Lorenz meant something closer to the opposite. His point was that the atmosphere is so sensitive to its starting conditions that nobody can trace the chain far ahead, because nobody can measure the world to the sixth decimal place. Look again at the first word of that title. Predictability. His answer, for long-range weather, was mostly no.

So a financial story that runs neatly from Tokyo to a Lagos tomato is, strictly speaking, anti-Lorenzian. It borrows the butterfly's glamour and ignores its warning. Markets are also harder than weather. Clouds do not read the forecast. Traders do, then act on it, and in doing so change it.

What meteorologists did with Lorenz is the useful bit. They stopped promising certainty and started issuing graded warnings against published criteria, so that "amber" means the same thing on Monday as it did last spring. The rest of this piece does the same. We will not forecast the tornado. We will say which warnings are in force, and on what evidence.

The long end asks for a pay rise

On 21 September the US 2-year Treasury yielded 4.76% and the 10-year 4.96%. Three sessions later the 2-year was at 4.87% and the 10-year at 5.18%.

The 2-year rose 11 basis points. The 10-year rose 22. The gap between them, the 2s10s spread, widened from 20 bp to 31 bp. That is a small bear steepener: bond prices fell and the long end fell harder. Nobody's lunch was ruined. It is still worth asking what the two maturities are saying.

The 2-year lives next door to the Federal Reserve and is mostly a bet on where short-term rates go next. The 10-year has a larger family: inflation, growth, federal borrowing, Treasury supply, fiscal credibility and the extra return investors demand for locking money away for a decade. That last relative is the term premium. When the 10-year outruns the 2-year, the market may be saying something ruder than "the Fed will stay tough". It may be telling the Treasury that it wants more money to own America's long-term debt.

Three sessions do not make a regime. But a bear steepener with the 10-year through 5% is precisely the shape the butterfly story predicts, which is why we note it rather than wave it away.

Then oil pulls up a chair

Brent above $100 complicates almost everything. It lifts headline inflation, freight and farm costs, and the cost of producing, moving, chilling and selling nearly every physical thing. It also traps central banks that would like to cut rates and would rather not explain, six months later, why inflation came back.

For Nigeria the arrangement is stranger. The country sells crude, and its citizens buy petrol and diesel at prices that ultimately follow crude. That is how a country can earn more from oil while many of its people become poorer because of it.

The dangerous hand, then, is not oil high on its own or Treasury yields high on their own. It is oil high, long yields rising and credit starting to care, all at once. That is when risks that looked separate turn out to share a cause.

Tokyo changes the arithmetic

For the better part of two decades Japan ran one of the great financial experiments. Money was nearly free, domestic bonds paid nearly nothing, and Japanese institutions had every reason to send savings abroad. Foreigners had every reason to borrow yen and spend them somewhere else.

Then Japan changed the rules. The BOJ's policy rate is now 1.25%, and the 10-year JGB yields about 3.1%, a number that would have looked like a misprint in 2019.

Picture a portfolio manager at a Japanese life insurer, a few floors above the pavements of Marunouchi. For most of her career the decision was trivial: America paid and Japan did not. Now a Treasury yields a little over 5% and a JGB about 3%. The Treasury comes with dollar risk, and hedging that risk costs roughly the gap between American and Japanese short-term rates. The JGB needs no hedge.

Run the rough sum and the answer is uncomfortable for Washington. If US short rates sit near the 2-year's 4.87%, hedging dollars back into yen costs something like 3.5 percentage points a year. Take that from a 5.18% Treasury and the hedged yield is about 1.7%, against roughly 3.1% for staying at home. The sum is crude; hedge costs depend on tenor, the cross-currency basis and how much currency risk an insurer is willing to leave open. The direction is not subtle. (Estimate.)

Japan does not need to dump a trillion dollars of Treasuries for this to matter. Markets are priced at the margin. Japanese investors only have to become less enthusiastic buyers, and a hedged buyer already has a reason to.

The yen has declined its part

The carry trade is simple and pleasant until it is neither. Borrow yen cheaply, convert into dollars, buy something that yields more, keep the difference. Danger arrives in two parts: your funding cost rises because the BOJ tightens, then the yen strengthens, so the currency you owe becomes dearer to repay. The natural response is to sell the foreign asset, buy back yen and go home. When enough people do that on the same afternoon you get August 2024, when a modest BOJ hike and a sharp yen rally gave Tokyo's stock market its worst day since 1987.

Here the story meets its first real problem. Despite a BOJ at 1.25%, the yen has stayed weak, at about 157.6 to the dollar. That is not an inconvenience to be explained away. It is evidence. The carry leg of the butterfly is not confirmed, and gate three is shut.

Everyone at the table is playing the players

Why should the US Treasury Secretary care about Japanese monetary policy? Because America sells an astonishing quantity of bonds and Japan owns an astonishing quantity of foreign assets. Scott Bessent has every reason to care what Japanese investors do next.

This is where Ben Hunt's poker framing earns its keep, and why the players in our hero illustration are sitting at a card table. Hunt's point is that investors cannot just play the cards; they have to play the players. The cards are inflation, jobs, oil, bond supply and earnings. The players are Bessent, the BOJ, Kevin Warsh at the Fed, Japanese insurers, hedge funds, pension funds and sovereign investors, each trying to guess what the others will do.

So the market does not simply forecast inflation. It forecasts Warsh's reaction to inflation, Bessent's reaction to yields, Tokyo's reaction to the yen and Japanese institutions' reaction to Tokyo. Then everyone tries to get there first. It is Lorenz's atmosphere, except the weather is also reading the weather forecast.

Credit is the lie detector

The US 10-year can rise for many reasons: inflation, growth, supply, fiscal worry, Japan, oil or all of them at once. Credit tells us whether higher rates have started to hurt people who actually have to borrow.

