Bank Capital Requirements: Basel and CET1 Explained

Introduction

A bank's balance sheet is mostly other people's money. For every dollar of shareholder equity, a large US bank carries roughly ten dollars of deposits, borrowings, and other liabilities. That is not recklessness, it is the shape of the business: banks make money by funding assets more cheaply than those assets yield, and leverage is what turns a thin spread into a respectable return. The problem is arithmetic. If assets are ten times equity, a loss of just over ten percent on the asset side erases the equity entirely, and the people left holding the bag are depositors, other banks, and eventually the government agencies that insure them.

Capital requirements exist to put a floor under that cushion. They are the single most important constraint on what a bank can do, and they explain things that otherwise look strange: why a bank with record earnings still cannot buy back as much stock as it wants, why two banks with identical assets can report very different capital ratios, and why some of the largest banks in the world are handing loan origination to private credit funds rather than keeping the loans themselves.

If you are interviewing for a financial institutions group, this is the framework behind almost every technical question you will get. Our companion piece on how to value a bank using tangible book value and return on equity covers the valuation side. This post covers the regime that decides how much book value a bank must carry in the first place. Everything below is labeled as in force or proposed, because the US rules are mid-reform and confusing the two is worse than not knowing the number.

Before the mechanics, here is the capital stack in one view.

Nobody sets a minimum equity ratio for a software company. Regulators set one for banks because three features of banking, none of which apply to an ordinary corporate, combine to make private failure a public problem.

Deposits Are Debt That Can Leave Overnight

The core oddity of a bank is the mismatch between the two sides of its balance sheet. Assets are long and illiquid: thirty-year mortgages, five-year corporate term loans, commercial real estate. Liabilities are short and instantly callable: checking accounts, savings deposits, overnight repo. That transformation of short-term money into long-term credit is genuinely valuable to the economy, and it is also structurally fragile, because there is no scenario in which a bank can sell a loan book fast enough to meet a simultaneous demand from every depositor.

Deposit insurance dampens the panic, but it does not remove the incentive problem. If depositors are insured, they have little reason to police how much risk the bank takes, which means the discipline has to come from somewhere else. Capital rules are that discipline: they make shareholders absorb the first losses so that insured depositors and the insurance fund are further back in the queue.

Leverage and the Systemic Externality

Leverage magnifies both sides of the outcome. A bank earning 1.1% on assets with ten times leverage produces a low double-digit return on equity, which is exactly what shareholders want. The same leverage means a modest deterioration in credit quality consumes a large share of the equity base. A 2% loss on assets is a routine bad year for a lender; against a 10% equity-to-assets ratio it removes a fifth of the bank's capital.

The externality is what makes this a regulatory matter rather than a private one. Banks fund each other, clear payments for each other, and hold overlapping exposures, so one failure transmits. A bank that fails imposes costs on institutions that made no bad decisions at all, and those costs are not priced into the failing bank's own funding. Basel exists because national regulators concluded that internationally active banks would otherwise compete by holding less capital than the system could bear.

Regulatory capital is not one thing. It is a hierarchy of instruments ranked by how reliably each absorbs losses while the bank is still operating. The Bank for International Settlements summary of the Basel III capital definition draws the key line: going-concern capital absorbs losses while the bank keeps trading, and gone-concern capital only absorbs losses once the bank has failed.

What Actually Counts as CET1

Common equity tier 1 is the top of the stack and the number everyone quotes. It is deliberately narrow: common stock and the related surplus, retained earnings, accumulated other comprehensive income, and a limited amount of qualifying minority interest in consolidated subsidiaries. Nothing in that list carries a coupon the bank is contractually obliged to pay, and nothing in it has a maturity date. That is the whole point. Losses reduce retained earnings automatically, without a negotiation, a default, or a bankruptcy filing.

Common Equity Tier 1 (CET1)

CET1 is a bank's highest-quality regulatory capital: common stock, retained earnings, and accumulated other comprehensive income, minus regulatory deductions such as goodwill and certain deferred tax assets. It absorbs losses immediately while the bank continues operating, which is why regulators measure a bank's strength primarily through the CET1 ratio rather than through total equity or total capital.

