Menu

THE SUPERVISION BLOG

Who pays the piper calls the tune: The need for and benefit of strong credit risk management

Blog post by Elizabeth McCaul, Member of the Supervisory Board of the ECB

Frankfurt am Main, 4 December 2020

Introduction

The end of the coronavirus (COVID-19) pandemic seems to finally be on the horizon. Over the last weeks, news about vaccines proving effective in late stages of trials suggests that we may eventually see an end to lockdowns and the related economic damage. But in the months ahead, before the vaccines usher in healthier days, we still face a reality of rising infections, lockdowns and economic disruption. As we look longingly towards a healthier future, it is worth reflecting on some lessons from the past on how banks can most effectively manage their main areas of risk, and particularly credit risk, through such a period of stress.

We have learned from the unhappy legacy of the great financial crisis that credit risk opacity, a pile-up of bad loans and lack of transparency amplifies the depth of the shock and hampers recovery, resulting in pro-cyclicality. When the pandemic finally recedes and the need for public support measures subsides, the impact on households and businesses will vary. One goal of our supervisory measures was to enable banks to continue providing credit and liquidity to the many businesses and households that will remain viable throughout the pandemic. Non-viable loans are a different matter. In order to maximise value and avoid cliff effects, banks’ credit treatment for non-viable businesses and households needs to be clear and transparent and their credit management must be strong to avoid loans piling up now. This will reduce the risk of procyclical effects in the months ahead as healthier days approach. If we delay – or worse, fail to act – we will have to pay the price of today’s inaction tomorrow. We would be wise to start preparing now so that we may enjoy the strongest possible conditions when those healthier days arrive.

With this in mind, today we sent a letter to the chief executive officers of the banks under our direct supervision, providing further guidance to management bodies on the importance of acting decisively to exercise strong credit risk management (see Letter to banks on identification and measurement of credit risk). In this blog post, I will discuss how this guidance fits into the overall supervisory strategy, which we have been pursuing since March and which is aimed at putting banks in the strongest possible position to support the post-pandemic economy. By strengthening credit practices now, we can improve the resilience of the banking sector, which in turn is a foundation for effectiveness of the transmission of monetary policy through the lending channel and of other support measures implemented over the last months. This is the best way to avoid debt overhang effects and procyclicality. I will discuss some concerns about credit risk caused by banks failing to accurately reflect reality in their credit treatment. In addition, I will highlight some best practices which we would like to see adopted more broadly.

Averting the procyclicality trap

The initial phase of the pandemic was marked by unprecedented spikes in macroeconomic uncertainty[1], lightning-speed implementation of social restrictions and economic support policies, and declining levels of business and consumer confidence[2]. The banking sector was faced with a dual threat: the potential for a sharp rise in risk aversion among lenders, and large-scale downgrades of borrower soundness exacerbated by a mechanistic application of existing rules and regulations. A deeply procyclical banking sector response to this dual threat would have disrupted the very life-blood essential supply of credit and liquidity to the economy and would have amplified the magnitude of the initial shock.

European and international authorities initially averted the procyclicality trap by deploying unparalleled and powerful monetary and fiscal support measures, moratoria and coordinated forward guidance. In March and April, ECB Banking Supervision, together with other EU institutions and international standard setters, issued a series of COVID-19-specific communications and guidelines to banks on the use of prudential and accounting frameworks in the exceptional environment that was taking shape.[3]

Two guiding principles underpinned these joint communications. First, the interpretation of prudential and accounting rules should not be mechanistic – particularly when determining the classification of exposures which are eligible for EBA-compliant moratoria and guarantee schemes, and when using forecasts for modelling purposes. And second, existing accounting and prudential rules should continue to be applied in order to identify any real unfolding credit deterioration as accurately as possible. Our guidance sought to assist banks in disentangling temporary financial difficulties from longer-lasting economic impacts indicative of significant increases in credit risk for certain borrowers and sectors most affected by the pandemic.

Against this backdrop, ECB Banking Supervision implemented the EBA Guidelines on moratoria and provided specific guidance to avoid excessively procyclical assumptions being made by banks when modelling expected credit losses and provisioning. We encouraged banks to derive estimates for their first-quarter reporting by exercising an informed judgement to update the March 2020 ECB staff macroeconomic projections to reflect the impact of the lockdown and massive public support measures that were being taken. We also asked banks to rely on the ECB’s projections from June 2020 onwards for the preparation of subsequent quarterly accounts. Finally, we encouraged banks to use their own long-term trends when estimating economic losses and provisions for time horizons beyond the scope of the ECB’s projections.

