The Credit Suisse Bondholders Were Right. The Money Stayed Gone.

On the weekend of 18 and 19 March 2023, a Swiss regulator sent Credit Suisse a letter, and by Monday's opening bell CHF 16.5 billion of its bonds were worth nothing.² Two and a half years later, a Swiss court ruled that the letter should never have been sent.¹⁰ The bonds are still worth nothing. Nobody has decided whether that changes.

Most of what got written about Credit Suisse's collapse in 2023 focused on the two facts that look unfair on their face: shareholders never got to vote on the deal, and bondholders who are supposed to rank above equity in a wind-down got zero while equity holders got paid. Both are true, and both are the wrong things to find shocking. The interesting part is what happened when the people who lost money went looking for the mechanism that was supposed to protect them, and discovered it worked exactly as written, except when the people operating it turned out not to have read it correctly either, at which point being right stopped being the same thing as being paid.

Start with the number everyone quotes as the scandal: Credit Suisse's Additional Tier 1 bonds, contingent convertible instruments known in the market as CoCos, went to zero while common shareholders received UBS stock worth roughly CHF 3 billion, about $3.25 billion at the time.¹⁹ Financial journalism spent a year calling this a reversal of the absolute priority rule, the doctrine that equity absorbs losses before subordinated debt does. It is a satisfying line. It is also not quite what happened.

AT1 bonds were never ordinary subordinated debt sitting one rung below equity in a liquidation queue. They carry a separate, freestanding contractual switch called a Viability Event clause, and Credit Suisse's version had two triggers: either the regulator determines a write-down is essential to prevent insolvency, or the bank receives an irrevocable commitment of extraordinary support from the public sector, full stop, whether or not insolvency is actually close.¹⁹ When either trigger fires, the write-down happens irrespective of the normal subordination ranking. It doesn't wait its turn in the waterfall. It bypasses the waterfall entirely.

So "AT1 got wiped while equity got paid" describes what happened, but "priority got reversed" describes a mechanism that was never in play. Nobody's place in line moved, because there was no shared line. Equity had its own instrument with its own terms, and AT1 had a kill switch equity was never wired into. The distinction matters for reasons beyond pedantry: a reversed priority rule is a scandal about fairness. A separate contractual switch that a regulator can flip is a design feature, one that institutional investors were paid extra yield to accept, and the actual question worth asking is whether it was flipped correctly.¹⁸

It wasn't, as it turned out. But get there in order.

AT1 paper was marketed and priced around a word the industry likes: loss-absorbing, a phrase that makes a bond sound like it does yoga. Buyers priced in the yoga. What they got, eventually, was the actual absorption, all at once, on a weekend, by decree.

The Federal Council's emergency ordinance of 19 March 2023, grounded in Articles 184 and 185 of the Swiss Constitution, the provisions covering foreign relations and internal and external security, waived the shareholder approval that Swiss merger law would ordinarily require from both companies.³⁴ The ordinance's own justification was that ordinary process was too slow for a systemically important bank in free fall, and that the vote could be dispensed with when a merger was deemed necessary to protect the national economy and financial system.⁴ Neither UBS's shareholders nor Credit Suisse's got a ballot on the transaction that determined what their shares would become.

What Credit Suisse's shareholders did get, sixteen days later, was an annual general meeting at a Zurich hockey arena, the kind of venue normally rented out for concerts, because no boardroom in the country could seat everyone who showed up angry.⁹ The bank pulled two items off its own agenda ahead of the meeting: a special bonus vote tied to the now-obsolete turnaround plan, and a dividend of roughly five cents a share. Five board members had already resigned before the vote. Chairman Axel Lehmann told the remaining shareholders, "I apologize that we were no longer able to stem the loss of trust that had accumulated over the years, and for disappointing you," and was re-elected with 55.67% of the vote, meaning close to 44% of shareholders looked at the man who had just lost them the company and voted no anyway.⁹

Notice what the ballot did and didn't contain. Shareholders were asked to weigh in on board composition and pay after the transaction that actually mattered had already been decided somewhere else, on a Sunday, by people none of them had elected for that purpose. The theatre of shareholder democracy ran on schedule for every item that no longer had stakes attached, and was suspended for the one item that did. Forty-four percent of the room still managed to register the correct verdict on the chairman, which suggests the electorate understood its situation better than the ballot it was handed.

