IFRS 9 | 3Q23 results update: holding steady for year-end?
As an update to our last IFRS 9 results blog on the 2Q23 results, and our special piece on the 2022-23 PRA IFRS 9 thematic feedback, this post gives a 3Q23 update on loan loss reserving trends and outlook for UK lenders.
Overall, the third quarter was uneventful in terms of credit performance, economic outlook, and ECL key performance indicators, with most key measures remaining flat.
While credit performance stayed strong there were continued signs of stress emerging at the margins of some portfolios at some lenders, but the picture is inconsistent. Industry-level leading indicators suggest that, for personal lending, there is some increase in credit stress compared to the more benign performance seen over recent years. While the absolute levels of these indicators are not concerning – at least not yet – their trajectory is worrying, as we show in the charts below.
The UK credit and economic environment has evolved over the year with expectations that rates may now have peaked, inflation is falling, real incomes are growing, fears of recession have receded, and the labour market still looks strong. In this context we cover several topical areas that lenders may wish to think about as we approach the year-end: mortgage refinancing risk, the negative impact of the rising cost of living on personal lending, risk in the commercial lending space, model accuracy and calibration, and the assessment of prepayment rates.
Absent any dramatic new news between now and reporting dates in February 2024, we think that the main questions for year-end are about level and lag. Any changes in ECL coverage will be prompted by changes in the expected level of loss risk given the competing influences of potential portfolio quality improvements (as set out in our last blog), forward-looking view of fundamental credit drivers, and deterioration in some leading indicators. To date there has been little in the way of a credit response to the higher base rate, inflation, and CRE/residential property price falls. Lenders will also need to assess the extent to which they think that any latent risk is lagged and how this relates to the lags already baked into their models and macro scenarios.
As usual, we have analysed the aggregate position of Barclays, HSBC, Lloyds Bank, NatWest and Santander UK as a proxy for the wider UK banking sector (on a “volume weighted” basis to prevent distortions from balance sheet size). Full detail is given below with accompanying charts to support our narrative.
1) Credit performance and economic outlook were both broadly stable in the quarter despite early warning indicators signalling an increase in stress in retail portfolios.
Source: company reports and Deloitte analysis
The proportion of loans in Stage 3 remained flat, with banks reporting good UK asset quality in the main. Nevertheless, some lenders began to show early signs of deterioration at the margins of some of their portfolios, but there appears to be no discernible trend so far.
The concerning upward trend in corporate insolvencies has stabilised. As noted in previous blogs, these mainly related to small businesses with limited-to-no bank lending, which do not show up on banks’ balance sheets. Personal insolvency levels are still looking good and even improved in the quarter, which is interesting given the level of concern around the impact of the rising cost of living on British consumers.
Source: Insolvency Service monthly statistics and Deloitte analysis
The base case economic outlook was broadly stable, with both unemployment and bank rate expectations flat on 2Q23. While the outlook for HPI fall-to-trough improved we consider this is likely because observed prices fell in the quarter with less distance left to go to reach the anticipated trough. The outlook for GDP also saw a marginal improvement. Interestingly, the range of expectations for HPI fall-to-trough still widened in the last quarter, illustrating the level of uncertainty baked into the outlook.
Source: Company reports, Deloitte analysis
However, there may be some early signs of stress emerging at the margins of retail portfolios with a steady quarter-on-quarter increase in the volume of mortgage new-to-arrears and applications for “breathing space”. The proportion of mortgages with forbearance has ticked-up slightly but is below pandemic levels; this is consistent with our conversations with stakeholders across the marketplace which indicate that the take-up of options available under the Mortgage Charter has been relatively modest to date.
Source: FCA MLAR statistics; Insolvency Service monthly statistics
Although the absolute level of mortgage arrears is still low by historic standards, with arrears now at pre-pandemic levels, the upward trend is still concerning. Furthermore, the most recent data release on mortgage arrears from UK Finance shows that the rate of increase in delinquency for buy-to-let lending is significantly higher than for owner occupier lending (albeit off a very low starting level).
Source: FCA MLAR statistics
2) ECL key performance indicators were broadly flat.
Cost of risk (i.e. annualised quarterly ECL P&L charge / gross loans and advances to customers) and ECL cover (i.e. ECL balance / gross loans and advances to customers) were broadly flat on 2Q23 reflecting resilient credit performance and stable economic outlook.
Source: company reports and Deloitte analysis
Source: company reports and Deloitte analysis
Beneath this average the range of movement in ECL cover from individual banks was slightly wider than at 1Q23 and 2Q23.
Source: company reports and Deloitte analysis
The proportion of loans in Stage 2 nudged down slightly, partly in response to small improvements in the economic outlook in 3Q23, and partly due to timing effects at some lenders with a lag from updates to the economic outlook at 2Q23.
Source: company reports and Deloitte analysis
3) Things to think about for year-end.
After a quiet 3Q23, reserving teams at many lenders are turning their attention to the main event of the year: December year-end results season. Much of the narrative over the end of 2022 and during 2023 has focussed on economic uncertainty and the expectation of an increase in defaults triggered by high interest rates, high inflation, and supply chain constraints. However, the situation has evolved and we highlight some of the topics that have interested us below.
Mortgage refinance risk has been a hot topic for some time now. The challenges here are threefold:
Firstly, as time has passed, the stock of mortgages still due to refinance from lower to higher rates has fallen, reducing the forward-looking balance at risk.
