Current system boundary · checked 8 October 2026
The FSB’s May 2026 private-credit review highlights deepening connections between private funds, banks and insurers. FSB primary report. This is consistent with examining the resilience of financing across institutions rather than treating a bank capital ratio as the complete system diagnosis.
The historical 2023 literature average and 2021–2022 bank sample below remain historical research inputs. No statement here establishes the current legal status of a Basel proposal or a current optimum for an individual bank. Capital definitions, implementation dates and jurisdiction-specific rules require the applicable current rulebook.
Higher bank capital can make the financial system safer, but the economic case for the next increase depends on the protection already created by other reforms and on where financial activity moves. An archived April 2023 review of post-crisis regulation argued that large-bank capital ratios were near the average optimum reported in the literature it surveyed. The useful policy conclusion is to evaluate interacting safeguards and financing channels together, rather than treating a higher capital ratio as a complete measure of financial resilience.
This report retains the historical review’s principal figures and disagreements. It does not describe the present legal status of any capital proposal, and it does not convert an average across research papers into a universal requirement for individual banks.
What the capital ratio measures
A risk-based capital ratio compares eligible loss-absorbing capital with risk-weighted assets. Risk-weighted assets are an adjusted measure of exposures, not simply the accounting balance sheet. A leverage ratio uses a different denominator and is intended to constrain balance-sheet expansion without relying entirely on risk weights.
The two constraints can bind at different times. A bank with apparently low-risk assets can be constrained by leverage requirements, while another with a smaller balance sheet can be constrained by credit or trading risk weights. The amount of capital is therefore only one part of the diagnosis; its definition, the denominator and the bank’s binding constraint matter.
Capital also performs a different function from liquidity. Capital absorbs losses. Liquidity allows obligations to be met when due. More of one does not automatically solve a shortage of the other. A bank can have positive long-run asset value and still face a funding run if it cannot obtain cash quickly enough. Conversely, temporary liquidity cannot repair a business whose assets are worth less than its liabilities.
Figure 01 · Qualitative comparison
Capital resilience has more than one measurement
| Lens | What it measures | What still needs examination |
|---|---|---|
| Risk-based capital ratio | Eligible capital relative to risk-weighted exposures | Risk weights, models and the treatment of losses |
| Leverage ratio | Capital against a broader exposure denominator | Composition, liquidity and concentration of exposures |
| Economic and funding view | Value sensitivity and the ability to meet cash demands | Accounting recognition, deposit behaviour and market access |
The historical numerical comparison
The archived review reported an average estimated optimal Tier 1 ratio around 15.5% across the literature it considered. It compared that with simple-average actual Tier 1 ratios of 15.5% in the fourth quarter of 2021 and 15.2% in the fourth quarter of 2022 for a selected group of 33 large banking companies.
Historical capital ratios and the sample used for comparison
| Historical figure | Reported value | Interpretation |
|---|---|---|
| Average optimum across selected literature | About 15.5% | A synthesis across models with different assumptions |
| Selected large-bank average, Q4 2021 | 15.5% | Actual historical sample average |
| Selected large-bank average, Q4 2022 | 15.2% | Actual historical sample average |
| Number of banking companies in the comparison | 33 | Sample scope, not the whole banking system |
The proximity of these figures supports further examination of marginal benefits. It does not establish that every sampled bank held the right amount of capital. A simple average conceals dispersion, business-model differences, stress exposures and variation in asset quality. It also gives equal numerical weight to institutions that may have very different economic footprints.
There is a separate comparability problem: papers can use different capital definitions, target populations and welfare functions. Averaging their optimal ratios does not remove those differences. A useful policymaker’s table would show the definition, sample, crisis-loss assumption and treatment of lending substitution beside each estimate.
The benefit channel: lower crisis probability and severity
More equity can absorb losses before depositors and other creditors are impaired. This can lower the chance of disorderly failure and reduce the external costs imposed on the wider economy. Those costs include disrupted payments, curtailed lending, asset fire sales and the loss of confidence that spreads beyond the original institution.
The benefit is not restricted to avoiding outright insolvency. A bank that can continue lending through a downturn may reduce the amplification of the initial shock. Credible loss-absorbing resources can also improve funding terms because creditors face less risk.
However, the benefit of a further capital increase depends on the starting point. Moving a fragile institution away from an obvious danger zone is not the same intervention as adding another percentage point to an already strongly protected balance sheet. The shape of the marginal-benefit curve matters more than the proposition that capital is beneficial in general.
The archived review questioned whether some models sufficiently incorporated the cumulative effects of stress testing, liquidity rules, supervisory changes, resolution arrangements and counterparty-risk reforms. If those measures already reduce crisis probability, adding capital may still help, but the incremental gain could be smaller than in a model that omits them.
The cost channel is not simply the price of equity
Equity generally requires a higher expected return than senior debt because equity bears losses first. But it does not follow that replacing debt with equity raises the overall funding cost by the entire difference between their initial required returns. Lower leverage can reduce the risk of both claims, and therefore their required returns.
A simplified weighted cost of funding is
where and are equity and debt weights, and their required returns, and the applicable tax rate in this simplified expression. Changing capital structure can change the weights and the required returns simultaneously. A calculation that changes only the weights can overstate the resulting increase.