Overall, our initial efforts were successful in averting an excessively procyclical lending response. The first quarter of 2020 closed with euro area banks reporting only moderate tightening of lending standards, as well as intentions to loosen in the following quarter – in stark contrast to the tightening that occurred in the immediate aftermath of the great financial crisis.[4] Lending to businesses increased throughout the first half of 2020, as firms drew on emergency liquidity lines and started benefiting from public loan guarantees.[5]

A scattered picture of credit risk management practices

The European economic and policy environment began to stabilise in mid-2020. The finalisation of public support programmes allowed banks, markets and public institutions to more effectively incorporate the impact of such schemes in their projections. The consensus was that macroeconomic projections would be less volatile in the future, as shown by the broad consistency between the June 2020 Eurosystem staff macroeconomic projections and September 2020 ECB staff macroeconomic projections, both of which cover a three-year period.

Once euro area banks had averted large-scale mechanistic asset reclassifications and had moved beyond the initial shock response phase, we expected them to focus on addressing challenges in relation to credit risk management and identification. We expected banks to conduct viability assessments and adequately differentiate loan level credit quality. Loans temporarily affected by the pandemic and subject to modifications that do not meet the criteria for EBA moratoria should at least be earmarked for forbearance, so they are under enhanced monitoring, facilitating identification of longer-term deterioration at an early stage. Credit management tools, such as restructuring, can then be applied in a timely manner. Loans demonstrating longer lasting, structural impact on creditworthiness need to be identified and, on the basis of sound credit analysis, warrant more advanced staging in the credit risk identification chain. This may also require higher provisions, in accordance with the ECB Banking Supervision’s guidance to apply prudential and accounting frameworks so that unfolding credit deterioration is accurately depicted. Such proactive credit risk management practices would help to contain the build-up of problem assets at bank and cliff effects once moratoria and other support measures are phased out.

With this in mind, we sent a letter to banks[6] in July asking them to prepare and gear up their operational capacity to manage credit risk deterioration. We stressed the importance of ensuring that early arrears are managed in a timely manner through efficient systems for debtor outreach and for designing bespoke debt restructuring solutions. We also highlighted the need for regular and detailed risk reporting to management bodies based on adequate early warning indicators and risk segmentation of the portfolios.

Since then, we have been observing differences in credit risk management approaches, the use of overly optimistic assumptions and some worrying cases where banks have relaxed their risk-modelling standards. While model overlays grounded in a sound analysis of underlying portfolios may indeed be appropriate given continued uncertainty, blanket application of intrusive model overlays that are not consistent with macroeconomic scenarios or subject to strict governance oversight give cause for concern.

The following sections describe the issues we have identified as requiring improvement, as well as examples of what ECB Banking Supervision considers to be prudentially sound practices.

Postponing the recognition of unlikeliness to pay

Many banks are not sufficiently engaged in identifying borrower-specific signs of unlikeliness to pay (UTP) that stretch beyond merely temporary financial difficulties caused by the pandemic versus cases of more fundamental structural deterioration in credit worthiness. The percentage of loans on the balance sheets of all euro area banks which have been classified as non-performing since August due to unlikeliness to pay seems low at just 1.3%[7] – especially given the EBA’s guidance to prioritise UTP assessments within the portfolios subject to EBA-compliant moratoria but more broadly on the whole loan book. More worryingly, this percentage has remained broadly stable since March, even as it becomes increasingly clear that certain borrowers especially within the most vulnerable sectors will face more fundamental structural challenges that will impact their credit worthiness.

Some banks are prevented from effectively differentiating between viable and distressed borrowers because they lack sufficient resources and expertise to conduct in depth assessments of individual creditors factoring in the situation and outlook of each sector. Others are failing to react to the fact that some of the indicators widely used to identify deteriorating credit risk, such as the number of days past due, have been rendered no longer fit for purpose in the context of the pandemic. This is usually because they have not yet updated their risk identification systems in a way that would enable them to properly differentiate among borrowers under moratoria.

Unreported forbearance

On a system-wide level as of August 2020, the reporting of forborne loans subject to COVID-19 measures other than the EBA-compliant moratoria remained quite low and stable at 0.4% of total loans. This may be partially due to deficiencies in the flagging of forbearance which already existed before the pandemic or, in other cases, due to relaxation of the criteria for flagging forbearance, such as the exclusion of borrowers’ temporary financial difficulties. Another explanation identified by our supervisory teams is that forbearance through measures other than the EBA-compliant moratoria is not always being flagged as forbearance. This happens either because such modifications are not deemed to reflect the borrower’s financial difficulties, or because normal forbearance triggers, such as temporary financial difficulties, have been removed or muted from banks’ systems in order to manage the cases that are covered by moratoria. Unreported forbearance is a concerning behaviour that undermines the timely and accurate monitoring and management of credit risk developments.

