WSBI-ESBG response to the BCBS consultation on reducing variation in credit risk-weighted assets
– constraints on the use of internal model approaches
WSBI (World Savings and Retail Banking Group)
ESBG (European Savings and Retail Banking Group)
Rue Marie-Thérèse, 11 - B-1000 Brussels
ESBG Transparency Register ID: 8765978796-80
Thank you for the opportunity to comment on the BCBS consultation on reducing variation in credit risk-weighted assets – constraints on the use of internal model approaches. We would like to share with you the following reflections that we hope will be appropriately taken into account by the BCBS.
The variability of risk weights across jurisdictions and across banks must not be viewed as a problem in itself because it should be reflecting real differences in risk profiles, i.e. justified variability. A different issue is that the variability occurs on one and the same exposure, i.e. unjustified variability. To the extent that different institutions' risk assessment of one and the same client do occur, the problem should be addressed through e.g. input floors and through further harmonisation of estimation methodology, rather than reverting to blunt solutions such as output floors.
Generally speaking, limitations on the application of risk models are, in our view, not in line with good and useful credit risk management. According to our opinion, internal models play an important role in risk management and in determining the capital adequacy. Exclusions and output floors to internal models act in the opposite way of what should be considered a prudent approach to risk management through the emphasis put on risk sensitivity. Such an approach is contrary to the risk management philosophy which is contained in the Basel III framework, the latter having implied huge efforts in order to improve and harmonise internal models. European banks have continuously invested in internal models since 2008 so as to allow a more prudent and better risk management approach. The current proposal would lead to a comeback of the Basel I framework (which is the case for exclusions and could be the case if output floors are operative/binding), although this framework was the one which led to the financial crisis due to regulatory arbitrage in view of reducing risk weights, causing adverse selection problems in the banking book, among other unexpected results. It is also important that the scope of use of internal models, parameter restrictions etc. are not subject to national supervisory discretion. Such discretion would in itself be a source for non-risk-based variation in risk weighting. There should be more harmonisation in these areas.
We believe there would be room for alternatives to reduce the variability of RWA while still allowing for some risk-differentiation. We indeed see the danger of some undesirable effects regarding banks' risk management in the BCBS's consultative document.
European banks provide finance to corporates through direct lending to a greater extent than in the US, where market finance plays a more prominent role in the funding of corporates and households. European banks are more sensitive to a regime shift towards a less risk based approach, and could, as a consequence, be forced to change their business model, decrease their direct lending to corporates and households and to shift towards a role as intermediaries. The ability of corporates, including specialised financing for key infrastructures and commodities, to raise funds might be significantly hampered.
What is more, WSBI-ESBG would like to refer to the following paradox that banking institutions could be faced with: the use of internal models for calculating banks' regulatory capital shall be restricted through the Basel revisions. However, in respect of insurance companies' regulatory capital and banks' provisions under IFRS 9 internal models seem to be promoted. This is particularly obvious in the case of financial conglomerates, which could be forced to apply different approaches to the same counterparty, depending on where they are located and leading to difficulties in explaining the situation to the various authorities.
Scope of use of internal models
The BCBS suggests evening out the variation in estimation of risks on corporate exposures through prohibiting the use of internal models entirely for very large corporates, and through limiting the use of internal models for semi-large corporates by forcing banks to switch from A-IRB to F-IRB for that latter category of exposures.
WSBI-ESBG supports the use of internal models in every portfolio. According to our point of view, it would be better if supervisory authorities make efforts to improve internal models instead of imposing standardised floors for all credit institutions, regardless of their business model. As indicated above, that might, in our view, negatively affect the risk sensitivity of the risk framework and deteriorate risk management practices. Moreover, it has to be added that the standardised approach for credit risk is still under revision, and thus, the effects of any floors remain uncertain.
In our opinion, it could indeed be worth exploring alternative possibilities instead of limiting the use of internal models. For example, decision-makers could propose allowing the use of external databases on the condition that these have to comply with certain requirements in order to be a reliable source, establish a list of compliant external databases (ECAIs are worth being mentioned in this respect), etc.
One of the main arguments in the international debate on the use of internal models has been the lack of data points for certain categories of exposures. For instance, so-called low default portfolios have been pointed out as being inappropriate for internal modelling, especially A-IRB, because the number of observations of defaults is "too low" to serve as a basis for internal modelling. WSBI-ESBG finds the reasoning less compelling in this particular case. A low number of observations of defaults of a certain portfolio may actually be a consequence of that portfolio being more stable and secure than others. The low number of default observations may be a consequence of the fact that lending to large and semi-large corporates is relatively secure. Moreover, banks and rating companies have appropriately solved this lack of default data through the use of models combining quantitative and qualitative data (huge amounts of financial data and deep knowledge of corporates by expert analysts) and expert criteria. It should be noted that models used for these portfolios have been previously validated, scrutinised and approved by supervisors. The outputs of these models are annually back-tested in order to show their performance.
