Southern Europe: IFRS 9 put to the test by the Covid-19 pandemic

The cost of risk^[1] in the Spanish, Italian and Portuguese banking systems has increased sharply as a result of the Covid-19 pandemic. Recording of provisions for impairment in 2020 was amplified by the forward-looking approach of the impairment model for financial instruments used in IFRS 9 Financial Instruments^[2]. Thus the banks recorded, at this stage of the economic shock, more additional provisions than if, as was the case under the previous IAS 39 regime, the impairment model had used an approach based simply on an analysis of days of a payment being past due. Moreover, banks have had to record provisions for impairment even if the associated financial assets will never default. Ultimately, excessive provisions for impairment will not only serve to cover future losses, they might also constrain bank lending, including lending that would have enabled certain borrowers to deal with difficulties which might have proved to be only temporary if credit had been available. The details of the application of IFRS 9 are therefore crucial during an economic shock, as they determine the scale of the increase in the cost of risk.

Applicable for annual periods beginning on or after 1 January 2018 (excluding transition period), IFRS 9 and its underlying principles have being tested by the Covid-19 pandemic. Provided that the health situation remains under control, we can cautiously begin to draw some lessons from this first use of IFRS 9 in the context of a major exogenous shock. The increase in the cost of risk for southern European banks has been all the greater as they are more largely focused on retail banking activities (deposit taking and lending to households and SMEs). However, the guidelines from the Single Supervisory Mechanism (SSM) and the European Banking Authority (EBA) have helped them limit, to some extent, the increase in their cost of risk. A strict automatic application of the IFRS 9 impairment model for financial instruments would have led banks to record an even greater increase in their cost of risk. Lastly, Spanish, Italian and Portuguese banks have an unprecedented aggregate capacity to absorb losses, whilst their non-performing loan (NPL) ratios might not increase as much as they have in previous crises.

Rise in cost of risks for southern European banks was amplified by their business model

The deterioration in the economic situation and in the solvency of banks’ debtors have led banks to record additional impairment provisions to cover future losses. This increase in the cost of risk has been amplified by the forward-looking approach of IFRS 9 impairment model for financial instruments. For Spanish, Italian and Portuguese banks, it has been further amplified by the fact that they are largely orientated towards lending activities and most notably lending to non-financial companies, including those most affected by the sanitary crisis.

The aggregate cost of risk doubled in 2020

Banks anticipated the deterioration of the situation

The economic fallouts of the Covid-19 pandemic have resulted in 2020, notably, in falls in GDP that were very pronounced in Spain (-11%) and somewhat smaller in Italy (-8.9%), with Portugal being in an intermediate position (-9.7%). This powerful exogenous shock resulted in a deterioration, at best temporary, in the position of companies and households. In order to deal with future losses, banks have recorded additional provisions for impairment. Banks’ cost of risk roughly doubled in 2020 in Spain (up 102.8%, to EUR 36 bn) and Portugal (97.8%, to EUR 2.5 bn). In Italy, it increased by as much as a factor of 2.5 (up 153%, to EUR 10 bn)^[3]. However, as a weighted average, these figures are less than one-third of those seen at the previous peak (EUR 97 bn in Spain in 2012, EUR 51 bn in Italy in 2013 and EUR 6 bn in Portugal in 2011^[4]). We should also note that 77% of the increase in the cost of risk in Spain during the second quarter of 2020 (Chart 1) was due to a EUR 12.6 bn adjustment in goodwill^[5] and deferred tax assets^[6] at Banco Santander. Even so, the peak in the cost of risk in Spain would nevertheless have fallen in the same period, even if we strip some of these exceptional items out of the aggregate cost of risk figure. The fact that the peak in Portugal also came in the second quarter of 2020 highlights the fact that the banks in these countries made substantial increases in impairment provisions in anticipation of a deterioration in the economic situation of their debtors. The Italian banking system saw a more gradual increase in its cost of risk over the course of 2020. This suggests that Italian banks made greater use of the margin of appreciation allowed to them under IFRS 9 (see below).

Deteriorated but healthy loans have increased, whilst non-performing loans have decreased

IFRS 9 impairment model for financial instruments is based on recognition of expected credit losses. Unlike the previous accounting standard (IAS 39 Financial instruments - Recognition and Measurement) which was solely based on incurred credit losses and the calculation of days of late payment, the materialisation of a credit event that provides an objective indicator of deterioration in the quality of a financial asset is no longer a necessary condition to record additional provisions for impairment. Under IFRS 9 it is not the defaults themselves that trigger the recording of loan loss provisions, but the mere expectations of loan losses (which, in general, precede actual defaults by some distance).