Differences in loan loss provisioning practices

Chart 1

Differences in loan loss provisioning practices across euro area banks

Horizontal axis: change factor for the Cost of risk.
Vertical axis: change factor of IFRS9 stage 2 ratio.

Source: ECB Banking Supervision data.
Notes: These circles do not disclose any official threshold used by the ECB. They are aimed at identifying areas of behaviour. Horizontal axis: Cost of risk in 2020H1 divided by Cost of risk in 2019H1. Example: Value of 2 means a 100% increase in cost of risk in 2020H1 vs. 2019H1. Vertical axis: IFRS stage 2 ratio in 2020Q2 divided by same ratio in 2019Q4. IFRS stage 2 ratio = stage 2 loans / total loans subject to staging. Example: Value of 1 means the IFRS9 stage 2 ratio is unchanged between 2020Q2 and 2019Q4.

We have also observed differences in loan loss provisioning practices across banks. The ratio of Stage 3 loans has remained broadly stable since the start of the COVID-19 crisis, partly as a result of moratoria freezing the 90 days past due count, as well as low UTP assessments and low forbearance classifications. Moreover, as we can see from Chart 1, Stage 2 classifications, which should anticipate credit risk deterioration, have also remained low for many banks. While some banks – those in the bottom-left hand side of the chart – seem to be in a wait-and-see mode in that low Stage 2 classifications are matched with low provisioning, others – those in the bottom-right hand side of the chart – are making provisions by means of overlays, albeit without clearly identifying the exposures subject to increased credit risk. Some banks matched the cost of risk increase with transfers to Stage 2, reflecting the use of adequate provisioning practices.

We can also see that some banks appear to have simply increased the probability of default thresholds normally applied to identify any significant increase in credit risk, instead of developing pandemic-specific policies to isolate merely temporary financial stress not affecting lifetime creditworthiness from more fundamental deteriorations in creditworthiness. Others have disregarded or excessively modified the official macroeconomic forecasts available for the purposes of IFRS modelling.

But not all our observations reflect the need to strengthen practices. We can see that some banks are actively making strides to identify heightened risks across vulnerable economic sectors more likely to suffer far longer when the pandemic ends. These banks are enhancing the monitoring of their sector exposures, and in some instances, reclassifying them. Given that our work and leisure habits may change in some ways more permanently, the pandemic may well have a structural and lasting impact on certain segments of the economy. Some banks are evaluating this trend using supplementary information sources, such as sectoral economic analyses of corporate portfolio segments, or data on employment status for loans to households. Other banks have developed robust internal management information and indicators to track what happens to borrowers once they exit payment moratoria and have established clear criteria to distinguish cases of merely temporary forbearance situations from longer lasting and significant deteriorations. This information is valuable for understanding repayment capacity trends and for back-testing and informing modelling and scenario planning assumptions.

Inaction today will cost us tomorrow

As I draft this blog post, we find ourselves at a very delicate juncture. The health and safety outlook is gloomy. Infection rates are on the rise in many places, threatening further economic deterioration over at least the short to medium term and until the recently discovered vaccines are administered to a significant portion of the population. In this context, the EBA has reactivated its Guidelines on eligible moratoria. However, where not renewed, these programmes will more imminently expire. If banks do not strive now to understand and accurately reflect creditworthiness in their portfolios, they will pay the price later on, possibly a higher one.

We would do well to keep the risks in mind.

First, as targeted forbearance and timely debt restructuring can maximise value recovery, foregoing these measures is likely to lead to higher bank losses at a later stage. Recognising losses only when public support measures expire fosters the likelihood of a cliff effect, with stronger and more abrupt deleveraging actions, diminishing the banking sector’s capacity for post-pandemic value creation. Procyclicality will ensue and will probably be amplified.

Second, insulating financial reporting and bank balance sheets from credit risk deterioration by postponing reclassification, IFRS staging decisions, and NPL coverage and write-downs will undermine markets’ and investors’ trust in the European banking system. If balance sheets are opaque, a strong market reaction can be expected. This will be reflected in higher funding costs caused by ratings downgrades and distrust among counterparties. At a time when banks should be positioned to fully support the economic recovery, lack of transparency results in uncertainty as losses start crystallising. Again, in this situation, procyclicality would be amplified.

