IFRS 9 | 4Q22 results update

As an update to our last IFRS 9 results blog, 3Q22 results update, published in November, this post gives an update on the loss reserving trends and outlook for UK banks as at 4Q22.

Happily, especially for those working in banks’ reserving teams, this set of results was calmer than 3Q22 with no late-breaking surprises.

Credit performance stayed strong, with Stage 3 broadly flat, albeit with some comments that there may be a small deterioration around the edges. Personal and business borrowers have been remarkably resilient given the stress they are experiencing.

Compared to 1H22, the deterioration in economic outlook led to an increase in Stage 2%, in ECL coverage, and in the cost of risk. As part of this, modelled ECL increased significantly but was mostly offset by banks releasing management judgment, leading to an overall small increase in ECL and coverage. The impacts of high inflation, supply chain issues, energy costs, debt refinancing and “cost of living crisis” on borrowers were big talking points of the results. With models often trained data from 2008/9, model risk is high and banks looked to management judgement to make sure ECL was adequate – although many took comfort from the higher outputs of the models as a rationale for unwinding “economic uncertainty” judgements.

One of the more interesting aspects of these results was a greater spread in the range of economic outlook and in how lenders considered the effects of higher base rates and inflation on borrowers.

The outlook statements from the large lenders seemed pretty relaxed about the credit environment, with most sticking to cost of risk predictions at or just above their through-the-cycle rates. Against the backdrop of a tight labour market, big increases in property prices since the end of 2019, and a stabilisation (and potential slight improvement) in outlook for 2023 over the first few months of this year, this seems reasonable for prime, high street banks. Whether all sectors of the lending market will be as robust remains to be seen.

As usual, we have analysed the aggregate position of Barclays, HSBC, Lloyds, 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). More detail below with a lot of charts to support the narrative…

P.s. we wish the teams at lenders with February and March year-ends good luck for their results; having been burned by being the first institutions to deal with Covid and then the conflict in Ukraine we reckon they deserve a break!

1. Credit performance stayed strong with Stage 3 assets as a proportion of total loans broadly stable.

Banks commented that credit performance stayed strong, albeit a couple of institutions noted that there were some increases in delinquency at the margin. Other default indicators stayed good too, with breathing space applications still low and personal insolvencies around the 2019 level.

The corporate insolvency rate was an outlier though, with a significant increase in insolvencies over the year and the period. This is interesting as it doesn’t seem to be translating into defaults for lenders. The pattern of insolvencies by sector hasn’t changed in the underlying data so it seems that these must be small companies with very little bank borrowing.

2. The economic outlook deteriorated leading to an increase in Stage 2 assets

We have plotted average weighted peak UK unemployment as a simple metric for tracking economic outlook. The downgrade in average expectations is intuitive… but the increase in the spread of estimates is interesting with quite different opinions about how the labour market will evolve.

The downgrade in HPI expectations was also noteworthy, especially for lenders with mortgage-heavy balance sheets. The outlook swung from an expectation of growth at FY21 to falls of c. 5-10% at FY22.

With little change in observed credit performance, this led to an increase in Stage 2 assets as the “forward look” downgrade worsened future default expectations, pushing more loans over the “significant increase in credit risk” PD thresholds and into Stage 2.

The labour market is still very tight. While there has been some press recently about a reduction in the number of people who are inactive, we think this is marginal from a loan loss reserving perspective.

The chart below looks at the cumulative change in the number of people who are employed, unemployed or inactive since the end of 2019. It shows that the number of those employed and unemployed is broadly the same as pre-Covid. The difference is the c. 0.6m people who have left the labour market and are now inactive. Given the benign performance of personal lending portfolios it seems that, like the increase in corporate insolvencies, these individuals do not hold significant levels of bank lending. Meanwhile the job vacancy rate remains very high, underpinning the labour market and personal lending credit prospects.

Mortgages and house prices have had a lot of press recently and the impact of the rapid rises in base rate in 2H22 has put a distinct chill on the market. Transactions have been holding up for the time being, but forward-looking indicators look weak. Mortgage applications have dropped significantly and house price growth has plummeted with Nationwide showing a YoY fall (albeit prices are up more than 25% since the onset of Covid, so this could be seen as unwinding of “froth” rather than a big problem). Surveyors also have a negative outlook for commercial property.

Mortgage lending and lending to businesses has fuelled balance sheet growth since the onset of Covid and the retrenchment in consumer lending seen in the first half of 2020 has not unwound. Banks have a cautious outlook on growth, expecting reduced demand for credit for all asset classes except for a small increase expected in remortgage activity. Given the current deep lows for consumer confidence, a return to pre-Covid appetite for consumer lending seems unlikely in 2023. Although the number and value of transactions on credit cards is above late-2019 levels (perhaps reflecting a transition away from cash), outstanding balances and “revolving” use are still low compared to pre-Covid levels.

3. The deterioration in outlook led to an increase in ECL coverage from 1.0% to 1.1%, and cost of risk of c. 30bp reflecting the small ECL build

Behind the average change in coverage, the increase in spread in economic outlook is reflected in the detail of individual firms’ changes in coverage with movements over FY22 ranging between -11% and +14%.

