Summary
This article from Daniel Murdoch, Director in Broadstone’s Banking & Credit Advisory team, explains why hybrid PD models still matter for IRB mortgage lenders, even though Basel 3.1 reduces some of the capital benefit. It highlights the delivery challenges around data, judgemental adjustments, MoCs and validation, and argues that IRB’s value is now shifting towards better risk insight, model transparency and strategic decision-making.
Why hybrid PD models still matter
Over the last few years, IRB mortgage lenders have been developing hybrid Probability of Default (PD) models to balance point-in-time (PiT) responsiveness with through-the-cycle (TTC) stability.
This has been driven by regulatory expectations and the continued capital benefits associated with IRB. Although those benefits remain positive, the introduction of the Basel 3.1 output floor will reduce the capital advantage of internal models and require firms to reassess their strategic value.
Driven by regulatory change, hybrid PD models emerged as the compromise between PiT and TTC approaches. In practice, delivery has proven more complex than expected. Data limitations, evolving portfolio strategies and inconsistent interpretations of regulatory design requirements have extended timelines and increased model risk.
In this article, we will explore why hybrid PD models still matter, where delivery and validation become difficult, and how Basel 3.1 is reshaping the strategic role of IRB for mortgage lenders.
Learn more about our Capital & Impairment Risk Advisory services.
Where hybrid PD model delivery gets difficult
The challenges of delivering hybrid PD models are significant. In particular, firms need to manage three interlinked issues:
- Scorecard design remains a strategic choice, requiring firms to balance stability with responsiveness while ensuring the model remains relevant for credit decisions and other uses, including IFRS 9.
- Data constraints, including limited histories, portfolio growth and product changes, require adjustments to ensure default rates reflect portfolio behaviour across downturn conditions.
- Measuring cyclicality is also difficult. Firms must distinguish genuine risk movements from changes in portfolio mix or underwriting and understand the implications for capital, particularly under Pillar 2.
Why validation becomes more complex
The requirement for appropriate adjustments and margins of conservatism (MoCs) adds further complexity.
They are essential for addressing data and model uncertainty, but they are often judgement-driven and difficult to apply consistently. This reduces comparability and increases variability in outcomes. Combined with constrained data, reliance on judgement and weak use-test alignment, MoCs can make results more difficult to justify.
The cumulative effect creates a greater need for model transparency. The combination of statistical models with judgemental adjustments can make model validity and performance more difficult to assess.
This creates a particular issue for validation teams. Historically, model validation relied heavily on quantitative statistical evidence to review models. However, hybrid PD models require a broader assessment of model validity, including whether judgemental adjustments and MoCs are appropriate, proportionate and sufficiently evidenced.
Identifying whether known limitations have been addressed may be relatively straightforward. However, assessing whether the scale of any uplift is appropriate is more difficult. By design, these adjustments are not typically data-driven and therefore depend on a clear rationale, robust governance and strong supporting evidence.
Benchmarking and sensitivity analysis can help, but they do not replace the need for clear justification and robust scrutiny of the judgement applied.
Read more: Broadstone acquires Rockstead and strengthen its Banking & Credit Advisory business
Why LGD modelling can amplify these challenges
These issues are not limited to PD modelling. They are often amplified in LGD modelling, where data limitations are more acute.
PRA engagement, including the IRB mortgages roundtable on 30 September 2024, has clarified modelling expectations but also highlighted the difficulty of evidencing robust downturn estimates. Limited downturn repossession and loss data can force reliance on proxies and judgement, increasing complexity and model risk, with MoCs playing a significant role.
This context is particularly important for firms with limited data that are considering a move from the standardised approach to IRB. It may be best to prioritise hybrid PD model development while monitoring regulatory developments. The PRA’s DP1/25, published in July 2025, signals a potential move toward a retail FIRB approach, which could reduce the need for complex LGD modelling and broaden IRB access to smaller firms.
Basel 3.1 changes the focus of IRB
Basel 3.1 changes the strategic focus of IRB because of the constraint introduced by the output floor.
Under the output floor, modelled capital requirements will be constrained by a floor based on 72.5% of standardised RWAs, reducing the relative benefit of internal models. While capital advantage remains, particularly for larger portfolios, the emphasis is shifting toward risk insight, pricing, stress testing and broader risk management usage.
Firms must understand what their models are measuring, particularly in relation to cyclicality, and how this translates into capital impacts. This requires scenario testing, decomposition and clear articulation of model drivers.
Key takeaways
- Mortgage lenders can still realise value from moving from the standardised approach to IRB after Basel 3.1, but the financial incentive is lower and generally requires larger exposures to generate a meaningful benefit.
- Basel 3.1 does not reduce the importance of IRB; it reframes its purpose. Firms that prioritise risk understanding and model usage over capital optimisation alone will realise the greatest value.
- Firms intending to start IRB programmes should prioritise PD modelling while the PRA decides whether to introduce a retail FIRB approach. If adopted, this could remove one of the major barriers currently facing smaller firms with limited data.
How Broadstone can help
Broadstone’s Banking & Credit Advisory (BCA) team supports lenders through every stage of IRB model development, validation and implementation. Whether you are assessing the strategic implications of Basel 3.1, developing hybrid PD models, or preparing for future regulatory change, our specialists can provide practical, proportionate support tailored to your portfolio and objectives.
