The document submitted for consultation by the EBA follows the Guidelines on managing non-performing loans published in March 2017 by the ECB banking supervision, later followed by the July 2017 European Council decisions which set out an action plan for non-performing loans. This was then followed by the consultation on the proposal to the ECB Addendum, later finalised in March 2018, and by the European Commission proposal which will now have to follow the European legislative process before becoming a Community law.
Despite following somewhat different approaches (at least in some cases), these measures can sometimes overlap, and this process does not always appear to be adequately coordinated.
In general terms, ABI points out the legal uncertainty caused by the quick succession of primary regulations, secondary regulations and guidelines on NPLs, which are not always coordinated and balanced; it does not facilitate compliance by the banking sector, nor does it support non financial firms or households. Furthermore, each new proposed rule has the potential to impact the pricing of NPL portfolios and risks jeopardize the development of a liquid and efficient secondary market for such asset class.
In particular, the EBA Guidelines – which introduce a specific quantitative threshold to determine the category of banks with a high level of NPLs, set at 5% of the NPL ratio (total non-performing loans on total loans) – seem to lack sufficient justification, especially given the differences in the timeframe of legal proceedings for loan recovery which continue to vary widely among Member States of the European Union.
The consultation sets a threshold in order to identify the Banks that are required to establish and apply a recovery strategy, the guidelines for which are detailed in chapters 4 and 5 (NPE Strategy and NPE Governance). This threshold is set at a 5% NPL ratio (Non-Performing Loans/Total Loans).
While it is true that the average level in Europe in 2016 was already close to 5%, in some countries this level has been well exceeded.
In our opinion, regardless of the threshold’s value, focus must be placed on the costs necessary to be able to apply the recovery strategies (included within the ICAAP, RAF and Recovery Plan) from as early as 2019 and finance the unplanned investments necessary in 2018 to ensure all the operational requirements set out in the chapter on Governance (particularly technical resources). Abi therefore asks for a gradual introduction of the operational requirements.
Moreover, we’ve also some concerns on the appropriateness of the threshold as proposed by the EBA. Following we propose an alternative procedure for calculating the threshold.
The EBA guidelines (like those of the ECB banking supervision ) set out stricter regulatory requirements in terms of governance, organisational structures and reporting for riskier banks, i.e. those with an high level of NPLs. One of the main differences between the EBA and the ECB banking supervision is that the EBA sets a specific threshold for banks to be classified as riskier:
• for the EBA, banks are classified as riskier if they have a gross NPL ratio above 5% (in total or across individual portfolios: e.g. mortgages, SMEs, specialised lending, etc.). Please also note that the EBA threshold would therefore not vary over time based on macroeconomic conditions, such as the economic cycle.
• On the other hand, for the ECB banking supervision, riskier banks are those with high NPLs (with no set threshold), in total or in individual portfolios.
Instead of setting a figure of 5%, ABI proposes that the threshold should be determined based on specific evaluations on each bank by the ECB banking supervision under its Supervisory Review and Evaluation Process (SREP). Indeed, the 5% NPL ratio is the current European average figure (close to pre-crisis figures in Italy), yet the possibility of reaching this figure in a short timeframe should be assessed by:
• considering the various existing risk scenarios of individual European jurisdictions, which result in non-uniform gross NPL ratios between countries (ranging from 1% to 47%); therefore, it should be taken into account the need to gradually work towards reaching the above threshold so that the burden of the high – direct and indirect – costs for banks and for the economy is not only on some intermediaries;
• considering that different exposures (to large enterprises, SMEs, residential mortgages, consumer credit, etc.) have risk characteristics that differ massively both structurally and cyclically.
• assessing the factors that affect the speed at which NPLs are removed from banks’ balance sheets (especially the timeframes for closing ordinary loan recovery procedures, which are known to vary widely amongst European jurisdictions).
In other words, from ABI’s perspective, the EBA’s proposal to introduce by 2019 a fixed threshold without distinction , i.e. a 5% NPL ratio, to discriminate against banks with high NPLs, should be revised to take into account specific factors for individual banks, which should be assessed by the Banking Supervision under its Supervisory Review and Evaluation Process (SREP), and:
1. to gradually introduce this threshold figure
2. to differentiate the threshold by different portfolio of exposure
3. to have the threshold vary over time according to the economic cycle
The approach and data we use to support our proposal are based on the following formula, which expresses the gross NPL ratio based on 3 parameters:
Gross NPL ratio = DR x EoDL x NPLTime
• DR = average default rate (of the portfolio)
• EoDL = average volume of NPL on the balance sheet, i.e. average volume of exposure from the time they become non-performing to the time they are removed from the balance sheet – due to write-offs or transfer to third parties – (which, for example, can be calculated as a simple average between 100% and the Loss Given Default, assuming the recoveries are in line with the removal from balance sheets)
• NPLTime = period between the time the exposure becomes non-performing and the time it is removed from the balance sheet (due to its return to being performing or write-off upon completion of proceedings or sale to third parties).
