We are asked, occasionally, what separates a structure that lasts from one that appears in a newspaper. The answer is never the practice. It is the construction. Below are four failures of construction, anonymised because their operators have suffered quite enough exposure without our adding to it, each reduced to the one pass its designer decided to skip.
A structure does not fail because the work it does is objectionable. A great deal of objectionable work is conducted, lawfully and permanently, by structures nobody has heard of, which is rather the point of retaining someone to build one. A structure fails because it was built badly, and it is almost always the same story: one pass in the design was dull, so the designer skipped it, and that pass was the one holding everything else up.
We keep these four on file for clients who are certain their own arrangement is sound. None of the operators below was undone by a regulator cleverer than they were. Each was undone by the pass they chose not to pay for.
The operator ran a customer-deposit business and a proprietary trading business, and joined them, through a related party, with a channel that let one draw on the other at will. There was no independent board, no segregation of client assets, and no reconciliation any third party was required to perform. The design worked precisely as long as no unusual number of customers asked for their money on the same morning, which is to say it worked until the first ordinary morning they did.
The defect was not the borrowing. Institutions lend against client balances lawfully every day. The defect was building a single entity in which the customer’s asset and the house position were the same dollar, so that one withdrawal wave took both and left nothing to return. A segregated structure fails slower, smaller, and in private. This one failed in a courtroom, which is the most expensive room a structure can fail in.
The balance sheet carried close to two billion in cash, held on paper in escrow accounts operated by third parties in an accommodating jurisdiction. The entire edifice rested on one assumption: that no person whose job was to confirm those balances would ever confirm them directly with the banks said to hold them. For years, none did.
A structure whose solvency depends on a confirmation nobody is compelled to perform is not a structure. It is a delay with a letterhead. The competent version of this arrangement never makes the load-bearing wall out of a step the process does not require, because a step that is optional will, on a long enough timeline, be taken by exactly one curious person. When the confirmation finally came, the two billion was not there, and on the evidence had never been. The design had one elegant property and one fatal one, and they were the same property.
The office held positions large enough to move the instruments they rested on, through swaps placed across several prime brokers, each of whom saw its own slice and none of whom saw the whole. Below the disclosure threshold, nothing had to be reported. Nothing was. The concealment, we will grant, was well executed.
It was also built for the one event that was never going to arrive and undefended against the one that always was: a move against the position large enough to trigger a margin call at every broker in the same hour. There was no buffer for a correlated unwind, because the designer had confused “nobody can see it” with “nothing can happen to it.” These are different properties of a structure. A professional builds for the second and treats the first as a pleasant courtesy. An amateur is briefly, expensively delighted by the first and has made no provision for anything else.
Capital raised for national development was moved outward through a chain of holding entities in cooperative jurisdictions, which is ordinary, and would have stayed unremarkable had the proceeds not then reappeared in a form visible from orbit. The layering was competent. The exit was not.
A structure designed to obscure the origin of funds is undone the instant those funds resurface as a superyacht, a feature film, and property in three of the most photographed cities on earth. The defect was a failure of discipline at the only point where discipline is not optional, which is the withdrawal. Everything upstream of the spending was defensible. The spending is what walked into the prosecutor’s office and introduced itself.
| The structure | The pass it skipped | What surfaced it | What a properly built structure does |
|---|---|---|---|
| Deposit-and-trading exchange | Segregation | Customer asset and house position were the same dollar | Fails slower, smaller and in private |
| Listed payments group | Failure design (who verifies, and when) | Solvency rested on a confirmation nobody was required to make | Never builds a wall out of an optional step |
| Single-family office | Failure design (what happens when it moves) | Concealment mistaken for safety; no buffer for a correlated unwind | Survives its own worst day intact |
| Sovereign development fund | Venue and characterisation of the exit | Flawless layering undone by an undisciplined withdrawal | Exits as quietly as it entered |
Note the column that does not appear: none of these was a failure of nerve, ambition or, whatever the coverage suggested afterward, character. Each was a failure of one specific, boring, skippable pass, and each operator skipped it for the same reason, which was that it was boring. We do not find this tragic. We find it avoidable, which is the more expensive category for the people who paid to have the thing built.
The structures we design do not appear on this page. That, rather than any question of who deserved what, is the whole of the service, and the only figure on it that matters is the number of our clients you have read about, which is none.