Core Path · L2.1.07

Landmark failures: SVB and Archegos

12 minCase studyInterest rate riskCounterparty risk

By the end you will be able to

  1. 1Explain the interest rate and funding failure at SVB (2023) and the counterparty losses from Archegos (2021).
  2. 2Show how rising rates create paper losses on safe bonds, and how a deposit run turns them into real ones.
  3. 3Explain how swaps hid Archegos's concentration, and why the banks that sold first lost least.
  4. 4Identify the governance and control failures in both cases, and link them to earlier failures.

Why it matters

One bank failed holding some of the safest bonds in the world. Several banks lost billions lending to a single client whose positions none of them could fully see. Neither story needed exotic products. Both needed risks that were visible and not acted on.

Core explanation

Two failures of the 2020s

L2.1.06: Landmark failures: LTCM and 2008 showed leverage, correlation and liquidity at work. These two cases add two more ingredients a risk manager must watch: interest rate risk in the banking book, and concentration, whether in one source of funding or one client.

Silicon Valley Bank, 2023

L1.4.06: The price-yield relationship introduced SVB as a lesson in the price-yield relationship. Here is the whole chain, from the deposit boom to the closure. Spot what failed at each step.

TimelineSilicon Valley Bank, 2020 to 2023

1 / 6 · spotted 0 of 6

2020–2021

A deposit boom

Silicon Valley Bank served technology start-ups and the venture capital firms behind them. As they raised record sums, the money arrived as deposits: the parent group's assets grew from about $71 billion at the end of 2019 to about $211 billion at the end of 2021 (Federal Reserve review). Loan demand could not keep up, so SVB put much of the money into long-dated US Treasuries and agency mortgage bonds at low fixed yields, many of them classed as held to maturity.

Spot it

What risk did this create?

Archegos, 2021

Archegos was not a bank, and it borrowed through swaps rather than loans. Like the hedge funds of L1.2.06: Hedge funds and private markets, it used leverage; unlike most of them, it kept its positions hidden even from its own lenders.

TimelineArchegos and its prime brokers, 2013 to 2024

1 / 6 · spotted 0 of 5

From 2013

A family office with swaps

Bill Hwang ran Archegos Capital Management as a family office. It built positions in shares through total return swaps with several prime brokers: the banks bought the shares, Archegos took the gains and losses and posted margin.

Spot it

Why did swaps keep Archegos's true positions out of sight?

SVBArchegos
Main riskInterest rates and fundingCounterparty and concentration
LeverageThin equity behind long bondsSwaps worth several times its capital
ConcentrationOne industry's uninsured depositsOne client, a few stocks
LiquidityDeposits left in hoursBlocks of shares could not be sold without crashing prices
ControlsHedges removed, no CRO for monthsLimits breached, margin cut, no action
Echo of the pastNorthern Rock's flighty funding (2007)LTCM's lenders blind to its totals (1998)

Try it

A simplified bank, built to show SVB's mechanics, not its actual balance sheet. Raise yields by 3 points first and look at the accounts: nothing seems wrong. Then let depositors withdraw 25%, then 40%. Finally switch on the hedge and run it again.

Try itRates rise, then depositors run

Open the bond pricer →

The bank: $10bn cash, $100bn of 10-year bonds bought at par with a 1.5% coupon, $90bn of loans; funded by $185bn of deposits and $15bn of equity.

Interest rate hedge

Equity at the start$15.0bn
Equity in the accounts after the run$15.0bn
Equity if every bond were valued at market$15.0bn
Bond price per 100100.00
Paper loss on the bonds$0.0bn
Withdrawn$0.0bn
Bonds sold (face value)$0.0bn
Loss made real$0.0bn
Equity in the accounts$15.0bn

Raise yields first, then let depositors withdraw.

Worked example

Paper losses, real losses and the race to the exit

Both examples are illustrative. The bank is the one from the Try it; the second example is a stylised Archegos with four prime brokers. Amounts in US$ billions.

A bank, a rate rise and a run

Step 1 of 5
  1. 1

    Yields rise 3 points

    The 10-year, 1.5% bonds bought at 100 now price at 76.26. The $100bn holding is worth $76.3bn: an unrealised loss of $23.7bn, 158% of the $15bn equity.

The race to the exit

Step 1 of 4
  1. 1

    The set-up

    A client holds $40bn of one stock through equal swaps with 4 banks: $10bn each. Each bank holds 20% margin, $2bn. The client's capital is the $8bn of margin: leverage of 5×. Each bank sees only its own $10bn.

