IFRS 9 | 1Q23 results update and “cost-of-living” focus

As an update to our last IFRS 9 results blog, 4Q22 results update, published in March, this post gives a 1Q23 update on the loss reserving trends and outlook for UK banks.

The 1Q23 results went smoothly, with economic outlook, ECL coverage, and Stage 2 and Stage 3 all remaining broadly flat on 4Q22. Lenders also reported credit performance was resilient in the main, although some banks did comment that they had seen some small upticks in arrears.

And so, everything seemed to be going nicely for much of 2Q23, with the Bank of England providing an optimistic view in May on prospects for the UK economy and labour market – an anticipated drop in house prices had not materialised, and there had also been a recovery in new finance costs. Then May and June’s CPI data landed, indicating that inflation would prove to be more stubborn than expected. Having previously indicated the rate tightening cycle had largely completed, market expectations for interest rates surged with ensuing turmoil in the mortgage market and fears over both the general economic outlook and the specific issue of increased refinance affordability stress for fixed rate mortgage deals maturing in the coming months.

The outlook has deteriorated over the last few weeks. The June CPI report reinforced the uncertain outlook for inflation and the anticipated improvement in 1H23 has been thrown into reverse as the risk profile of the industry climbs. Model risk, for example, is elevated in a high inflation, high interest rate environment and, for the time being, lenders must consider whether their models are operating effectively. This includes their ability to assess the risk of rising mortgage refinancing costs. However, with arrears still largely under control, any resulting Post Model Adjustments (PMAs) to patch model outputs will remain very judgmental at 2Q23.

1Q23 ECL results were broadly flat on 4Q22

Figure 1: ECL Coverage %¹

Figure 2: Stage 3%²

Figure 3: Stage 2%³

Cost of Living model adjustments – to release or not to release?

The headline rate of CPI has come down from its peak but is still high. This erodes debt service capability for both households and businesses through cost inflation (e.g. fuel, food), and higher finance costs. Individuals have seen their real disposable income hit hard in the last twelve months, while businesses are finding their margins squeezed. Yet arrears levels are at record lows, although there are some early signs of worsening around the margins.

From a loan loss reserving perspective, the question lenders are grappling with is: “do I think there will be credit losses arising from the “cost-of-living”/inflation increase that are not captured by my models?”

We have seen a broad spectrum of answers to this question, from an optimistic view that borrowers are more resilient than expected and the impact will be minimal, to a more pessimistic view that borrowers have just been eating through their liquidity and a credit stress is coming. However, to date, even those with a more pessimistic viewpoint don’t think this potential stress will have a material impact on the heart of their portfolios but will rather be at the edges, affecting more marginal borrowers with worse affordability and indebtedness. We have seen a wide range of outputs on “cost-of-living” and inflation PMAs in 4Q22, with some dependence on asset classes and portfolio risk profile and concentration.

While some firms were starting to wonder if they should release PMAs that had been booked at 4Q22 to cover this risk – prompted by sustained borrower exposure to higher inflation but offset somewhat by limited delinquency rates and a strong labour market – the recent change in outlook, with likely higher rates and stubborn inflation, suggest that risk and uncertainty in this area have increased again. However, in the absence of any noteworthy credit stress to date, this remains a judgement call for lenders, at least for the time being.

Household income and corporate margins are under pressure…

The story around the decline in real income (versus nominal), the associated squeeze on corporate margins and the rising cost of finance are illustrated in the charts below.

As we can see real wages have been struggling to keep up with the inflation increase and the gap between nominal and real income has been growing substantially since the end of 2021:

Figure 4: Average weekly earnings annual growth rate in Great Britain, seasonally adjusted⁴

In the corporate sector, the results from the most recent Deloitte CFO survey show that despite a better outlook and recent recovery in margins, operating costs have increased substantially since the start of Covid and remain on historically high levels:

Figure 5: Operating margins and costs⁵

Average interest rates on mortgages and unsecured personal loans have been growing significantly, in line with the BOE interest rate increase:

Figure 6: Average interest rates on personal loans and morgages (fixed)⁶

But with ~1.3 million mortgage holders yet to refinance this year, monetary policy changes have been slow to impact the wider market. We may therefore see affordability shocks in the future on both households and businesses as this works through the system, with the spectre of rising debt levels for businesses, and especially SMEs, who have substantially higher levels of debt than before Covid.

