Why a Few Severe Model Errors Matter More Than Many Minor Ones
A new ECB Working Paper argues that supervisory capital add-ons linked to internal ratings-based models are driven mainly by a small set of high-severity deficiencies. For borrowers and property buyers, the key lesson is indirect: risk-sensitive bank models do
The central finding: severity matters more than quantity
Banks use internal ratings-based (IRB) models to estimate capital requirements in a risk-sensitive way. These models operate within rules set by banking regulation, but they can still contain imperfections, including oversimplifications and incorrect assumptions.
The ECB Working Paper finds that a small group of high-severity model deficiencies accounts for the majority of the counterfactual risk-weighted asset burden associated with supervisory capital add-ons. In other words, the number of findings alone does not explain the largest impact. The seriousness of particular deficiencies matters more.
This distinction is important for supervision. Addressing every issue in the same way would not necessarily be the most effective use of limited supervisory resources. The paper’s result points instead towards prioritising deficiencies that can materially distort the assessment of risk.
Why internal models can create model risk
An internal model is a structured way of translating data and assumptions into a risk estimate. Its output may be affected when the underlying data is handled incorrectly, when an assumption is too simple, or when the model does not adequately reflect the risk it is intended to measure.
The paper describes this as model risk. It is particularly consequential when model outputs are used to assess a bank’s solvency. If risks are underestimated, the resulting capital assessment may not fully reflect the exposure identified by supervisors.
The paper traces the underlying non-compliances to a subset of articles in the Capital Requirements Regulation (CRR), mostly concerning how banks handle data relevant to IRB models. The practical implication is that data governance is not a technical detail at the edge of banking supervision. It can influence how risk is measured at the centre of the capital framework.
What a supervisory capital add-on is doing
A supervisory capital add-on can be understood as an adjustment applied when supervisors identify shortcomings that make a bank’s model-based risk measurement insufficiently reliable. The source focuses on the burden that these add-ons would impose in a counterfactual analysis, rather than presenting them as a new charge for individual mortgage borrowers.
That distinction matters. A capital add-on is a measure at bank level. It is not the same thing as a change in a borrower’s mortgage rate, loan-to-value ratio or approval decision. The source does not establish a direct mechanical effect on any particular household or property transaction.
It does, however, show why banks’ internal risk systems are subject to supervisory scrutiny even when the models are designed to be risk-sensitive.
What this means for someone planning a property purchase
The immediate lesson for a household is not that an ECB working paper predicts a specific mortgage price or lending decision. No such borrower-level conclusion is supported by the source.
A more useful iNSTRID interpretation is to separate three layers of a mortgage decision:
- Your own affordability: income, regular expenses, reserves and the payment you could withstand under less favourable conditions.
- The property and loan structure: purchase price, financing amount, collateral and loan-to-value.
- The bank’s risk framework: internal models, regulatory requirements and supervisory assessment.
The third layer can affect how a bank manages capital and risk, but a borrower cannot infer a concrete outcome from the existence of model add-ons alone. A robust household plan should therefore avoid relying on the most optimistic borrowing capacity shown by a single calculation. Keeping room in the budget is a practical safeguard, regardless of how a bank’s internal model is calibrated.
This is general financial planning guidance, not a conclusion about any individual bank or loan.
The broader supervisory message
The paper addresses a gap in the literature on model risk and proposes a more targeted use of scarce supervisory resources. Its argument is straightforward: if most of the counterfactual burden comes from a small set of severe deficiencies, identifying and correcting those deficiencies may be more valuable than treating all findings as equally important.
For banks, that places emphasis on the quality and compliant handling of IRB-relevant data, as well as on the assumptions built into risk models. For supervisors, it offers a basis for focusing attention where model weaknesses have the greatest potential effect on risk measurement.
For the public, the result is a reminder that a capital ratio based on sophisticated models is not automatically free from uncertainty. The reliability of the result depends on the data, assumptions and regulatory controls behind it.
Key takeaways and FAQs
Does a supervisory capital add-on directly increase my mortgage rate?
Not necessarily. The source discusses bank-level supervisory adjustments connected to IRB model deficiencies. It does not establish a direct effect on an individual borrower’s mortgage rate, approval or loan terms.
What is an IRB model?
An internal ratings-based model is a bank’s internal framework for estimating risk and capital requirements within the rules of banking regulation.
Why do a few deficiencies matter so much?
The ECB Working Paper finds that a small set of high-severity deficiencies is responsible for most of the counterfactual risk-weighted asset burden linked to supervisory capital add-ons.
What to take away
The paper’s key message is not that internal models are unusable. It is that model quality depends on details that can materially change the measurement of risk. A few severe deficiencies may matter more than a long list of minor findings. For a prospective borrower, the sensible response is to treat bank affordability calculations as one input—not a guarantee—and to build a purchase plan that remains workable beyond the most favourable scenario.
This article is informational and does not replace financial, investment or mortgage advice.
