Black Wednesday 1992: What the Treasury Papers Say It Cost
Some links pay us a referral fee; each one says so. Disclosure
Update alert · free
An email when the currency and market crises numbers change
When a rate, rule or filing behind this page changes: what changed, the one number that matters, and the source to check it yourself.
Quick answer
On Wednesday, September 16, 1992, the Bank of England spent the day buying pounds, raised its lending rate from 10% to 12% at 11:00 am and announced 15% at 2:15 pm, and at just after 7:30 pm the Chancellor took sterling out of the European Exchange Rate Mechanism (ERM) and cancelled the 15%. HM Treasury's own papers, released on February 9, 2005, say it sold about $28 billion of reserves that day (nearly $40 billion over August and September 1992) and lost nearly £800 million on its reserves operations in those two months. The much-quoted £3.3 billion is something different: the Treasury's 1997 paper calls it the “opportunity cost” of the intervention and the later rebuilding of the reserves, measured at the end of February 1994. George Soros's fund was widely reported to have made about $1 billion on the other side; no Bank or Treasury document we found names any winner. Figures as of November 16, 2026, all from the documents listed below.
Key takeaways
- The Bank of England's own bulletin dates the day: sterling at DM 2.7794 by 8:00 am, MLR (Minimum Lending Rate, the Bank's lending rate) set at 12% at 11:00 am, a rise to 15% announced at 2:15 pm to take effect the next day, and sterling's suspension from the ERM announced just after 7:30 pm, with the 15% rescinded. By 9:30 am on September 17, 1992 the rate was back at 10%.
- The Treasury says it sold some $28 billion of reserves on September 16, 1992, taking intervention over August and September 1992 to nearly $40 billion, and that its net reserves went from nearly $24 billion to minus $16 billion. That is a swing of about $40 billion (our arithmetic).
- Two different costs sit in the same papers. The loss relative to sterling on reserves operations in August and September 1992 was nearly £800 million, which the paper calls “the true cost to the taxpayer”. The opportunity cost, against a counterfactual of not intervening, was around £2.4 billion measured to October 1, 1992 and around £3.3 billion measured at the end of February 1994.
- The same paper shows the opportunity cost depends on the date you stop counting: £1.9 billion measured to April 1997 against keeping $24 billion of reserves, £4 billion on its preferred counterfactual taken to April 1997. The Guardian's 2005 report quotes a former prime minister saying figures of £13 billion, £20 billion and £27 billion had been bandied about.
- No Bank of England or Treasury paper we read gives Soros's profit. The roughly $1 billion comes from a 1994 magazine report reproduced in a congressional hearing record; Soros's own words in the same record are that sterling would have left the ERM “whether or not I speculated against it”.
CSV · 55 rows
Black Wednesday, September 16, 1992: the day by the hour, rates and the pound, UK reserves and the Treasury's cost estimates
55 rows from the Bank of England's November 1992 Quarterly Bulletin, three HM Treasury papers released on February 9, 2005, the House Banking Committee hearing record of April 13, 1994 and a Guardian report. Source and document location on every row.
The story in one paragraph
Britain had tied the pound to the German deutschmark inside the ERM, a system in which the pound could trade only above a floor set against the deutschmark, and the Bank had to buy pounds to hold it there. The bulletin records the pound barely more than one pfennig above its absolute limit of DM 2.7780 on August 26, 1992, the day the Bank moved to “overt buying of sterling for deutschmarks”. On September 16, 1992 that defence ran out of road in about eleven and a half hours. The paradox the documents show is that the government announced rate rises totalling five points in one day to protect a price, and the market did not take them as a reason to stop selling. By the evening the Bank said defending the parity was “prohibitive”.
