• For Irish mortgages, modelled RWA densities are considerably higher than other European countries, with average risk-weights in Dec-18 (36%) almost 2.5 times the EU average (15%).
• Current strategies to address high NPE levels will reduce the gap; however, this could be offset by conservative adjustments that may be introduced following the outcome of the ECB TRIM exercise.
• The divergence in RWA density across individual Irish mortgage portfolios is significant. This will cause inconsistencies in minimal capital requirements set by the Supervisor.
• Banks must endeavour to increase understanding of the strengths and limitations of internal models and ensure regulatory capital requirements accurately reflect their underlying credit risk. Senior Management must ensure that any excess variability when compared with other European banks is understood and, where possible, addressed through model redevelopments.

The risk to future earnings has been a key focus for the European Central Bank (ECB). It has featured prominently in recent supervisory priority lists as part of Business Model Assessments. Supervisory concerns persist on banks’ ability to generate sustainable profits – especially given the ongoing low interest rate environment. This was highlighted in a speech by the Chair of the Supervisory Board of the ECB in February 2019.

A recent publication by the Department of Finance (DoF) examines these concerns in a European context. The report describes how few European banks are meeting current shareholder expectations of a 10% return on equity (ROE); the average ROE is c. 6.7%. The lack of growth throughout Europe – combined with the low interest rate environment – are identified as the primary contributors.

The Irish Retail banks are not exempt from the challenge of generating sustainable levels of profitability. The vulnerability is even more pronounced when recent one-off gains from legacy impairment releases are excluded.  Despite the overall improvement in the Irish economy over the past number of years, low ROE remains a concern. 

In conjunction with the income and cost challenges, minimum capital estimates are also adversely affecting ROE. Of particular relevance are Risk Weighted Assets (RWAs) – which determine the regulatory capital requirements under the Pillar 1 framework.

For Credit Risk, banks’ internal models calculate RWAs in accordance with the assumed risk profile of their underlying loan portfolios. This paper considers the outputs from these models in the case of Irish mortgage portfolios.

For their mortgage portfolios, all five Irish retail lenders have Advanced Internal Ratings Based (IRB) permission. This means that key credit risk parameters (Probability of Default (PD) and Loss Given Default (LGD)) are modelled internally from each Bank’s individual default and loss history. Key regulatory requirements include:

• The PD must be calibrated to a long-run average measure minimising any pro-cyclicality within the estimate. This ensures capital levels aligned to grades within credit rating systems remain stable through both benign and stressed conditions.
• The LGD must be estimated with a downturn measure – which is representative of adverse economic conditions.

A summary of the total mortgage risk-weights for the five main retail lenders for both June 2018 and December 2018 is illustrated in Table 1.

In a European context, the average risk-weight in Dec-18 (36%) is almost 2.5 times the EU average (15%). This is partially driven by two key factors:

1. Ireland’s ongoing high percentage of Non-Performing Exposures (NPE): NPEs attract higher risk weights when compared to Performing Exposures, which adversely skews the overall RWA figure for mortgage portfolios.
2. Incorporation of Ireland’s most recent default and loss experience into Advanced IRB models: EBA guidelines on PD and LGD estimation require banks to have a representative mix of “Good” and “Bad” years in their reference datasets. The Irish mortgage data history contains strong adverse bias from the recent financial crisis. The extreme volatility from the recession to the current benign conditions is also likely to increase sensitivity in both the PD calibration and the downturn LGD estimate.

So Irish banks must allocate higher levels of regulatory capital to mortgage exposures than European equivalents. However, there is little uniformity across the individual Irish bank RWA measures. At Dec-18, the standard deviation (11%) and range (22%) across the banks is notable.

While some allowance can be made for portfolio specifics (e.g. credit quality variations, vintage distribution), the high variability of RWA estimates appears to be disproportionate. Given the likely similarities in behavioural and historical experiences within ROI mortgage portfolios, a lower distribution of risk-weights would be expected.

Variances of this magnitude can have a substantial impact in portfolio performance. Banks with higher risk-weights must allocate greater levels of regulatory capital to address their credit risk. If pricing is relatively standardised across the asset class, higher capital will result in lower ROE for investors.

The basic trend between the two snapshot dates indicates RWA reductions – both in absolute levels and volatility. The average total risk weight has reduced from 38% (Jun-18) to 36% (Dec-18). In addition, standard deviation and range measures have also reduced over the 6-month period – by 2% and 10% respectively. This indicates some progress towards harmonisation.

In the European context, the expectation is that the gap between Irish banks’ RWAs and the EU average is likely to reduce. This will be driven by a number of factors, including:

• The on-going initiative in Irish banks to reduce NPLs: with portfolio sales likely to reduce the skewed impact of higher non-performing risk-weights; and
• Updated re-calibration of long-run average measures: the recent improvement in the Irish economy should increase the number of “Good” years in the reference dataset. This should offset some of the bias towards the previous more stressed years.

The ECB Targeted Review of Internal Models (TRIM) may help decrease the gap, given its primary purpose is to reduce unwarranted variability in RWA measurements across European banks. However, initial indications from the DoF paper are that the impact of TRIM may further increase Risk-Weights for Irish banks. This is due to potential introduction of additional margins of conservatism to address underlying limitations in their reference datasets. However, it still remains unclear precisely how TRIM will ultimately impact the gap with the rest of the European banks.

Finally, the introduction of the standardised RWA floor under the proposed Basel IV regulations could alter the landscape dramatically. While it is unlikely to significantly affect the required RWA density for Irish mortgages, the impact across European banks is likely to be more significant. 

The DOF paper focuses on the impact on pricing of higher Risk-Weights (i.e. higher Risk Weights require greater margins to ensure pricing hurdles can be met). However, the impact of the more elevated RWA estimation is much wider.

1. Higher RWA estimates result in higher minimum regulatory capital requirements (in absolute terms) under the Supervisory Review and Evaluation Process (SREP). In turn, this will also affect all related thresholds across the capital adequacy spectrum, including capital triggers and tolerances for Risk Appetite and Recovery Plan Frameworks.
2. The high RWA levels may adversely impact the observed performance of Irish banks in EBA Stress Testing exercises. One of the key assumptions is that RWA levels are floored at the starting position. If initial RWAs on Irish mortgage portfolios continue to be high relative to European peers, this conservatism will be carried through the forecast. Therefore, the final stressed capital position will not accurately reflect the true impact of the scenarios.
3. With Basel IV scheduled to introduce a new standardised RWA floor for mortgage portfolios, the benefit of maintaining internal models becomes less apparent. However, it is essential that data driven solutions continue to be prominent within banks’ risk management practices.

Therefore, the Irish Retail banks must continue to closely monitor RWA levels for their mortgage portfolios. Senior Management must ensure that any excess variability when compared with other European banks is understood and, where possible, addressed through model redevelopments. 

We continue to support our clients to enhance their understanding of the strengths and limitations of internal models. KPMG can help with the implementation of a robust risk modelling framework, through solutions such as:

• Interpretation of RWA calculations and projections under both base and stress scenarios;
• Assessment of the suitability of reference datasets used in model development;
• Evaluation of the appropriateness of thresholds set for capital adequacy; and
• Implementation of independent model validation programmes, using both qualitative and quantitative techniques, to reduce model risk.

These solutions help ensure that minimum regulatory capital estimates are appropriately aligned with the underlying credit risk.

For further information on mortgage weighted risk assets in Ireland, please contact Adrian Toner, Regulatory Advisory Director, via this form.