Where Money Came From · Lesson 35 · 13 min read

When a Peg Breaks

Two Afternoons That Are Still Being Studied
Kanin D. Alldredge
Legal & Quantitative Analyst / Author / Educator, The Currency Classroom
Free

1Five Ways to Break

Pegs do not fail randomly, and they rarely fail without warning. They fail through identifiable mechanisms. The indicators for each are usually visible in public data beforehand.

There are five. Most real collapses involve more than one.

MechanismWhat Happens
Reserve depletionThe money to defend the rate runs out
Fundamental misalignmentThe declared rate drifts far from economic reality, usually through inflation
Speculative attackParticipants bet against the rate at a scale the defense cannot absorb
Voluntary abandonmentThe institution chooses to stop, before being forced to
Credibility withdrawalBelief in the commitment disappears, and with it the commitment
Figure 35.1: The five ways a fixed rate breaks. Most real collapses involve more than one.
Figure 35.1The five ways a fixed rate breaks. Most real collapses involve more than one.

The first three are what most people picture. The last two are more interesting. Between them they explain the event that mattered most to ordinary traders.

2Reserve Depletion Feeds Itself

The most straightforward mechanism, and the one with the nastiest shape.

A central bank defending an overvalued currency spends reserves buying its own currency. Reserves are finite. Every day of defense costs some of them.

The problem is what happens as the market watches the balance fall.

reserves fall → confidence falls → selling rises →
reserves fall faster
Figure 35.2: Falling reserves feed on themselves, and the loop accelerates.
Figure 35.2Falling reserves feed on themselves, and the loop accelerates.

It is a loop, and it accelerates. So the level of reserves tells you less than the direction and the speed.

Figure 35.3: The direction and speed of reserves tell you more than their level.
Figure 35.3The direction and speed of reserves tell you more than their level.

There is an asymmetry here that matters for Section 5. Holding down a currency that wants to rise is a different problem. The central bank sells its own currency, which it can create without limit. It does not run out. It accumulates, and accumulation has no natural ceiling.

Figure 35.4: Propping up a falling currency runs out of reserves. Holding down a rising one does not.
Figure 35.4Propping up a falling currency runs out of reserves. Holding down a rising one does not.

3The Asymmetry That Invites an Attack

Speculative attacks look like aggression. They are better understood as arithmetic.

Consider what a participant risks by betting against a peg they believe is unsustainable.

If the Peg HoldsIf the Peg Breaks
You pay the cost of carrying the positionYou gain the entire size of the devaluation
A known, limited, ongoing expensePotentially very large, and it arrives at once
Figure 35.5: Betting against a misaligned peg risks a small cost for a potentially large gain.
Figure 35.5Betting against a misaligned peg risks a small cost for a potentially large gain.

Limited downside, large upside. When the conditions are right, that bet is not reckless. It is the sensible response to the numbers.

Which is why a visibly misaligned peg attracts pressure. The misalignment does not just weaken the rate. It gives others a profitable reason to push.

4Black Wednesday

On September 16, 1992, the Bank of England attempted to defend sterling's place in Europe's exchange rate mechanism against exactly that arithmetic.

Several participants had concluded the pound was held above what the British economy justified. They also judged that Britain's weak conditions made sustained high interest rates politically impossible. One reported position was around ten billion dollars short of sterling.

The Bank used two of the three tools from Lesson 34, at extraordinary scale, in a single day.

The DefenseWhat Happened
Reserve interventionBillions of pounds of reserves spent, at a later estimated cost of about £3.3 billion
Interest ratesRaised from 10 percent to 12 percent, then 15 percent announced but never applied, in one day
Figure 35.6: Britain's interest rate defense on September 16, 1992.
Figure 35.6Britain's interest rate defense on September 16, 1992.

Neither worked. And the second one failed in a revealing way.

Rates that high would obviously damage an already weak economy, and nobody believed they could last. The defense was so extreme that it advertised desperation. It undermined confidence rather than restoring it.

Figure 35.7: A defense that extreme advertised desperation and undermined confidence.
Figure 35.7A defense that extreme advertised desperation and undermined confidence.