The day, hour by hour, from the Bank's own bulletin
The Bank of England's Quarterly Bulletin for November 1992 includes an article, “Operation of monetary policy”, that logs the day. Times are London time; the rate named is the Minimum Lending Rate (MLR), the rate at which the Bank lent to the money market.
| Time on September 16, 1992 | What the Bank records | Bulletin page |
|---|---|---|
| By 8:00 am | Sterling had fallen to DM 2.7794; in the following hour the Bank undertook very heavy overt intra-marginal intervention, with help from the Bundesbank and the Banque de France | p. 392 |
| Shortly after 10:00 am | Interbank rates had risen to 12%-11 1/4% overnight, effectively discounting a 2% rise in rates | p. 392 |
| 11:00 am | With sterling still at its lower ERM limit, MLR announced at 12%, with no indication of how long it would apply | p. 392 |
| 12:00 noon | Money-market shortage revised to £900 million | p. 392 |
| 12:45 to about 2:00 pm | One-month rates went from 14%-13 1/2% to 15 1/2%-15% and then 18% (bid); the rates agreed that day for overnight Thursday and weekend money reached 100% and 180% | pp. 392-393 |
| 2:15 pm | A rise in MLR to 15% announced, to be effective the following day; the clearing banks deferred any decision and none raised base rates to 15% | p. 393 |
| 3:00 pm to 4:00 pm | The FTSE-100 had risen 72 points above its low after the first MLR announcement and closed 8.3 points up on the day | p. 393 |
| Just after 7:30 pm | The Chancellor announced sterling's suspension from the ERM and rescinded the decision to raise MLR to 15% | p. 393 |
| 8:00 am, September 17, 1992 | The Bank confirmed the 15% rescinded | p. 393 |
| 9:30 am, September 17, 1992 | MLR reduced to 10% | p. 393 |
Two details the headlines usually drop. First, the second rate rise never took effect: it was announced for the next day and cancelled that evening. Second, the Bank's own explanation for stopping was cost, not collapse: the bulletin says the prospective cost of defending the existing parity was “prohibitive”, and that devaluing inside the ERM was not realistic because a sustainable new rate could not be set amid the uncertainty before the French referendum.
In our arithmetic, 8:00 am to just after 7:30 pm is about 11 and a half hours, and the two rate announcements were three hours and fifteen minutes apart.
The pound and the rates before and after
| Date | Sterling in deutschmarks | What the source says |
|---|---|---|
| September 9, 1992 | DM 2.7860 | Sterling had fallen back to this level (Bank bulletin, p. 391) |
| September 14, 1992 | DM 2.8131 | Afternoon level, with the lira at the top of its ERM grid (Bank bulletin, p. 392) |
| September 15, 1992 | DM 2.7812 | Close, after sterling fell through the day (Bank bulletin, p. 392) |
| September 16, 1992, 8:00 am | DM 2.7794 | Heavy intervention begins (Bank bulletin, p. 392) |
| Bottom of the ERM band | DM 2.7780 | As stated in the Treasury's 1997 paper |
| September 18, 1992 | DM 2.6100 | Close, 6% below its ERM floor; $1.7435, 5.6% lower against the dollar than on September 16 (Bank bulletin, p. 394) |
| September 30, 1992 | DM 2.5005 | A new low (Bank bulletin, p. 394) |
| Mid-1995 | almost DM 2.20 | Low against the deutschmark (Treasury paper, 1997) |
| End of April 1997 | close to ERM levels | Treasury paper, 1997 |
From the bottom of the band to the September 30, 1992 low is a fall of about 10.0% (our arithmetic: 2.5005 divided by 2.7780). On rates, the Bank's lending rate was cut to 10% on the morning of September 17, 1992 and to 9% on September 22, 1992, according to the bulletin, so the 12% and 15% of Black Wednesday were in force for hours, not days.