Real banks face additional frictions. Tax treatment, guarantees, deposit franchises, issuance costs, market structure and management incentives can prevent a perfect offset. The practical question is the size of the remaining cost and how it is transmitted to borrowers, counterparties or shareholders.
The archived review acknowledged research suggesting that safer banks experienced lower costs of debt and equity after reform announcements. It also discussed studies concerned about lending rates and output. Retaining both channels is essential. Removing the offset would bias the case against capital; ignoring frictions would make the costs disappear too easily.
Figure 02 · Conceptual framework
More capital affects both resilience and funding economics
How the parts connect
- Higher equity fundingloss absorptionLower creditor exposure
- Lower creditor exposureclaim risk changesRequired return adjustment
- Lower creditor exposurecrisis channelResilience benefit
- Higher equity fundingfunding mixCredit economics
- Required return adjustmentrisk-adjusted funding pricesCredit economics
Financial activity can migrate
When a bank changes the price or availability of credit, borrowers may seek financing elsewhere. Non-bank lenders, securities markets and private funds can substitute for some bank activity. That substitution can preserve credit supply and increase competition, but it may also move leverage, liquidity risk and underwriting weaknesses outside the original regulatory perimeter.
The archived review emphasised that many models did not fully capture the growth and behaviour of non-bank intermediation. A narrow bank-only measure can show an improvement even while vulnerabilities accumulate elsewhere. Equally, a model that assumes all reduced bank lending vanishes can overstate the economic cost if borrowers have viable alternatives.
A complete assessment distinguishes activities by substitutability. A large issuer with market access has options that a smaller borrower may lack. A committed bank line is not identical to a fund that can change deployment priorities. Financing available in normal conditions may also disappear under stress.
The relevant outcome is the resilience of credit provision across the system. It is not enough to show that one regulated segment has become smaller or better capitalised.
Trading liquidity is a separate transmission channel
Capital requirements can influence dealers’ willingness to hold inventory and intermediate risk. If balance-sheet capacity becomes more expensive, market-making terms may change. Wider bid-offer spreads, lower depth and a higher liquidity premium can raise financing costs even when formal bank lending is unaffected.
This does not mean every increase in capital reduces liquidity by the same amount. The effect depends on the activity, the binding constraint, competition, netting, hedging and the availability of other intermediaries. It can also differ sharply between normal markets and stress.
Corporate bonds illustrate the interaction. They can replace bank loans, cushioning a reduction in lending. But if the same reform reduces dealer capacity, the bond alternative may become more expensive or less reliable. A model that counts substitution benefits without considering market liquidity can be incomplete.
The archived review therefore argued for examining bank capital, trading inventories, non-bank participation and financing costs as a connected system. This is a more demanding exercise than applying one lending-rate elasticity to the entire economy.
Figure 03 · Conceptual framework
Risk can migrate beyond the regulated bank balance sheet
How the parts connect
- Bank constraint changescredit supply responseBorrower response
- Bank constraint changesdealer liquidity channelSystem resilience review
- Borrower responsesubstitutionNon-bank or market finance
- Non-bank or market financedifferent risk locationNew system exposures
- New system exposuresfollow the full chainSystem resilience review
Interest-rate risk and the recognition of losses
The historical discussion of 2023 banking stress distinguished the proximate problem of interest-rate risk and funding pressure from a simple shortage of nominal regulatory capital. It examined whether accounting and capital treatments made some losses less visible in headline ratios.
One issue was the treatment of accumulated other comprehensive income on certain securities. Differences in whether those valuation movements flow into regulatory capital can make similar economic exposures appear different in reported capital measures. That is a measurement and incentive question as well as a question about the required ratio.
Stress scenarios also matter. A framework concentrated on a downturn with falling rates can fail to expose a balance sheet vulnerable to a different combination of rising rates, valuation losses and deposit withdrawals. Adding a percentage point of capital without examining the scenario design may leave the most informative weakness unaddressed.
The lesson is not that capital is irrelevant to a run. Losses, confidence and liquidity can reinforce one another. The lesson is that policy should identify the mechanism before choosing the remedy.
How to compare competing policy estimates
A useful comparison starts with a common baseline. Each estimate should disclose which reforms are already embedded, which institutions are covered, how risk weights are treated and whether the analysis concerns a transition or a permanent state.
It should then identify the crisis channel. Does the model reduce failure probability, loss severity or both? Does it capture correlated distress and fire sales? Are the assumed social costs of crisis consistent with the sample and time period?
The financing channel requires equal care. The analysis should show how funding costs change, how much is passed through, which borrowers can substitute and how non-bank and market-liquidity effects enter the calculation. A headline output-loss estimate without those assumptions is difficult to interpret.
Finally, the estimate should be treated as conditional. A range of model results is more informative than a falsely precise single ratio, particularly when their assumptions differ. Observed averages can provide a reference point, but they cannot replace institution-level stress evidence or system-wide analysis.
The economic conclusion
The historical review’s call was to consider whether additional capital requirements were justified when large-bank averages already stood near its surveyed literature’s estimated optimum. That call is preserved as an argument from its time, not endorsed as a current calibration.
The stronger general principle is that resilience comes from the interaction of capital, liquidity, risk measurement, supervision and the structure of financial intermediation. A good reform improves the system’s capacity to absorb shocks while recognising where costs and risks are transferred. It should be assessed by that outcome, rather than by whether one visible ratio increases.