The 15 percent rate never took effect. By that evening Britain had suspended its membership and sterling fell, and the next day rates were back at 10 percent. The reported profit on the largest position against it was about a billion dollars.

Figure 35.8: Black Wednesday in four numbers.
Figure 35.8Black Wednesday in four numbers.

5The Fourth Mechanism

Now the event that matters most to anyone trading with leverage, and it is almost always described wrongly.

For more than three years the Swiss National Bank held a floor under EUR/CHF at 1.2000. It would not let the euro fall below that against the franc, and it defended the level successfully throughout.

On the morning of January 15, 2015 the pair was trading around 1.2010. There was no attack underway. There was no failed defense. There was no crisis.

Figure 35.9: EUR/CHF held above the 1.2000 floor for more than three years, then fell roughly 30% within minutes.
Figure 35.9EUR/CHF held above the 1.2000 floor for more than three years, then fell roughly 30% within minutes.

And the SNB was not running out of anything.

What People AssumeWhat Was Actually True
Reserves were exhaustedReserves were accumulating, toward roughly 80 percent of Swiss GDP
A speculative attack overwhelmed the defenseNo coordinated attack was underway
The SNB was forced outThe SNB chose its moment

Recall the asymmetry from Section 2. The SNB was holding down a currency that wanted to rise. That meant creating francs to buy euros, which it could do without limit. Capacity was never the constraint.

The problem was the European Central Bank, which was preparing a large asset purchase program. That would have forced the SNB to expand its balance sheet faster and faster, with no endpoint, holding euro assets whose value the program itself would push down.

Figure 35.10: The SNB stopped because continuing would have cost more than stopping.
Figure 35.10The SNB stopped because continuing would have cost more than stopping.

So the SNB judged that continuing cost more than stopping. It stopped on its own terms, rather than waiting to be overwhelmed.

The floor was not broken. It was withdrawn.

Figure 35.11: The floor was withdrawn by choice, not broken under pressure.
Figure 35.11The floor was withdrawn by choice, not broken under pressure.

6Credibility Cuts Both Ways

The fifth mechanism explains why the announcement did what it did.

A credible institution needs less intervention, because belief does the work. If the market is sure a floor will hold, it stops testing it, and the defense becomes cheap. More than three years of success had made the SNB's commitment almost unquestioned.

Which means the floor was not being held up by reserves that morning. It was being held up by belief.

And belief does not erode. It switches off.

Figure 35.12: Belief in a commitment does not fade gradually. It switches off.
Figure 35.12Belief in a commitment does not fade gradually. It switches off.

The moment the commitment was withdrawn, there was nothing left to be credible about. The pair fell roughly thirty percent within minutes. It had no reason to be near 1.2000 and never had, beyond the commitment itself.

Figure 35.13: Credibility can calm a crisis when it is created and cause one when it is removed.
Figure 35.13Credibility can calm a crisis when it is created and cause one when it is removed.

7What It Did to Traders

This is why January 15, 2015 is studied rather than simply remembered.

For more than three years EUR/CHF near the floor was one of the calmest things on any screen. Tiny daily ranges, almost no movement, week after week.

A trader sizing a position by that daily range was making a specific and understandable error.

That calm was not the pair's volatility. It was the floor, seen from above.

Remove the constraint and the genuine volatility appears instantly, because it had been there all along, held down.

The practical conclusion is one line, and it applies to any pair near an institutional commitment. Here it is.

Size to the gap, not to the range. The relevant question is not how far the pair moves on an ordinary day. It is how far it would move if the thing holding it stopped.

Figure 35.14: Near an institutional commitment, size positions to the gap, not the daily range.
Figure 35.14Near an institutional commitment, size positions to the gap, not the daily range.

8The Same Pattern, in Crypto

Everything in this lesson transfers directly to something from Lesson 11, and the mapping is exact.