What the Treasury papers say it cost
On February 9, 2005 HM Treasury released ten sets of papers on the ERM under the Freedom of Information Act, after a request from the Financial Times dated January 4, 2005, according to the Treasury's covering note, as published by The Guardian. Four of them deal with cost. We read the scans on the Internet Archive's copy of the Treasury page; the 1997 paper is the one that sets out the measures, and the table below uses its own words.
| Measure | Figure | How the paper words it | Date of paper |
|---|---|---|---|
| Reserves sold on the day | about $28 billion | “We sold some $28bn of reserves on Black Wednesday (16 September 1992), taking total intervention in August and September 1992 to nearly $40bn.” | August 6, 1997 |
| Net reserves, start of August to end of September 1992 | nearly +$24 billion to -$16 billion | “We held nearly $24bn of net reserves at the start of August 1992, moving down to -$16bn by the end of September.” | August 6, 1997 |
| Loss relative to sterling, August-September 1992 | nearly £800 million | “the loss relative to sterling was nearly £800m on reserves operations in August and September 1992” | August 6, 1997 |
| Same loss accumulated to April 1997 | just over £1 billion | “taking the total accumulated loss since August 1992 to just over £1bn by end-April 1997” | August 6, 1997 |
| Opportunity cost to October 1, 1992 | around £2.4 billion | “intervention in August and September 1992 cost us around £2.4bn” | August 6, 1997 |
| Opportunity cost to April 1, 1993 | £3.2 billion | the “headline figure in the Bank's original analysis” | August 6, 1997 |
| Opportunity cost, intervention plus rebuilding reserves, end of February 1994 | around £3.3 billion | “probably the most sensible measure of the cost of the 1992 intervention and subsequent rebuilding taken together” | August 6, 1997 |
| Same measure taken to April 1997 | £4 billion | “that cost would have risen to £4bn when measured at April 1997” | August 6, 1997 |
| Bank's estimate, 1993: value of reserves had there been no intervention | about £3 1/4 billion higher by April 1993 | “The actual loss could have been as large as £5 billion.” The note says it was reduced by two offsets | December 10, 1993 |
| Return to intervention in the second half of 1992 | -£246 million in the third quarter, -£1,013 million in the fourth | table of returns to intervention, in £ million | August 2, 1993 |
Three things the table makes plain.
£3.3 billion is not a cheque. The Treasury paper separates the “profit and loss relative to sterling”, which it calls “the true cost to the taxpayer”, from the “opportunity cost”, which compares what happened with a counterfactual: keeping $24 billion of reserves in the first version, running the reserves down after leaving the ERM in the one it prefers. The cash loss in the two months was nearly £800 million; the £3.3 billion is the comparison measured at a chosen date.
The cost moves with the date. In the paper's tables, the opportunity cost against no intervention was £2.4 billion at October 1992, £3.2 billion at March 1993, £3.5 billion at February 1994 and £1.9 billion at April 1997, as the pound recovered. The paper says that last figure goes beyond the cost of the 1992 intervention and rejects it as the right measure. Whoever quotes one number is choosing a date.
The pound's fall hurt because of which side the Treasury was on. The August 1993 note says that a sterling depreciation makes a negative contribution to the return when net reserves are negative. After September 16, 1992 the net reserves were minus $16 billion, so the pound's fall cost the Treasury money.
What the documents show about Soros, and what they don't
George Soros's fund, the Quantum Group, is the name attached to the day. The record here is thinner than the legend.
- No official paper names a winner. The Bank's bulletin and the Treasury papers we read deal with the authorities' own position. They do not say who took the other side or how much anyone made.
- The $1 billion is a press figure. In the record of the House Banking Committee's April 13, 1994 hearing on hedge funds, a U.S. News & World Report article of April 18, 1994 is reproduced. It says: “Quantum made $1 billion in a few days during 1992's European currency crisis, when the British pound was devalued.” That is that magazine's report, not a fund document.
- Soros's own words, from September 29, 1993. In a speech to the Aspen Institute Berlin, reproduced in the same hearing record, he said: “Sterling would have left the ERM whether or not I speculated against it.” That is his claim, made as a participant; the Bank's bulletin points the same way when it says the cost of defending the parity was prohibitive, but it does not mention him.
