Reader question: How does the Basel standardised approach turn an ordinary bank exposure — a loan, mortgage, guarantee or undrawn commitment — into risk-weighted assets, and where can the calculation go wrong?
The short answer is that the standardised approach is a classification-and-weighting algorithm. It does not ask a bank to estimate every borrower’s probability of default inside an internal portfolio model. Instead, it maps each exposure into the correct regulatory class, determines the exposure amount, converts eligible off-balance-sheet amounts with a credit conversion factor, recognises permitted credit-risk mitigation, assigns the prescribed risk weight, and then aggregates the resulting risk-weighted assets.
What this page does — and does not do
This is a mathematics and computational-risk article about how a regulatory calculation works. It is not a lending recommendation, a capital-management instruction, or financial advice. The Basel Committee publishes international standards; binding requirements come from each jurisdiction’s implementation. A correct implementation must therefore record the rule version and local regulatory perimeter rather than assume that a Basel text is automatically law everywhere.
The algorithm in one line
For a simple exposure, the core operation is:
RWA = exposure amount after applicable CCF and credit-risk mitigation × prescribed risk weight.
For a portfolio, the bank sums this quantity across exposures and combines it with the relevant treatments for counterparty credit risk, securitisation and other categories that have their own rules.
The simplicity of the multiplication is deceptive. Most serious errors happen before the multiplication: in classification, exposure measurement, conversion, collateral recognition, property treatment, rating use or rule-version selection.
Step 1: classify the exposure before calculating anything
The standardised approach has different treatments for sovereigns, public-sector entities, banks, covered bonds, corporates, retail exposures, real estate, subordinated debt, defaulted exposures and other categories. An exposure cannot be assigned a meaningful risk weight until its regulatory class is known.
This is a computational lesson in type safety. A mortgage is not merely “a loan with property.” The algorithm must test whether the real-estate requirements are met, whether repayment is materially dependent on cash flows generated by the property, whether the property is residential or commercial, and whether the relevant loan-to-value conditions and valuation rules are satisfied. A corporate exposure cannot be treated as regulatory retail merely because the borrower is small.
Step 2: determine the exposure amount
For an on-balance-sheet loan, the starting exposure is generally its regulatory exposure amount after the adjustments required by the framework. For an off-balance-sheet promise to lend or support a customer, the notional amount is not always the amount used for RWA. The algorithm first converts that promise into a credit-equivalent exposure.
In the current Basel Framework, CRE20 applies a 100% credit conversion factor to specified direct credit substitutes, 50% to note-issuance and revolving-underwriting facilities and certain transaction-related contingent items, 40% to commitments unless they qualify for a lower factor, 20% to qualifying short-term self-liquidating trade letters of credit, and 10% to commitments that are unconditionally cancellable under the stated conditions. National supervisors may require a higher factor where practical constraints weaken a bank’s ability to cancel.
A compact representation is:
credit-equivalent exposure = off-balance-sheet notional × CCF.
If a $2 million undrawn commitment receives a 40% CCF, the converted amount is $800,000 before any further applicable treatment. If the same contractual label is incorrectly mapped to a different CCF category, the error propagates directly into RWA.
Step 3: apply eligible credit-risk mitigation without inventing protection
Collateral, guarantees and credit derivatives can change the regulatory exposure or risk weight only when they satisfy the framework’s eligibility, legal certainty, maturity, currency and operational requirements. The mathematical idea is straightforward: recognised protection can move part of an exposure toward a different effective loss profile. The regulatory implementation is deliberately more restrictive than “there is some collateral, therefore risk is lower.”
A robust engine therefore stores the unmitigated exposure and the mitigation transformation separately. That makes it possible to answer two audit questions: what was the exposure before protection? and exactly which rule allowed the protection to reduce the capital calculation?
Step 4: assign the prescribed risk weight
Once the exposure class and adjusted amount are known, the standardised approach applies a prescribed risk weight or a rule for deriving one. The appropriate weight can depend on variables such as counterparty category, external rating where permitted, due-diligence requirements, loan-to-value ratio, whether repayment is materially dependent on the property, or whether an exposure is in default.
For qualifying residential real-estate exposures where repayment is not materially dependent on property cash flows and the whole-loan approach applies, the current Basel table assigns risk weights of 20%, 25%, 30%, 40%, 50% and 70% across successive LTV bands of up to 50%, 50–60%, 60–80%, 80–90%, 90–100% and above 100%. Other real-estate cases use different treatments. Copying this table onto every mortgage would therefore be a category error.
A worked example: why LTV is an algorithmic variable
Suppose an illustrative qualifying residential mortgage has a loan exposure of $400,000 and an eligible property value of $600,000. Its LTV is:
LTV = 400,000 / 600,000 = 66.7%.
If the exposure satisfies the conditions for the whole-loan residential treatment described above and is not materially dependent on property cash flows, 66.7% falls in the 60–80% band. The corresponding 30% risk weight gives:
RWA = 400,000 × 30% = $120,000.
This example is deliberately narrow. Change the property eligibility, lien structure, valuation basis, dependence test, jurisdictional implementation or exposure status, and the correct treatment may change. The arithmetic can remain flawless while the regulatory answer becomes wrong.
Why the standardised approach is still an algorithm
“Standardised” does not mean “non-computational.” A production implementation resembles a rule engine with typed inputs, decision branches and validation constraints. Typical inputs include exposure class, counterparty type, balance-sheet or off-balance-sheet status, notional amount, CCF category, external rating where applicable, collateral and guarantor attributes, property value, lien information, LTV, default status, currency and maturity.
