Archegos: learning lessons from the past

Credit Suisse suffered losses of close to US$5.5 billion from the default of Archegos Capital Management. In the latest in a series of articles, Bob Currie reports on the role played by a lack of effective risk technology and failure to learn lessons from past defaults

The market value of Archegos Capital Management’s investment holdings fell sharply during the week of 22 March 2021, driven by a slide in the price of a number of single-name tech stocks, particularly ViacomCBS, in which Archegos had a large and heavily leveraged position. This triggered a chain reaction that resulted ultimately in the hedge fund’s default.

A 170-page report by a Credit Suisse Group special committee makes detailed and critical observations about the engagements of the investment bank, and specifically its prime services division, with Archegos in the lead up to the fund’s closure.

The report, analysed in a series of articles in Securities Finance Times, presents the findings of a Credit Suisse Group Special Committee, on behalf of its board of directors, into the circumstances that led to the Archegos’s collapse and the large financial losses and reputational damage sustained by the investment bank (subsequently referred to as the ‘Special Committee report’).

This event resulted in combined losses of more than US$10 billion for the prime services divisions of global investment banks including Credit Suisse (CS), Morgan Stanley, Nomura, and UBS.

Learning lessons

Among the major conclusions to emerge from this report, covered in the first part of this article published in SFT Issue 284 on 17 August 2021, the investigation found that CS had failed to learn from past risk situations.

For example, in March 2020, CS sustained a US$214 million loss following the default of Malachite Capital Management, another hedge fund, which was a client of CS’ equity derivatives business. Like Archegos, Malachite’s transactions were statically margined (for further discussion, see SFT Issue 284). Both buy-side firms had losses substantially exceeding their potential exposure (PE) and stress scenario limits.

Although circumstances leading to the Malachite default in March 2020 differed in significant ways from those that precipitated Archegos’s collapse, there were also commonalities (see box opposite). Some remedial action had been implemented after Malachite, but other fundamental issues that CS had failed to tackle also re-emerged with Archegos.At the heart of this failure, the Special Committee finds that “CS failed to address a culture that encouraged excessive risk-taking and injudicious cost cutting, as well as a complex and siloed organisational structure that impeded the swift identification, understanding and escalation of risk” (p 30).

This culture prevailed, even though Malachite apparently generated relatively modest revenue — US$6.9 million over the term of the relationship — relative to the risk borne by CS (p 15). The Special Committee drew a similar conclusion with regard to the Archegos default 12 months later.

Following the investigation, led by its Internal Audit team, CS took steps to ensure that its equity derivatives division did not service other clients with profiles similar to Malachite. This process also triggered a programme known internally as Project Copper, intended to improve the bank’s ability to identify early warning of a default event and to reinforce controls and escalation procedures, with a primary focus on OTC derivatives transactions.

Among other changes, Project Copper created a new committee, the Investment Bank (IB) Counterparty Oversight Committee (CPOC), co-chaired by the IB chief risk officer and its chief operating officer, along with senior IB executives including the global head of equities.

At the instruction of the CRM team (see box below) Archegos was one of a small number of counterparties that were discussed at the initial CPOC meeting held in September 2020. Significantly, neither of the co-heads of the prime services division — the division within CS, containing prime brokerage and prime financing sub-units, which had the largest exposure to Archegos — were CPOC members and neither was asked to attend the meeting.

The committee report notes that, in practice, CPOC did not discuss Archegos again for a further six months, not until 8 March 2021, and by this time the client’s risk exposure had surged dramatically (for a more detailed analysis, see SFT 284).

In the meantime, during the autumn of 2020, Archegos’s swap business was migrated from one UK-based CS entity, Credit Suisse Securities (Europe) Ltd (CSSEL), to another, Credit Suisse International (CSi), with the transfer completing in December 2020. This, the report notes, was partly an attempt to wind down CSSEL. However, in so doing, it also transferred the client’s activity to an entity (CSi) that had a “higher stress scenario appetite” — a tolerance for larger risk exposure to the hedge fund counterparty (p 17).

The CRM reportedly backed a case for an accommodative approach, indicating that Archegos was a “significant relationship” for the prime services division and, given that Archegos was also doing business with other prime brokers across the Street, a sudden margin increase could potentially result in “irreversible damage to the client relationship” (ibid).

Mixed messages

Following an annual credit review of Archegos at the end of 2020, CRM took the decision to downgrade Archegos’s credit rating from BB- to B+, thereby placing the hedge fund in the bottom third of CS’ hedge fund counterparties by credit rating (p 18).

But, even while downgrading Archegos’s credit rating, CRM recommended raising Archegos’s potential exposure limit from US$20 million to US$50 million — advice that was a departure from the bank’s standard position that a B+ rated hedge fund should typically have maximum PE of no more than US$10 million.