The "extraordinary support" that satisfied the AT1 trigger had two components. The Swiss National Bank extended emergency liquidity assistance to Credit Suisse backed by a federal default guarantee, and the government separately gave UBS a Loss Protection Agreement covering up to CHF 9 billion of losses on a portfolio of Credit Suisse's non-core assets, with UBS absorbing the first CHF 5 billion itself before the state guarantee would ever be drawn.⁶ Parliament's finance delegation approved emergency credit commitments totalling roughly CHF 109 billion the same weekend the AT1 bonds were written off.⁵

On 11 August 2023, less than five months later, UBS terminated the Loss Protection Agreement without ever drawing on it, having paid the state a flat CHF 40 million fee for coverage it turned out not to need, and Credit Suisse fully repaid its emergency liquidity loan the same day.⁷⁸ The crisis that justified zeroing CHF 16.5 billion of someone else's money had, by its own resolution, wound itself down inside two quarters.

Read charitably, an emergency that resolves itself in five months and costs the state one CHF 40 million insurance premium it never has to pay out was, in hindsight, closer to a bridge loan than a national trauma. Bern got its premium back the moment the danger passed. The AT1 holders whose money financed the appearance of that danger did not get a comparable refund, because their instrument doesn't have an undo button once the switch has been thrown. One party to this arrangement bought insurance it didn't end up needing. The other party was the insurance, and there's no clause that gives it back its premium just because the building didn't burn down.

Roughly 3,000 bondholders filed around 360 separate appeals against FINMA's decree with Switzerland's Federal Administrative Court. On 1 October 2025, in the pilot case, the court ruled.¹⁰ Its finding was unambiguous on both of FINMA's stated grounds. The contractual viability event had not actually been triggered, because Credit Suisse remained adequately capitalized and the support it received addressed a liquidity problem, not a capital one, which is a different thing under the bond's own definitions.¹⁰ And the provision of the emergency ordinance that FINMA leaned on to write down the bonds anyway was, in the court's words, unconstitutional in several respects, including how it delegated what amounted to expropriation power without the safeguards the constitution requires for that.¹⁰

That is about as complete a vindication as an administrative court hands out. It is also, notably, not a check. The press release announcing the ruling states that the court has not yet decided the question of reversal, meaning whether the write-down itself gets undone or compensated, and that the other 359-odd suspended cases wait on that question too.¹⁰ FINMA announced within two weeks that it would appeal to the Federal Supreme Court. UBS said it would appeal as well.¹³¹²

Sit with that sequence for a second. A regulator is currently asking Switzerland's highest court to overturn a ruling that the regulator acted unlawfully, in a matter where the entity it wronged, the free-standing Credit Suisse, no longer exists to enforce the win. That is a genuinely confident use of a losing position, the institutional equivalent of getting a parking ticket dismissed for lack of jurisdiction and appealing anyway because you'd rather the ticket stood.

Six days after the AT1 ruling, a different court handed down a different verdict to a different class of Credit Suisse stakeholder. A shareholder had sued the state for roughly CHF 140,000, arguing the emergency ordinance had left him extensively undercompensated on his holdings, and on 7 October 2025 the Federal Court dismissed the claim in full.¹⁵ Two Credit Suisse constituencies, one week apart, took the same emergency ordinance to two different Swiss courts on two different theories. Bondholders got told they were right and given nothing. The shareholder got told he was wrong and given nothing. Both outcomes end at the same number.

There is one route around this that has nothing to do with Swiss law, and it belongs only to some of the bondholders. At least nine notices of investor-state dispute arise from bilateral investment treaties Switzerland has signed with other countries, and one group, 184 Japanese investors, filed a formal arbitration request with the International Centre for Settlement of Investment Disputes arguing that the write-down breached Switzerland's free trade agreement with Japan.¹⁶¹⁷ These claims don't run through the Federal Administrative Court, don't depend on Article 5a of the emergency ordinance being found unconstitutional, and aren't waiting on the Federal Supreme Court's calendar. They run through an entirely separate forum, in front of arbitrators who never have to defer to a Swiss regulator's read of Swiss constitutional law, because the claim isn't about Swiss constitutional law. It's about a treaty.