Type of StablecoinIts Currency EquivalentWhy
Fiat-backedA currency boardReserve assets mechanically back the peg
AlgorithmicA credibility-based commitmentNo reserve backstop. The peg holds because people believe it will.
Figure 35.15: Each type of stablecoin maps onto a currency regime from this series.
Figure 35.15Each type of stablecoin maps onto a currency regime from this series.

Lesson 11 made the point that a stablecoin's dollar is a company's promise rather than a country's. This lesson gives you the framework for what that promise rests on.

A fiat-backed stablecoin sits in the same category as Hong Kong's arrangement. Reserves back every unit, and the question worth asking is what the reserves actually are and who verifies them.

An algorithmic one sits in the same category as the SNB floor. The peg holds because participants believe the mechanism will restore it. When TerraUSD collapsed in May 2022, its dollar was an institutional price rather than a market one. It failed exactly as this lesson describes. Belief went, and there was no reserve underneath.

What This Series Was For

Seven lessons ago the question was what money actually is. The answer was a structure of trust with four pillars, and one of them moving faster than the others.

Then where money comes from, and that most of it is lent into existence rather than printed. Then gold, where an exchange rate was arithmetic until it met something governments cared about more. Then Bretton Woods, designed by people who had lived through the alternative, carrying a contradiction from day one. Then the afternoon it ended, with no announcement when the replacement arrived.

And then the regimes operating now, and how they end.

Every monetary commitment is held by an institution that could choose to stop.

That is the thread. Gold in 1914. Sterling in 1931, and again in 1992. The dollar in 1971. The franc in 2015. Different centuries, different instruments, and the same four steps from Lesson 31 arriving at the same place.

Figure 35.16: Every one of these commitments was held by an institution that could choose to stop.
Figure 35.16Every one of these commitments was held by an institution that could choose to stop.

The next series asks the question this one keeps circling. The dollar sits at the center of the system by historical accident rather than design. What is that position actually worth, and what would it take to lose it?

Three Things to Remember

Try This

Fifteen minutes. One chart tells most of this story by itself.

Words From This Lesson

TermWhat It Means
Reserve depletionRunning down the foreign reserves used to defend a rate. Self-reinforcing, because falling reserves invite more selling.
Fundamental misalignmentA declared rate that has drifted far from what the economy justifies.
Speculative attackLarge-scale betting against a rate, made rational by limited carry cost against a potentially large devaluation gain.
Voluntary abandonmentAn institution ending a commitment on its own terms, before being forced to.
Credibility withdrawalThe loss of belief in a commitment. Instant and complete, and it does not require a failed defense.
Gap riskThe distance a price can travel with no trading in between, leaving exit instructions unfilled at their stated level.
Institutional priceA price held where it is by a commitment rather than by supply and demand.

Test Your Knowledge

Eight questions. Answer them all before checking the key on the following pages. If a question gives you trouble, the answer key tells you which section to revisit.

1. Why is reserve depletion described as self-reinforcing?
  1. A. Reserves are automatically replenished
  2. B. Falling reserves reduce confidence, which increases selling, which depletes reserves faster
  3. C. Central banks must spend reserves at a fixed daily rate
  4. D. It only applies to currency boards
2. What makes a speculative attack rational rather than reckless?
  1. A. Limited carry cost if the peg holds, large gain if it breaks
  2. B. Attacks always succeed
  3. C. Central banks rarely defend their pegs
  4. D. Regulators permit it
3. What happened when the Bank of England raised rates to 12 percent and announced 15 percent in one day in 1992?
  1. A. Confidence was restored and the peg held
  2. B. It succeeded but the cost was too high
  3. C. Interest rates were not used in the defense
  4. D. The move was so extreme it undermined confidence rather than restoring it
4. What was the state of Swiss National Bank reserves before January 15, 2015?
  1. A. Accumulating, toward roughly 80 percent of Swiss GDP
  2. B. Nearly exhausted
  3. C. Roughly unchanged for three years
  4. D. Not publicly disclosed
5. Why did the SNB abandon the floor?
  1. A. It was overwhelmed by a speculative attack
  2. B. The Swiss government ordered it
  3. C. Continuing would have cost more than stopping, so it chose its own moment
  4. D. Its reserves ran out
6. Why did EUR/CHF move so violently when the floor went?
  1. A. A speculative attack was already underway
  2. B. Trading was suspended and reopened lower
  3. C. The floor was held up by belief, and belief switches off rather than eroding
  4. D. The ECB intervened simultaneously
7. How should a position near a central bank floor be sized?
  1. A. To the pair's recent average daily range
  2. B. To the gap risk, meaning the distance to a fully broken level
  3. C. To the stop loss distance, since that defines the loss
  4. D. Position size is irrelevant near a floor
8. Which currency arrangement is an algorithmic stablecoin most like?
  1. A. A currency board
  2. B. A crawling peg
  3. C. A free float
  4. D. A credibility-based commitment with no reserve backstop