When a bank assesses that the credit risk associated with a financial asset has increased significantly since its initial recognition (e.g. when granted or purchased), it must record additional provisions for impairment. As a result, the onset of an economic shock will immediately bring a sharp increase in the cost of risk for banks, so that they can then, at some future point, cover any deterioration in the quality of their financial assets. This increase will be all the greater when the shock is substantial and was unexpected or insufficiently anticipated (an exogenous shock like, as it happens, the Covid-19 pandemic).

Under IFRS 9, where there has been a significant increase in credit risk on a financial asset since its initial recognition, it will be transferred from Stage 1 (healthy asset) to Stage 2 (deteriorated asset) or Stage 3 (impaired asset) as appropriate. As the quality of the asset deteriorates, additional provisions for impairment must be made. In the event of a significant reduction in credit risk, the bank can recognise reversals of provisions matching those made to reflect its deterioration. This symmetrical approach under IFRS 9 sets it apart from IAS 39, under which the reversal of provisions was also possible but only for assets in default where the amount of the loss proved to have been overestimated.

The reductions in Stage 3 assets between the fourth quarter of 2019 and the fourth quarter of 2020 (7.5% in Spain and 10.2% in Portugal) were all the greater given the substantial increases in Stage 2 assets (19.1% and 22.8% respectively). In Italy, the fall in Stage 3 assets was twice the amounts of those of Spain and Portugal, at 21.6%, whilst the increase in Stage 2 assets was three times as large, at 67.7%. Over and above the continued cleaning up of bank balance sheets, which contributed to the reduction in Stage 3 assets, the diverging trends between such assets and those in Stage 2 suggests that the increase in the cost of risk for southern European banks has indeed been amplified by the forward-looking approach of IFRS 9. The previous passive approach, based on incurred credit losses, would probably have resulted in a smaller increase in Stage 2 assets. The gap with the previous accounting standard is all the greater as Stage 2 did not even exist as part of IAS 39. Taken as a whole, financial reports from southern European banks tend to support this interpretation^[7]. Lastly, the forward-looking approach of IFRS 9 impairment model led the share of total assets in Stage 2 to increase through to the fourth quarter of 2020 (7.1% in Spain, 13.7% in Italy and 11.6% in Portugal, see Chart 2).

Retail activities amplify banks’ exposure to an increase in their cost of risk

Banks with a predominance of retail activities have historically had a higher cost of risk

The dispositions of IFRS 9 relative to impairment apply to financial assets recognised at their amortised cost, like bank loans, or at their fair value through other comprehensive income^[8], such as a corporate bond that is held only temporarily. Banks with a significant focus on retail activities tend to have a higher intrinsic cost of risk than more diversified banks.

Between 2017 and 2019, loans to retail clients reached an average total of 60% of bank balance sheets in the three countries (60% in Spain, 58% in Italy and 62% in Portugal). Over the same period, cost of risk levels, standardised by total assets, showed a wider dispersion: higher in Portugal (0.60% on average) and Spain (0.55%) and more contained in Italy (0.38%). By way of comparison, the fifteen largest banks in the euro zone – including three in Spain and two in Italy – which are primarily diversified banks, had significantly lower ratios: loans to retail clients represent an average between 2017 and 2019 of 51% of total assets for a standardised cost of risk of 0.22%^[9].

In 2020, the cost of risk, standardised by total assets, at Spanish banks and to a lesser extent those at Portuguese and Italian banks, was significantly above the average seen between 2017 and 2019 (1.01%, 0.74% and 0.71%, respectively). This meant that it remained higher than for the fifteen largest euro zone banks mentioned above (0.44% in 2020). These differences between the ratios for different banking systems tend to highlight the fact that, in the event of shocks that affect mainly non-financial corporations and households, the increase in the cost of risk is automatically greater for banks exposed to these client groups. Lastly, the relative increase in the cost of risk could be even greater for the smallest banks, which are generally pure retail banks and which are not included in our sample.

Southern European banking systems are particularly exposed to the branches most affected by the sanitary crisis

New loans, which are healthy by definition but which nevertheless require provisions for impairment under IFRS 9, contribute, albeit marginally, to an increase in a bank’s cost of risk. Thus, non-financial corporations (NFCs) which drew heavily on their authorised credit lines at the beginning of the health crisis (Chart 3) contributed to the increase in the cost of risk. New loans to NFCs cumulated over 12 months reached in June 2020 levels not seen since 2015 in both Spain (EUR397bn) and Portugal (EUR37bn). However, before their decrease, these figures remained well below record levels (EUR1,022bn in 2007 for Spain and EUR67bn in 2008 for Portugal). Once again the Italian banking system sets itself apart from the systems in Spain and Portugal, as new loans to NFCs (EUR496bn in February 2021) have yet to show any sign of slowing down and thus is closing in on its record level of EUR682bn in 2009.