In February 2021, the business began taking steps to move Archegos to dynamic margining. However, internal calculations (provided by the software engineering team responsible for transitioning clients to dynamic margining platforms) estimated that under CS’s existing prime brokerage dynamic margining framework, this would require the fund manager to post an additional US$3 billion in margin, raising the total margin for its swaps portfolio to US$4 billion.

Fearing the impact this would have on the business relationship, the head of Prime Services Risk (PSR) proposed applying a more forgiving dynamic margining framework for Archegos — one that would create an average margin of 16.3 per cent, requiring the client to post an additional US$1.3 billion on implementation. A written proposal was sent to Archegos on 24 February, but it was ignored by the hedge fund, the report says, despite repeated follow up from CS (p 20).

Archegos default

The value of Archegos’s concentrated positions dropped rapidly in the week of 22 March 2021, reversing the upwards momentum that these positions had experienced for several months leading up to its default. The price of its largest position, ViacomCBS, fell 6.7 per cent on 22 March and further in the days that followed. US$600 million in excess margin that the fund manager had with CS on 23 March was eliminated by these movements in the market and CS requested a further US$175 million in variation margin (VM), which Archegos paid (p 22).

With the valuation of another significant long position, Tencent Music Entertainment Group, falling 20 per cent on 24 March, CS took a decision to issue a US$2.7 billion VM call the following day. Responding to this request, Archegos’s COO told CS’ co-heads of prime services and its head of equities that evening that the buy-side firm did not have sufficient liquidity to meet margin calls the following day to CS or its other prime brokers.

In the course of 25 March, CS’s legal department prepared Event of Default notifications, setting into play CS’s contractual right to demand full return of all outstanding balances in case of a default, including any failure to pay margin when it was required (p 125).

By the evening on the 25th March, Archegos informed its prime brokers on a group call that, although it had US$9 to $10 billion in equity, it had US$120 billion in gross exposure (US$70 billion long, US$50 billion short). It asked its prime brokers to enter into a standstill agreement, whereby these PBs would not push Archegos into default while it took steps to wind down its positions. The hedge fund indicated that it was committed to making all PB and swap counterparties whole by liquidating assets to cover the shortfall owed to each dealer. However, it said that it needed to liquidate assets carefully so as to “not tip the market”.

In the face of this request, CS indicated that it remained firm in its commitment to issue a termination notice, which it sent by email the same evening and by hand delivery on 26 March, specifying 26 March as the termination date.

On the morning of 26 March, CS was approached by Archegos and told that Goldman Sachs was executing block sales of some ADR positions and invited CS to be involved. CS subsequently participated in three Goldman Sachs-led block trades, selling stock in Baidu, Tencent and Vipshop Holdings (p 127).

Alongside this, CS indicates it executed algorithmic trading on that day designed to stay within 2-3 per cent of average daily volume. It ultimately sold slightly more than US$3 billion notional on 26 March, approximately US$1.25 billion being sold through Goldman Sachs block sales.

Archegos and its PBs held another call on 27 March, including CS, Morgan Stanley, Goldman Sachs, Nomura, UBS, Wells Fargo and Deutsche Bank. The fund manager again asked for a forbearance agreement.

Without Archegos in attendance, these PBs discussed the potential for a managed liquidation, where PBs would send positions for review to an independent counsel or other trusted third party and lenders would be offered a percentage range within which they would be allowed to liquidate their overlapping positions.

CS, Nomura and UBS indicated their interest in moving ahead with this managed liquidation. Several other banks, including Deutsche Bank, Morgan Stanley and Goldman Sachs, declined to be involved (p 128).

The Special Committee report indicates that, on 28 March, CS stepped into a managed liquidation agreement with UBS and Nomura. Following this agreement, CS liquidated approximately US$3 billion through block sales of overlapping positions on 5 April and US$2.2 billion on 14 April 2021. It liquidated other Archegos positions through open-market algo trading — and by 22 April 2021 it had liquidated 97 per cent of its Archegos exposure (p 128).

CS indicates, in line with its public reporting, that it lost US$5.5 billion as a direct outcome of the Archegos default and the subsequent unwind. Drawing on public information, it indicates that Nomura lost close to US$2.5 billion, Morgan Stanley approximately US$1 billion and UBS lost US$774 million.

Deutsche Bank, Goldman Sachs and Wells Fargo are reported to have suffered “immaterial losses” (p 129)

Plain sight

As reported in the preceding section of this article, Credit Suisse’s internal investigation into the Archegos failure arrived at the following damning conclusion: “no one at CS — not the traders, not the in-business risk managers, not the senior business executives, not the credit risk analysts, and not the senior risk officers — appeared to fully appreciate the serious risks that Archegos’s portfolio posed to CS (p 129).

These risks, it notes, were not hidden. They were in plain sight from at least September 2020, when CPOC first met and CS senior leaders discussed Archegos’s concentrated, long-biased, volatile equity swap positions. “Yet, no one at the bank acted swiftly and decisively to try to mitigate the risks posed by Archegos. And when CS finally took steps to mitigate the risks, the actions it took were ineffective, too little and too late”.

