Too Big to Fail — The Credit Suisse Test —

I. The principle and its paradox

Too Big to Fail is not a legal principle. It has never been enacted as such in any constitution, no civil code contains it, no founding treaty consecrates it. It is a pragmatic fear — the fear that the collapse of a large enough entity would drag down, in its fall, a portion of the real economy inhabited by human beings who had nothing to do with the decisions that led to that collapse. And it is precisely because it is not a legal principle that it is so dangerous: it operates in the shadows of the law, suspending the law when the law becomes inconvenient.

The legal architecture governing commercial entities rests, in all major Western traditions, on a fundamental principle: the legal person is a fiction distinct from those who compose it. The company is not its shareholders. The bank is not its depositors. The holding is not its subsidiaries. This principle — articulated with surgical clarity by the House of Lords in Salomon v Salomon in 1897 — was designed to protect: it separates risks, it localises liability, it allows collective action without collective ruin. B exists so that A and C do not merge.

Too Big to Fail contradicts this architecture from end to end. When a state decides to save an entity on the grounds that its disappearance would be too costly, it implicitly writes what Salomon forbids: it makes C — the taxpayer, the citizen, the sovereign — bear the consequences of decisions made by A, crossing B as if it did not exist. The fiction is preserved, but the principle that justified it is erased. One invokes Salomon when B is profitable. One suspends Salomon when B fails. Two measures, one single legal fact.

II. Credit Suisse 2023: the 72-hour test

The Credit Suisse case is not another Lehman Brothers. It is more instructive, because it is more recent, more deliberate, and more revealing of the structural logic of the system. In March 2023, the Swiss federal authorities did not allow the market to operate, did not open insolvency proceedings, did not apply the ordinary rules of creditor hierarchy. They engineered, in seventy-two hours, a forced merger between two private entities — compelling UBS to absorb Credit Suisse — while simultaneously guaranteeing nine billion francs of potential losses on public funds.¹

The Swiss government passed an Emergency Ordinance on 16 March 2023, granting FINMA the power to bypass shareholder general meetings and to order the write-down of AT1 instruments. The bailout was implemented by administrative fiat, bypassing both parliament and the shareholder assemblies of the affected banks.² The sixteen billion dollars of AT1 bonds — instruments contractually senior to equity in the standard creditor hierarchy — were annulled entirely. Shareholders, who should have lost everything first, received 3.25 billion dollars in UBS shares. Goldman Sachs described the operation as the largest loss ever inflicted on AT1 investors since the birth of that asset class after the 2008 financial crisis.³

This is the Too Big to Fail test in its pure form: not a bailout softened by political necessity, but a sovereign legal decision to suspend the legal order to preserve a financial entity. The fiction won. The law yielded. And the order of losses was not determined by law, contract, or any creditor hierarchy — it was determined by administrative decree, in a weekend, without democratic deliberation.

III. The accountability question: who paid, who answered?

After the dust settled, the question of individual accountability produced the most revealing answer of all: essentially none. Credit Suisse had accumulated, over more than a decade, a record of compliance failures that would be remarkable even for a minor institution. In June 2022, it became the first major Swiss domestic bank to receive a criminal conviction — for failing to prevent a Bulgarian cocaine trafficking ring from laundering money through its accounts. That conviction was subsequently annulled on appeal in 2025, on the procedural ground that the individual employee initially found guilty had died in April 2023, making it impossible to examine the bank's liability without violating the presumption of innocence.⁴ The bank's criminal record thus disappeared with the death of a subordinate.

On the American front, Credit Suisse Services AG pleaded guilty in August 2025 to conspiring to conceal more than four billion dollars from the IRS across at least 475 offshore accounts, paying more than 510 million dollars in penalties.⁵ The entity paid. No individual director was named in the proceedings. The legal person absorbed the sanction. The human beings who designed, supervised, and profited from these structures continued, for the most part, their careers.

The shareholders of Credit Suisse — the human beings who had benefited from dividend distributions during the years of excess — received 3.25 billion dollars in UBS shares in the rescue. They were not wiped out. The AT1 bondholders — institutional investors who had extended credit in good faith under instruments contractually ranked above equity — received nothing. Thousands of them have since filed proceedings before Swiss courts and under bilateral investment treaties.⁶ It is the victims who litigate. The procurators remain silent.

The answer to the question 'who paid to their last penny' is therefore precise — and the precision is damning. The AT1 bondholders paid in full — wiped to zero by administrative decree. The Swiss taxpayer guaranteed nine billion francs in potential losses. And the shareholders? The formulation that they 'paid partially — at a fraction of market price' is legally accurate but economically false, and the distinction matters enormously to any theory of commutative justice. A shareholder who purchased stock at twenty francs, received dividends for a decade, and sold at fifteen francs during the collapse has not suffered a net patrimonial loss. The dividends extracted during the prosperous years may well exceed the capital loss realised on exit. What he loses is a hoped-for future gain — not an effectively invested patrimony that disappears. The informed shareholder — the one who reads the warning signs, sells before the floor, waits out the crisis, and buys back at the trough — loses nothing at all. He has extracted value during the good years and externalised the risk of collapse onto others: the AT1 holders, the employees made redundant, the taxpayer standing as guarantor. At no point has he put a single franc of his pre-existing patrimony on the table to pay for the consequences of a management he implicitly endorsed by holding his position and cashing his dividends. This is the structural irresponsibility of the shareholder model as it currently operates worldwide: the principle of limited liability — conceived to encourage investment — has become an instrument of systematic immunisation. Those who benefit most during the ascent contribute least during the descent. Commutative justice, in its most elementary formulation, requires that he who received during the good years contribute in proportion to what he received — not merely to the residual value of his portfolio at the moment of collapse.⁷ The directors, the architects of the risk, the compliance failures, the decade of scandal — they paid nothing in criminal or civil terms that was commensurate with the damage produced. The FINMA regulator itself, which had oversight responsibility throughout, issued no proceedings against the individuals responsible.⁸