The increase in the cost of risk also depends on banks’ exposure to the economic branches most affected by health protection measures and more specifically mandatory closures^[10]. These measures have accentuated (or in some cases created) difficulties for certain debtors. The high level of exposure of southern European banks to the branches most affected by the economic fallouts of the sanitary crisis therefore helps to explain the scale of the increase in their cost of risk. Moreover, this exposure has increased since the beginning of the Covid-19 pandemic, whilst, on average, it has fallen in the rest of the European Union (Chart 4). The Bank of Spain has suggested that the increase in the outstanding amounts of loans could be notably a consequence of the introduction of public support measures which were focused on the branches of the economy most affected by the pandemic^[11]. However, these public support measures were not sufficient to offset the effects of the increase in credit risk due to the Covid-19 pandemic on the cost of risk. Lastly, the increase of the latter may have been exacerbated by the situation of NFCs themselves, which as a whole, was more fragile at the onset of the sanitary crisis than it had been in the lead-up to previous economic downturns^[12].

Health protection measures also hit the solvency of households. However, loans to households consist largely of mortgages: their share in total lending to households^[13] (as an average over 2020) was particularly high in Portugal (79%) and Spain (73%), and slightly less in Italy (61%). However, the real estate sector, which was the cause of previous economic downturns in southern Europe, is not for the time being one of the branches most affected by the economic consequences of the Covid-19 pandemic. Moreover, the default risk amongst mortgage borrowers is, in general terms, lower than amongst consumer credit borrowers for example. The lending process means that mortgage loans are generally granted to a population of borrowers which, on average, is less exposed to the risk of unemployment than the general population. In addition, the near-automatic inclusion of a mortgage security limits the expected credit losses and thus the cost of risk relating to this client group (always provided, of course, that there is no significant fall in the value of the mortgaged properties).

The forward-looking nature of IFRS 9 has exacerbated the increase in the cost of risk

IFRS 9 is based on a set of principles which give banks some margin of appreciation when calculating their cost of risk (see Box 1). Excessively automatic application of IFRS 9 could have had the effect of constraining bank lending at a time when borrowers needed it most. From the same point of view, governments in southern European countries introduced various support measures, some of which benefited from preferential accounting treatment.

The flexibility of IFRS 9 does not fully neutralise its forward-looking nature

Any increase, even marginal, in credit risk on a large number of assets can produce a significant increase in the cost of risk

At each reporting date, and at least once a year, banks must first estimate if the credit risk associated with a financial asset has increased significantly since its initial recognition^[14]. They must then calculate the related expected credit losses. The assessment of a significant increase in credit risk on a financial asset is based on changes in the default risk that it could face through its residual life. Banks must use all information available to them on “past events, current conditions and forecast of future economic conditions” when making this analysis. Where possible, banks should use other criteria than simply the number of days of late payments^[15].

Because of the forward-looking nature of IFRS 9 impairment model, banks are liable to record additional impairment provisions, including for financial assets that will never fall into default. During periods of growth, recording provisions on financial assets where the substantial increase in credit risk is only temporary is less of a constraint for banks as they can absorb more easily thanks to higher net operating income. In addition, between two reporting dates, reversals of provisions on certain financial assets can offset the effect on the cost of risk of new provisions on other assets, where the possible deterioration in quality has not yet materialised. Under circumstances like the Covid-19 pandemic, the cost of risk relating to assets that will ultimately never fall into default is likely to increase substantially, which accentuates the temporary reduction in banks’ net income.

As the health crisis affected all economic agents, albeit to a varying degree, even a marginal increase in expected credit losses for a large number of financial assets whose deterioration will only be temporary can have significant consequences for the cost of risk. When other elements of the banks’ income statements are not sufficient to cover the increased cost of risk in full, it will start to erode their capital, limiting their capacity to lend, particularly to non-financial corporations, whose difficulties could be only temporary if they had access to additional lending.

The authorities have published guidelines to avoid an excessive increase in the cost of risk

Having recognised that IFRS 9 was likely to result in “inappropriate volatility in [banks’] financial statements”^[16], the European Securities and Markets Authority (ESMA), the European Banking Authority (EBA), the International Accounting Standards Board (IASB) and the ECB published in March 2020^[17] a set of guidelines setting out their expectations for the application of IFRS 9 in the context of the Covid-19 pandemic. Banks were thus encouraged to make full use of the flexibility afforded to them by the accounting standard for the two steps of the calculation of their cost of risk (1: assessment of significant increase in credit risk on a financial asset since its initial recognition; and 2: calculation of the amount of expected credit losses). This flexibility of assessment left at the banks’ discretion was intended to allow them to “faithfully reflect the specific circumstances of the Covid-19 outbreak” (ESMA, 2020) by adjusting to the circumstances the factors and underlying assumptions used in their calculations of their cost of risk.

