This post continues our occasional series of investigations into what the future might hold for UK credit risk, and the implications for banks accounting estimates for Lifetime Expected Credit Losses (LECL).
In our previous article, our estimates suggested that the uplift in LECL with respect to year-end 2019 had fallen for the first time since the onset of COVID, but still suggested some significant LECL uplifts with respect to year-end 2019. Notwithstanding the tricky and non-linear impact of stage allocation, banks' half-year disclosures showed commensurately large increases in loan loss impairment.
As of 31 July 2020, our LECL estimates have increased by up to 10% (secured) and 20% (unsecured) month-on-month. Despite an uptick in the house price index, the default risk on the basket of UK companies that drives our Credit Cycle Index (CCI) increased by 20% month-on-month. This brings our LECL estimates broadly into line with where we were at April month-end.
As data describing the current state of the economy become observable (e.g. our CCI, unemployment, GDP, etc), we no longer have to guess where we are now. This greatly simplifies the task of forecasting the future. By shifting the focus to when the system might mean revert, to a first order approximation a forecast writes itself. Comparability between forecasts can be tested by simply checking the number of months before the forecast reaches equilibrium.
In our model, a reasonable proxy for entering "normal" conditions is when the base case predicts Z>-1. The 31-March prediction of 30 months had fallen to 8 months at 31-May and now stands at 17 months.
The Office for Budget Responsibility (OBR) April reference scenario has 2021 GDP level exceeding 2019 GDP level, and unemployment not returning to its 2019 level before the end of the scenario horizon in 2024. This scenario was updated in July, with a higher peak unemployment in H2 2020, and GDP levels not recovering until 2022.
The Bank of England (BOE) August financial stability report predicts peak unemployment in H2 2020, and a more-gradual recovery that also doesn't see 2019 GDP levels until around 2022.
Amongst banks' half-year presentations, internal card spend and current account turnover data suggests that after an initial sharp decline, there has already been a bounce-back. If the upwards trend continues, then pre-COVID levels could recover before the end of 2022.
Banks' half-year disclosure reveals a range of uplifts in balance sheet impairment that are reasonably aligned to our top-down estimates. However, it is important to remember that banks only recognise lifetime losses for contracts that have experienced a significant increase in credit risk (SICR), with respect to initial recognition (IFRS 9's "stage 2"). This causes a non-linear step change in impairment when triggered, from 12-month to lifetime cash shortfalls. Variability in firms' SICR triggers hampers comparability between firms, as well as with our analysis. Thus, it is plausible that banks towards the lower end of the range could yet experience increases, as and when loans meet their internal SICR triggers. Equally, banks at the upper end of the range could plausible book releases if loans' SICR classifications turn out to be "false positives".
Appendix: Full Results at 30 June 2020
The LECL uplifts with respect to the 31 December base case are presented below. These should be interpreted in the context, that they are conditional on the model's assumptions. We continue to highlight that those assumptions tend to deteriorate under extreme or tail conditions.
Benchmark Secured Portfolio
The table below summarises LECL uplifts with respect to the 31 December 2019 base case, for secured lending:
Benchmark Unsecured Portfolio
The table below summarises LECL uplifts with respect to the 31 December 2019 base case, for unsecured lending: