What it means
Banking rules require every lender to hold capital against the risk that borrowers do not repay. The simplest method assigns fixed risk weights set by regulators, but larger banks can apply to use their own internal estimates instead, which usually produces a lower and more risk-sensitive number.
There are two internal ratings-based levels. Under the foundation approach the bank estimates only the probability of default and accepts supervisory values for everything else; under the advanced approach it estimates loss given default and exposure at default as well.
Those three inputs drive everything. Probability of default is the chance the borrower fails within a year, loss given default is the share of the exposure not recovered after collateral and workout, and exposure at default is what will actually be owed when trouble arrives, including undrawn commitments likely to be drawn.
The commercial appeal is capital efficiency. A bank that can demonstrate strong recoveries on well-secured lending may justify a much lower risk weight than the standard rules would impose, releasing capital to support more lending or higher returns to shareholders.
The cost is heavy supervisory scrutiny. Approval requires years of clean default and loss data, independent validation, board-level governance, and use of the models in real lending decisions rather than only for capital reporting.
Post-crisis reforms also introduced an output floor that limits how far internal models may fall below the standardised result.
In practice
Real-world examples.
Example
A large retail bank uses advanced IRB for its residential mortgage book, where decades of loss data show very low losses on low loan-to-value lending. Its modelled loss given default is far below the supervisory figure, so mortgage capital requirements fall materially and pricing on prime mortgages becomes more competitive.
Example
A corporate lender builds internal estimates for exposure at default on revolving credit facilities, showing that distressed borrowers typically draw a high share of undrawn limits before failing. The model raises the capital held against committed but undrawn lines, and the bank responds by repricing commitment fees.
Example
A mid-sized bank applies for advanced IRB approval and is told its default history for specialised lending is too short and too clean to support the estimates. It withdraws that portfolio from the application, uses the foundation approach there, and continues gathering loss data for a future submission.
Formula
Calculation
Expected loss = Probability of default (PD) x Loss given default (LGD) x Exposure at default (EAD)
Capital requirement = Risk-weighted assets x Minimum capital ratio
Worked example. A bank holds a $20,000,000 exposure to a mid-sized manufacturer. Its internal models estimate a one-year probability of default of 1.5% and, given the security held over plant and receivables, a loss given default of 35%.
Expected loss = 1.5% x 35% x $20,000,000 = $105,000 a year.
Under the foundation approach the bank would have been required to use a supervisory loss given default of 45% for this type of senior claim.
Foundation expected loss = 1.5% x 45% x $20,000,000 = $135,000.
The difference of $30,000 a year reflects the security the bank can actually prove it realises.
The effect on capital is larger. Suppose the standardised approach assigns a 100% risk weight, while the bank's approved model produces a 62.5% risk weight for this exposure, and the minimum capital ratio is 8%.
Standardised: risk-weighted assets = $20,000,000 x 100% = $20,000,000, so capital = $20,000,000 x 8% = $1,600,000.
Advanced IRB: risk-weighted assets = $20,000,000 x 62.5% = $12,500,000, so capital = $12,500,000 x 8% = $1,000,000.
The model releases $600,000 of capital against this single exposure, which is why banks invest heavily in obtaining and keeping approval.Case study
Seen in the real world.
This example is illustrative and Ashcombe Commercial Bank is fictional. Ashcombe, an invented lender with a $14,000,000,000 loan book, ran a four-year programme to move its corporate and small business portfolios from the standardised approach to advanced IRB. The business case rested on an estimated 18% reduction in risk-weighted assets and a corresponding release of about $190,000,000 of capital.
The project cost more than expected. Reconstructing eleven years of default and recovery records absorbed most of the budget, and the supervisor rejected the first submission because collateral valuations in the workout data could not be traced to independent sources.
In this fictional account the second submission succeeded, but with a supervisory add-on and a conservative floor on loss given default for unsecured small business lending. Ashcombe achieved a 12% reduction in risk-weighted assets rather than 18%, and the chief risk officer's summary to the board was that the greater benefit was the credit decisioning discipline the models forced on the front line, not the capital release itself.
Watch out
Common mistakes.
- Assuming advanced IRB always means less capital. Portfolios with genuinely poor recoveries or thin data can attract higher requirements under internal models than under the standard risk weights.
- Treating expected loss and capital as the same thing. Expected loss is covered by provisions and pricing, whereas regulatory capital is sized for unexpected loss, the tail beyond the average.
- Building models purely for regulatory reporting. Supervisors apply a use test, and models that do not genuinely drive lending, pricing and limit decisions will not keep approval.
Questions
People also ask.
Which banks can use advanced IRB?
Only institutions with explicit supervisory permission, which in practice means large banks with long, clean data histories and strong model governance.
What is the output floor?
It is a post-crisis rule that caps how far internally modelled risk-weighted assets may fall below the standardised calculation, set at 72.5% under the Basel III reforms.
How does advanced IRB differ from the foundation approach?
Foundation banks model only the probability of default and take regulator-set values for loss given default and exposure at default, while advanced banks estimate all three inputs themselves.
From the founder's library
Take it further with the book.
Build your financial confidence beyond this definition. Shihan's full-length guide, Accounting Fundamentals, takes the same plain-English approach and turns it into a complete, practical playbook for non-finance managers, business owners and students - with chapter-end quiz answers and presentation slides included.
25% off with code MMHQ25, applied at checkout. Priced in USD - checkout may show the equivalent in your local currency.
View the book and save 25%