Any automatic application of models unsuited to the sanitary context could have increased banks’ cost of risk well beyond the actual increase seen in 2020. New lending would have been constrained, whilst at the same time, demand from debtors would have increased in order to cope with their difficulties, sometimes temporary but which a drying up of their sources of financing would have aggravated. The guidelines therefore sought to allow banks to distinguish in more detail between financial assets where the significant increase in credit risk was only temporary and those where the increase in credit risk was irrevocable. It nevertheless remained an accounting requirement to identify and treat appropriately the financial assets whose credit quality has in fact deteriorated. In any event, the forward-looking approach of the impairment model gave rise, in 2020, to a higher cost of risk than if this had been calculated solely for the financial assets which were effectively in default. Over the economic cycle as a whole, the gap between the calculated cost of risk and the losses incurred should narrow thanks to this latitude in assessment.

Banking systems in southern Europe made use of the flexibility given by these guidelines. To an extent, the effects of the guidelines are to be found in the increase in the volume of financial assets put into Stage 2 (Chart 5). An assessment of a significant increase in credit risk on financial assets based only on automatic triggers would almost certainly have resulted in Spanish and Portuguese banks recording a much higher increase in their Stage 2 assets, particularly in the immediate aftermath of the first round of lockdown measures. However, the continued cleaning up of bank balance sheets (loans write-offs, disposals, etc.) does not allow us to draw conclusions on trends in Stage 3 assets.

The moderating effect of government support measures

Moratoria do not necessarily result in an increase in the cost of risk

In response to the outbreak of the Covid-19 pandemic, the governments of southern Europe, in line with many others, introduced a range of measures designed to limit the impact of the pandemic on their economies^[18]. Debt repayment moratoria, either automatic or on demand by the borrower, were one of the most popular support measures. On 30 June 2020, a date that represents more or less the peak of lockdown measures in Europe, outstanding amounts of bank loans to households and NFCs throughout the EU^[19] that were covered by a moratorium were valued at EUR 893 bn, or some 7.5% of these banks’ total assets^[20]. In Spain, 9.7% of bank loans to households and NFCs, or EUR 187 bn, were covered by a moratorium. This figure was as much as 13.3% in Italy (EUR 168 bn) and even 22.3% in Portugal (EUR 44 bn), amongst the highest levels in the EU. Across the EU, 60% of moratoria were to the benefit of NFCs and the remaining 40% to the benefit of households. Small and medium-sized enterprises (SMEs) made greater use of moratoria than NFCs as a whole. This was the case in Spain (6.8% of SME loans covered by a moratorium, compared to 5.4% for NFCs as a whole), Italy (25.3% and 14.7% respectively) and Portugal (34.6% and 29.3%).

When a financial asset is subject to payment delays because of a deterioration in its quality, this suggests, as a minimum, that its credit risk has increased significantly since its initial recognition. It must therefore be transferred to Stage 2 or Stage 3 as appropriate. A moratorium can significantly alter the nature of a financial asset and/or the terms of the contract and thus will affect the risk of default over its residual life, thus justifying its transfer to a different stage. Moratoria must therefore be taken into account by banks in the calculation of their cost of risk. In its guidelines of 2 April 2020, the EBA stipulates that moratoria, whether legislative or simply contractual, are not necessarily an indication of a significant increase in the default risk on a financial asset. However, such moratoria must have been granted by banks to a large community of borrowers. They must not be in response to specific criteria (notably an increase in credit risk) but should address temporary difficulties, such as liquidity constraints, for borrowers. Unless shown otherwise, moratoria that do not comply with EBA guidelines are by default considered to be indications of a significant increase in credit risk, requiring the recognition of additional provisions. ESMA takes the view that moratoria do not result in a significant reduction in the net present value of financial assets if the extension of the payment schedule is the only change made to the terms of the contract. Changes in interest rates are nevertheless allowed provided that their only effect is to compensate the bank for the consequences of the extended payment period.

The high proportion of bank loans covered by moratoria (Chart6), suggests that an automatic application of IFRS 9, not taking account of circumstances, would have led to a significantly larger increase in the cost of risk for southern European banks.