The Deductions That Surprise People

The deductions are where candidates lose points, because the accounting definition of equity and the regulatory definition diverge sharply. Goodwill and other intangibles come out in full, net of associated deferred tax liabilities, on the logic that an acquisition premium cannot pay a depositor. Deferred tax assets that depend on future profitability, particularly those from net operating loss carryforwards, are deducted entirely, because a bank that is losing money cannot use them.

Beyond those, several items are deducted only above threshold amounts, which is why capital reporting is more complicated than a single subtraction:

Accumulated other comprehensive income is the deduction question that actually moved markets. Banks in categories III and IV, broadly those between $100 billion and $700 billion in assets, may currently elect to exclude unrealized gains and losses on available-for-sale securities from regulatory capital. Silicon Valley Bank made exactly that election, which is how a bank could report a comfortable capital ratio while sitting on enormous unrealized bond losses. Removing that option for large regionals featured in the 2023 endgame proposal and it is back on the table: the March 2026 revised standardized approach proposal would eliminate the AOCI opt-out for categories III and IV, phased in over five years from 1 January 2027. Proposed, not final, as of July 2026.

Additional Tier 1 and Tier 2

Below CET1 sit two thinner layers. Additional tier 1 is perpetual preferred stock and contingent convertible instruments: no maturity, fully discretionary coupons the bank can switch off without triggering a default, and a contractual trigger that writes the instrument down or converts it into common equity if the CET1 ratio falls below a set level. It is going-concern capital, but conditionally so. The 2023 Credit Suisse rescue demonstrated the mechanism at full scale when the Swiss regulator ordered roughly $17 billion of AT1 written to zero inside the UBS takeover, a reminder that these instruments are not senior debt with a better coupon.

Tier 2 is gone-concern capital: subordinated debt with an original maturity of at least five years (amortizing for regulatory purposes over its final five years) plus a limited amount of loan loss allowance. It protects depositors in a failure but does nothing to keep a struggling bank alive, which is why regulators care far more about the CET1 ratio than the total capital ratio. Because tier 2 is subordinated debt, its pricing behaves much like the credit spectrum described in our guide to the line between investment grade and high yield bonds, with the added wrinkle that regulatory treatment, not just credit quality, drives issuance.

A capital ratio is capital divided by something. If that something were simply total assets, every bank would have an incentive to load up on the riskiest loans available, since a Treasury bill and a subordinated construction loan would consume identical capital. Basel's answer is to scale each exposure by an estimate of its riskiness before adding it up.

How a Risk Weight Works

A risk weight is a multiplier applied to an exposure amount. Under the standardized approach currently in force in the United States, codified at 12 CFR Part 217 Subpart D, US Treasury securities and central bank reserves carry a 0% weight, agency mortgage-backed securities carry 20%, prudently underwritten first-lien residential mortgages carry 50%, general corporate exposures carry 100%, and the unsecured portion of exposures more than ninety days past due carries 150%. Off-balance-sheet items such as undrawn revolver commitments are first converted into a credit-equivalent exposure using a credit conversion factor, then risk weighted like any other asset.

Risk-Weighted Assets (RWA)

Risk-weighted assets are a bank's exposures scaled by regulatory risk weights that reflect how likely each is to generate losses. A $1 million Treasury holding at a 0% weight adds nothing to RWA, while a $1 million unsecured corporate loan at a 100% weight adds the full amount. Capital ratios are expressed as capital divided by risk-weighted assets, so the risk profile of what a bank holds, not just the size of its balance sheet, determines how much capital it must carry.

Worked Example: From Assets to Risk-Weighted Assets

Take a hypothetical mid-size US bank with $200 billion of on-balance-sheet assets. Applying the weights above:

That totals $113.5 billion of credit RWA from the balance sheet. Now add $40 billion of undrawn corporate revolver commitments with a maturity beyond one year: a 50% credit conversion factor turns those into $20 billion of credit-equivalent exposure, which at a 100% risk weight adds $20 billion. Total RWA is $133.5 billion, against $200 billion of assets, giving an RWA density of roughly 67%.

Density is the number worth remembering. A trading-heavy or mortgage-heavy bank might run at 40% density, while a commercial lender might run above 70%. Two banks can hold the same assets and report different ratios purely because of what those assets are.