- What he said about risk, to Congress. His statement says his firm had operated for more than 25 years without a single occasion when it could not meet a margin call, and that “We do use borrowed money”.
What this means for your money
You will probably never hold a currency peg, but you may hold a promise that looks like one: a fund that redeems at a published NAV, a tender offer at a stated price, a redemption plan. The Black Wednesday papers are a clean case study in how those promises work and fail.
- A defended price is only as good as the reserves behind it. The Bank bought pounds all day, with the Bundesbank and the Banque de France helping, and moved net reserves by about $40 billion in two months (our arithmetic). A fund's reserves are its cash, its credit lines and its sale of assets. Where we can, we read them in the filing: see REIT redemptions 2026 for which NAV REITs paid every request and which suspended, and the ranking of non-traded NAV REITs for how they compare.
- The defender raises the price of waiting, and the price of staying. The Bank's rate rises made borrowing pounds, which is how a bet against them is financed, more expensive. In our arithmetic, borrowing £1 billion for one day costs about £329,000 at 12% a year and about £411,000 at 15%. A one percent fall in the pound on that £1 billion is £10 million, about 24 days of 15% interest. A defence has to hold for longer than the other side can afford to wait; it did not, and the Bank said so itself. (This is a hypothetical position for illustration, not any real fund's.)
- When a defended price breaks, it can break fast and in one direction. Sterling closed two days later 6% below its floor and the pound lost about 10% against the deutschmark by the end of the month. Investors in a fund with a suspended redemption plan, or a tender offer that pays only part of the requests, see the same shape: a stated price, limited buyers, then a gap. Our page on what happened to GPB Capital walks through a case where redemptions were suspended and investors have since been paid only in part, and our list of real estate crowdfunding failures collects others.
- Who gets paid first is a matter of position, not of fairness. In the Treasury's accounts the taxpayer was on the losing side of the trade by construction, with negative net reserves. In many funds repurchases are paid in order of request or pro rata, and some investors rank above others in the structure; the filing says which. You can read for the repurchase limit, the sources of cash and the date the fund last paid every request.
- Read the document, not the headline. The headline number for Black Wednesday has been £3.3 billion, £13 billion, £20 billion and £27 billion, depending on who counted and when. The document says which it is and why.
This is analysis of public documents, not investment, legal or tax advice.
FAQ
Sources, read on November 16, 2026: Bank of England, Quarterly Bulletin, November 1992, “Operation of monetary policy” (bankofengland.co.uk, quarterly-bulletin 1992 Q4); HM Treasury papers released under the Freedom of Information Act on February 9, 2005, via the Internet Archive's copy of the Treasury's FOI disclosures page (snapshot of February 14, 2006): “The cost of Black Wednesday reconsidered” (August 6, 1997), “The cost of Intervention” (December 10, 1993) and “Calculating the cost of Black Wednesday” (August 2, 1993), read as scanned images and transcribed by us for the passages quoted; the Treasury's covering note to the Financial Times request (January 4, 2005), as published by The Guardian; The Guardian, “Treasury papers reveal cost of Black Wednesday” (February 9, 2005); and the hearing record Risks that Hedge Funds Pose to the Banking System, House Committee on Banking, Finance and Urban Affairs, 103rd Congress, April 13, 1994 (Serial No. 103-131; Internet Archive identifier risksthathedgefu00unit), including George Soros's statement, his Aspen Institute Berlin speech of September 29, 1993 and the U.S. News & World Report article of April 18, 1994 reproduced in it. The Bank's bulletin was read as machine-extracted text of a scanned print, with spacing errors in the extraction repaired by us; page numbers are the bulletin's printed pages. We did not read the 1992 bulletin's separate General assessment article or the Bank's or Treasury's monthly reserves series for 1992, and for the August 1993 and December 1993 papers we read only the passages cited, not the other six papers of the release. The one-day interest figures, the share of intervention done on the day, the swing in reserves and the fall in the pound are our arithmetic. This is analysis of public documents, not investment, legal or tax advice.