Two bondholders can hold identical CHF-denominated notes, purchased through the same underwriter on the same settlement date, and end up in front of completely different tribunals with completely different odds, for a reason that has nothing to do with the bond documentation either of them signed. It has to do with which government their holding vehicle is domiciled in, and whether that government happens to have a treaty with Switzerland that covers this kind of expropriation. A Swiss retail holder gets the Federal Administrative Court, an appeal to the Federal Supreme Court, and, eventually, maybe, a ruling on restitution. A Japanese institutional holder gets a seat at ICSID in Washington, an ocean away from the regulator that wronged them, arguing a treaty claim that doesn't care what a Swiss court decides about Swiss constitutional delegation doctrine. The seniority printed on the bond turned out to matter less than the passport attached to the buyer.

Almost nobody reading this will ever hold an AT1 bond or get caught in a sovereign's Sunday-night rescue ordinance. But the mechanism generalizes further than the instrument does. Any contract that gives one counterparty a unilateral trigger, self-certified, irreversible once pulled, is running the same experiment Credit Suisse's bondholders just lived through, whether it's a lender's material-adverse-change acceleration clause, a majority investor's ability to force a recapitalization at a distressed valuation, or a board's emergency bylaw provision that lets a written consent substitute for a shareholder meeting when things move fast. The fairness of the clause is not the useful question. The useful question is what you get back if the trigger turns out, on review, to have been pulled wrongly.

Before you sign anything with a switch like that in it, or before you rely on one already sitting in your documents, get specific answers to five things:

• Who certifies that the trigger condition is met? If the party pulling the trigger also decides whether the conditions for pulling it exist, you are relying entirely on their judgment being reviewed later, not checked first.
• Does the action complete before anyone can review it? A merger that closes, shares that get issued, or an instrument that gets written down cannot generally be un-closed, un-issued, or un-written just because a court later disagrees with the decision. Ask what happens to the transaction itself while a challenge is pending, since the paperwork alone tells you very little.
• What remedy does a successful challenge actually produce? A finding that the other side acted unlawfully is not the same as an order to pay you back. Read for the difference between a declaration and a restitution mechanism, and assume you'll only get the first one unless the second is spelled out.
• Does your own jurisdiction give you a forum the other side doesn't control? A treaty right, an arbitration clause naming a neutral seat, or incorporation somewhere with different remedies than your counterparty's home courts can matter more than anything printed in the substantive terms.
• How long does that forum realistically take, measured against how fast the harm becomes permanent? A write-down happens in a weekend. A ruling on whether the write-down was lawful took two and a half years and still hasn't answered the money question. Size your risk to the slower number, not the faster one.

None of this is a reason to refuse every clause that hands a counterparty a unilateral trigger; sometimes speed genuinely is the point, and a lender or a regulator or an acquirer will not sign without one. It is a reason to stop treating "is this fair" as the operative question and start asking "if this is wrong, what do I get back, and on what clock." Switzerland has now spent CHF 16.5 billion answering that question for its own bondholders: a published finding of unlawfulness, an appeal, and, so far, zero francs.

In cash terms, yes. Shareholders received roughly CHF 3 billion in UBS stock while AT1 holders received nothing.¹⁹ But equity holders never had a vote on the transaction that set that number, and a shareholder who sued the state arguing the compensation was too low lost outright on 7 October 2025.¹⁵ Getting paid something is not the same as having had a right that was honored; it just means the number the state and UBS agreed on for you happened to be greater than zero.

Not yet, and not automatically even if it is eventually confirmed unlawful. The Federal Administrative Court's October 2025 ruling annulled FINMA's decree but has not yet decided whether bondholders are entitled to reversal or compensation, and both FINMA and UBS are appealing the underlying finding to the Federal Supreme Court.¹⁰¹³¹² Separately, some bondholders are pursuing investor-state arbitration claims that don't depend on the Swiss court process reaching a favorable conclusion at all.¹⁶