So far the answer is: not broadly. Broad US high-yield spreads were 280 basis points at the latest reading, about 4 bp tighter than at the end of July. At 280 bp the average junk borrower is being treated like a slightly risky friend, not a suspect.

Walk down the quality ladder and the tone changes. CCC-and-lower spreads were 1,112 bp, 106 bp wider than on 30 July, and 37 bp of that arrived in the last two sessions. The strongest junk issuers are still being shown to their table. The weakest are being asked for a much larger deposit at reception.

This is the klaxon in this piece, and we want to be precise about what it is sounding for. When the bottom of the market widens by a hundred basis points while the index barely moves, lenders are usually separating borrowers who can refinance from borrowers who cannot. It does not tell us the storm is coming. It tells us whose roof leaks first if it does.

And the sirens did not stop at our snapshot. The 25 September closes, published on 28 September, took CCC spreads to 1,128 bp and broad high yield to 293 bp, a 13 bp jump in a single session. Broad credit is still a long way from our 400 bp amber line. It is no longer quite asleep.

The reason the spread matters more than the headline yield is that a borrower pays both:

Suppose Treasuries yield 4.5% and a company pays a 300 bp spread. It borrows at about 7.5%. Now let Treasuries rise to 5.25% and the spread widen to 500 bp. The new cost is about 10.25%. Treasuries moved 75 bp; the company's funding cost moved 275 (Figure 4, left). That is the point at which macroeconomics acquires a cash-flow statement.

Who breaks first

Probably not a reinsurer, whatever the more cinematic versions of this story say. Higher yields raise the income insurers and reinsurers earn as their bond portfolios roll over. Reinsurance becomes dangerous when several bad things arrive together: falling asset values, a heavy catastrophe year, claims inflation, collateral calls and credit losses. That makes the sector a possible amplifier. It does not make it the obvious first casualty.

The first break is more likely to be dull. A leveraged company that borrowed at 5% and finds the refinancing rate is 10%. A private-credit vehicle whose borrowers can no longer cover their interest. A leveraged fund on the wrong side of a margin call. A sovereign that happily issued dollar bonds at 7% and discovers that the next buyer wants 10%. Crises rarely start with the most famous institution. They start with whoever has the nearest maturity date.

Then Nigeria sits down holding the one card that helps

Nigeria has something very useful in this particular hand: oil. With Brent above $100 the country should earn materially more dollars than it would at $70, which supports the current account, the reserves and, in principle, the naira. Gross reserves stood at $55.25bn at our cut-off and were rising. That is the shield.

So the interesting test is not whether the naira weakens. Currencies weaken; it is part of the job. The test is what the naira does while Nigeria is enjoying $100 oil. If reserves keep rising and the currency stays orderly, the shield is working and the contagion thesis weakens. But suppose Brent is $110, reserves start falling, Eurobond spreads widen, the CBN is selling dollars and the naira keeps sliding anyway. Then the capital account is overpowering the commodity windfall, and we would be worried out loud.

Watch the spread, not the yield

Suppose a Nigerian Eurobond yields 8% while the matching Treasury yields 5%: a spread of 300 bp. If the Treasury moves to 5.5% and Nigeria to 8.5%, the spread is still 300. Nigeria's borrowing cost rose, but nobody became more frightened of Nigeria. America simply got dearer.

Now let the Treasury move to 5.5% and Nigeria to 10.5%. The spread is 500 bp, and those extra 200 bp are the interesting part. That is global stress turning into Nigerian stress.

The Debt Management Office publishes Nigerian Eurobond prices and yields every day, so this is a test rather than a figure of speech. It is also the one gate we cannot yet score: we have not built a maturity-matched daily spread series for this piece, and the dashboard below says so rather than guessing.

Cardoso has trenches before he needs the MPR

The CBN recently reset the monetary policy rate to 23%. That looks like a large easing until you remember that Nigeria still runs very tight cash reserve requirements and keeps a drawer full of liquidity tools.

So if global stress does reach the naira, Governor Olayemi Cardoso's first move need not be 23% to 27%. There are trenches before that: FX liquidity, open-market operations, managing banking-system liquidity, and simply pausing further easing. Only if the naira keeps weakening and inflation turns up again does the policy rate need to move, and the word to watch is crawl. The risk is less an emergency hike than 23% to 24% to 25%, meeting after meeting.

The first thing to break may be dinner

This is where the story leaves the Bloomberg terminal. The first Nigerian thing to break may not be a bank, the naira or an insurer.

It may be a household budget.

That is the transmission in its plainest form:

The sovereign can earn more dollars at every step while the citizen ends up poorer. Economists sometimes call this a paradox, because "two people have two different balance sheets" sounds insufficiently academic. It is only arithmetic. The same barrel lands on two ledgers, and the central bank has to answer both.

Why we publish our tripwires in advance

Aesop's shepherd boy is remembered as a liar. Read the fable as a forecaster and his deeper failure was calibration. He raised the alarm with no threshold, the village learned to ignore him, and when the wolf finally came nobody moved.

Weather services learned this lesson the expensive way. A warning is only worth something if it is rare enough to be believed and its criteria are published before the weather arrives, so nobody can draw the target around the arrow afterwards. So here are ours. Amber means the condition is present and deserves attention. Red means the link is transmitting. The thresholds are Seven Gates analytical tripwires, not official market definitions, and we are publishing them before we need them.

Figure 6. Sources: Bank of Japan; Federal Reserve H.15 via FRED; ICE BofA via FRED; CBN; DMO; NBS; Reuters. Snapshot dates vary between 24 and 28 September 2026. Thresholds and status: Seven Gates Research. “Watch” means within striking distance of amber without crossing it.