The outputs are not only final RWA. A good system should also return the intermediate lineage: selected exposure class, converted exposure, mitigation recognised, final exposure amount, risk-weight band, rule citation and calculation date. Without that lineage, a number may be impossible to reproduce.
Assumptions hiding underneath the calculation
- Classification is correct. Regulatory classes are treated as mutually coherent even when commercial product labels are ambiguous.
- Inputs are current. Property values, ratings, default status and commitment features have not silently become stale.
- Legal protection is enforceable. A guarantee or collateral arrangement is not counted merely because it exists in a database.
- Rule version is controlled. The calculation uses the framework and local implementation effective for the reporting date.
- Aggregation is complete. Exposures are neither omitted nor counted twice when systems feed the regulatory engine.
Evidence polarity: what supports the result, and what argues against it?
Evidence for a result includes a traceable contract, reconciled exposure amount, verified counterparty and property attributes, a documented exposure-class decision, an applicable CCF, valid mitigation evidence and a direct mapping to the current rule text.
Evidence against confidence includes missing lien data, a property valuation outside permitted conditions, a guarantee with uncertain enforceability, a rating that does not satisfy due-diligence requirements, an undrawn line mislabelled as cancellable, unexplained overrides, or a result that cannot be independently reconstructed from source data.
This polarity matters because regulatory calculation is not a one-way search for a number. Good control actively looks for facts that would force a different branch of the algorithm.
Failure modes and counterexamples
Wrong class, right arithmetic. A retail exposure classified as corporate can produce a perfectly calculated but incorrect RWA.
CCF drift. An older implementation may retain a historical conversion factor after the framework changes. Unit tests that only compare totals may miss the rule-version error.
Bad LTV denominator. Using an ineligible or stale property value can move an exposure into the wrong risk-weight band.
Mitigation double counting. The same collateral benefit can accidentally be reflected in more than one upstream system and then again in the capital engine.
External-rating automation without due diligence. A rating is not a magic lookup key. The framework includes operational and due-diligence conditions, and local rules govern how external ratings may be used.
False comparability. Two banks with identical commercial loans may report different standardised RWA if the legally relevant collateral, jurisdiction, exposure classification or regulatory implementation differs.
Diagnostics: how to test the weak links
- Recalculate a sample exposure independently from raw source fields and compare every intermediate step.
- Run boundary tests just below and above LTV bands, CCF categories and rating mappings.
- Compare on-balance-sheet totals with the general ledger and off-balance-sheet totals with commitment systems before applying weights.
- Re-run the portfolio with credit-risk mitigation removed; unexpected increases or decreases can expose sign and mapping errors.
- Track unexplained movements in RWA by separating volume, mix, rating, LTV, CCF, mitigation and rule-change effects.
- Maintain negative tests: intentionally feed an ineligible guarantee, stale property value or incompatible exposure type and verify that the engine rejects it.
What would falsify confidence in the calculation?
Confidence should fall sharply if the same exposure cannot be reproduced from its stored inputs; if small irrelevant metadata changes alter the result; if changing a valid LTV across a regulatory boundary fails to change the selected band; if an off-balance-sheet item ignores its CCF; if ineligible collateral reduces RWA; or if the engine’s rule citation does not match the effective reporting date.
At portfolio level, large unexplained differences between regulatory disclosures, finance balances and source-system exposure totals are also falsifiers. The goal is not to force the numbers to agree. It is to explain why they should or should not agree.
Alternatives answer different questions
The standardised approach is not a general theory of credit risk. Internal-ratings-based capital, expected-credit-loss accounting, economic-capital models, stress tests and loan-pricing systems ask different questions and use different assumptions. A standardised risk weight should not be interpreted as a borrower’s literal probability of default or as a forecast of loss.
That distinction is especially important when comparing this page with the Basel IRB credit-risk algorithm. The IRB framework transforms estimated PD, LGD, EAD and maturity through a nonlinear portfolio model. The standardised approach instead places much more of the structure in prescribed exposure classes and weights.
How this connects to the surrounding mathematics
Start with the broader bank-capital model. For undrawn facilities, connect the CCF step to revolving-credit utilisation and exposure at default. Then compare the standardised calculation with IRB RWA. The two meet again in the Basel output floor, where standardised RWA becomes a constraint on modelled RWA.
Verification and update triggers
A defensible implementation should preserve the reporting date, Basel chapter/version, national rule source, exposure-class decision, source-system identifiers, CCF, mitigation method, risk weight and reviewer or automated-control outcome. Re-check the logic when the Basel Committee changes CRE20 or related credit-risk-mitigation chapters, when a jurisdiction changes implementation, when property or rating rules change, when new product types enter the book, or when reconciliations reveal unexplained shifts.
Primary and high-quality references
- Basel Committee on Banking Supervision, CRE20 — Standardised approach: individual exposures.
- Basel Committee on Banking Supervision, Basel Framework, consolidated framework and current effective-version notes.
- Bank for International Settlements, Revised credit risk framework — Executive Summary.
- Basel Committee, DIS40 — Credit-risk disclosure templates, including post-CCF and post-credit-risk-mitigation exposure reporting.
Educational boundary: This article explains regulatory mathematics and control logic. It does not determine how any particular bank should classify an exposure, calculate binding local capital requirements, extend credit, or make an investment decision.