Standardized Approach Versus Internal Models

The largest banks do not use a single approach. Under the current US framework, the biggest institutions calculate ratios under both a standardized approach and an advanced internal-models approach, and the binding requirement is whichever produces the lower ratio, so the standardized calculation acts as a floor under the modeled result. The March 2026 endgame re-proposal, discussed below and not in force, would replace the advanced approaches for credit and operational risk with a revised standardized framework, ending the use of internal models for credit risk while retaining models in parts of the trading book. The stated aim is comparability: if two banks hold the same loan, their capital charges should not diverge because their internal models disagree.

Once you have the numerator and the denominator, the ratios follow mechanically. There are four to know, and knowing which one binds in which situation is more useful than reciting all four.

CET1, Tier 1, and Total Capital

The three risk-based ratios use the same denominator and progressively wider numerators:

The tier 1 ratio adds additional tier 1 to the numerator, and the total capital ratio adds tier 2 on top of that. The Basel minimums, in force in the United States since the 2013 capital rule took effect, are 4.5% CET1, 6.0% tier 1, and 8.0% total capital. Those numbers are almost never the operative constraint, because buffers sit on top of them, but they are the floor beneath which a bank is considered undercapitalized and subject to prompt corrective action.

Worked Example: A CET1 Ratio Under Pressure

Return to the bank above with $133.5 billion of RWA, and give it $15 billion of CET1 capital.

That is a CET1 ratio of 11.24%. Suppose its requirement is the 4.5% minimum plus a stress capital buffer of 3.2%, a total of 7.7%. Headroom is 3.54 percentage points, which on $133.5 billion of RWA equals roughly $4.7 billion of distributable excess capital.

Now stress it. A credit downturn produces $3 billion of losses, cutting CET1 to $12 billion, and $10 billion of corporate loans migrate from performing to past due, moving from a 100% weight to 150% and adding $5 billion to RWA, which rises to $138.5 billion. The new ratio is 12.0 divided by 138.5, or 8.66%. Headroom has fallen from 3.54 percentage points to 0.96, a reduction of roughly three quarters, even though the bank remains above its requirement and has lost only 1.5% of its assets.

The Leverage Ratio as a Backstop

Risk weighting has an obvious failure mode: it depends on regulators correctly estimating risk. When those estimates are wrong, as they were for sovereign debt before the euro crisis, banks can accumulate enormous exposures that consume almost no capital. The leverage ratio exists as an unweighted backstop, dividing tier 1 capital by total exposure with no risk adjustment at all.

The generic US tier 1 leverage minimum is 4.0%. Large banks additionally face the supplementary leverage ratio, which uses a broader exposure measure including off-balance-sheet commitments and derivatives, with a 3.0% minimum. Global systemically important banks face an enhanced version on top. A final rule adopted by the agencies in November 2025 and effective 1 April 2026, summarized in the OCC's bulletin on the modified eSLR standards, replaced the flat 2% enhanced buffer with a buffer equal to half the firm's method 1 surcharge, and capped the requirement for depository subsidiaries so their total sits no higher than 4%.

Capital ratios are the fastest way to expose a shallow FIG answer: Work through CET1, risk-weighted assets, and bank balance sheet questions with full worked solutions on the practice platform, start practicing interview questions for free and find the gaps before an interviewer does.

Here is the point most explanations skip: the 4.5% CET1 minimum is not what any large bank actually manages to. Buffers sit above the minimum, and because breaching a buffer triggers automatic restrictions rather than failure, the buffer stack is where the real constraint lives.

The Buffer Stack Above the Minimum

Three buffers sit on top of the CET1 minimum, all of which must be met with CET1 rather than lower-quality capital.

The capital conservation buffer is a fixed 2.5% for most banks and has been fully phased in since 2019. For US bank holding companies above $100 billion in assets it is replaced by the stress capital buffer, a firm-specific number derived from the annual supervisory stress test with a floor of 2.5%. That is the mechanism by which stress testing translates into a hard requirement rather than an advisory exercise.