4. Increases in modelled provisions were mostly offset by releases in PMAs

The small increase in coverage is interesting but, given the significant swing in economic outlook, a bigger increase might have been intuitive. However, as can be seen in the chart below, model output went up in 2H22, consistent with the downgrade in outlook, but was mostly offset by decreases in management judgement, including items relating to economic uncertainty and inflation/base rate/supply chain risks.

This can be seen in a change in mix of ECL over the year too – an increase in Stage 1 and Stage 2 modelled ECL, an almost complete run-off of Covid PMAs over the year, and a reduction in other PMAs.

There was a wide divergence in views across banks (including our wider portfolio of clients) relating to the impacts of higher finance costs and inflation. The two ends of the spectrum are characterised below:

More optimistic: borrowers are more resilient than expected. No change in delinquency has been seen so far and the inflation bubble has peaked. Borrowers are likely to make it through the stormy weather. Models are picking up the risk sufficiently with higher risk facilities having appropriately higher coverage and no management judgement is needed.

More pessimistic: borrowers have been using up the liquidity they stockpiled during Covid and it will run out at some point. Inflation is a cumulative game and, even though the peak may be approaching, stuff still costs a lot more, hitting both businesses and household debt service capacity. There is a credit issue coming with increases in financing costs a particular concern for mortgage borrowers and exposure to energy costs for business borrowers. Models are not able to sufficiently capture the increased risk – it’s not in the training data – and management judgement is needed to mitigate the model risk.

The “cost of living” crisis is uneven in its impact, hitting less affluent households harder. These dynamics, including the higher impact on tenants as compared to home owners, are covered brilliantly in regular ONS analysis. While the largest banks’ customers tend to be fairly wealthy and their products more “vanilla”, smaller more niche segments of the lending market may be more impacted.

The same is true for businesses and sector vulnerability, for example exposure to energy costs and discretionary spending.

In our year-end work we have approached this set of risks in two ways: firstly challenging whether a firm’s econometric default rate models are able to adequately address the risk of higher inflation and base rates at portfolio-level and, secondly, whether the models are able to adequately address the risk for the most marginal borrowers/sectors, where the model will not be optimised (assuming built at portfolio-level) and the losses are likely to occur first.

With almost no lenders having seen a noteworthy change in delinquency yet this area is purely judgmental… but events over 2023 will likely yield the answer.

5. Outlook

The outlook statements released as part of the 4Q22 results were benign, with banks sticking to expected cost of risk rates at-or-around their previously published through-the-cycle rates, reflecting their view that most of their customers will make it through the current period of economic turbulence.

The outlook seems to have stabilised too, with no further deterioration in consensus (and a possible slight improvement since the year-end if you squint at the chart) and a view that the peak in inflation may have passed (see Chart 1 in Andrew Bailey’s recent speech).

Data in the credit conditions survey is consistent with this. Lenders’ expectation of forward-looking default rates is still elevated but not at crisis levels. While consumer, medium business, and large business default rate expectations have improved, mortgage and small business data points are staying higher.

Regarding mortgages, refinance risk has eased slightly since year-end. The yield curve and mortgage pricing have dropped and, with likely reduced volumes in the market, competition may erode margins to the benefit of borrowers. While the consensus seems to be that house prices will fall in 2023, expectations don’t look sufficiently bearish to present a significant reserving threat given the growth in equity over the last couple of years. Refinance risk remains high as many households will continue to feel the pressure from managing higher interest rates alongside the cost of living pressures. The latest Bank of England estimates of the market-implied policy rates (Bank of England Monetary Policy Report February 2023) may indicate some respite as interest rate expectations came down from their peak at Q323, indicating a potential easing in refinance risk.

While the refinance shock for some Buy-to-Let landlords will be significant over the coming couple of years, new rents have been growing strongly (hometrack estimate 12% in 2022 overall and 17% in London), increasing cash rent cover since origination and providing some refinance buffer. Tenant affordability is becoming increasingly stretched and the pricing dynamic seems to continue to favour landlords, with rental demand remaining strong and supply reducing as landlords exit the market. However, at some point there must be a limit to the pricing elasticity of tenants given the stretch to affordability and their general elevated vulnerability to cost of living pressures.

The outlook on energy prices is also encouraging (to some extent): wholesale gas prices have been falling and recent press suggests the government may improve the level of support of domestic energy. However, there is a realistic prospect that energy costs will go up for businesses and households when the current support schemes expire (assuming that new schemes are less generous), affecting debt service capacity.

Overall, the credit outlook seems better at some points in 4Q22, and most high-quality borrowers seem likely to weather the storm, underpinned by a tight labour market and moderating expectations of any economic downturn in 2023.

However, some deterioration in credit performance and losses seems inevitable and will likely be focussed on borrowers in a more marginal debt service position, smaller businesses (including those that took on extra debt over Covid), and sectors that are energy intensive or have exposure to discretionary spending or construction/real estate. These risks are brought out very well through the stories in Andrew Bailey’s recent speech on the cost of living.

We also need to recognise that an increase in delinquency and losses will be part of a gradual normalisation of the credit environment from the abnormal lows of the last couple of years.

You can also check out our previous blogs or contact one of the team who would be happy to discuss any of the topics covered.