Answer Key

Each answer explains why the correct choice is correct, why the others are not, and where to reread.

Question 1 · correct answer B

It is a loop. Visibly falling reserves reduce confidence in the defense, which brings more selling, which drains reserves faster. The direction of travel matters more than the level.

A describes something that does not happen, since reserves must be earned or borrowed. C invents a fixed spending rule. D reverses it, as currency boards are the arrangement least exposed to this.

Review: Section 2, Reserve Depletion Feeds Itself

Question 2 · correct answer A

The loss if the peg holds is the ongoing cost of carrying the position. The gain if it breaks is the full devaluation. Limited downside against large upside makes the bet sensible when conditions support it.

B is wrong because many attacks fail and the attacker pays. C is wrong because central banks defend vigorously, as 1992 showed. D confuses legality with the arithmetic that motivates the trade.

Review: Section 3, The Asymmetry That Invites an Attack

Question 3 · correct answer D

Rates that high were plainly unsustainable for a weak economy. The severity of the defense advertised desperation and reduced confidence rather than rebuilding it.

A is wrong because sterling left the mechanism that evening. C is wrong because rates were the second tool deployed. B is wrong because the defense did not succeed at any cost.

Review: Section 4, Black Wednesday

Question 4 · correct answer A

Reserves were rising, not falling. The SNB was holding down a currency that wanted to strengthen, which meant creating francs to buy euros, and that has no mechanical limit.

B is the common assumption and the opposite of the facts. C is wrong because accumulation was rapid. D is wrong because reserve data is published.

Review: Section 5, The Fourth Mechanism

Question 5 · correct answer C

An approaching European asset purchase program would have required accelerating balance sheet expansion with no natural endpoint. The SNB judged that continuing cost more than exiting and chose its timing.

A is wrong because no attack was underway that morning. B is wrong because the decision was the central bank's. D contradicts the reserve position, which was growing.

Review: Section 5, The Fourth Mechanism

Question 6 · correct answer C

More than three years of successful defense meant the floor was sustained by belief rather than by active intervention. Once the commitment was withdrawn, there was nothing left to believe in, and the pair found its own level at once.

A is factually wrong, since the pair was quiet that morning. B describes a trading halt rather than what occurred. D introduces an intervention that did not happen.

Review: Section 6, Credibility Cuts Both Ways

Question 7 · correct answer B

Compressed volatility near a commitment reflects the constraint rather than the pair. The relevant distance is how far the price would travel if the commitment ended.

A sizes to the managed surface, which is the error that emptied accounts in 2015. C treats a stop loss as a guarantee, which gap risk makes false. D abandons the decision rather than making it correctly.

Review: Section 7, What It Did to Traders

Question 8 · correct answer D

An algorithmic stablecoin has no reserve backstop. Its peg holds because participants believe the mechanism will restore it, which is the same structure as a credibility-based currency commitment.

A describes a fiat-backed stablecoin, where reserves mechanically back each unit. C describes a price set by the market, which a peg is not. B describes a rate adjusted gradually on a schedule.

Review: Section 8, The Same Pattern, in Crypto

Keep going

Where Money Came From is written by Kanin D. Alldredge for The Currency Classroom, and it is free to read. There is more free from CNS: start here.

Start free with CNS →

Or see CNS Pro →