Secondly, the payment shock for fix-to-fix transfers has been at a high in 2H23 (the chart below shows the difference in the average quoted mortgage rate at a given date compared to the rate two or five years prior to that, with that difference being an approximation of the payment shock at that point). It is still too early to tell how these transfers will perform and to make reliable predictions about the performance of coming transfers. However, we anticipate that transfers in 2H23 will probably need at least six months of outcome data to form a reliable view of their likely credit outcome.
Source: BoE quoted rates data, Deloitte analysis
Thirdly, the forward-looking view of interest rates has changed with expectations now flat-to-down rather than showing an increase followed by a fall. Depending on the lag in lenders’ macroeconomic models this may lead to a reduced forward-looking default response (i.e. the model may think that the risk from higher rates has passed). However, lenders may consider that the risk is still yet to emerge, leading to a divergence between model and management expectations.
Source: BoE OIS data
Lenders should consider the level of refinance risk in their portfolios and the extent to which it is covered by their models, particularly given the likely change in the shape and level of the input economics.
Retail “cost-of-living” risk is broadly considered as the credit risk on personal lending portfolios arising from the reduction in debt service capacity from high inflation. It is worth noting that, for mortgage borrowers, this effect is usually small in comparison to the impact of a refinance event. At 4Q22 and 2Q23 the picture around real wage growth was pretty dismal, with an expectation that erosion of real incomes, and therefore debt service capacity, would lead to a credit response.
The position at 4Q23 is a little different. Nominal wage growth is strong and real wage growth may have crept back into positive territory for the second half of the year. However, the OBR November 2023 forecasts indicate that real household disposable income per person will reach its 1Q22 level again in 2027-28 so the recovery in spending power will likely be fairly slow.
Source: ONS Annual growth in total pay (excluding bonuses)
The effects of high inflation are felt unevenly across the income spectrum, with lower income households being hit much harder than higher income households. This is shown very well in our new favourite data visualisation, the Citizens Advice cost-of-living data dashboard. Of course, as discussed in previous blogs and in the BoE’s Financial Stability Reports, the distribution of bank debt (and savings) tends to sit with higher income deciles giving lenders some protection from this dynamic which, in combination with portfolio quality dynamics, may explain why the credit response to this risk has been limited so far with lenders mostly reporting good performance in consumer lending portfolios.
Lenders should consider the extent of this risk and whether it is captured by their models.
Commercial sector risk also continues to be a concern. While credit performance has been strong to date, concerns remain about the potential lagged emergence of defaults from factors including increases in finance costs, the impacts of supply chain disruption, price falls in the CRE market, and some idiosyncratic sector-specific risks. The extent to which the credit impact of these events may have been cushioned by liquidity built up over Covid is not clear. Lenders should consider whether their models are picking up these risks and if model risks may arise from choices around segmentation and the difference between current credit conditions and those reflected in the model build data.
There has been a far bit of press coverage and academic literature on the increase in “zombie” companies since the Global Financial Crisis (i.e. businesses with consistently low profitability and interest coverage ratios). It is interesting that the change in rate environment does not yet seem to have led to any clear-out of “zombies” on bank balance sheets. This could be because of improved underwriting and risk-management practices since the GFC or perhaps a long hangover from Covid-era liquidity hoarding.
Lenders should continue to consider sector-specific risks in their commercial lending portfolios and the extent to which these are covered by their models.
Model accuracy and model calibration become important in an environment where default rates are rising. With an increase in mortgage arrears, lenders should ensure that their Probability of Default models are appropriately calibrated. Default emergence may be limited to certain segments of the portfolio and lenders should consider how this compares to calibration segmentation to make sure any emerging risk is captured. Furthermore, conventional calibration processes are inherently lagged because of their outcome periods; lenders should consider taking a view of early delinquency emergence to ensure that calibration procedures reflect contemporaneous events.
Prepayment profiles have also changed for some products (most notably mortgages); while the main point of impact here is on income recognition via EIR, there is also an impact on EAD and PD survival.
Lenders should check the consistency of their prepayment assumptions between EIR and ECL and consider whether they give an appropriate forward-looking view.
4) Outlook
Banks’ outlook statements suggest they are expecting a fairly benign 4Q23.
We interpret lenders’ ECL position over the last few quarters as signalling an expectation that given the prevailing economic outlook, with GDP positive but sluggish and the labour market in good health offset by higher-for-longer rates and potentially sticky inflation, most borrowers will be fine from a credit perspective. However, there will likely be some increase in credit events for more marginal, indebted obligors with lower debt service capacity and lenders have already factored this in to their ECL coverage.
The latest data from the Bank of England’s Credit Conditions survey supports this for commercial lending, where the trend of expected default stress has improved. But the retail chart is less comforting with expectations for mortgage defaults worsening and the consumer credit line quite volatile, albeit with a downward trend following the highs of the Covid-19 pandemic (if you squint a bit).
Source: Bank of England Credit Conditions Survey
Source: Bank of England Credit Conditions Survey
Absent any dramatic new developments between now and reporting dates in February, the main questions for year-end are about re-evaluating level and lag.
Any changes in coverage will be prompted by changes in the expected level of default risk given the competing influences of potential portfolio quality improvements (as set out in our last blog), the forward-looking fundamental credit drivers, and deterioration in some leading indicators. To date there has been little credit response to higher base rates, inflation, and CRE price falls. However, lenders will still need to assess the extent to which they think any latent risk is lagged and how this relates to the lags already built into their models and macro scenarios.
We hope you found this quarterly assessment useful. Please don’t hesitate to contact one of our team who will be happy to discuss any of the topics covered here. You can find previous blogs in this series here:
