Reader question: Two securities can be backed by the same pool of loans, yet one attracts a modest regulatory risk weight while another can reach 1250%. How does the Basel framework turn a tranche’s position in the loss waterfall into bank capital?
The key idea is that securitisation capital is not determined only by the average credit quality of the underlying pool. It also depends on where the bank’s tranche starts absorbing losses, where it is fully exhausted, how thick the tranche is, how long the exposure remains at risk, and which Basel approach is available. The framework therefore combines pool-level credit risk with tranche geometry.
At a high level, the calculation route is:
underlying pool risk + tranche attachment/detachment + maturity/rating inputs → SEC-IRBA, SEC-ERBA or SEC-SA → risk weight → RWA.
What this page owns — and what it does not
This page owns the public regulatory-capital transformation for securitisation exposures. It does not replace the underlying Basel standardised credit-risk calculation, mortgage-backed-security cash-flow projection, or the Basel output floor. Those are adjacent owners.
This is prudential mathematics and computational education. It is not a rating opinion, transaction recommendation, legal interpretation or personalized financial advice.
Why securitisation needs a different capital algorithm
Suppose a 100 million loan pool is split into three tranches:
- equity: first 0–3% of pool losses;
- mezzanine: losses from 3–7%;
- senior: losses above 7%.
The same underlying pool supports all three securities, but their loss exposure is radically different. The equity tranche is hit by the first loss. The senior tranche is untouched until junior protection is exhausted.
A simple “underlying pool average risk weight × tranche balance” would ignore that structure. Basel therefore models the tranche’s attachment point A and detachment point D.
Attachment point A
The attachment point is the fraction of pool losses that can occur before the bank’s securitisation exposure begins to absorb principal loss.
For a tranche covering pool losses from 3% to 7%:
A = 0.03.
Below 3% cumulative pool loss, the tranche is protected by more junior positions.
Detachment point D
The detachment point is the pool-loss fraction at which the tranche is completely exhausted.
For the same 3–7% tranche:
D = 0.07.
At 7% cumulative pool loss, the tranche has lost 100% of its principal, ignoring recoveries and structural details already reflected in the defined waterfall.
Tranche thickness
The tranche thickness is:
T = D − A.
For the 3–7% tranche:
T = 0.07 − 0.03 = 0.04.
A thinner tranche concentrates a given slice of portfolio loss into a smaller amount of capital. All else equal, thinner non-senior tranches can therefore be more sensitive to model and rating changes.
The Basel approach hierarchy
The Basel securitisation framework does not use one formula for every position. Subject to eligibility, information and jurisdictional implementation, the main routes are:
- SEC-IRBA — internal-ratings-based approach for an eligible IRB pool where required inputs are available;
- SEC-ERBA — external-ratings-based approach where the jurisdiction permits it and an eligible external or inferred rating is available;
- SEC-SA — standardised approach using the standardised capital requirement of the underlying pool and delinquency information.
If the required approach cannot be used and no permitted fallback applies, the position can receive a 1250% risk weight.
Why 1250% appears in Basel
A 1250% risk weight produces an 8%-of-RWA capital requirement equal to the full exposure amount:
8% × 1250% = 100%.
So, before considering the bank’s actual capital stack, 1250% is economically similar to requiring one unit of regulatory capital for each unit of exposure.
It is therefore a useful diagnostic boundary: when information, due diligence or method eligibility collapses, Basel can move the exposure to an extremely conservative treatment rather than allow missing data to reduce capital.
SEC-IRBA: start from KIRB
SEC-IRBA begins with the capital requirement that the underlying pool would have generated under the IRB framework had the assets not been securitised:
KIRB = underlying-pool IRB capital requirement as a fraction of pool exposure.
The approach then combines KIRB with:
- A — attachment point;
- D — detachment point;
- p — a supervisory parameter driven by portfolio characteristics;
- other prescribed inputs such as maturity and, where relevant, effective number of exposures and pool LGD.
The supervisory formula transforms these inputs into a capital requirement per unit of securitisation exposure. Multiplying by 12.5 converts that capital requirement into a risk weight before applying the exposure amount.
Why the location of KIRB relative to A and D matters
Think of KIRB as a pool-level capital-loss benchmark.
If a tranche lies entirely below KIRB, it is structurally junior to a large part of the pool’s capital requirement and can attract very high capital. If it lies well above KIRB, substantial subordination protects it.
The formula smooths this relationship, but the geometry is the important intuition:
pool risk determines the loss scale; A and D determine where the tranche sits on that scale.
SEC-SA: KSA and delinquency W
SEC-SA performs a similar structural transformation when the underlying pool is treated under the standardised credit-risk framework.
Basel defines:
KSA = 8% × weighted-average standardised risk weight of the underlying pool.
For example, if the weighted-average underlying risk weight is 100%:
KSA = 0.08.
SEC-SA also uses:
W = delinquent underlying exposures / total underlying exposures.
Basel’s definition of delinquent exposures includes specified 90-days-past-due, default, bankruptcy, insolvency, foreclosure and similar conditions.
Why delinquency W matters
A pool with the same original risk weights can deteriorate after securitisation.
W gives SEC-SA a direct observable measure of pool stress. The approach combines KSA and W into an adjusted pool capital input before applying the supervisory tranche formula.
If delinquency information is unavailable for too much of the pool, the framework becomes conservative. Under the Basel standard, if the bank does not know delinquency status for more than the permitted threshold, the securitisation exposure receives 1250%.
SEC-ERBA: rating, seniority, maturity and thickness
SEC-ERBA uses an eligible external rating or inferred rating as the main credit-quality input.
For long-term ratings, Basel risk weights depend on:
- rating grade;
- whether the tranche is senior;
- tranche maturity;
- for non-senior tranches, tranche thickness.
This means two tranches with the same rating can receive different risk weights if one is longer-dated or structurally thinner.
Maturity increases exposure to model and migration risk
A one-year tranche and a five-year tranche with the same current rating do not expose the bank to the same amount of future credit migration and structural uncertainty.
Basel therefore adjusts SEC-ERBA risk weights across maturity. In the current framework, tranche maturity is floored at one year and capped at five years for this purpose.
Maturity can be determined from contractual cash-flow timing where conditions are met or from final legal maturity under the prescribed method.
A maturity engine must use the securitisation rules, not a generic bond maturity
For a commitment or liquidity facility, the period of potential exposure can extend beyond the commitment itself because a future draw may expose the bank to assets with longer maturities.
A generic “maturity date minus today” calculation can therefore be wrong for securitisation capital.
Simple, Transparent and Comparable securitisations
Basel and IOSCO developed criteria for simple, transparent and comparable (STC) securitisations. Where a transaction satisfies both the STC criteria and the additional capital-treatment requirements, the framework can apply more favourable risk weights.
This is not a statement that STC securitisations are risk-free. It is a regulatory recognition that simpler structures, better transparency and comparable asset characteristics can reduce some structural and modelling uncertainty.
Risk-weight floors matter
The framework includes minimum risk weights so that even very senior, highly rated positions do not fall toward zero capital solely because of model outputs.
For example, the Basel SEC-ERBA table currently shows a 15% minimum for many non-STC senior securitisation exposures and can allow a 10% floor for qualifying senior STC exposures.
The exact floor depends on approach, tranche and STC status, so a rules engine should not apply one universal number.
Caps prevent securitisation from creating nonsensical capital in some cases
Basel also includes caps on securitisation capital in specified circumstances.
One intuition is that a bank holding a very senior securitisation position should not always have to hold more capital than it would have needed for the relevant underlying exposures simply because they were securitised.
But caps have eligibility conditions and do not rescue a position that fails due-diligence or other operational requirements.
Due diligence is part of the algorithm
A bank cannot safely treat securitisation capital as a pure spreadsheet formula.
Basel requires banks to understand material structural features and risk characteristics of their securitisation exposures and underlying pools. Failure to meet due-diligence requirements can trigger a 1250% risk weight.
This creates a public computational lesson:
missing understanding is itself a risk input.
Traditional versus synthetic securitisation
In a traditional securitisation, assets or their cash flows are transferred into a structure whose tranches absorb different layers of loss.
In a synthetic securitisation, the assets may remain on the originator’s balance sheet while credit risk is transferred through guarantees, credit derivatives or funded credit protection.
Both can fall under the securitisation framework, but recognition of risk transfer has specific operational conditions. A model must not assume that adding a credit derivative automatically grants capital relief.
Resecuritisation is different again
A resecuritisation contains at least one securitisation exposure in the underlying pool and retranches that risk.
Because model uncertainty and tail dependence can compound, Basel applies more conservative treatment to resecuritisation exposures. A security master therefore needs to distinguish ordinary securitisation from resecuritisation before selecting parameters.
Worked structural example
Assume a 1 billion underlying pool with a mezzanine tranche from 4% to 9%.
Then:
A = 4% = 0.04.
D = 9% = 0.09.
Thickness = 5% = 0.05.
The tranche principal is approximately:
1 billion × 5% = 50 million.
If pool losses reach 3%, the tranche has not yet absorbed principal loss.
If pool losses reach 6%, losses have penetrated 2 percentage points into the tranche, or 40% of its 5-percentage-point thickness.
If pool losses reach 9%, the tranche is fully exhausted.
The regulatory capital formula does not simply use these realised-loss states, but A and D encode exactly this structural position.
Inputs and outputs
A robust securitisation-capital engine can require:
- jurisdiction and Basel implementation version;
- exposure type and securitisation/resecuritisation flag;
- underlying-pool balances and capital treatment;
- IRB eligibility and KIRB inputs where applicable;
- standardised underlying risk weights and KSA;
- delinquency ratio W;
- attachment point A;
- detachment point D;
- tranche maturity;
- seniority and thickness;
- eligible external or inferred rating;
- STC status;
- due-diligence and operational-requirement flags;
- exposure amount after permitted credit-risk mitigation.
Outputs can include selected approach, capital requirement per unit, risk weight, RWA, applicable floor/cap, 1250% fallback reason and validation diagnostics.
Evidence polarity: what supports confidence?
Evidence for a correct result includes A and D that reconcile to the legal waterfall, underlying pool balances that reconcile to trustee or servicer data, KSA/KIRB that reproduce the uns securitised capital calculation, ratings that satisfy operational requirements, maturity calculated under the securitisation convention, and approach selection that follows the jurisdiction’s hierarchy.
Evidence against confidence includes A greater than D, tranche thickness that does not reconcile to principal, missing delinquency data, a senior-tranche flag inconsistent with deal documentation, an expired rating, STC status with no supporting criteria, or capital that falls after the pool deteriorates without a structural reason.
Counterexample: same rating, different capital
Two BBB tranches can receive different SEC-ERBA risk weights because one is senior and one non-senior, because their maturities differ, or because the non-senior tranches have different thickness.
Therefore “rating → one Basel risk weight” is false for securitisations.
Counterexample: same pool, different capital
A first-loss tranche and a senior tranche reference the same underlying assets, but A and D place them at opposite parts of the loss waterfall.
Therefore “pool risk → one risk weight for every tranche” is also false.
Counterexample: a model-friendly pool can still receive 1250%
If a bank lacks required information, fails due diligence, or cannot use the necessary approach, a mathematically attractive estimate of expected loss does not override the regulatory fallback.
Method eligibility is a hard gate, not another soft model input.
Counterexample: capital relief does not prove economic risk disappeared
A securitisation can transfer enough risk to qualify for regulatory recognition while the originator still retains servicing, liquidity, reputational or residual economic exposures.
Regulatory capital treatment and total economic risk remain different questions.
Weak links in implementation
Waterfall mapping error. Senior and junior balances are mapped incorrectly into A and D.
Pool denominator mismatch. Tranche balances and underlying balances come from different dates.
Approach hierarchy error. SEC-SA is used even though SEC-IRBA is required and available, or SEC-ERBA is used where local rules do not permit it.
Maturity shortcut. Final legal maturity is replaced by a generic duration field without checking Basel conditions.
Delinquency gap. Unknown W exceeds the permitted threshold but the model still calculates SEC-SA.
STC stale flag. A transaction remains marked STC after criteria cease to be met.
Resecuritisation misclassification. A retranched securitisation is treated as an ordinary securitisation.
Floor/cap ordering error. A cap is applied before a mandatory minimum or in a case where the cap is not allowed.
Diagnostics: how to test the engine
- A/D geometry test: require 0 ≤ A < D ≤ 1.
- thickness reconciliation: D − A should reconcile to tranche principal divided by pool principal, subject to deal-specific adjustments.
- loss penetration test: a 3–7% tranche should be untouched below 3% pool loss and fully exhausted at 7%.
- approach-selection test: toggle IRB eligibility, ratings availability and jurisdiction to verify the hierarchy.
- SEC-SA data test: push unknown delinquency above the permitted threshold and require conservative fallback.
- maturity test: enforce the one-year floor and five-year cap where the Basel rule applies.
- rating-maturity test: longer maturity should not accidentally receive the one-year SEC-ERBA table value.
- STC test: qualifying and non-qualifying tranches should use the correct floors/tables.
- 1250% test: due-diligence failure should override otherwise favourable model inputs.
- RWA arithmetic test: exposure × risk weight must reproduce reported securitisation RWA.
What would falsify confidence?
Confidence should be withdrawn if tranche geometry cannot be reconciled to legal documents; if the model produces A ≥ D; if required delinquency or rating inputs are stale; if a due-diligence failure does not trigger the prescribed conservative treatment; if the selected approach contradicts the current local rulebook; or if an independent implementation using the same inputs produces a materially different risk weight.
Alternatives and limits
Economic models can estimate tranche expected loss using Monte Carlo simulation, copulas, CreditRisk+ style portfolio distributions or loan-level cash-flow models. Those tools can be useful for pricing and risk management.
Basel securitisation capital has a different purpose: it creates a prudential, comparable regulatory treatment with model constraints, floors, operational requirements and conservative fallbacks. A pricing model is therefore not a substitute for the Basel capital algorithm.
How this connects to the surrounding knowledge estate
The underlying pool can be measured under the standardised credit-risk algorithm. Loan and mortgage cash flows connect to MBS cash-flow projection. Portfolio loss distributions connect conceptually to CreditRisk+ portfolio loss modelling. The resulting RWA ultimately feeds output-floor and capital-ratio calculations.
Verification and update triggers
Preserve the Basel/local rule version, transaction legal documents, pool snapshot, tranche balances, ratings, maturity method, STC evidence, KIRB/KSA inputs and due-diligence status. Revalidate after pool replenishment, amortisation, rating migration, tranche write-down, restructuring, STC-status change, legal amendment or regulatory implementation change.
Primary and high-quality references
- Basel Committee on Banking Supervision, CRE40 — Securitisation: general provisions.
- Basel Committee, CRE41 — Securitisation: standardised approach (SEC-SA).
- Basel Committee, CRE42 — Securitisation: external-ratings-based approach (SEC-ERBA).
- Basel Committee, CRE44 — Securitisation: internal-ratings-based approach (SEC-IRBA).
- Basel Committee and IOSCO, Criteria for identifying simple, transparent and comparable securitisations.
Educational boundary: This article explains public Basel securitisation-capital mechanics. It does not determine any bank’s legal capital requirement or the investment merit of a securitisation.