L10 - BIS

A.1 Why regulate banks? Tier 1 background

Banks are special: they uniquely combine payments, lending, and deposit-taking, and act as the conduit for monetary policy. They also rely on two structural features that create fragility:

• Leverage — a thin equity base supports a large balance sheet.
• Maturity transformation — funding short-term (deposits, callable on demand) while holding long-term, illiquid assets (loans).

This matters because banking crises cause permanent GDP losses (output never returns to the pre-crisis trend — the GFC scar is still visible in US and EU27 real GDP). Public backstops (deposit insurance, lender of last resort) stabilise the system but create two problems regulation must counter:

• Failure to internalise systemic costs — an individual bank ignores the spillover its failure imposes on the system.
• Moral hazard — knowing a backstop exists, banks take more risk than they otherwise would.

So regulation exists to force banks to hold buffers (capital, liquidity) that internalise these costs ex ante.

A.2 What is capital? Tier 1 — core exam topic

Common Equity (going-concern capital)

The residual claim on a bank: the difference between what it owns/is owed (assets) and what it owes (liabilities). It is the shareholders’ claim after all other funders are accounted for.

Three things it is not:

• Common equity is funding, not an asset.
• Common equity is not cash.
• Common equity is not “held” — you cannot point to it on the asset side.

With positive common equity, depositors and senior funders are comfortable that assets cover their claims, so they keep funding the bank — it remains a going concern.

The key exam property (Exam: Rep Task 14A): common equity is the first form of funding to bear a loss, and it does so instantaneously. It moves up with every profit and down with every loss, with no liquidation, write-down trigger, or process required. That instantaneous, automatic adjustment is exactly what makes it the first loss absorber — distinguishing it from subordinated debt (which needs a process) and from deposit-insurance-style protections.

Return expectations (Exam: SS Task 14A): because equity holders sit at the bottom of the hierarchy (first loss, residual, no fixed claim), they bear the most risk and therefore demand the highest expected return. The ordering of required returns mirrors the loss-absorption hierarchy:

$$r_{\text{common equity}}>r_{\text{hybrid / AT1}}>r_{\text{subordinated debt}}>r_{\text{senior debt / deposits}}$$

Subordinated Debt (gone-concern capital)

Debt that ranks junior to depositors and senior funders, so — like equity — it can absorb losses before them. But unlike common equity it requires a process of liquidation to absorb a loss, and liquidation carries substantial costs: administrative costs plus fire-sale losses on assets.

The loss waterfall in liquidation: losses hit common equity first, then subordinated debt, and only finally senior liabilities / depositors.

Hybrid Capital

A form of subordinated debt that shares features of common shares. A typical instrument is perpetual (no maturity date) and lets the bank cancel interest payments (like cancelling dividends). Some classes are written off or converted to common shares at a specified trigger event — these are the contingent convertibles (CoCos / AT1, see Part C).

The capital stack (Tier 2 — useful framing for the whole topic):

$$\underset{\text{Total Capital}}{\underbrace{\stackrel{\text{Tier 1 — going concern}}{\overbrace{\text{CET1 (common equity)}+\text{AT1 (hybrid)}}}+\stackrel{\text{gone concern}}{\overbrace{\text{Tier 2 (subordinated debt)}}}}}$$

A.3 What is liquidity? Tier 1 background

Funding Liquidity

Having cash when you need it to meet repayments to liability holders (depositors) or clients (loan-commitment / credit-card drawdowns). It can be generated from either side of the balance sheet:

• Asset side — selling assets.
• Liability side — raising new funds (additional borrowing).

Liquid Asset / Market Liquidity

An asset is liquid if it can be rapidly converted to cash (sale or repo) with no loss of value. This usually requires a deep and active market in which it trades — referred to as market liquidity. All else equal, a bank with more liquid assets copes better with depositor repayment requests.

Note the distinction between funding liquidity (can I meet my obligations?) and market liquidity (can I sell this asset without moving the price?). They interact in a crisis: when market liquidity dries up, funding liquidity does too.

A.4 How much capital is enough? Tier 2

Two regulatory metrics, each a floor:

$$\text{Leverage ratio}=\frac{\text{Capital}}{\text{Leverage Ratio Exposure}}\ge 3$$ $$\text{Risk-based ratio}=\frac{\text{Capital}}{\text{RWA}}\ge 8$$

Usually you will have to compute the RWA/LCR required minimum capital so you do 3% * Leverage Ratio Exposure

Leverage Ratio Exposure vs. Risk-Weighted Assets (RWA)

• Leverage Ratio Exposure — a measure of bank size: total assets plus off-balance-sheet items converted to on-balance-sheet equivalents. It does not discriminate between the riskiness of assets. (This is the “measure of bank size” distractor in the LCR MC question.)
• RWA — nominal exposures converted to risk-weighted exposures, using either standardised risk weights (set by regulation) or banks’ internal models.

