This article continues our occasional series of investigations into what the future might hold for lifetime expected credit losses (LECL) in banks' loan portfolios.
In our initial analysis of the subject we presented a benchmark model and assumptions whose output suggested that, as of 31 March 2020, LECL values were likely to have increased by a significant multiple with respect to year-end 2019. This is of course conditional on the modelled assumptions themselves holding - something that we highlighted cannot be fully relied on during tail events.
As at the end of April 2020, LECL estimates remained significantly above year-end 2019, but had fallen significantly. We posed the question of whether the combined strategic response (of all agents, combined) might have "changed the result by looking at it" - i.e. once it became clear that the overall dynamic system might not revert to an equilibrium state for some time, perhaps the policy response (as well as other agents' responses) encouraged it in a particular direction?
As at the end of May 2020, our LECL estimates fall by a further 30% and 20% in secured and unsecured lending, respectively. The uplifts with respect to year-end 2019 are presented below.
Results at 30 May 2020
Benchmark secured facilities
Benchmark unsecured facilities
In some cases the uplifts are less than half what our model was predicting at the end of March. The principal driver is that the marked-implied Credit Cycle Index (derived from UK listed corporates) has continued its return towards benign conditions. Under the autoregressive framework, forecasts that are conditioned on the current and previous CCI values have gained momentum. Markets seem to expect larger asset returns and/or lower volatility of asset returns, than one and two months ago.
In light of recent state interventions (most notably, the imposition of lock-downs to restrict transmission of COVID-19) it seems apt to revisit the problem of free will versus determinism: To what extent do future states of the world arise from the aggregate actions of individual acting of their own free will, versus a planned deterministic path? (Note: We do not propose to discuss how agents or individuals should behave).
Perhaps the combined policy (including lock-down, monetary and fiscal) responses drove the state of the world towards something more-deterministic? The assumptions of EDF model would certainly respond well to lower expectations of volatility in asset returns. But what about EDF models' asset returns, which are surely suppressed by both an unconstrained pandemic and a lock-down?
The philosopher Gottfried Leibniz claimed (and Voltaire satirised in Candide), that despite the existence of evil and suffering, the actual world is the best of all possible worlds. Considering the range of LECL scenarios presented at the end of March, we leave readers with the question - is the emerging outturn, the best of all possible LECLs?