The 1985 agreement and the power of signalling
The G5 finance ministers and central bank governors met at the Plaza Hotel in New York on 22 September 1985 and put out a communiqué saying, among rather a lot of other things, that exchange rates should play a role in adjusting external imbalances and that the participants stood ready to cooperate more closely to that end. Behind the language sat a commitment to sell dollars and buy their own currencies.
The amounts were trivial. One reconstruction of the operation puts the total intervention commitment at about $10 billion, split roughly $3.2bn from the United States, $3.0bn from Japan, $1.8bn from West Germany and $1.0bn each from France and the UK; another puts cumulative intervention at about $11 billion through the fourth quarter of 1985. A third account has up to $18 billion authorised with about $10 billion actually deployed over roughly six weeks. Against global foreign-exchange turnover estimated at around $200 billion a day in 1985, $11 billion spread over sixty trading days works out at roughly 0.1% of daily turnover. The $10 billion figure is a little over 0.2% of US GDP at the time.
What that tiny sum accompanied was enormous. had risen about 44% into 1985 and then fell about 40% between 1985 and 1987 against the major currencies. Against the yen the fall was about 51%, from roughly ¥239 to ¥121; against the Deutsche about 46%. Between February 1985 and September 1986 the dollar went from about ¥260 to ¥155 and from about DM 3.40 to DM 2.00, while sterling rose from $1.10 to $1.47. The G5 reconvened at the Louvre on 22 February 1987 to stop it.
There is a widely circulated claim — Wikipedia's, and explicitly flagged there as uncited — that most of the devaluation was caused by the $10 billion the central banks spent. Take that with a shovel of salt. Nobody has produced a credible empirical split between order flow and expectations, but the arithmetic above is why most analysts land on the second: five governments announcing in unison which way they wanted a currency to go was the trade, and the money was the receipt. CEPR's retrospective Lessons of the Plaza Accord and Yoichi Funabashi's Managing the Dollar are the standard accounts if you want the diplomacy.
The myth of the Plaza-induced Lost Decade
The popular version of this story, especially popular in Tokyo, is that Plaza killed Japan: the yen doubled, exports died, the bubble inflated and then the 1990s happened. Revisionist economists think this gets the causation wrong at the crucial step.
Their argument is that the appreciation was the shock and the Bank of Japan's response was the disaster. The official discount rate was cut from 5.0% to 2.5% by February 1987, and it then stayed at 2.5% until May 1989. Two and a half years of very cheap money, long after the initial adjustment shock had passed, is what funded the credit expansion into property and equities. The shock also exposed institutional vulnerabilities in Japan's financial system, which is a large part of why the same communiqué produced such different results in different countries.
The control experiment is West Germany. It signed the same document, watched its currency appreciate 46% against the dollar, and did not get a bubble, because it did not pair the stronger currency with the same prolonged accommodation. Appreciation alone does not inflate asset prices. Appreciation plus a central bank that keeps rates on the floor long after the emergency has passed does.
The Mar-a-Lago Accord blueprint
There is no Mar-a-Lago Accord. There is a paper.
In November 2024, Stephen Miran — then a senior strategist at Hudson Bay Capital, later a Federal Reserve governor — published A User's Guide to Restructuring the Global Trading System. It argues that the dollar's reserve-currency role makes it structurally overvalued, that the overvaluation hollows out US manufacturing, and that correcting it is a precondition for narrowing the trade deficit. It styles a coordinated depreciation deal as the "Mar-a-Lago Accord", in deliberate echo of 1985. The stated objective, as one summary of the paper puts it, is "to achieve a significant devaluation of the dollar … while maintaining the hegemony of the dollar as a global currency". Miran himself wrote that doing this would require "careful planning, precise execution, and attention to measures to minimize adverse consequences".
The toolkit is not a signing ceremony. It is tariffs used as both revenue and leverage, on the theory that much of their burden gets shifted onto foreign exporters through currency offset; bilateral currency agreements; a willingness to intervene in support of trading partners' currencies; and, at the sharper end, possible charges on foreign official holdings of Treasuries. Commentary describing this as a Trump policy is running ahead of the evidence. The paper had no formal status with the administration, and no multi-country dollar-depreciation treaty has been signed. What it is, is a roadmap that a lot of subsequent behaviour happens to match.