Example (Exam-style): a $100$ CHF residential mortgage with a $30\%$ risk weight.

$$\begin{array}{rl}\text{Leverage-ratio min capital}& =3\%\times 100=3\text{ CHF}\\ \text{Risk-based min capital}& =30\%\times 100\times 8\%=2.4\text{ CHF}\end{array}$$

Here the leverage ratio binds (it asks for more capital). The two metrics are designed as complementary backstops: the risk-based ratio is risk-sensitive but model-dependent (and gameable); the leverage ratio is crude but model-independent and catches the case where internal models understate risk.

Tier 2 stylised facts from the historical charts: capital/asset ratios fell from ~50% in the 1840s to single digits by the late 20th century; since 2007–09 the trend reversed upward. Average RWA density of large banks declined over 1994–2024 even as leverage rose — part of what motivated the Basel III output floor (Part B.3), because stakeholders lost faith that low reported RWAs reflected genuinely low risk.

A.5 How much liquidity is enough? Tier 1 — core exam topic

Two regulatory liquidity metrics:

$$\text{LCR}=\frac{\text{Stock of HQLA}}{\text{Total net cash outflows over next 30 calendar days}}\ge 100\%$$ $$\text{NSFR}=\frac{\text{Available Stable Funding (ASF)}}{\text{Required Stable Funding (RSF)}}\ge 100\%$$

ASF = portion of banks funding sources expected to be reliable over the horizon
RSF = amount of stable funding needed to support the banks asset holdings and certain off balance sheet exposures
NSFR → manages “can we still be funded over the next year” where LCR is “next 30 days”

Liquidity Coverage Ratio (LCR)

Banks must hold enough High-Quality Liquid Assets (HQLA) to exceed estimated total net cash outflows over a 30-day stress period. (Exam: SS Task 14B — the correct MC option is precisely “HQLA exceeds net cash outflows over a 30-day period”; the 60-day, NSFR, and bank-size options are distractors.)

• HQLA: e.g. government securities and central bank reserves.
• Outflow rates are stressed assumptions, e.g. ~$3\%$ on stable insured deposits, rising for less stable / corporate / operational deposits.

LCR assumes specific outflow rates → 2023 turmoil exposes limitations in capturing extreme, rapid outflows.

The LCR was introduced after the 2008 financial crisis exposed how quickly liquidity can evaporate. Its core purpose is to ensure banks can meet short-term obligations without central bank emergency support.

Advantages:

• shock absorption → real liquidity buffer
• transparency & comparability: standardised across jurisdictions
• Reduced Moral Hazard → less reliance on central banks
• systemic stability → dampens fire-sale dynamics like in 2008
  Disadvantages:
• higher cost of HQLA portfolios → lower yield
• 30-day horizon is arbitrary
• HQLA definition gaming → Hold-to-maturity assets not being marked-to-market hides losses when rates fall
  • Long-term HTM treasury bonds → 15-17B unrealised losses
• LCR doesn’t capture structural funding risk → role of NSFR
• higher outflow rates for uninsured deposits are not modelled

Net Stable Funding Ratio (NSFR)

The longer-term companion to the LCR: requires a stable funding profile relative to the composition of assets, measured as available stable funding over required stable funding.

• e.g. corporate loans with >1y residual maturity: $\text{RSF}=85\%$; funding with >1y maturity: $\text{ASF}=100\%$.

ASF (liabilities & equity) = how stable your funding sources are

• each liability gets a factor assigning how likely it sticks around under stress
  • LTD (maturity > 1 year) → 100%
  • stable retail deposits 95%
    RSF (asset requirement) = how much stable funding each asset requires
• Cash & central bank reserves → 0%
• short-term interbank loans (< 6 months) → 10%
• …
• illiquid / other assets 100%
  → high RSF = asset is long-dated or illiquid
  • a $ 100m 25 year mortgage cannot be sold overnight in a crisis
  • thus the $ 100m are gone → need liabilities that are long term
  • for example 5 year bonds → the investors can’t demand their money back overnight, so the bank can fund the $ 100m taken for the mortgage from those bonds it issued

Three-way contrast to memorise (these are each other’s MC distractors):

• LCR — short-term (30-day) survival, asset-side liquidity buffer.
• NSFR — long-term structural funding stability.
• Leverage ratio — bank size / capital adequacy, not liquidity at all.