The countercyclical capital buffer is a macroprudential dial regulators can raise when credit growth looks excessive and cut when conditions deteriorate, releasing capital into a downturn. The US rate has been 0% since the framework was established, and the Federal Reserve has repeatedly affirmed it there, in contrast with the UK and several European jurisdictions that run positive rates through the cycle.

The G-SIB surcharge applies only to the largest, most interconnected banks and scales with a systemic importance score built from size, interconnectedness, cross-jurisdictional activity, complexity, and reliance on short-term wholesale funding. In force since 2016, US surcharges under the method 2 calculation currently run from 1.0% to 4.5% depending on the firm.

G-SIB Surcharge

The G-SIB surcharge is an extra layer of CET1 capital that global systemically important banks must hold on top of the standard minimum and buffers. The amount scales with a supervisory score measuring the bank's size, interconnectedness, cross-border activity, complexity, and short-term wholesale funding. Its purpose is to offset the funding advantage very large banks receive from an implicit expectation of government support, and to make failure less likely at the institutions where failure would be most damaging.

Stack the pieces and a US G-SIB with a 2.5% stress capital buffer and a 3.0% surcharge faces an effective CET1 requirement of 10.0%, not 4.5%. Add the management buffer boards hold above the requirement so they never come close, and reported ratios in the 13% to 15% range make sense.

What Happens When a Bank Dips Into Its Buffers

Falling into the buffer zone is not a breach in the sense of a covenant default. Regulators designed buffers to be usable, which is the entire logic of stacking a usable cushion on top of a hard minimum. What happens instead is automatic and graduated: the further into the buffer a bank falls, the smaller the share of earnings it may distribute as dividends, buybacks, or discretionary bonuses.

The schedule set out in 12 CFR 217.11 caps the maximum payout as a percentage of the previous four quarters of eligible retained income. A bank holding its full buffer faces no restriction at all. Once inside the top quartile of the buffer the cap is 60% of eligible retained income, then 40%, then 20%, and a bank with less than a quarter of its buffer remaining may distribute nothing.

Stress Testing Sets the Real US Number

The Federal Reserve's annual supervisory stress test projects each large bank's losses, revenues, and capital under a hypothetical severely adverse scenario, and the peak-to-trough decline in the bank's CET1 ratio (plus four quarters of planned dividends) becomes its stress capital buffer, subject to the 2.5% floor. That is why the effective requirement is firm-specific: a bank with concentrated commercial real estate exposure draws a larger buffer than a payments processor with the same balance sheet size.

The 2026 stress test results published in June covered 32 large banks under a scenario assuming unemployment rising to 10%, a 30% decline in house prices, a 39% fall in commercial real estate prices, and a 58% drop in equity prices. In aggregate the banks absorbed more than $708 billion of projected losses, and the group's CET1 ratio fell from an actual 12.8% at the end of 2025 to a minimum of 11.2% before recovering to 12.7%. A decline of only 1.6 percentage points is historically mild. Note one wrinkle: the Board voted in February 2026 to hold stress capital buffer requirements at their existing levels until they are recalculated in 2027, once it has worked through revisions to its stress testing models, so the 2026 results did not reset requirements the way results normally would.

Capital answers whether a bank is solvent. Liquidity answers whether it can meet withdrawals tomorrow, and the two are not the same question: a bank can be perfectly solvent and still fail because its assets cannot be turned into cash fast enough. Basel III added two liquidity standards for the first time, both of which are in force for large US banks.

The liquidity coverage ratio requires a bank to hold enough high-quality liquid assets, mainly cash, reserves, and government securities, to cover thirty days of net cash outflows under a stress scenario, with a 100% minimum. Outflow assumptions vary by funding type: insured retail deposits are assumed to run off slowly, uninsured corporate deposits and wholesale funding much faster, which is precisely why the deposit mix on a bank's liability side matters as much as the loan mix on its asset side.

The net stable funding ratio works over a one-year horizon rather than thirty days, requiring available stable funding to be at least 100% of required stable funding. In effect it forces banks to fund long-dated illiquid assets with long-dated stable liabilities rather than overnight borrowing. The US final rule took effect on 1 July 2021 and applies to the largest institutions. Together the two standards constrain the same maturity transformation that makes banking profitable, which is why liquidity requirements, not just capital requirements, shape which businesses a bank chooses to run.