The proposal to restrict the use of A-IRB for certain corporate exposures and other changes to the A-IRB approach (parameter floors and parameter estimation practices) should be carefully evaluated before going forward with extensive changes such as forcing banks to go from A-IRB to F-IRB for certain categories of exposures (corporates with revenues above EUR 200 million). The cliff effects mentioned by the BCBS are in themselves an argument for not going forward with the proposal.
In summary, in WSBI-ESBG's opinion, refining the IRB approach is a better way forward to achieve the BCBS's goal to reduce variation in risk weighted assets:
Risk sensitivity is crucial and internal models are important for understanding risks in banks and for supervisors;
Internal models are also the best measure for reflecting the risk of a bank's portfolio of exposures;
The removal of IRB would reduce the motivation to improve risk methods and management.
Based on this, WSBI-ESBG would like to propose increasing the level of harmonisation for internal methods. Concrete initiatives in the European Union, e.g. the draft RTS on the definition of default, developed by the EBA, could be taken on board in the work process in Basel going forward, additional works of harmonisation could be done by banks.
Output floors would become the binding constraint for banks with a large proportion of low risk assets on the balance sheet, reducing the incentive to keep those assets on the balance sheet. Against the background of our proposal on parameter floors below, there should, in our opinion, be no need for output floors. Most of the WSBI-ESBG members who use the IRB approach would face drastically higher capital requirements by Standardised Approach-based floors and they would likely become restrictive in new lending, given current credit portfolios and a calibration of the floor indicated by the BCBS. What is more, WSBI-ESBG believes that a leverage ratio already fulfils the function of a backstop.
Furthermore, the outcome of the QIS regarding the proposed changes has to be considered before any decision on the final framework
Overall we believe that a revision of the internal risk weighted framework, with the aim to reduce variation in credit risk weighted assets, should not result in higher capital requirement for the banking sector in total as well as introducing a permanent output floor should not increase the capital requirement compared with the situation today.
Exposures to banks, other financial institutions and corporates
We question the basis for the proposed EUR 200 million revenue limit and the EUR 50 billion asset size limit and believe that BCBS should carefully calibrate such limits based on quantitative studies of historical default rates above such limits, number of companies in different regions above such limits, etc. Generally, WSBI-ESBG would at least suggest an increase in the revenue limit.
Furthermore, it could be considered to establish common criteria to identify Low Default Portfolios. For instance a more stringent standard of default observations for modelling, e.g. in line with the UK PRA's benchmarks and complemented with the use of external data where necessary or, where such benchmarks are not met, F-IRB for large corporates, higher EUR-thresholds and risk assessment to guide treatment of large corporates' subsidiaries could be contemplated.
Parameter floors and modelling
The need for temporary floors on Probability of Default (PD), Loss Given Default (LGD) and Maturity (M) in order to safeguard adequate levels whilst the implementation of the above mentioned harmonisation efforts is being completed seems understandable to us. Such floors should be temporary, and the BCBS should explicitly state in the final paper that the LGD and M factor floors will eventually be removed.
In our view, exposure level parameter floors significantly reduce the risk sensitivity of the IRB approach and can incentivise banks towards higher risk lending. Moreover, parameter floors should be applied on portfolio level – keeping the possibility for risk sensitivity on exposure level. In particular regarding the LGD, we think that a portfolio level approach is the more suitable option.
If we are not mistaken, the EBA is still working on its standards and guidelines for risk parameter estimations. Additionally, the ECB is also preparing its "expectations" as a supervisor on risk modelling.
Based on that, WSBI-ESBG is concerned that the BCBS might introduce risk modelling requirements that could interfere with the developments and outcomes at the EBA and ECB levels.
In respect of the PD, there seems to be a never-ending, philosophical discussion on Point-in-Time (PIT) and Through-the-Cycle (TTC) models. One WSBI-ESBG member points out that, in fact, they have "hybrid" models, i.e. they are neither entirely TTC nor entirely PIT models. Asking for full TTC models and calibrations would, indeed, have a severe impact in terms of new methodologies and developments.
However, the main question is whether these full TTC models are useful for management purposes. If not, how could one fulfil the use test for IRB applications? In order to be approved to use internal models for regulatory proposals, a banking institution first of all has to demonstrate that it uses those models for managing purposes. Full TTC are not the best option for managing purposes. Therefore, if for regulatory purposes full TTC were mandatory, it would be very difficult to comply with the above mentioned requirement.
With regard to the LGD, WSBI-ESBG believes that it could be a good idea to fix a floor for the downturn add-on. However, in our view, most of the variability in the context of the LGD stems from modelling criteria. So perhaps a good way to address the excessive variability would be to define common criteria (as the EBA is trying to do) for those elements that affect the LGD most: e.g. discount factor, definition of downturn, criteria to account for direct and indirect costs, treatment of open cycles.
In respect of the maturity of a facility, the latter is not the outcome of a model, but a straight calculation from facility characteristics. We would appreciate using the actual average maturity of the facility also in the F-IRB approach (instead of a fixed maturity of 2.5 years).