…but borrowers are potentially more resilient than in 2008

There is good reason to believe that today’s borrowers are more resilient than their forebears who lived through the Global Financial Crisis (GFC). UK households are in a better position than before the GFC, with debt-to-income lower than in 2008 (155% in 2008 versus 131% in 2022⁷). UK corporates are also similarly more robust, with debt-to-earnings ratios substantially lower than pre-GFC, despite increasing during the Covid pandemic (377% in 2008 versus 322% in 2019 and 350% in 2021⁸).

Affordability controls for retail lending introduced after the GFC should mean that borrowers have more headroom to absorb some of these impacts compared to the past, and banks have not repeated the excesses of pre-GFC lending in the corporate space leaving balance sheets stronger than pre-2008. Lastly, as the FPC has shown, the distribution of bank debt and cost-of-living and inflation vulnerability is not evenly spread. Individuals and businesses who are less exposed to the worst effects of the crisis hold the majority of debt.

This supports the reasoning from prime lenders that any stress arising from the current situation is likely to be contained to certain segments of banks' balance sheets and is also supported by their coverage ratios being near 4Q19-levels.

Model risk – portfolio level

Model risk is elevated for estimating ECL under high interest rate and high inflation conditions.

For retail lending, macroeconomic models are typically trained on data from the GFC. The structural differences compared to today’s environment – when inflation was broadly flat, base rates decreased in 08/09 and risk appetites were generally less constrained – mean that those models may now be operating outside of their design tolerances.

For corporate lending, models more often include longer-run data, in some cases from the early ‘90s recession, which is helpful given the coincidence of high inflation and rising base rates, but the world has changed a lot in the last 30-years and those trends may not be reliable for the modern economy.

Macroeconomic models are typically built at portfolio-level too and, in an environment of heightened model risk, judgmental reviews of the plausibility of outputs and benchmarking are useful controls for managing model risk.

Model risk – sub-portfolio level

Even if models are working well at portfolio-level this may not be the case under the surface. Stress will not be felt evenly across a portfolio and understanding how well the modelling framework differentiates more marginal segments in the tail (e.g. lower debt service capacity, higher leverage) is important. This is illustrated by BoE analysis showing mortgage arrears for different debt service ratios with a non-linear response of risk to affordability¹⁰.

Emerging approaches

Given the lack of defaults to date, the approaches adopted at 2022 year-end relied on judgmental assumptions and often revolved around stressing debt service capacity to infer default risk. Some firms also looked deeper to understand whether the assessments of Significant Increase in Credit Risk (SICR) in stage allocation should be adjusted with some facilities moved into Stage 2, but this was less common.

Mortgages

The main topic for mortgages was risk from refinancing of fixed rate deals. Models are typically not designed to deal with this risk, so many firms resorted to alternative analysis to assess adequacy and support PMAs if required. The range of approaches was wide, as were the size of PMAs.

In such cases, one approach we have seen that helps account for this risk is toderive a relationship between affordability and default rates, to estimate the impact on affordability of both general inflation and loan refinance costs (often using existing new lending affordability models) to infer a revised default rate. Some lenders have also stressed the inferred PDs for some marginal segments to reflect the risk that the historic affordability/default relationship may understate current risks for these sub-segments.

Another approach uses an affordability assessment to identify the population of most marginal customers who are then subjected to a PD uplift and/or moved to Stage 2 and lifetime ECL.

On owner-occupied portfolios, lenders have generally used assumptions to update expectations of customers’ income since origination. Similar logic has been applied to Buy-to-Let, based on the estimation of updated Interest Coverage Rate (ICR) positions as the measure of affordability. Buy-to-Let is complicated by the difficulty of looking through to the position of tenants, although the current strength of tenant demand is comforting.