LCR — importance, advantages, disadvantages (Exam: Rep Task 14C asks exactly this):

Why it matters / advantages. It is a standardised, comparable, simple ratio that forces every bank to pre-fund a 30-day liquidity buffer of high-quality assets, directly targeting the maturity-transformation fragility that causes runs.

Disadvantages — and the March 2023 evidence that exposed them:

• Assumed outflow rates were far too low. SVB lost ~85% of deposits in 2 days; the LCR’s stressed retail/corporate outflow rates (≈3–40% over 30 days) badly underestimated digital, social-media-driven runs.
• The 30-day horizon is the wrong clock for a run that plays out in hours/days.
• It misses drivers like prepositioning, local/intraday liquidity needs.
• Buffer usability problem: banks are reluctant to dip below 100% (operational capacity limits, and the market signal of disclosing buffer use can itself accelerate a run), so the buffer is less usable than intended.
• HTM securities with unrealised losses may be unsellable / hard to repo in stress, so “liquid” assets are less liquid than assumed.

B. Why global standards? Tier 1 (the "why"); Tier 3 (the history)

Why global cooperation? (Exam: SS Task 14C)

Safeguarding financial stability is a global collective responsibility. Core arguments to give in the exam:

• Banking is cross-border; a failure in one jurisdiction spills into others (the founding example: Bankhaus Herstatt, 1974).
• Common standards create a level playing field and prevent regulatory arbitrage (banks migrating to the laxest regime).
• Getting a complete picture of a global bank requires consolidated supervision, which only works with cooperation.
• Financial trilemma (Schoenmaker, 2011): you cannot simultaneously have (1) financial stability, (2) financial integration, and (3) national financial policies — at most two. Cross-border banking forces a choice, and cooperation is how you keep stability while integrated.

Tier 3 — origin of the Basel Committee. Established 1974 by the G10 governors after Herstatt’s failure. The 1975 Concordat set the basis for supervisory cooperation to avoid gaps, with the aims that (i) no banking establishment escapes supervision and (ii) supervision is adequate and consistent across jurisdictions. The BCBS today spans 45 institutions from 28 jurisdictions, ~$165trn of banking assets. It is the rare body that is simultaneously global, a standard-setter, and non-political.

Consolidated Supervision (Tier 3)

Supervising the banking group as a whole, not entity-by-entity. The parent’s “investment in subsidiary equity” nets out against the subsidiary’s equity on consolidation, so the consolidated balance sheet shows the true aggregate assets, liabilities and equity — preventing double-counting of capital and hidden leverage inside the group.

Tier 2/3 — Basel I → II → III timeline:

• Basel I (1988) — minimum capital/RWA of $8\%$ by end-1992; first risk-weighted approach. Market Risk Amendment (1996) added a capital charge for market risk and allowed internal value-at-risk models.
• Basel II (2006) — the three pillars: Pillar 1 (minimum requirements for credit, market, operational risk; internal models permitted), Pillar 2 (supervisory review), Pillar 3 (disclosure / market discipline).
• Basel III (2010, post-GFC) — stricter quality/quantity of capital (central role of CET1); the capital conservation buffer (restricts payouts when breached); the countercyclical buffer (built up in credit booms); the leverage ratio; LCR and NSFR; extra requirements for systemically important banks.
• Basel III finalisation (2017) — revised credit/CVA/operational-risk standards, revised leverage ratio + G-SIB leverage buffer, and the
  • output floor (caps how far internal models can cut risk-based requirements below the standardised approach).
  • Key objective: reduce excessive RWA variability, because at the GFC peak stakeholders lost faith in reported risk-weighted ratios.

C. Lessons from crisis episodes Tier 1 (2023 causes); Tier 3 (framework)

Tier 3 — the lessons framework. Micro: governance/risk-management failures and misaligned incentives; weak regulation (quality+quantity of capital); weak supervision and market discipline; over-reliance on internal models and a single resilience metric; under-appreciated risk concentrations; inconsistent implementation. Macro: missing macroprudential element; interconnections / contagion / NBFI; tension between simplicity, risk sensitivity and comparability; the regulatory cycle.

