The macro environment and related accounting and reporting impacts
Mark Gregory, Chief Economist, Ernst & Young LLP
Mark discussed several challenges relating to ECL modelling in the current macroeconomic climate.
Estimates of ECL appear low given the macroeconomic situation
IFRS 9 was implemented at the end of an unusual economic period for most European banks as the global financial crisis and the subsequent Eurozone led to a period of relatively slow growth and low inflation. This resulted in major central banks pursuing low interest rate policies.
Mark cautioned that the external environment is changing with more significant risks, weighted to the downside. These include the impact of the trade disputes that have emerged in the last 12 to 18 months, the risk of the adoption of more populist domestic economic policies signaling fiscal expansion and Brexit. Central banks have also started to highlight the risks that climate change could pose to financial stability. Policy makers are not necessarily in a good position to deal with potential shocks because of high public debt to gross domestic product (GDP) levels in several European countries, increased corporate indebtedness in the USA and China, and low interest rates. Within Europe, while many countries have stabilized since the 2008 financial crisis, structural weaknesses remain.
Against this backdrop, the relatively low impact of applying Multiple Economic Scenarios (MES) to estimate ECL is surprising. While reviews of individual bank estimates may appear reasonable when we look at them overall, the situations seem out of synch with the economic backdrop. Mark went on to highlight some of the possible causes of these findings.
Most European banks used three scenarios, with UK banks tending to provide an additional, Brexit-specific scenario. This scenario may be merged into business as usual, base case scenarios in the future depending on how Brexit progresses. In general, base case forecasts tended to be broadly in line with consensus.
Across Europe, the overall comparisons suggest that downside scenarios may be too benign on average. European growth and forward-looking indicators have weakened significantly in recent months and in the coming year, we will be interested to see how downside scenarios evolve.
Probabilities and scenario weights
Many banks used symmetrical probabilities for upside and downside scenarios. This is surprising given that the consensus amongst economists is that risks are skewed significantly to the downside. Equal weights tend to mute the downside risks and is another area that requires attention in the coming year.
Mark also identified wide variations in the degree of change of key model assumptions such as house prices, unemployment relative to the probabilities and weights used. More work is required to ensure the assumptions and weights are consistently derived and accurately reflect potential impacts.
Reversion to mean assumptions
Reversion to mean assumptions shapes the impact of upside and downside scenarios. While many banks have assumed reversion to mean based on previous market cycles, Mark challenged participants to consider whether this is reflective of the current market environment if a downturn were to occur. With the previously mentioned limited tools currently available for policy makers to respond to a crisis, there is a risk therefore that the impact of a downturn could be more significant and long-lasting than previous market cycle data indicates.
Since 2008, interest rates in Europe have been low. ECL models developed using economic data from the last ten years only are, therefore, at a risk of not adequately catering for ECLs in the event of volatile or higher interest rates.
Credit risk models
While beyond the scope of the MES review, Mark raised questions about the scale of impact generated when the MES outputs are put into credit risk models. Some of the impacts seem low which may reflect the historical data used to calibrate models or the form of the models themselves. There are numerous adjustments in this part of the process, but we do need to ensure these are consistent with the economic logic underpinning the specification of the MES.