One of the key assumptions is the time horizon over which lenders are assessing the risk. This has typically varied between one and five years. This is a balance though, since while mortgage costs (and refinance risk) are expected to moderate over time, and customers’ nominal incomes will likely grow, the risk is still likely to persist.

Consumer

Without the spectre of mortgage refinance risk to worry about, we saw a lower level of affordability PMAs for consumer lending at 4Q22. Approaches were more skewed to making portfolio-level adjustments rather than isolating more marginal segments of the book and making facility-level adjustments.

In some instances, lenders have used variations of their existing macroeconomic models to capture inflationary and affordability effects, for example by swapping in real drivers to replace nominal ones.

Given the worries about mortgage refinancing, some lenders also considered potential contagion effects from consumer lending customers who also have a mortgage.

Commercial

Inflationary pressures impact corporate borrowers differently given their size, complexity and sector, as well as their ability to pass costs on to clients and to manage risks through treasury functions. The complexity and relative sophistication of larger corporates should mean inflationary risks are better understood and therefore mitigated by more effective risk management, compared to less complex SME borrowers, who are unlikely to hold hedging arrangements and are therefore more exposed to increases in costs and rates. This is consistent with the latest results from the Bank of England Credit Conditions Survey for 1Q23¹¹, which shows that lenders' expectations of forward-looking default rates, although improved from the Covid period, remain more elevated for small and medium-sized enterprises than for larger corporates.

The pandemic and subsequent recovery impacted sectors differently, resulting in varying levels of liquidity and gearing, with sectors that have higher gearing now potentially more susceptible to refinance risk in an increasing interest rates environment. Sector performance has been further complicated by material changes in consumer spending patterns and cost increases, and in particular in energy costs. An analysis produced by the ONS shows that some sectors – in primis discretionary spending – have been more impacted by energy prices and rising staff costs than others:

Figure 7: Proportion of UK businesses affected by energy price and staff cost rises, by industry, September 2022 to January 2023¹²

More judgmental approaches have been used in the commercial and SME space, generally based on assessing a sample or sub-sample of customers for expectation of liquidity shortfalls and determining the resulting increase in risk across the portfolio, in some instances with sectorial considerations when existing models have been deemed inappropriate to capture differential risks. The need for adjustments is likely to be more limited for corporates, where the use of watchlists in portfolio risk management allows firms to capture emerging risks more effectively.

Conclusions

The outlook has deteriorated over the last few weeks, and the improvements witnessed earlier in 1H23 have been thrown into reverse as risk rates begin to climb again. Model risk is high in a high inflation, high interest rate environment and, for the time being, lenders must consider whether their models are operating effectively, including their ability to assess the risk from mortgage refinance. However, in the absence of compelling arrears emergence, any resulting PMAs to patch the model outputs will remain very judgmental at 2Q23.

_________________________________________________________________________________________

1. Company reports, Deloitte Analysis (Barclays, HSBC, Lloyds, NatWest and Santander UK)
2. Company reports, Deloitte Analysis (Barclays, HSBC, Lloyds, NatWest and Santander UK)
3. Company reports, Deloitte Analysis (Barclays, HSBC, Lloyds, NatWest and Santander UK)
4. ONS – Average Weekly Earnings in Great Britain: June 2023 - Monthly Wages and Salaries Survey
5. Deloitte CFO Survey, April 2023
6. BOE Interactive Database
7. Households (S.14): Debt to Income ratio: percentage: NSA - Office for National Statistics (ons.gov.uk)
8. Financial Stability in Focus: The corporate sector and UK financial stability | Bank of England
9. Financial Stability Report - July 2022 | Bank of England – 1.4 Debt vulnerabilities
10. Financial Stability Report June 2017 | Issue No. 41 (bankofengland.co.uk) – Chart A.6 (Households with high debt-servicing ratios (DSRs) are more likely to experience repayment difficulties)
11. Credit Conditions Survey - 2023 Q1 | Bank of England
12. ONS – Business Insights and impact on the UK economy