The March 2023 banking turmoil Tier 1 — core exam topic

The most significant system-wide banking stress since the GFC: five banks with ~$1.1 trillion in assets failed, merged or were liquidated (Silicon Valley Bank, Signature Bank, Credit Suisse, First Republic, plus SVB UK), amid a broader crisis of confidence and extensive public support.

Key causes to cite (Exam: Rep Task 14B — the correct MC answer is “high levels of uninsured deposits and unrealized interest-rate losses”):

• Unrealised interest-rate losses — banks held long-dated securities (often at amortised cost / HTM) that fell in value as rates rose; selling to raise cash would crystallise the loss. This is an IRRBB failure.
• High uninsured-deposit concentration — uninsured depositors run fast, and a narrow client base (e.g. SVB’s tech/VC concentration) runs together.
• Speed and scale of outflows far exceeded LCR/NSFR assumptions — social-media-driven, digital runs (SVB lost ~85% in 2 days vs. the LCR’s 30-day, 3–40% assumptions).
• Underlying it all: fundamental risk-management shortcomings, poor risk culture, weak board oversight, and failure to respond to supervisory feedback, plus supervisors’ limited ability/willingness to enforce prompt action.

Tier 2 — Additional Tier 1 (AT1) lessons. AT1 instruments provided loss absorbency at failure / point of non-viability (PONV) but generally no meaningful loss absorbency in advance — going-concern features (cancelling distributions, automatic principal write-down) were not activated in almost all cases:

• Credit Suisse — AT1 written down because authorities decided to grant public support.
• SVB / Signature — cancellation not triggered, given the speed and nature of failure.
• First Republic — only cancelled AT1 dividends after a $30bn deposit injection.

Market participants therefore price AT1 as a PONV instrument; going-concern triggers and write-down/conversion features play little role in investor and rating-agency assessments. AT1 design space: equity-classified (no principal-loss-absorbing mechanism, e.g. preferred shares) vs. liability-classified with PLAM (CoCos), with $5.125\%$ or $7\%$ CET1 triggers leading to conversion-to-CET1 or write-down (permanent/temporary).

Tier 2 — other 2023 takeaways: deposit outflows were larger/faster than LCR/NSFR assumed; non-LCR drivers (prepositioning, intraday) can be material; HTM securities with unrealised losses are hard to monetise in stress; operational capacity and disclosure stigma impede buffer use; Pillar 2 monitoring tools are broadly fit for purpose but their value depends on frequency, granularity and scope.

D. Recent trends and open issues Tier 3 — interesting, unlikely tested

• Rising public debt — AE/EME government debt projected to climb steeply to 2050; large shares of outstanding debt mature within 3 years, creating rollover/funding risk that feeds back into bank balance sheets (sovereign exposure).
• Growth of non-bank financial intermediation (NBFI) — NBFI assets now exceed bank assets globally; growth concentrated in investment funds, pension funds, insurers, MMFs, hedge funds. Banks’ domestic and cross-border claims on NBFIs are rising, increasing interconnection and contagion channels (four stress scenarios: NBFI losses hitting banks; conglomerate step-in risk; NBFIs withdrawing risk transfer; NBFIs pulling funding — the dash-for-cash).
• Private credit & AI — private credit (~$3trn) is growing fast with a rising share lent to AI firms, whose credit risk (5y CDS) has climbed; banks connect via NAV loans, subscription lines, portfolio financing, synthetic risk transfers, CLOs.
• Cyber / frontier AI — frontier models could change the speed, scale and distribution of cyber incidents (the “Last Ones” simulated network-attack benchmark). (Footnote of personal interest: the slide’s chart tracks Claude Opus 4.6/4.7 and a “Mythos Preview” model among the frontier systems.)
• Cryptoassets — bank exposures rising fast but still ~$0.20\%$ of total assets; BCBS prudential + disclosure standards take effect 1 January 2026.
• Stablecoins & the three tests of money — market cap still limited (~$300bn, dominated by USDT/USDC). Stablecoins are assessed against the benefits of sovereign money:
  • Singleness — trading at par. Holds under normal conditions; large deviations in stress (USDC fell to $0.88$, others to $0.95/0.96$).
  • Elasticity — the system’s ability to expand balance sheets/credit on demand; stablecoins are rigidly reserve-backed and lack this.
  • Integrity — AML/CFT. Stablecoins circulating peer-to-peer on public blockchains move outside the issuer’s KYC boundary, creating integrity gaps.
• Buffer usability — the recurring open question (echoing the LCR disadvantage above): regulatory buffers only help if banks are actually willing to use them.