The above formula essentially summarises the trend of the NPL ratio by NPL generation and makes it possible to:
• determine the maximum time the NPL are recorded on the balance sheet in line with a pre-set NPL ratio target, given a certain average default rate
• determine the implicit minimum NPL ratio: in the current default rates, considering the minimum average time the NPL remain on the balance sheet.
Shown below we outline the 3 objectives set out above.
1. GRADUAL APPROACH TO REACHING THE THRESHOLD
Based on Formula 1, by applying the European average data to the 2.1% default rate and 42.4% LGD, and therefore a 71% EoDL, we can see that the NPL ratio of 5% can only be achieved if the NPL remain on the balance sheet for an average of no more than 3.4 years (41 months).
Formula 1: DR * EoDL * NPLTime = gross NPL ratio
where NPLTime = gross NPL ratio /(DR *EoDL)
so NPLTime = 5% (2.1% * 71%) = 3.4 years
The 5% threshold is therefore compatible with the NPL remaining on the balance sheet for a maximum period of 41 months. This timeframe appears to be inconsistent with the internal NPL management – especially in jurisdictions where ordinary debt recovery procedures far exceed this timeframe – and would run the risk of having dangerous effects on the economy. In fact, to avoid reaching the 5% threshold, banks:
1. may be encouraged to try to reduce default rates by applying more selective lending policies, which could penalise small and medium-sized enterprises which have structurally higher average default rates;
2. may be encouraged to try to decrease the average time NPL remain on the balance sheet by making a fast and comprehensive transfer of NPL to third parties, which would potentially entail a large cost (for banks, their customers and the economy as a whole), due to the gap between the book value and the NPL market prices.
Both these strategies could lead to a decrease in lending to households and non financial firms (especially SMEs), with negative effects for customers, banks and the economy as a whole.
To mitigate these effects, ABI proposes to introduce the NPL ratio threshold gradually. In line with the framework shown above (NPLTime = 3.4 years), we would suggest to set a four-year phase-in period.
However, rather than the threshold being 5% for all portfolios, it should be determined based on the various degrees of implicit risk in the different business exposure.
2. DIFFERENTIATING THE THRESHOLD BY EXPOSURE PORTFOLIOS
We have created Table 1 below, based on Formula 1 and using the average 2013-2017 data of 9 large European AIRB groups.
On the one hand, Table 1 reiterates that for all lending to households and firms in Europe, in order to achieve a NPL ratio of 5% NPL can remain on the balance sheet no longer than 41 months (NPL-time = max 41 months). On the other hand, it shows that the implicit NPL-time can vary greatly among portfolios and is particularly low for exposures to SMEs (25 months on average), while it is higher for residential mortgages to households (56 months) and for exposures to large enterprises (58 months).
TABLE 1 – By portfolio exposure
(average figures 2013-2017; based on Pillar 3 figure released by 9 large EU AIRB banking groups;)
Default rate LGD NPL-TIME implicit in a NPL ratio level of 5%
Total 2.1% 42.4% 3.4 41
1.Corporate Exposure 1.5% 42.7% 4.5 55
- Specialized financing 5.7% 38.3% 1.3 15
- Exposure to SMEs 2.8% 48.4% 2.4 29
- Exposure to other
corporations 1.5% 41.7% 4.9 58
2.Retail exposure 1.9% 41.2% 3.7 44
- Exposures to SMEs
secured by residential
property 3.4% 42.3% 2.1 25
- Exposures to private individuals secured
by residential property 1.6% 31.6% 4.7 56
- Other exposure to SMEs 3.2% 67.7% 1.9 22
A fixed 5% threshold for all portfolios implies maximum periods for NPL to remain on the balance sheet, which vary among portfolios. It may disincentives developing exposures where this timeframe is lower (as is usually the case for lending to SMEs in its various forms or specialised lending) in favour of other exposures (lending to large companies or household mortgages), the economic impacts of which are easily predictable.
ABI therefore proposes a threshold system that is differentiated by portfolio, which would be determined based on the average risk level – e.g. over a four-year period – of the individual exposure. The figure would then be subject to the gradual approach outlined in paragraph 1.
