Currency peg collapse
A Socratic walk-through of currency peg collapse — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #Why can a currency peg hold steady for years and then break in a single week?
A pegged exchange rate is the flattest line in finance. It does not wobble, it does not drift, and then one Wednesday it is gone. Britain left the European exchange rate mechanism on 16 September 1992; Thailand floated the baht on 2 July 1997; Argentina's one-peso-one-dollar convertibility ran from 1991 and ended in January 2002. In each case the flat line gave no warning, and the end took days. Something in the setup must produce that shape — a slow illness does not usually kill in a week.
Reasoning it through
REASONING #Start with what a peg is. Not a law about the exchange rate but a standing offer: the central bank promises to buy its own currency at a fixed price from anyone selling, paying in foreign reserves. The rate stays flat because the offer is honoured, not because the market agrees with the price.
So the constraint is inventory. Every defence spends reserves, and reserves are finite. Now the step that does the real work. Suppose reserves are draining slowly and predictably, because the country is printing money faster than its peg allows. Naively the peg should end on the day the last reserve is sold, with the drain visible all the way down. That is not what happens, and the reason is that traders can do the arithmetic too.
Ask what the currency would be worth if it floated today. Call that the shadow rate. Early on it is stronger than the peg, so selling to the central bank at the pegged price is a bad trade — you hand over something worth more than you get. As reserves fall, the shadow rate weakens, and there is a precise moment when it arrives exactly at the peg. One instant later, buying the remaining reserves at the pegged price is profitable, and any trader who waits loses to those who did not.
So the whole remaining stock goes in a single transaction. Krugman set this out in 1979, and its consequence is the answer to the question: the attack happens strictly before reserves would have run out, at a date fixed by an arbitrage condition rather than by exhaustion, and it is discontinuous by construction. Nothing gradual precedes it because everything gradual was already priced.
That accounts for pegs undone by visible bad policy. Does it account for the ones that were not? Test it against 1992: Britain was not printing recklessly, and its reserves were not visibly draining for years. So the first account is incomplete, and the second — associated with Obstfeld — changes what is being defended. A government defends a peg by raising interest rates, and high rates cost it something real: unemployment, mortgage pain, bad banks. It abandons the peg when that cost exceeds the value of keeping it. Now the loop: if traders expect abandonment, they demand higher rates to hold the currency, which raises the cost of defending, which makes abandonment more likely. The expectation is an input to the thing expected.
That gives two internally consistent outcomes for identical fundamentals: everyone believes the peg holds, defence stays cheap, and it holds; everyone believes it breaks, defence becomes ruinous, and it breaks. The switch needs no news at all — which is why the flat line carries no information about how close the end is.
Underneath both sits a structural fact rather than a story: with open capital markets, a country cannot simultaneously fix its exchange rate and run monetary policy for its own economy. A peg is the choice to give up the second, and it stays comfortable only while the policy it imports happens to suit. Britain's problem in 1992 was importing the monetary stance of a Germany managing reunification, into a recession.
Which account fits which episode is argued rather than settled, and the empirical difficulty is severe: the second story predicts collapse without deteriorating fundamentals, but a peg about to break also causes fundamentals to look bad, so the evidence is contaminated at exactly the moment you want to read it.
The analogy
THE ANALOGY #Think of a shopkeeper who promises to change any amount of tokens for silver at a fixed rate, keeping a chest of silver behind the counter. Nobody queues while the rate looks fair. The instant people work out that the chest will not cover the tokens outstanding, the whole queue forms at once — not because the chest is empty, but because being last in it is the only way to lose.
the shopkeeper cannot change the price of silver, whereas a central bank can make holding tokens temporarily lucrative by raising rates — so a real defence is a bidding contest, not a fixed stock, and the second kind of collapse is the bank deciding the bidding is not worth winning.
Clarifying the model
THE MODEL #Two refinements connect the halves. The reserve stock is not really the constraint; the cost of defence is, and reserves are only its most legible form. And the "single week" is not the crisis — it is the settlement of a question long open but never expressed in the price, because a defended peg by definition cannot express it.
The common misconception is that speculators cause the collapse. What they do is choose the date. In the first account the terminal date is set by policy long before the attack, which merely arrives at it early; in the second, the vulnerability is created by the government's own willingness to abandon, which speculators test rather than manufacture. A peg with no cost of defence cannot be attacked profitably.
Also worth naming as wrong: that a peg is safe if reserves exceed some ratio of imports or of money supply. Such rules are useful and none of them binds an attack, because what a defender needs is not reserves greater than a stock but greater than what everyone might sell — and that quantity is set by belief.
The queue resembles the collection's explanation of bank runs, and the second account is genuinely the same multiple-equilibrium structure. The difference worth holding onto is the first account, which has no counterpart there: a peg can be doomed on fundamentals, with a computable end date, and still show a perfectly flat line until the day traders reach it.
A picture of it
THE PICTURE #How to readStart at Credible, the flat line, and note that the observable exchange rate is identical in the first three states — that is the point. Movement rightward is driven by the gap between the peg's implied policy and the one the country needs. The edge into Attacked is not a slide but a threshold crossing: the moment the currency's free-market value reaches the pegged price, waiting stops paying. Two edges leave Attacked and only one is reversible, which is why a return arrow to Credible exists and none leaves Floated.
What became clearer
WHAT CLEARED #The flatness and the suddenness are the same fact. A peg's price is an offer being honoured, so the exchange rate cannot report the growing gap between the peg and what the currency is worth — pressure accumulates somewhere invisible, in reserves or in the political tolerance for high interest rates. When it finally registers, it registers as a jump, because the trade that ends a peg is one nobody can afford to make second.
Where to go next
ONWARD #- Why currency boards and full dollarisation are harder to attack, and what they cost in exchange.
- How capital controls change the arithmetic, and why they buy time rather than safety.
Key terms
TERMS #| Term | What it means |
|---|---|
| Shadow exchange rate | the rate the currency would trade at if the peg were dropped at this instant. |
| Speculative attack | the concentrated sale of a pegged currency to the central bank at the pegged price, once that price favours the seller. |
| Impossible trinity | fixed exchange rates, free capital movement, and independent monetary policy cannot all be had at once. |
Every term the collection defines is gathered in the glossary.