Regulatory lens

Under the US tailoring rules of 2019, banks of SVB's size faced lighter liquidity and capital requirements than the largest banks, and could leave unrealised losses on their available-for-sale bonds out of regulatory capital. Losses on held-to-maturity bonds stay out of capital at every bank. The Basel Committee's report on the 2023 banking turmoil (October 2023) and the US authorities' own reviews have led to proposals to tighten rules for mid-sized banks and to look harder at interest rate risk in the banking book, which has its own Basel standard. After Archegos, supervisors pressed prime brokers to collect fuller information on clients' total positions and to margin concentrated positions more heavily. L2.6.04: Hedging interest rate risk and M2.7: Introduction to regulation return to both.

Sparks

SparkOne bad month in 2023
SVB closed on 10 March 2023 and Signature Bank two days later. In Switzerland, Credit Suisse, weakened by years of losses including Archegos, was taken over by UBS on 19 March. On 1 May, First Republic Bank failed too. SVB, Signature and First Republic were three of the four largest bank failures in US history.
SparkThe same lesson, 23 years apart
In 1999 the President's Working Group warned that LTCM's lenders had not seen its total positions or demanded enough margin. In 2021 Archegos's lenders made the same mistake. Risk lessons fade unless they are built into limits and controls.

Key takeaways

  • SVB: long fixed-rate bonds lost value as rates rose; a concentrated, mostly uninsured deposit base ran, and selling bonds made the losses real.
  • Paper losses are still losses: a bank can be insolvent at market value while its accounts look healthy.
  • Archegos: swaps spread across several prime brokers hid one client's huge, concentrated, leveraged positions.
  • When a concentrated position must be sold, the first seller loses least: margin must allow for how hard the exit will be.
  • In both cases the risks were visible inside the firms; the failure was that warnings and limit breaches did not lead to action.

Reaction

Check what stuck

Question 1 · objective 1

Which combination best explains SVB's failure?

Question 2 · objective 2

A bank has $10bn of equity. Its held-to-maturity bonds are worth $12bn less than their book value. How large is the unrealised loss as a percentage of equity?

%

Question 3 · objective 3

A bank has a $10bn swap with a client and holds $2bn of margin. The client defaults and the bank sells after the stock has fallen 45%. What does the bank lose, in $bn?

$bn

Question 4 · objective 3

Why could no single bank see the full size of Archegos's bets?

Question 5 · objective 4

What control failure do SVB and Credit Suisse's handling of Archegos have in common?

Glossary

Held to maturity
Held to maturity: bonds a bank says it will keep until they are repaid. They are carried at cost, so falls in their market price do not show in profit or in regulatory capital for most banks.
Unrealised loss
A fall in the market value of an asset you still hold. It becomes a realised loss when you sell.
Uninsured deposits
Deposits above the deposit insurance limit, US$250,000 per depositor per bank in the US. Their owners can lose money if the bank fails, so they run first.
Family office
A firm that manages the money of one wealthy family. It faces lighter disclosure rules than funds that manage other people's money.
Total return swap
A total return swap: the bank buys the shares and passes all their gains and losses to the client, who pays a fee and posts margin. The client has the exposure without owning the shares.
Prime broker
A bank that lends to, trades with and holds assets for hedge funds and similar clients.
Margin
Collateral a client posts against a position, to cover losses if it cannot pay.
Concentration risk
Too much exposure to one counterparty, one asset or one source of funding, so a single event can cause a large loss.

Deep Dives & Origins

Deep Dives

  • Board of Governors of the Federal Reserve System, Review of the Federal Reserve's Supervision and Regulation of Silicon Valley Bank, April 2023 (the four key takeaways at the start). Beginner.
  • Paul, Weiss, Rifkind, Wharton & Garrison LLP, Credit Suisse Group Special Committee of the Board of Directors: Report on Archegos Capital Management, 29 July 2021. Intermediate.

Origins

  • Board of Governors of the Federal Reserve System, Review of the Federal Reserve's Supervision and Regulation of Silicon Valley Bank, April 2023.
  • Federal Deposit Insurance Corporation, press releases of 10 and 12 March 2023, and Options for Deposit Insurance Reform, May 2023.
  • US Government Accountability Office, Bank Regulation: Preliminary Review of Agency Actions Related to March 2023 Bank Failures, GAO-23-106736, April 2023.
  • SVB Financial Group, Form 10-K for 2022 and Form 8-K of 8 March 2023.
  • Basel Committee on Banking Supervision, Report on the 2023 banking turmoil, October 2023.
  • Paul, Weiss, Report on Archegos Capital Management, July 2021, as above.
  • US Securities and Exchange Commission, complaint against Archegos Capital Management and Sung Kook (Bill) Hwang, April 2022.

Module complete

You have finished M2.1: What risk is. Try the Lab Test: case problems on risk and loss, uncertainty, risk and return, why firms hedge, and the great failures.

Watch this lesson

Video version

Prefer to watch? The same lesson as a 12-minute video will sit here, made from this page's script.