Archegos: How a Single Portfolio Cost Banks Ten Billion Dollars

*A family office almost nobody had heard of accumulated concentrated positions of such size that its unwinding inflicted billions in losses on some of the world's largest banks.*

In March 2021, a family office called Archegos Capital Management, which managed the private wealth of a single individual and was almost unknown outside a narrow circle, collapsed over the course of a few days. The unwinding of its positions inflicted losses on its lenders that ultimately exceeded ten billion dollars, with one major Swiss bank absorbing something in the region of five billion, a blow that contributed materially to the institutional difficulties that followed it.

The structure was straightforward in principle and remarkable in scale. The fund held highly concentrated positions in a small number of companies, and it held them through arrangements with banks in which the bank purchased and held the actual shares while the fund received the economic exposure. This meant the fund could obtain the benefit of an enormous position without appearing as the owner of the shares, and without the disclosure obligations that a holder of that size would ordinarily face.

The leverage was very substantial. The fund's own capital supported positions many times larger, and the effect of this arrangement was that relatively modest movements in the underlying shares translated into very large movements in the fund's equity. When the shares rose, the returns were extraordinary. When they fell, the same arithmetic operated in reverse, and it did so with a speed that left no time for anything.

The feature that made the situation genuinely dangerous, however, was that each lending bank could see only its own exposure. The fund had established similar arrangements with several institutions simultaneously, and no single bank knew the full extent of the positions the fund held elsewhere. Each therefore believed it was extending credit against a position of manageable size, when in aggregate the fund's exposure to certain companies represented a very substantial portion of those companies' outstanding shares.

When several of the underlying holdings declined in March 2021, the fund could not meet the demands for additional collateral that its arrangements required. The banks then faced a situation with no good options. Each held actual shares, purchased on the fund's behalf, which it now needed to sell. They all needed to sell the same shares, in enormous quantity, at the same time, and each knew the others were in the same position.

What followed was a scramble. Some institutions moved to sell quickly and absorbed relatively contained losses. Others, having attempted a coordinated and orderly unwinding, found themselves selling into a market that had already been informed, by the earlier selling, that an enormous quantity of stock was about to arrive. The prices they achieved were correspondingly poor. The losses fell most heavily on those who moved most slowly, which is the ordinary arithmetic of forced selling.

The episode illustrates a number of things at once, and the first is simply what concentration and leverage do when combined. Either alone is survivable. Concentration without borrowing produces volatility, and an investor can hold through it. Leverage on a diversified position produces risk, but the diversification provides some cushion. The combination removes both defences simultaneously, and the resulting structure can be destroyed by an ordinary adverse movement in a small number of securities.

The second concerns the visibility of risk, and it is the feature that made this episode distinctive. Each bank had performed its own assessment and had reached a reasonable conclusion given the information it possessed. The information it possessed was incomplete in a way it could not detect. This is a general property of risk that is assembled across multiple counterparties who do not communicate, and it means that a risk assessment can be entirely competent and entirely wrong at the same time.

The third concerns the speed with which such a structure resolves. The fund did not decline over months. It was destroyed in days, because the mechanism of collateral demands operates on a timescale measured in hours, and because the fund's positions, once known, invited other participants to trade ahead of the selling that was certain to come. There was never any possibility of waiting for the holdings to recover, because waiting was not among the options the structure permitted.

For an ordinary investor, the lesson is not about the exotic arrangements involved, which few will ever encounter. It is about the underlying shape, which is entirely available to anyone. An investor who concentrates their holdings and borrows against them has constructed, in miniature, the same structure, and it will behave the same way. The specific instruments differ; the arithmetic does not, and the arithmetic is what determines the outcome.

One further aspect of the episode deserves mention, which concerns the effect on the companies whose shares were involved. The fund's positions had been so large relative to the available stock that they had contributed materially to supporting the prices of those companies, and the forced unwinding drove several of them down very sharply over a period of days. Ordinary shareholders in those companies, who had never heard of the fund and had no relationship with it whatever, watched substantial portions of their holdings' value disappear for reasons that had nothing to do with the businesses they owned. This is a form of risk that no amount of analysis of a company can anticipate, because it does not originate in the company. It originates in the balance sheet of someone else entirely, and its only real defence is the ordinary one of not holding so much of any single thing that another party's collapse can seriously damage you.

At VESTFY™ Archegos is presented as a demonstration that the two most dangerous decisions an investor can make are dangerous chiefly in combination. Concentration expresses confidence in one's judgement. Leverage removes the ability to survive being wrong. An investor who does both has arranged matters so that a single error, of the kind that everyone eventually makes, is sufficient to end them.