It concludes that the business should never have put on such large, concentrated positions with Archegos, particularly without securing adequate margin. It indicates that the prime services business was aware that Archegos’s portfolio was highly concentrated and also that its portfolio holdings with CS were severely undermargined (p 131).

For example, at a CPOC meeting on 8 March 2021 a representative from the bank’s client risk management (CRM) team indicated that Archegos was an outlier in the prime financing swap book with gross market value (GMV) exposures of US$20 billion, when compared to the next largest client of US$5 billion. CRM also noted that Archegos had a net-long bias at this time of more than US$7 billion, compared to the next-largest long-biased client at US$1.5 billion (p 20).

The CPOC also flagged up Archegos’s single issuer concentration, including a US$3.3 billion position which represented more than 8 per cent of the outstanding free float. At this meeting, CPOC discussed potential difficulties in liquidating Archegos’s portfolio, should this become necessary, given the size and concentration of these positions.

Additionally, the business missed multiple opportunities to “right-size” Archegos risk. Prime services saw its relationship with the fund manager as “significant”, contributing revenue of US$16 million in 2020 and this was expected to rise to US$40 million in 2021 on the basis of appreciation in Archegos’s positions. The business was intent on growing its relationship with the hedge fund, recognising that it was competing for this business with multiple other prime brokers (p 133). Consequently, the business pushed for an “accommodative approach” to risk, refusing to take forceful steps and remaining reluctant to have difficult conversations with and about Archegos.

Risk employees failed to appreciate Archegos’s true risk, despite a series of red flags, says the Special Committee report. (p 138) CRM learned from Archegos that other prime brokers were charging higher margins and were dynamically margining Archegos’s swaps portfolio. CRM also had access to non-public information confirming that Archegos had concentrated exposure to the same single-name positions with other prime brokers as it held with Credit Suisse (p 27).

“This information — that Archegos maintained similar positions across the Street, that CS was the only PB using static margining and that CS’s margin rates were lower than those of other PBs — should have sounded alarm bells,” says the report (p 138).” The CRM employees that were privy to this information did not appear to appreciate its importance and failed to escalate the information to more senior risk or business managers.

A similar myopia applies to Archegos’s limit breaches. Rather than viewing the breaches as symptomatic of a much larger problem, CRM and the prime services business were focused on eliminating the breaches without considering the larger significance.

Consequently, the CRM did not insist that the business take immediate steps to reduce Archegos’s exposure — by having Archegos sell down its concentrated positions and reduce its portfolio — and instead resorted to “superficial fixes” such as raising Archegos’s PE limit from US$20 million to US$50 million and raising Archegos’s scenario limit from US$250 million to US$500 million (p 139).

Technology and data failures

The Special Committee finds that the prime services business did not prioritise investment in technology that would have assisted in managing risk more effectively. The PSR had identified a relatively inexpensive fix for bullet swaps that, subject to the client’s consent, would have automatically recalculated IM based on the current mark-to-market value of the portfolio. This fix would have cost approximately US$150,000 — a relatively small investment given the size of Credit Suisse’s subsequent losses against its Archegos exposure — but the business did not prioritise or fund this (p 146).

It was also slow to adopt automated technology that would have enabled it to dynamically margin swaps. Despite the systems limitation, the Special Committee says that dynamic margining could have been agreed with Archegos at any time and implemented manually — even though this would have been cumbersome and time consuming for CS to operate.

CS already had an automated means to margin swaps dynamically (through cross-margining with prime brokerage) when these were conducted in the same legal entity. However, the systems for dynamically margining were inadequate within CS for clients such as Archegos whose swaps and PB accounts were held with different legal entities (see SFT Issue 284 for further discussion).

Notwithstanding, by September 2020 CS had developed the capacity to dynamically margin swaps, regardless of whether the client had a PB account or in which legal entity this account was held. However, the Special Committee finds that Archegos was not on the list of clients that the business had prioritised for transition to dynamic margining, despite the increasing risks presented by this client (p 147).

More generally, the report finds that risk divisions had struggled with “fragmented, ineffective technology” during the year leading up to the Archegos default, along with data quality issues that impaired its ability to assess counterparty and portfolio credit risk in a timely manner. “These issues were well known, but not sufficiently addressed before the Archegos default,” says the report.

For example, the systems employed by the CRM team only displayed a counterparty’s aggregate portfolio and not individual position data. This impeded CRM’s ability to see full details of the risks presented by a client. Further, CRM only received verified data for scenario limit breaches on a monthly basis — although it could access “rough” scenario data twice per week. “This posed ‘undue risks’, given that Prime Services had a number of hedge fund clients with volatile positions that could change significantly intra-monthly,” concludes the report (p 147-8).