None of this is academic. Capital is the scarcest input at a bank, and the way it is allocated explains most of what large banks have done strategically over the last decade.

Balance-Sheet-Light Business Models

Every business inside a bank competes for the same capital, and the internal scorecard is return on allocated capital. A lending business that generates a 3% margin but consumes RWA at full weight can be less attractive than an advisory business that consumes almost none. That arithmetic is why the mix at large banks has shifted toward fee income: mergers and acquisitions advisory, underwriting, payments, custody, and above all asset and wealth management, which produces recurring fees against a negligible capital charge.

Our breakdown of how investment banks make money across their divisions shows how differently these businesses behave. Read it through a capital lens and the pattern is obvious: the divisions banks have expanded fastest are the ones whose revenue does not require holding assets. It is also why banks fight so hard over the market risk and operational risk components of Basel proposals, since operational risk charges fall on precisely the fee businesses that were supposed to be capital-light.

Why Banks Partner With Private Credit Funds

The most visible consequence of capital rules is the rise of private credit. A direct lending fund financed largely with equity and modest leverage faces no CET1 ratio, no stress test, and no buffer stack, so it can hold a leveraged loan at a cost of capital that a bank subject to a full risk weight plus a G-SIB surcharge cannot match. Our explainer on how direct lending funds are structured and deployed covers the vehicle itself.

Rather than compete on a playing field tilted against them, banks increasingly partner. Three structures dominate:

Significant risk transfer has grown fast enough to attract supervisory attention, and the Basel Committee published a dedicated paper on synthetic risk transfers in February 2026 examining whether the capital relief banks claim matches the risk actually transferred. The concern is circularity: if the bank also lends to the fund buying the protection, the risk has moved a shorter distance than the capital treatment implies. That question sits at the center of the debate covered in our piece on what stress in private credit means for the wider market, and it is a genuinely open supervisory issue rather than a settled one.

Working through the capital stack is one piece of a FIG interview: Download the 160-page PDF, access the IB Interview Guide covering accounting, valuation, and the technical questions that come up alongside bank regulation.

Labeling the state of play correctly is worth more in an interview than memorizing a coefficient, because the US framework is genuinely in flux and the direction of travel reversed.

What Is In Force Today

The binding rules today are the 4.5% CET1, 6.0% tier 1, and 8.0% total capital minimums; the 2.5% capital conservation buffer and the firm-specific stress capital buffer; the 0% US countercyclical buffer; G-SIB surcharges under method 2; the 4.0% tier 1 leverage and 3.0% supplementary leverage minimums, with the revised enhanced supplementary leverage standards effective 1 April 2026; and the LCR and NSFR.

What Is Proposed but Not Final

On 19 March 2026 the Federal Reserve, OCC, and FDIC issued a re-proposal of the Basel III endgame rules for category I and II firms, alongside a revised standardized approach proposal reaching a wider set of banks and a proposal to recalibrate the G-SIB surcharge published in the Federal Register on 27 March 2026. Where the 2023 version would have raised large-bank capital substantially, the agencies describe the 2026 package as modestly reducing requirements for large banks and moderately reducing them for smaller ones. Comments closed on 18 June 2026 and no final rule has been adopted as of this writing. Also proposed rather than final: eliminating the CET1 deduction for mortgage servicing assets in favor of a 250% risk weight, revised risk weight grids for residential real estate driven by loan-to-value, and the Federal Reserve's changes to its stress test models.

Capital questions are popular with financial institutions group interviewers precisely because they separate memorization from understanding. The recall version, "what is the CET1 minimum," takes five seconds. The versions that actually differentiate ask you to reason.

Expect some combination of these:

That last question is the one worth rehearsing, because it forces you to work both sides of the ratio, which is exactly what the worked example above demonstrates. If you are still building context on what these teams actually do day to day, our overview of the financial institutions group and how its coverage works sets the scene.

Get comfortable with the arithmetic and the vocabulary follows. Build one small bank balance sheet by hand, apply the weights, compute the ratio, then stress it and watch both the numerator and the denominator move. That single exercise covers most of what a FIG interviewer can reasonably ask, and it gives you something far more valuable than a memorized figure: an intuition for why a bank behaves the way it does.