This blog looks at the year-end 2020 loss reserving trends and outlook for UK banks.
We have analysed the aggregate position of the UK's 5 largest banks as a proxy for the wider UK banking sector.
We start off with inputs (portfolio profile, credit performance, economic outlook), move on to IFRS 9 outputs (impairment charge, ECL balances, ECL loan cover, Stage 2 proportion) and finish with outlook.
The data for the banks is at Group-level, covering their global portfolios, but we focus on some of the UK market dynamics as this is relevant for the majority of the exposures (c. 70%).
IFRS 9 Inputs
1) Reported loan balances are slightly down over 2020…
After growing 5% in 1Q20, loan balances shrank slightly with a 1% to 2% reduction in each of the following quarters.
2) …but there are some interesting movements under the surface for UK lending assets.
Mortgages increased by a steady 3% (£40bn) over the year… but credit cards saw a big pay-down in 2Q20, with balances down 20% (£11bn) by 4Q20, and other consumer lending coming off by c. 10% as customers retrenched. Lending to corporates increased in 1Q, with the initial “dash for liquidity” and drawdown of £34bn, with a further £12bn in 2Q20 and some small net repayments thereafter. Loans to unincorporated (i.e. smaller) businesses kicked off in 2Q20 with a c. £8bn increase by 4Q20. These figures include the value of government support loans outstanding at year end (BBLS £44bn, CBILS £20bn, CLBILS c. £5bn).
3) After being a big issue at 2Q20, Covid-19-related payment holidays have now largely gone away.
Covid-19-related payment holidays have been a hot topic this year and, after reaching 10% of the total loan book at end 2Q20, they have run down to 2% at year-end. Banks report that the overwhelming majority of customers exiting from a payment holiday have resumed payments. However, more recent and repeat payment holidays seem to be higher risk than the initial cohorts. We also see this trend at smaller lenders. Forbearance balances, which exclude Covid-19 payment holidays, have increased 10% over 2020, but the picture across firms is mixed from small reductions to much larger increases.
4) Observed credit performance has been benign…
Corporate insolvencies and bank write-offs have been running below the 2019 trend since the onset of the crisis. This is consistent with the observation at many banks that credit performance has been benign in 2020, with flows to default and loss at low levels – a trend that is mostly attributed to the successful effect of government intervention and banks’ own customer support rather than an improvement in credit risk fundamentals. However, there is likely to be some impact, particularly on shorter duration portfolios, from banks’ criteria tightening early in the crisis.
5) …and defaults have not started to emerge yet.
Despite the economic turbulence since 1Q20 there has been a relatively small change in Stage 3, with defaults not emerging and the expected credit risk still latent. However, on the wholesale side, banks are seeing a few single name issues and some Covid-19-related restructuring, but nothing systemic (yet).
6) Banks’ economic outlook has improved since June 2020.
Banks’ forward-looking economic scenarios have improved since 2Q20, reflecting their view on the net impact of new events: the impacts of Brexit becoming less unknown, new lockdowns, new strains, vaccine rollout success etc. Expectations for the timing of the trough keep moving out as government support is extended to match restrictions to inhibit the spread of the virus. It’s hard to sum up banks’ economic outlooks in one metric… but we have tried to do this by looking at scenario-weighted peak UK unemployment expectations made at 2H20 and 4Q20. In addition to a drop in the peak rate from 9.4% to 8.2% as forward-looking expectations improved, the level of uncertainty has moderated shown by a narrower range of expectations.
IFRS 9 Outputs
7) After the big impairment charges in 1Q20 and 2Q20 the cost of risk has been low.
The reaction to the crisis in 1Q20 was a scramble, with banks making the best effort they could to put a loss estimate on the evolving crisis and booking some big impairment charges. This continued in 2Q20, but with more considered assessments of the risk set against a backdrop of continually deteriorating outlook. This led to even bigger charges at 2Q20 and cost of risk (i.e. impairment charge / drawn balances) peaking at 1.7%. Cost of risk has returned to a level marginally above the pre-crisis position in 3Q20 and 4Q20.
8) ECL balances and coverage stepped up in 1Q20 and 2Q20 and have been broadly flat since.
The movement in ECL balances mirrors the pattern of the P&L charge – significant increases in 1Q20 and 2Q20 and then a slight decline in 3Q20 and 4Q20. Loan coverage also shoes the step up, reaching 1.6% in 2Q20, but plateauing thereafter (rather than falling slightly) because of the concurrent slight decline in loan book balances.
9) Stage 2 has fallen since peaking at 2Q20.
The movement in Stage 2 shows a slightly different pattern. It mirrors the run up, with the deteriorating economic outlook at 1Q20 and 2Q20 pushing more loans into Stage 2 as the models predicted an increasing level of future significant increase in credit risk (i.e. loans are good now but are expected to deteriorate as the scenario unfolds). The reason for the subsequent run-down is less clear. Some firms show flat Stage 2 as a % of book, with others showing a decrease, focussed on their wholesale portfolios. This may be because they overshot earlier in the year… but management commentary is not clear on this and focusses on the YoY movement which is up significantly.
10) IFRS 9 inputs suggest an improvement in risk… but ECL is flat… what’s going on?
Despite the improvement in the economic outlook and run-down of Covid-19-related payment holidays, which would point to a reduction in credit risk, loan cover (i.e. ECL / drawn balance) was flat at 3Q20 and 4Q20. It may be the case that, given the high level of uncertainty, banks are reluctant to release provision with the risk of having to book it again if the outlook deteriorates. There are some signs that the models are releasing ECL and firms may be compensating by increasing model overlays, with several lenders talking about “economic” or “uncertainty” overlays… but the disclosure on Post Model Adjustments doesn’t support this with no clear relationship between change in Post Model Adjustments and change in ECL in 2H20 (i.e. the dots are all over the place in the chart). However, the position is opaque because the nature of PMAs changed over 1H20 (e.g. compensating for over-prediction from downswing in GDP and then under-prediction from the upswing) and there is variation in the categorisation of judgement and inclusion in disclosure (which is understandable given the judgement needed to classify different IFRS 9 decisions as “overlays”).
All the banks caveat their outlooks for impairment charge: the current situation is unprecedented and uncertain. But, in the absence of another big problem, they all point to P&L charges returning to materially lower levels than in 2020, with a couple of banks venturing that cost of risk will fall back to pre-pandemic levels at or below their through-the-cycle rate (usually in the range of 30-40bp).
This suggests that the banks think they’ve booked all the provision they need to… and now must wait for the defaults and losses to emerge and “eat” the reserves they’ve raised. Of course, this “pig in the python” of losses will cover the catch-up from artificially benign conditions in 2020 and the credit consequences of the crisis. The question is when will the risk emerge and, if the outlook continues to improve, when will they have the confidence (or be challenged by those charged with governance and their auditors) to release excess risk.
Charts 1, 3, 4, 6, 7, 8, 9, 10: company reports, Deloitte analysis
Chart 2: BoE statistical data covering series: RPQB68D, RPQB69D, RPQB78D, RPQB73D, RPQB74D, RPQB75D; gov.uk information on gvt lending schemes:
Chart 5: The Insolvency Service, monthly insolvency statistics, Table 1, total company insolvencies; BoE statistical data TFHA Write-offs of loans by monetary financial institutions (excluding central bank) £ millions, not seasonally adjusted