One comparison worth resisting. The 1985 dollar peak everyone cites is a nominal trade-weighted peak, not a real effective exchange rate. The BIS real effective index for the US stood at 109.35 in July 2026 on a 2015=100 basis. Whether that is above or below February 1985 in real effective terms is not something we can establish from what we have, and anyone telling you today's dollar is "as overvalued as 1985" is probably comparing two different series. On the current account, the US deficit was around 3.6% of GDP in 1985 and has run around 4% in recent years; we do not have a clean 2026 figure.
Why it matters to this crash
Because the machinery has already been switched on, and it has been re-plumbed in a way that changes where the stress lands.
On 31 July 2026 the United States and Japan carried out their first joint yen intervention in 28 years, driving the yen up from a near 40-year low to 157.40 within two days. Treasury Secretary said the US "would not hesitate to participate in further coordinated intervention to correct the yen's significant undervaluation", and told CNBC: "We will do whatever it takes to support them in a way that helps the American economy, the American taxpayer." A Reuters photograph at a Camp David cabinet meeting caught his notepad reading "Task pending: buy Japanese yen, 5–10.000 billion [dollars]". Reuters' own read on 3 August was that the exercise fits the White House's long-stated goal of weakening the to shrink the trade deficit. Analysts in Tokyo reached straight for the phrase: a new Plaza Accord.
The genuinely important change is not the diplomacy but the funding. Japan's Ministry of Finance spent ¥9.7885tn on yen-buying in 2022 (¥2.8382tn on 22 September, ¥5.6202tn on 21 October, ¥1.3301tn on 24 October) and ¥15.3274tn in 2024 (¥5.9187tn on 29 April, ¥3.5089tn on 1 May, ¥5.8998tn on 2 May). Those dollars came from the foreign exchange account, with US government securities sold as needed. That is the classic doom loop: yen weakness forces intervention, intervention forces Treasury sales, Treasury sales push US long yields up, higher yields push the yen weaker again.
That link has now been cut. On 3 August 2026 Bessent announced a repurposed COVID-era facility letting the Bank of Japan borrow dollars from the Federal Reserve against its Treasury holdings — more than $1tn of them — as collateral, using rather than sales. Japan's top currency diplomat, Atsushi Mimura, called it "a US-Japan currency union". We wrote it up on 18 August and returned to it later that day: one of the most-cited crash channels in the has been quietly plumbed around. Japan can defend the yen without becoming a forced seller of US government bonds. Read that as risk moved, not risk removed.
What would make this dangerous
The 1985 template only looks clean in hindsight because the Louvre meeting stopped the slide seventeen months later. The failure modes are specific and watchable.
A Bank of Japan that overcorrects. This is the 1986 mistake in modern dress: a currency shock met with domestic monetary policy calibrated to the shock rather than to the economy, and held there too long. Futures were pricing a 100% chance of a BoJ hike to 1.25% by the end-October meeting as of our 18 August coverage. An error in either direction, too fast or too slow, is the thing to watch, and the collateral damage this time lands in Japanese funding markets rather than Treasuries.
Intervention that stops working. The yen gave back about half its post-intervention gains within weeks, with USD/JPY back around 159 on the latest quote available to us and officials seen as ready to act well before 162.80, the level that triggered the last round. Plaza's authority came from the market believing five governments meant it. A round of joint intervention that the market fades is the moment the signalling channel — the entire mechanism, as the $10 billion arithmetic shows — stops being available.
The collateral chain going the wrong way. The means is taking Japanese Treasury collateral instead of the market taking Japanese Treasury supply. If the dollar value of that collateral falls while the facility is drawn, the arrangement stops being a neat piece of plumbing and starts being a credit exposure between two central banks, negotiated in public, with a president who has already publicly backed it.
And everyone under-prices: Miran's paper canvasses charges on foreign official holdings of Treasuries. Nothing of the sort has been enacted. But it sits in the same document as the currency that is visibly being followed, and the largest foreign holder of those Treasuries is the country the United States just intervened alongside.