The expiry of moratoria could provide more information on the deterioration in the quality of certain financial assets

Despite being awarded indiscriminately, in practice moratoria were focused primarily on those borrowers hardest hit by the Covid-19 crisis. Thus, the banks estimated that at 30 June 2020, some 17% of financial assets covered by a moratorium were in Stage 2, a figure twice as high as for assets not so covered. Meanwhile, the share of Stage 2 assets still covered by a moratorium at 31 December 2020 (26.4%) was significantly higher than the share for which the moratorium had expired (20.1%). The future trend in the cost of risk at southern European banks will depend, amongst other things, on financial assets for which the moratorium has yet to expire. This could be a particular issue in Portugal, where 88% of moratoria granted were still in force at 31 December 2020. This proportion was lower in Spain (32%) and Italy (65%), but these levels are still amongst the highest in the European Union (35%).

Given that some moratoria, by providing breathing space to certain borrowers, will have hidden the deterioration in the quality of certain financial assets, their expiry could now reveal it. Moreover, the continuation of the sanitary crisis and lockdown measures has warranted the extension of moratoria (which were introduced to address exactly such conditions). As a result, in December 2020, the EBA renewed the preferential treatment of moratoria to 31 March 2021. This renewal came after the expiry of preferential treatment on 30 September 2020, which in turn followed a three-month extension in June 2020. The preferential accounting treatment is intended to encourage banks to increase the number of moratoria granted to borrowers, and thus protect the latter from excessive financial pressure. The effects of moratoria and their expiries will not be fully visible in southern European banks’ income statements for a number of quarters.

By limiting expected credit losses, government guarantees encourage banks to lend more

When the credit risk associated with a financial asset has increased significantly since its initial recognition (i.e. when it is transferred to Stage 2 or Stage 3) banks must “recognise a loss allowance for expected credit losses”^[21]. As expected credit losses are calculated across the entire residual life of the financial asset (rather than over the next 12 months, as is the case in Stage 1), the introduction of government loan guarantees reduces the cost of risk. The measures introduced by southern European governments have thus helped reduce expected credit losses, as, on a weighted average basis, the share of total government guaranteed loans in Stage 2 in the fourth quarter of 2020 was around 10% (Table1). At this point, Spanish banks had issued the largest amount of guaranteed loans in southern Europe. The share of total loans represented by government guaranteed loans was nevertheless similar in Spain and Portugal. The share of new loans covered by government guarantee was also very similar in the two countries. The most generous government coverage came in Italy, with around 87% of new loans benefiting from a guarantee. Proportionally, therefore, southern European governments were the largest issuers of government guarantees on new lending. Such guarantees were very much targeted at supporting economic activity, with non-financial corporations receiving 95% of government guaranteed loans, only 2% of which were renegotiated loans.

By reducing expected credit losses, government guarantees (whether full or partial) loosen the link between a deterioration in the quality of financial assets (transfers to Stages 2 or 3) and an increase in the cost of risk (Chart7). More generally, the cost of risk of each bank was affected, most notably in 2020, by the allocation of its lending book to branches most affected by the pandemic, by the dates and nature of the lockdown measures and by specific features of national insolvency law, which can result in substantial differences in the time to realise the collateral and thus its value at the time of its realisation. An increase in the cost of risk at a bank, or in its coverage ratio^[22], cannot be interpreted solely as a reflection of an effective deterioration of its loan book or of a conservative approach, and this is all the truer under IFRS 9.

In the fourth quarter of 2020, more than 85% of new loans with a government guarantee in southern Europe had a maturity of over 2 years (Chart8). The greater the average residual maturity of government guaranteed loans, the longer the reduction in the cost of risk for banks will last. However, such loans still make up only a marginal proportion of total loans. As a result, although their effect on the cost of risk might last for several years, it will remain limited.

Towards a limited decline in bank solvency and a managed increase in non-performing loans

Internal capital generation by southern European banks fell back sharply in 2020. Their capacity to absorb losses is nevertheless at a much higher level than it was immediately prior to the onset of previous crises. Moreover, the increase in non-performing loans could remain relatively controlled, thanks in particular to a continued cleaning up of bank balance sheets.

Banks have the capacity to absorb further increases in the cost of risk, despite lower net income

The cost of risk has reduced banks’ capacity to generate capital internally

The reduction in net operating income, combined with an increase in the cost of risk, has put a squeeze on net income at southern European banks. Thus the ability of these institutions to allocate profits to reserves has been restrained. Net income for the largest banking groups in Spain and Italy^[23] contracted on average by 40.4% and 19.8% respectively in 2020. In Portugal, the scale of the fall in net income at banks was such that on average they recorded losses equivalent to 37.4% of their 2019 net income. Thus, the return on average assets was negative in the fourth quarter of 2020 in both Spain and Portugal (at -0.26% and -0.02% respectively, from 0.52% and 0.36% respectively in the fourth quarter of 2019). The Italian banking system once again set itself apart from its Iberian peers with a return on average assets that just about stayed positive in the fourth quarter of 2020 (0.03%, from 0.42% a year earlier). It was thus closer to the figure for the European Union as a whole (0.39% in 2019 and 0.13% in 2020).

