Previous researchfootnote  indicates that firms using the SA to credit risk may have a reduced ability to compete compared to firms using the IRB approach, which allows firms to model their risk-weighted assets (RWAs).footnote  This is in part because internal modelling can often produce lower RWAs, which reduces the amount of capital required against those assets and therefore could potentially impact lending costs relative to the SA.footnote  footnote 
This issue was particularly acute in the residential mortgage market, where a large gap between risk weights (RWs) calculated using the SA and the IRB approach was identified, especially for lower loan to value (LTV) mortgages. PRA staff research found this may also incentivise firms using the SA to concentrate in riskier exposures, raising concerns around the safety and soundness of these firms.
New stylised analysis suggests this gap could narrow in the next few years. Charts A and B illustrate current mortgage RWs (Average IRB RWs (2021) and SA RWs (2021)) in the owner-occupier and buy-to-let (BTL) markets, and the potential impact of certain policies. They are based on regulatory data, supervisory intelligence, firms’ disclosures, and staff calculations.
- (a) The 'Average IRB RWs (2021)' line follows the methodology used to produce the IRB Benchmark in the PRA’s Statement of Policy for setting Pillar 2 capital.
- (b) The IRB RWs represented in the charts are adjusted for expected loss (EL). The final risk weight is calculated as (RWA + 12.5*EL) divided by exposure value at each LTV band.
There are two key drivers of the narrowing gap:
- The PRA’s proposed implementation of the Basel 3.1 standards would introduce a more risk-sensitive SA framework for credit risk. If implemented, for typically smaller firms using the Basel 3.1 SA this will decrease RWs for owner-occupier mortgages with LTV ratios below 80% (SA RWs (2021) versus Basel 3.1 SA RWs (2025) lines, Chart A). This alone narrows the existing gap at the 0%–50% LTV band by almost 45%.
- Changes to IRB modelling of mortgage risk, stemming from hybrid modelling and the IRB roadmap which are primarily aimed at reducing unwarranted variability in RWAs across banks, are likely to somewhat increase average IRB mortgage RWs, all else being equal.footnote  These changes include a revised definition of default and 'hybrid' probability of default models. The shaded area shows a stylised landing zone for average IRB RWs, based on currently available estimates of how IRB reforms may affect residential mortgage RWs at different LTV bands and in current economic conditions.
Bringing these together, the gap between SA and average IRB RWs for residential mortgages will narrow. For owner-occupier mortgages with LTV ratios below 50%, it could narrow from around 5.5x to between 2.5x and 1.5x by 2025, depending on firms’ responses to IRB policy changes. A broadly similar pattern would be observed all along the risk spectrum, including in the 70%–90% LTV range where many new mortgages are issued. For BTL mortgages with LTV ratios below 50%, the gap could narrow from approximately 4.5x to between 3x and 2x (Chart B). It is worth noting that for BTL with LTVs greater than 80%, both the current and future (Basel 3.1) SA have lower RWs than the average RW under the IRB.
Chart B: Stylised analysis of SA and average IRB risk weights for buy-to-let mortgages, pre and post reforms
- See footnotes (a) and (b) of Chart A.
This analysis is not a forecast of residential mortgage RWs. There is uncertainty around hybrid IRB RWs as hybrid models still need to be approved and data are limited. Moreover, LTV is not the only driver of credit risk in hybrid IRB models and macroeconomic conditions may affect hybrid IRB RWs.
While uncertain, the analysis does suggest a smaller SA-IRB gap that may facilitate competition to a greater degree in the mortgage market in the future. This will be an important factor for the PRA in designing a simpler prudential regime for small banks and building societies, in line with its secondary objective to facilitate effective competition between firms. As set out in The Strong and Simple Framework: Liquidity and Disclosure Requirements for Simpler-regime Firms consultation paper, ‘The PRA is making the planning assumption that the Basel 3.1 Pillar 1 approach to credit risk would be the starting point for designing the simpler-regime risk-based capital framework. This is because the PRA considers the proposed Basel 3.1 Pillar 1 approach to credit risk to represent an improvement over the current capital rules; they would be more risk sensitive and would facilitate competition without compromising safety and soundness’, as this analysis indicates. The PRA intends in Phase 2 to focus on simplifications to Pillar 2 and buffer requirements for Simpler-regime Firms.
This post was prepared with the help of Marco Schneebalg, James McGoay, Ala Marar and Faith Bannier.
This analysis was presented to the Prudential Regulation Committee in January 2023.
Share your thoughts with us at BankOverground@bankofengland.co.uk
This includes the Competition and Markets Authority review of competition in the retail banking market (2016).
Each asset class is assigned a risk weight according to how risky it is judged to be. These risk weights are then applied to a bank’s assets, resulting in RWAs. For more information, see Quarterly Bulletin 2013 Q3, ‘Bank capital and liquidity’.
The study also cites other competition barriers for smaller banks, including informational disadvantages and higher funding costs given more limited access to stable and lower cost retail deposits.
Risk weights are not the only driver of capital requirements for SA versus IRB firms. Other factors contribute to how big the ‘capital gap’ is. For example, the refined methodology to Pillar 2A allows firms that meet specified criteria to compare SA credit risk capital requirements to those derived from a typical IRB model. When making an overall assessment of the adequacy of their total capital requirements, the firm and the PRA can then consider that comparison and assess whether there is any excess conservatism in some aspects of the existing Pillar 1 SA.
‘IRB roadmap’ refers to the policy set out in PRA Policy Statement 7/19 – ‘Credit risk: The definition of default’; March 2019 on the definition of default and PRA Policy Statement 11/20 – ‘Credit risk: Probability of Default and Loss Given Default estimation’, May 2020 on Probability of Default and Loss Given Default estimation.