The risk no UHNW client measures in their own bank

Anyone who ran the finances of a British technology company in March 2023 remembers the weekend, since on the Friday the Bank of England announced its intention to place Silicon Valley Bank UK into a bank insolvency procedure and by seven o’clock on Monday morning the bank had been sold to HSBC for a pound, its customers told they could bank as normal. Payroll went out. The episode is usually recalled as a story about a failing lender, yet its more instructive lesson concerns the procedure, because the resolution regime Parliament built after 2008 treats the continuity of banking services as an objective in its own right; on that weekend continuity was, quite deliberately, delivered first and the accounting settled afterwards.

I have sat through a particular meeting often enough, in St James’s as in Lugano, to recognise it before the coffee arrives: a family office serving an industrial family, usually in its second or third generation with two branches that communicate largely through their advisers, has to decide where to place the proceeds of a divisional sale or an exceptional dividend, and since the existing bank is sound but slow someone proposes a small private bank in Zurich or Geneva, relatively young, with a relationship manager who answers the phone on Saturdays and an execution speed the large houses no longer offer. The due diligence is then conducted conscientiously, in the way we all know (capital ratios, the standing of the shareholders, the quality of the portfolio manager and the fee schedule), yet nobody asks what would happen to the family’s assets if the bank lost its licence without failing, not out of carelessness but because the scenario does not exist in anyone’s mental map. A bank, to its client, either stands or falls.

Since February that scenario has had a name and an address in central Zurich, for on 6 February FINMA withdrew the licence of MBaer Merchant Bank, having found serious and systematic failings in its controls against money laundering, failings that had allowed sanctioned clients to circumvent asset freezes; the bank appealed, the Federal Administrative Court granted suspensive effect and for three weeks business carried on, until a notice from the US Treasury’s Financial Crimes Enforcement Network at the end of February made resistance untenable and the appeal was withdrawn. The merits of the violations, which are a matter for supervisors and courts, do not concern me here; what concerns me is what happened to the clients afterwards, because it belongs to a category of risk that appears in no due diligence questionnaire I have ever seen.

MBaer was not insolvent; on the contrary, it held roughly CHF 70 million of equity against just under CHF 5 billion of client assets spread across some 700 relationships, which is why the liquidation FINMA ordered is not a bankruptcy but an ordinary winding up under the Swiss Code of Obligations, a route for which Switzerland has no genuinely comparable recent precedent. And yet, around six months after the licence was withdrawn, according to reporting in the Zurich financial press only about CHF 30 million of those five billion had been paid out, so that on paper nobody has lost a franc, the accounts reconcile and the assets are all there; they simply do not move.

The mechanics repay close attention, link by link, because that is where the risk actually sits: when a bank loses its licence it also loses direct access to SIC, the interbank payment system for Swiss francs run on behalf of the Swiss National Bank, since an account there is reserved for licensed institutions; in MBaer’s case the formal exit came on 20 March. Without direct access it therefore needs correspondent banks willing to route payments for an institution in liquidation that has attracted the attention of American authorities, banks which the liquidators announced they had found only on 22 May; in the meantime clients were told they could withdraw at most CHF 100,000 each, in francs only and only to an account in their own name at another Swiss bank. Then comes verification, which is the real bottleneck, since every payment undergoes a double KYC review, first by the bank’s remaining compliance team and then by the liquidators’ own, including files that had already been thoroughly re-examined in previous years under FINMA’s eye. No link in this chain is insolvent, and any one of them can stop everything on its own.

Set beside the SVB UK weekend the comparison is uncomfortable, for a lender that could not survive on its own was absorbed over two nights so that its customers would notice nothing, whereas a lender with its capital intact has left its customers waiting half a year. I do not draw from this a verdict on Swiss supervision, whose procedure answered in this case to a different priority, but rather a narrower point that is more useful to the client: continuity is a property of the procedure rather than of the bank, and which procedure applies is decided by events the client does not control.

Put more clinically, the UHNW client assesses a bank as a balance sheet, whereas the risk that struck these 700 accounts concerns the bank as a node in a network, so the question the client asks is “can this bank fail?”, a question nobody in London or Zurich has underestimated since March 2023, the month of SVB and of Credit Suisse. The question the client does not ask is a different one, namely: “on the day this bank can no longer move my assets, who decides when and how quickly they will move again?”. The first measures the probability of losing the wealth, the second the probability of not being able to use it; hence two risks with entirely different distributions, since the second depends on parties with whom the client has no contract whatsoever: the central bank whose system conditions access to payments, the correspondent banks deciding whether to take on someone else’s reputational risk, a foreign authority able to render an institution untouchable with a notice published in Washington and the liquidator, who answers to the supervisor rather than to those who are waiting.

I have watched many families choose a small bank for precisely the qualities that expose it to this second risk, yet I have never seen one ask how many correspondent banks it relied upon or what would happen if one of them withdrew; and the scale that makes the personal relationship possible, the manager who knows the grandchildren’s names and the decision taken in an afternoon, is the same scale that narrows the exits when something jams, because a large bank that loses a correspondent has others while a small one sometimes does not. It is not a law, since there are very solid boutiques whose correspondent networks are sturdier than those of many mid-sized institutions; it is rather a correlation the client should verify instead of assuming, all the more so in a market that is contracting, where the number of Swiss private banks tracked by industry analysts fell from 156 in 2010 to 79 by the end of May this year and where the same analysts expect some of the smaller ones to become entry points for international groups. Each change of ownership, then, can redraw a bank’s correspondent network or its compliance appetite without the client ever being told.

There is a second, less visible layer, which concerns the incentives of the procedure itself: a liquidation ordered by a supervisor is built to protect a financial centre from reputational contagion rather than to give clients back their time, which means the client’s time is not a variable for which anyone in the chain is accountable. The liquidator is paid on a time basis and reports to FINMA, so much so that when the press asked him about the delays he referred the questions to the authority that supervises him, which in turn firmly rejected any suggestion that there might be an interest in prolonging the process. I take that answer at face value, not least because slowness requires no bad faith: it needs only an architecture in which every actor minimises its own risk and nobody is measured on anyone else’s time. According to unconfirmed reports the mandate has been budgeted over a horizon of roughly ten years, while clients have, in principle, an all but unconditional claim to the immediate return of their assets; it is a claim, precisely, and not access.

One variable cannot be settled from public information, namely whether clients with spotless files are emerging from the process faster than the rest, since if they are the case reduces largely to a bank that chose its clientele badly. Even on the most charitable reading, though, something does not add up, because the review is being repeated for every single relationship and because, according to the Zurich press, those hit hardest have not been the large fortunes but the small companies that ran their operating cash through the bank, a distinction that anyone who remembers what the SVB UK weekend meant for companies holding most of their cash in one place will recognise at once. Capital can afford to wait a few months; a payroll cannot.

This is where the family office from that meeting ought to reopen its map, for spreading liquidity across several banks to reduce counterparty risk does only half the job; the other half lies in knowing which part of the family’s wealth must remain available in all circumstances and on which third parties its mobility depends, information that rarely coincides with the list of banks in the consolidated report. A family that has not separated the money that can wait from the money that cannot, and has never asked its bank what happens on the day it can no longer move the family’s assets, will thus discover, when the problem arrives, that the risk it measured with such care and the risk it actually runs are two different things. Nobody will ever send it a rating for the second.