Choosing this approach rather than a fixed 5% threshold would provide the following benefits:
1. the threshold would be achieved gradually;
2. the threshold would be differentiated by portfolio and the firms with structurally higher default rates (such as lending to SMEs) would have a higher threshold; this is the only way, comparing different business lines, to create an incentive for lending to SMEs – something that the EBA’s current approach does not provide, with the risk of there being a lack of lending to SMEs in a period of recession.
One last aspect to be considered is the need for the threshold to be variable according to the cycle, rather than fixed over time.
3. VARIABLE THRESHOLD ACCORDING TO THE CYCLE
To take the developments of the economic cycle into account, ABI proposes that, from the year that they come into effect (2022, based on our assumption), thresholds should be recalculated every four years, applying the average default rate and EoDL figures recorded over the last four years (e.g. between 2018 and 2021).
In this way, the threshold would vary according to the cycle, i.e. in periods of recession the threshold would tend to be higher and vice versa.
In summary, therefore, according to ABI the definition and calibration of the NPL ratio threshold should be based on the following principles:
Determining the NPL ratio threshold:
- by the Banking Supervision
- for each individual bank as part of the SREP
Setting the threshold:
- different for each business portfolio
- gradual introduction over time
- variable over time
According to the chapter 6 on Forbearance, before granting any forbearance measures, banks should assess borrower’s creditworthiness or in general borrower’s ability to pay in the future (at the end of the forbearance measures). In Italy, laws issued by Government or Agreements signed with Associations of consumers or enterprises, provided mandatory forbearance measures (e.g the moratoria on payment of instalments) for banks in case of specific events regarding to the borrower (e.g. loss of jobs, death, earthquake). As a matter of fact, it should be provided that in case of law, regulation or Institutional Agreement where forbearance measures are mandatory, banks should be exempted from the requirements of Chapter 6.
Chapter 9 deals with Collateral valuation. The document proposes that the revaluations be conducted by specialised technicians for exposures with a gross book value of more than Euro 300,000.
“Indexed valuation should not be used to update the valuation of immovable property collateral for a NPE, the gross carrying amount of which is larger than Euro 300,000 or a lower amount defined by the competent authority.”
Given also past experience, the limit of Euro 300,000 seems to be very conservative and results as being excessively penalising in the cost/benefit ratio since it would increase fixed NPL management costs without modifying recovery capacity. A threshold of around Euro 1 million could constitutes a more efficient balance.
There are no particular comments regarding the guidelines in Chapter 9 on the valuation of movable property collateral.
Nevertheless, it should be noted that if Leasing activities were to also fall within a banking group’s business, there would be transactions where there is an asset on which the recovery value is estimated but which does not constitute a “traditional” collateral.
EBA Guidelines consider the banking NPLs as unique, without any distinction between the different technical forms of financing. According to this approach, the collateral to bank exposures is always defined with respect to the traditional" type of collateral, whose enforcement requires specific judicial processes (eg foreclosure of capital goods, release of premises in the case of real estate, etc.) and processes for realizing the value of the asset subject to public procedures (eg judicial auctions). These processes for the realization of collateral are long and expensive and in most cases lead to realizable values that are significantly lower than market values, with a strong negative impact on the final loss.
On the opposite, leasing contracts, by virtue of direct ownership by the "collateral" institute, benefits from short time of disposal of property collateral as well as less costly recovery procedures of the financed assets. Moreover, once the leasing company has repossessed the collateral, it can sell it under normal market conditions without the need for judicial procedures, thus succeeding in obtaining a normally higher price and reducing the loss associated with the NPE’s recovery process.
In addition, with the entry into force of the new Italian leasing law, leasing companies cannot determine the value of the assets and the sale price in the event of a dispute, but must rely on the results of objective appraisals, shared with the customer. This adds robustness and reliability to the process of evaluation of property collateral and to the estimating cash flows from property collateral liquidation.
On the grounds of these considerations, we think that leasing transactions, both on movable and immovable assets, should be considered "secured". We also think that, for the same reasons, an ad hoc, higher threshold should be set for leasing exposures, so as to properly determine whether the institutions fall among the “high NPE” level ones.
This threshold should consider the higher reliability of the recovery process of collateral linked to leasing exposures.
For the same reasons, we moreover think that the requirements for valuation of movable collateral in the case of leasing collateral should differ from the one proposed for “traditional” collateral.
In order to keep the estimation cost down, it would for example be advisable to keep the technical valuations of such assets outside the collateral valuation rules. For example, the valuations of leased asset could be conducted by employees instead of third parties, with the application of indexed re-valuations, such as the asset deterioration tables. This could be done keeping in mind, as amply exposed above, that leasing industry is characterized by the presence of a high degree of technical specialization as regards either managed or repossessed movable, as well as immovable, property collateral."