The overall capacity of southern European banks to absorb losses is at a record high level

Bolstered by a continuous increase in their solvency ratios over the past ten years, southern European banks now have a much greater capacity to absorb losses than they did in 2008. Overall, between 2008 and 2020, Spanish and Portuguese banks increased their capital by around one-third (34% and 38% respectively) whilst the increase was more modest for Italian banks at 16%^[24]. The smaller increase here was due, amongst other things, to a much less significant reduction in risk-weighted assets in Italy (-23%, against -31% in Spain and -42% in Portugal). On this point, the SSM believes that, all other things being equal, southern European banks should be capable of absorbing additional impairment of their loan portfolios to branches most affected by the pandemic (mining and quarrying, manufacturing, wholesale and retail trade, transport and storage, accommodation and food service activities and arts, entertainment and recreation). The SSM calculates that Portuguese banks have the biggest cushion: they could bear additional impairment of 39% before their regulatory minima (Pillar 1 and Pillar 2 requirements, set at 5.5%) came under threat^[25]. Italian and Spanish banks have smaller, but nonetheless substantial cushions (21% and 25% respectively) before their solvency ratios reach the critical thresholds (5.6% and 5.4% respectively). According to the SSM, this would correspond to levels of impairment significantly greater than the peak levels seen for non-financial companies in Italy, Ireland and Portugal after the sovereign debt crisis that hit the euro zone in 2010-2012.

The ‘management buffer’ of capital built up by banks in excess of regulatory requirements, due in particular to the implicit requirements of the market, would on its own be enough to absorb a weighted average impairment of 10% on their portfolios of loans to the branches most affected by the Covid-19 pandemic (8% in Spain, 10% in Italy and 15% in Portugal, Chart9). The ability of banks to use these capital buffers nevertheless remains uncertain due to other capital requirements (for instance the minimum requirement for own funds and eligible liabilities (MREL) and total loss-absorbing capacity (TLAC)), implied market requirements and, above all, the need to compensate shareholders for the non-payment of dividends^[26] following the ban on payments and then the cap of 15% of cumulative profit for 2019-2020.

Transitional measures that allowed banks to include temporarily part of the provisions recognised on initial adoption of IFRS 9 (mainly for Stage 2 assets, which did not exist under IAS 39) in their CET1 equity have been extended for two years “to mitigate the potential impact that a sudden increase in expected credit loss provisions could have on institutions’ capacity to lend to clients at times when it is most needed”.^[27] Indeed banks which initially elected not to apply the transitional measures were given an opportunity to reverse that decision. These measures helped to limit the effects of the Covid-19 pandemic on regulatory solvency ratios at banks. The SSM and national supervisory bodies have gone even further by temporarily removing certain components of the combined requirements (capital conservation buffer, countercyclical capital buffer, G-SIB surcharge, systemic risk buffer) and using their discretion to restrict dividend policies, whilst banks were already authorised under Basel III to operate with capital ratios below the combined requirements^[28]. It is however unlikely that banks would take the risk of being penalized by the market for reducing their capital ratios below its implied requirements.

Non-performing loan ratios might not increase as much as they have in the past

Stage 2 assets do not necessarily become non-performing loans

The definition of non-performing loans, in the sense of the ECB’s guidelines^[29], is fairly close to that of Stage 3 assets. However, the former is based on an automatic approach, using primarily days of late payment, whilst the latter is based on a forward-looking approach. Thus there may be a degree of divergence between the trend in non-performing loans and that in Stage 3 (impaired) assets. The relationship between non-performing loans and loans in Stage 2 (deteriorated) is even looser. At most, an increase in the volume of assets in Stage 2 can be seen as an advanced indicator of an increase in non-performing loans.

As with the analysis of a significant increase in credit risk on a financial asset, legislative or contractual moratoria which are compliant with the EBA’s guidelines^[30] do not necessarily require the loan to be classified as non-performing. Loans covered by a moratorium do not automatically satisfy the ‘unlikeliness to pay’ criterion of the definition of non-performing loans^[31]. Provided that it is not an emergency restructuring measure as defined in the EBA’s guidelines on the definition of default^[32], the introduction of a moratorium is not necessarily a sign of expected payment default (unlikeliness to pay in the sense of Article 178(3)(d) of the Capital Requirements Regulation, or CRR^[33]). These structures thus help explain the continued fall in the non-performing loans ratios in southern Europe during 2020 (Chart10).