With regard to credit conversion factors (CCF), WSBI-ESBG would like to ask for some clarifications on what is precisely meant by a "12-months fixed horizon" (page 11). Besides, it is not entirely clear to us why this approach would be better suited than other methodological approaches. We would prefer an approach where defaults can occur at any moment of the following 12 months, not exactly in 12 months.
WSBI-ESBG does not agree with the BCBS's argument when stating that exposures to equities should be subject to the standardised approach. According to the proposal "[...] it is unlikely that banks will have specific knowledge concerning the issuer over and above public data." This is obviously not true in many cases. As a matter of fact, banks might have even better data and knowledge concerning the issuer than the markets. It should not be forgotten that banks sometimes have significant holdings in listed companies and even a seat on their board of directors, which enables them a prompt and direct access to the best available information.
If internal models were indeed banned in respect of equities, the BCBS's proposal of applying a 250% capital charge to all such exposures would not be convincing in terms of a risk-sensitive capital framework. More buckets would clearly be needed.
The BCBS proposes to remove the IRB approaches for specialised lending that use banks' estimates of model parameters, leaving only the SA and the IRB supervisory slotting approach. This is based on the presumption that banks are unable to produce robust estimates on such exposure in the absence of available data. We would like to underline that deep data pooling already exists (e.g. S&P, GCD).
For portfolios where the risk sensitiveness is not always best modelled by statistical models based on the basis of historical default and loss data, theoretical models based on asset values or on future cash flows generated by the asset financed are more relevant. Theoretical models are based on deep data basis covering decades of asset values (e.g. aircrafts or vessels), commodities and economic data (oil and gas prices, electricity prices, etc.), provided by external appraisers and enabling, for example, the assessment of the LGD of object finance or project finance. These theoretical models are risk sensitive as they enable taking into account both the possible underlying asset volatility of value or cash flows generated, and the specific structure of the transaction with the leverage and contractual structure, e.g. existence of off-take (sale) contract, political risk after mitigants, etc. These models use well documented modelling techniques. They are annually back tested and are flexible, i.e. could be easily recalibrated, if needed in a harmonisation process.
We therefore propose to keep internal models for Specialised Lending, with harmonisation works and enrichment of pooled data with additional relevant information enabling the enlargement of back testing of these models.
At the end of the day, the financing of the real economy would again be negatively affected if the BCBS proposal was implemented. It does not reflect the low average risk of these asset classes (lower than unsecured corporate loans) with more than half of the LGD observed below 10% which require risk sensitive models.
Specialised Lending represents massive amounts, (EUR 496 bn currently under A-IRB with EU banks [source: EBA]) and relates to key infrastructures and commodities which could not easily be financed otherwise, given that these types of assets require bespoke specialised financings.
About WSBI (World Savings and Retail Banking Institute)
WSBI brings together savings and retail banks in all continents and represents the interest of circa 6,000 financial institutions with total assets of USD 14 trillion and serving some 1 billion customers in 80 countries worldwide (2013 figures). As a global institution, WSBI focuses on international regulatory issues that affect the savings and retail banking industry. It supports the aims of the G20 in achieving sustainable, inclusive and balanced growth and job creation around the world, whether in industrialised or less developed countries. WSBI favours an inclusive form of globalization that is just and fair, supporting international efforts to advance financial access and financial usage for everyone. It supports a diversified range of financial services that meet customers' transaction, savings and borrowing needs responsibly. To these ends, WBI recognizes that there are always lessons to be learned from savings and retail banks from different environments and economic circumstances. It therefore fosters the exchange of experience and best practices among its members and supports their advancement as sound, well-governed and inclusive financial institutions.
World Savings and Retail Banking Institute - aisbl
Rue Marie-Thérèse, 11 ￭ B-1000 Brussels ￭ Tel: +32 2 211 11 11 ￭ Fax : +32 2 211 11 99
Info@wsbi-esbg.org ￭ www.wsbi-esbg.org
About ESBG (European Savings and Retail Banking Group)
ESBG brings together nearly 1000 savings and retail banks in 20 European countries that believe in a common identity for European policies. ESBG members represent one of the largest European retail banking networks, comprising one-third of the retail banking market in Europe, with 190 million customers, more than 60,000 outlets, total assets of €7.1 trillion, non-bank deposits of €3.5 trillion, and non-bank loans of €3.7 trillion. ESBG members come together to agree on and promote common positions on relevant regulatory or supervisory matters.
European Savings and Retail Banking Group – aisbl
Rue Marie-Thérèse, 11 ￭ B-1000 Brussels ￭ Tel: +32 2 211 11 11 ￭ Fax: +32 2 211 11 99
Info@wsbi-esbg.org ￭ www.wsbi-esbg.org
Published by WSBI-ESBG. June 2016.
 Nevertheless, it should be noted that these differences, even in respect to the same client, could be justified for differences in the guarantees between both banks, client positions and historical data. The relevant point is whether the differences are justified or not.