The interesting term here is the collateral. Handing Treasuries to the US Treasury and taking dollars back to sell for yen [13] finances Japan's position in the American bond market rather than retiring it, and the roughly $87 billion drawn [3] is about 7.9 percent of the roughly $1.1 trillion Japan holds [10][1]. The seller the long end was worried about is still there, funded rather than gone, which makes the facility look less like currency policy than like support for the Treasury bid wearing a foreign-exchange label.
The price is doing some work too. Satyajit Das puts the slide from 102 to 164 since end-2020 at 60 percent [2], which is the yen-per-dollar arithmetic (164 divided by 102 is 1.608, so the dollar bought 60.8 percent more yen at the low); measured the other way, a yen holder buying dollars lost 37.8 percent [2]. Both are correct, and the distance between them is roughly the distance between the Japanese importer's problem, which is energy priced abroad [5], and the carry trader's ledger.
That $87 billion is being asked to steady a much larger position. Das sizes the carry trade at anywhere between $500 billion and more than $4 trillion depending on definitions [11], which puts the intervention at between 2.2 and 17.4 percent of the thing it is meant to stabilise [3], and the honest reading of a range that wide is that nobody can size the position to within a factor of eight [4]. August 2024 is the reference case, when the Nikkei 225 fell 12.4 percent as the trade came off and asset prices fell globally with it [12].
This is probably wrong, but I read the operation as being about the calendar rather than the level: it buys Washington time before Japanese capital, which is exported at scale, starts preferring domestic yields [9] and turning up as thinner demand at Treasury auctions [10]. Yves Smith's account of Bessent's motive, either fiscal orthodoxy or the placation of a president who wants lower rates while running tariffs, deficits and an energy fight with Iran [16][18], matters less than the observation that neither motive produces a rate decision.
The counter-thesis is Plaza, where the risk turned out to be overshoot rather than futility. The G-5 got more than it asked for, the yen ran up far enough to require the Louvre Accord to walk it back, and the export rebalancing never arrived [15]. Overshoot is the outcome that would hurt here, because fast appreciation unwinds carry faster than drift does, and the 32 percent of Japanese equities held by foreign investors [7] is the second tripwire in the same trade.
What would prove this desk wrong: a BoJ rate rise that holds the yen without a second round of buying, and a repo repaid instead of rolled. Yves Smith reports that nobody expects the intervention to last more than a very short time [18]. The size makes that hard to argue with.