The EBA^[34] and the SSM, in particular, feared that the expiry of moratoria would give rise to a significant increase in non-performing loans due to an almost automatic increase in those with payment being past due more than 90 days (the ‘past due’ criterion of the definition of non-performing loans). However, in the second quarter of 2020, the share of non-performing loans in the total of financial assets covered by a moratorium was slightly lower than their share in the total loan portfolio (2.5% against 2.9%). The EBA attributed this – relatively low – difference to the fact that moratoria concerned mainly healthy assets. Moratoria that have not expired are nonetheless likely to cover a growing share of assets for which a significant increase in credit risk will present a more definitive nature.

The cleaning up of bank balance sheets will probably continue despite the pandemic

Despite the Covid-19 pandemic, over the next few quarters the volume and ratios of non-performing loans in southern European banking systems could, as in 2020, continue the downward trend that began in 2016. In 31 December 2020, they reached their lowest level since this data series began in the third quarter of 2014. Innovations in the securitisation of non-performing loans could help remove these from bank balance sheets, despite a drop in the appetite of secondary markets^[35]. Lastly, regulatory requirements on minimum coverage levels for non-performing loans are likely to continue to force banks to remove ‘old’ non-performing loans from their balance sheets. Eventually, this same constraint will apply to ‘new’ non-performing loans. The ‘Quick Fix’ banking package put in place will nevertheless give an additional period of up to seven years after a loan is classified as non-performing before a bank is required to cover 100% of its gross book value with provisions^[36]. It gives banks more time to record provisions that might be greater than actual expected losses and that banks would probably not have had recognised in the absence of this regulatory requirement.

***

The impairment model for financial instruments introduced by IFRS 9 in 2018 has been put to the test by the Covid-19 pandemic. Its forward-looking approach resulted in a doubling of the cost of risk, as a weighted average, for Spanish and Portuguese banks between 2019 and 2020. Despite what appears to be a more flexible application of the accounting principles of IFRS 9, the increase in the cost of risk at Italian banks was even greater, as it increased by a factor of 2.5 over the year as a whole. This was due primarily to a much steeper rise in loans in Stage 2 in Italy (loans that are still healthy but where the associated credit risk has increased significantly since initial recognition). As with the banking systems in Spain and Portugal, the Italian system has much greater capacity to absorb losses now than it had at the onset of previous economic shocks, thanks to the increase in capital and the cleaning up of balance sheets; this said, it is still in a less favourable position relative to its two peers.

The forward-looking approach of the impairment model results in a considerable increase in banks’ provisioning at the time of the initial shock, but provisions tend to be lower thereafter. With this in mind, the SSM and EBA have set out guidance on the application of the IFRS 9 principles specific to its use for the first time during an economic shock. Excessively automatic application could have increased the difficulties experienced by certain debtors. Meanwhile, against a background of economic recovery in 2021, the banks of southern Europe are likely to see a cost of risk that is lower than in 2020. However, in the absence of a renewal of support measures, the resulting expiry of moratoria could reveal a deterioration in certain debtors’ positions that would slow, but not stop, the process of reducing the cost of risk.

^[1] The cost of risk corresponds to impairment provisions recognised over a period, less reversals of provisions. It is equal to the sum of net allocations to impairment, recovery of amortised receivables, losses on irrecoverable loans, impairment recognised during the period and net allocations to impairment on tangible and intangible fixed assets.

^[2] Commission Regulation (EU) 2016/2067 of 22 November 2016.
See in particular Humblot T, 2018, The impacts of IFRS 9 first-time adoption on southern European banks, Conjoncture, BNP Paribas for a more detailed presentation.

^[3] Representative samples of the Spanish, Italian and Portuguese banking systems have been constructed on the basis of those used by the ECB and EBA in order to facilitate comparison with data from these sources. Due to a lack of published data, some banks were removed from our sample, whilst others were added to ensure that the sample remained representative (see table in Appendix). Underlying net income, which excludes some exceptional items, was used in 2020 for Banco Santander in Spain and UBI in Italy in order to give a more accurate picture of the two countries’ banking systems.

^[4] The move from IAS 39 to IFRS 9, coupled with structural changes in the Spanish, Italian and Portuguese banking systems, limits the scope of historical comparisons, which are provided more to give orders of magnitude than to offer a basis for direct comparison.

^[5] Difference between the bank’s net book value and its market valuation. A negative gap is considered as ‘badwill’ or negative goodwill.

^[6] Losses not set off against taxable income in the year in which they arose. These give rise to a credit with the tax authorities which may be deducted from the bank’s future tax charges.

^[7] See, for example, page 448 of the Santander group’s 2020 annual report.

