Post-mortem on the July 31 intervention, four weeks out

The yen broke back through 160 to the dollar on Monday, its weakest level in a month, and the number matters less than what it confirms: more than half of the gains purchased on July 31 are gone. That operation was not routine. It was the first coordinated US-Japan yen-buying intervention since 1998, with Tokyo committing something in the neighborhood of $75 billion and Washington adding a smaller but symbolically loaded $5-10 billion of its own. The dollar was pushed from above 163 down toward 155 in the days that followed. Four weeks later, it is back above 160, and the drift did not require a crisis to get there — it required one hawkish paragraph from Kevin Warsh at Jackson Hole on Friday, the same speech that repriced September hike odds on this side of the Pacific, and the yen gave up half a summer's worth of official effort in a single session.

Diagnosis: the intervention treated a symptom.

Someone on this desk needs to say plainly what the July 31 communique did not: buying yen does nothing to the interest-rate differential that is selling it. The Bank of Japan's policy rate sits at 1.0%. The Fed's sits at 3.5%-3.75%, with fed funds futures now pricing better than a 60% chance of a hike to 3.75%-4.00% at the September 16 meeting. That is not a gap intervention can close; it is a gap intervention can only paper over, and only for as long as nobody tests it. Warsh tested it with a single sentence about having "work to do," and the test failed within seventy-two hours. Every basis point Washington adds to that differential this fall is a basis point Tokyo has to fight with reserves instead of policy, and reserves run out; policy differentials, left alone, do not resolve themselves out of politeness.

Pattern recognition: this is not new, and it is not free.

The July 31 action was itself a response to a level — 160 — that has repeatedly proven to be where verbal warnings escalate into rate-checking, and rate-checking escalates into actual intervention, and the pattern has now run at least three full cycles this year without ever addressing the underlying driver. Strategists on the street are already gaming out the next trip wire: another defense somewhere near 161, a harder one in the 162-163 band if the drift continues, with most of them adding the same caveat, that any fresh action buys weeks, not months. Call that what it is: an institution running the same play a fourth time and expecting it to hold longer because the last one was bigger. It will not. Size bought four weeks last time against a static rate gap. It will buy less against a widening one.

The risk nobody on this desk likes to model.

A 2.5-to-2.75-point policy gap does not just leak slowly through the spot market. It funds a carry trade — borrow yen at 1%, hold dollar assets paying north of 3.5% — that has a well-documented history of unwinding all at once rather than gradually, most memorably in August 2024, when a much smaller BOJ move briefly knocked a percentage point off global equity indices in a single session as leveraged yen shorts were forced out simultaneously. The current setup is larger, not smaller: the gap this cycle is wider, the intervention already spent is bigger, and the trigger sitting on the calendar — a Fed decision three weeks out that markets themselves cannot agree whether to price as a hike or a hold — is more consequential than anything that provoked the 2024 episode. Nobody is forecasting that outcome here. Everybody should be pricing that it is no longer a tail case.

Cost accounting nobody wants to put in writing.

$75 billion of reserves is not an abstraction; it is capacity that does not exist the next time this desk needs it, and the next time is arriving faster than the last one did. If September delivers the hike the futures market is now leaning toward, the differential widens further at precisely the moment Tokyo's ammunition is depleted and its credibility from the last round is already half spent, evidenced in real time by a currency that just gave back the majority of what that ammunition bought. The uncomfortable comparison, and it should be made explicitly rather than left implied: Washington's own Treasury is running a parallel operation on the long end of its curve, expanding buybacks to lean against a 5.2% thirty-year yield it cannot otherwise control without touching the deficit that is driving it. Both governments are now spending real money to manage the visible symptom of a rate structure neither is willing to change at the source. Both operations will be judged, fairly, by how long the number stays put after the checks clear.