Third, as the not-so-distant past has shown us, inaction can give rise to a syndrome characterized by bank balance sheets clogged with high levels of deteriorated assets for prolonged periods. This in itself undermines economic recovery. We went through this experience in the last ten years, and the lessons should be clear by now. Post-financial crisis data for euro area banks shows that banks with higher levels of NPLs are likely to have a lower capacity to generate income. They also face higher funding costs and tend to lend less.[8] Furthermore, asset quality deterioration has a statistically significant negative impact on the transmission of monetary policy through the lending channel. In particular, banks with materially impaired balance sheets do not adjust their loan pricing in response to policy rate changes. Similar dynamics have contributed to the segmentation of the monetary policy transmission mechanism in the euro area since the great financial crisis.[9] We must avoid repeating the same mistakes this time around and avoid replacing the pandemic with the abovementioned syndrome.

It is not only an issue of credit supply; demand is also affected. NPLs are associated with corporate and household debt overhang. Corporate debt overhang in particular has been associated with weaker investment and delayed recoveries in the euro area, as economic agents overly exposed to non-viable debt are less responsive to positive shocks and stimuli and remain reluctant to engage in economic activity.[10]

Conclusion

In March and April, we advised banks to make use of the flexibility available under existing accounting standards to avoid excessive procyclicality, while at the same time continuing to identify and report structural asset quality deterioration and address the build-up of NPLs. We reminded banks not to reclassify broad swathes of borrowers automatically when implementing EBA-compliant moratoria without regard to true financial soundness and we promoted the use of realistic macroeconomic scenarios that would reflect the temporary nature of the pandemic.

Since then, not all banks have implemented our guidance to the same extent. In order to avoid longer-term procyclical effects, it is important that risk is adequately assessed, measured and managed, and that banks are operationally prepared. The goal of the public support measures and moratoria and of the guidance given to recognise the temporary nature of the shock was to foster borrower survival during the pandemic, not to mute idiosyncratic deterioration in credit quality unrelated to the pandemic or to ignore longer-lasting effects on creditworthiness generated by structural changes to the economy ushered in by the pandemic. Banks need to be rigorous in putting in place the tools and processes to differentiate between vulnerable and less vulnerable borrowers. Early recognition, management and measurement of vulnerable exposures is fundamental in preventing severe cliff effects at a later date – this was true for the last financial crisis and will be true as the current one unfolds.

Postponing proactive credit risk identification, measurement and management will delay and could amplify the procyclical reaction of the banking sector. It risks generating strong cliff effects at the least opportune moment, when most economic and balance sheet measures are withdrawn. In the medium term, it risks leaving both the banking system and its customers unable to proceed with much needed structural reform and engage in economic recovery. Last of all, postponement will thwart efforts to mitigate procyclicality and maximise leverage from policy transmission mechanisms through the lending channel when most needed.

Transparent data on borrower health needs to be made available to management and the markets. Adequate credit risk management marked by strong governance and the accurate reflection of risk allows for prudent balance sheet management and can even offer a strategic advantage. This will help to avoid procyclicality and the adverse impact on coordinated policy transmission mechanisms aimed at ensuring the pandemic does not turn into a financial crisis.

Our communication today reinforces this message in addition to the urgent need for banks to start their preparations immediately. The recovery is less likely to be protracted if we take steps now to gain clarity, quickly manage true weaknesses in the loan book and provide transparency on the credit effects of the pandemic. We are looking forward to the healthy days ahead.

[1]ECB analysis shows that indicators of macroeconomic uncertainty and forecast disagreement reached levels approximately eight times higher than during the great financial crisis. Economic Bulletin Issue 6/2020, ECB, Chart A, page 62.
[2]ECB analysis documents that indexes of business and consumer confidence fell by within two to five times their standard deviation between February 2020 and May 2020. Financial Stability Review, ECB, November 2020. Chapter 1, Chart 1.1.
[4]The ECB’s April 2020 euro area bank lending survey documents the net percentage of banks reporting a tightening in credit standards to enterprises at 4%, which compares to 60% in response to the financial and sovereign debt crises
[5]Financial Stability Review, ECB, November 2020, Chart 1.10.
[7]ECB supervisory reporting.
[8]See “A Strategy for Resolving Europe’s Problem Loans”, Staff Discussion Note, IMF, September 2015.
[9]See, for example, Byrne, D. and Kelly, R.(2019), “Bank asset quality and monetary policy pass-through”, Working Paper Series, No 98, ESRB, June.
[10]See footnote 8.