^[8] “Total other comprehensive income” in the definitions of IFRS 9.

^[9] Generally, size and diversification are linked. Reaching critical mass often requires sufficient diversification of banking activities.

^[10] The ECB noted in May 2020 that the areas of the economy worst affected were mining and quarrying, manufacturing, wholesale and retail trade, transport and storage, accommodation and food service activities and arts, entertainment and recreation.

^[11] Banco de España 2021, Financial Stability Report, Spring.

^[12] Ari, A., Chen, S. and Ratnovski, L., 2020, Covid-19 and non-performing loans: lessons from past crises, Research Bulletin (71), ECB

^[13] And to non-profit organisations providing services to households.

^[14] Article 5(5)(9) of Commission Regulation (EU) 2016/2067 of 22 November 2016, IFRS 9, Financial instruments

^[15] Article 5(5)(11) of Commission Regulation (EU) 2016/2067 of 22 November 2016, IFRS 9, Financial instruments

^[16] ECB, 2020, Guidance on the use of forecasts to estimate the ECL during the Covid-19 pandemic, 1 April 2020

^[17] Respectively, ESMA, 2020, Public statement – Accounting implications of the Covid-19 outbreak on the calculation of expected credit losses in accordance with IFRS 9, 25 March 2020; EBA, 2020, Statement on the application of the prudential framework regarding Default, Forbearance and IFRS 9 in the light of COVID-19 measures, 25 March 2020; EBA, 2020, Guidelines on legislative and non-legislative moratoria on loan repayments applied in the light of the COVID-19 crisis, 2 April 2020; IASB, 2020, Accounting for expected credit losses applying IFRS 9 Financial Instruments in the light of current uncertainty resulting from the covid-19 pandemic, 27 March 2020; ECB, 2020, Identification and measurement of credit risk in the context of the coronavirus (COVID-19) pandemic, 4 December 2020

^[18] See in particular the monitoring provided by the European Systemic Risk Board (ESRB): Policy measures in response to the COVID-19 pandemic

^[19] These banks account for approximately 80% of total banking system assets across the European Union.

^[20] EBA, 2021, First evidence on the use of moratoria and public guarantees in the EU banking sector, November 2020, Thematic note (31)

^[21] Article 5(5)(1) of Commission Regulation (EU) 2016/2067 of 22 November 2016, IFRS 9, Financial instruments

^[22] Ratio of provisions to non-performing loans

^[23] Representative samples of the Spanish, Italian and Portuguese banking systems have been constructed on the basis of those used by the ECB and EBA in order to facilitate comparison with data from these sources. Due to a lack of published data, some banks were removed from our sample, whilst others were added to ensure that the sample remained representative (see table in Appendix). Underlying net income, which excludes some exceptional items, was used in 2020 for Banco Santander in Spain and UBI in Italy in order to give a more accurate picture of the two countries’ banking systems.

^[24] By way of comparison, capital in the EU banking system as a whole increased by 60% over the same period.

^[25] ECB, 2020, Financial stability review, May

^[26] Investors are likely to increase temporarily their implied requirements in order to ensure that they will receive a dividend later in compensation for those they are forced to forgo today.

^[27] Regulation (EU) 2020/873 of the European Parliament and of the Council of 24 June 2020

^[28] Quignon, L., 2021, Du caractère utilisable des coussins de fonds propres bancaires, Revue Banque (855)

^[29] More accurately, non-performing exposures. See: ECB, 2017, Guidance to banks on non-performing loans.

^[30] EBA, April 2020, Guidelines on legislative and non-legislative moratoria on loan repayments applied in the light of the Covid-19 crisis

^[31] Annex V, Paragraph 145 of Commission Implementing Regulation 2015/227 of 9 January 2015 amending Implementing Regulation (EU) No 680/2014 laying down implementing technical standards with regard to supervisory reporting of institutions

^[32] Paragraph 49, EBA, 2016, Guidelines on the application of the definition of default under Article 178 of Regulation (EU) 575/2013

^[33] Regulation (EU) 575/2013 of the European Parliament and of the Council of 26 June 2013

^[34] EBA, 2021, First evidence on the use of moratoria and public guarantees in the EU banking sector, November 2020, Thematic note (31)

^[35] On 22 March 2021, Banca IFIS securitised an NPL portfolio including of 69% of non-secured loans backed by orders of assignment where recovery through compulsory enforcement was already at an advanced stage. This portfolio received the best rating ever given to this type of product. See, in particular, Banca IFIS, 2021, Banca Ifis develops the first securitisation in Italy of NPLs assisted by orders of assignment

^[36] Regulation (EU) 2020/873 of the European Parliament and of the Council of 24 June 2020