Response by the Riksbank to the European Commission’s consultation on possible further changes to Directives 2006/48/EC and 2206/49/EC (the Capital Requirement Directive)
Office of the Director General
Financial Stability Department
Dnr: 2009-682-AFS
This is a joint response by Finansinspektionen (the Swedish Financial Supervisory Authority) and Sveriges Riksbank, hereinafter the “Swedish authorities”, to the Commission’s consultation on possible further changes to Directives 2006/48/EC and 2206/49/EC (the Capital Requirement Directive).
THROUGH-THE-CYCLE EXPECTED LOSS PROVISIONING ('DYNAMIC PROVISIONING')
General remarks
Swedish authorities support the objective of reducing potential pro-cyclical effects of financial regulation by introducing counter-cyclical measures. However, we do not believe that dynamic provisioning is the right way forward.
First of all, accounting rules should strive to give a true view of the firm’s position by reflecting the fair value of the firms’ assets and liabilities. A transparent framework will increase investor confidence and decision making, which will be positive from a financial stability perspective. To facilitate investor confidence accounting standards also need to be a long term stable framework. Hence, any changes should be aimed at increasing the fair value applications and be thoroughly analysed before implemented. Also political pressure for changes implemented with an accelerated timeline should be avoided as it would risk resulting in a less objective and more volatile framework.
Refining the accounting framework, i.e. introducing a prudential focus, in itself may seem rather harmless. However, allowing for specific concerns to be met risks opening up for other concerns to be accommodated, leading the whole framework farther away from transparency.
Changing accounting rules should only be considered if it is absolutely clear that the effects of greater transparency and market valuation will be the achieved, that there will not be any negative side effects that are not accounted for and that the same effects cannot be obtained by introducing changes to prudential regulation, e.g. reserving and/or dynamic capital requirements.
Second, dynamic provisioning incorporates technical difficulties which remain to be solved. Long time series are required, and we are not at all convinced that there is sufficient data for guaranteeing robustness of the model.
Third, the model relies on the assumption that losses in the next business will be the same on average as in the last business cycle, which will not necessary be the case.
Fourth, the Spanish provisioning method is not based on the characteristics of each credit portfolio, but applies “one size fits all”. This risks reducing incentives for prudent risk management, and will unjustly distribute the “burden” among banks. One may note that Spain recognizes the existence of the problem, and this is why the Bank of Spain has a maximum ceiling for the size of the buffer.
Last, there is an issue of timing. The designer of the Spanish model has explicitly pointed out that the dynamic provisioning needs to be introduced in upturns. What may happen in practice if introduced in a downturn, (when there may not be enough profits to draw from) is that banks will create reserves by reducing their capital, thereby putting pressure on equity capital. There is therefore ample time for a thorough analysis.
Moreover, other methods to counteract procyclical effects of financial regulation are currently analysed in the international for a, e.g. capital buffers and leverage ratios. At this point we do not have a clear understanding of the joint impact of those measures. Thus, before rushing to a decision on a specific model on provisioning it seems logical to await the results from CEBS work on downturn PDs and from the Basel Committee’s quantitative impact assessment of the aggregated effects including all measures
proposed.
In addition, for level playing field reasons it is important to signal that we need to look for a global solution.
Question 4: The Commission services suggest that the through-the-cycle value adjustment should not count as regulatory capital (see ANNEX 1, suggested amendment to Article 57). Do you agree?
As already stated, Swedish authorities do not support the introduction of dynamic provisioning. If however the EU decides to push for such method, we agree with the Commission’s proposal that the through-cycle expected loss provisioning should not count as regulatory capital. According to regulatory capital requirements expected losses should be treated as business costs and either charged to the profit and loss account or provisioned against, while unexpected losses should be covered by banks regulatory capital. We find this distinction useful and fail to see any logical reason for diverging from this approach.
Question 5: Should off-balance sheet items be captured under the formula for through-the-cycle expected loss provisioning, given that 'provisions' for off-balance sheet items are not recognised in all relevant accounting standards? Should only assets subject to an impairment test be subject to through-the-cycle expected loss provisioning? (See ANNEX 1, suggested Article 74a (2).
Given that the present financial crisis partly resulted from (or at least was amplified by) the fact that the regulatory regime did not take off-balance sheet items sufficiently into account Swedish authorities believe that all existing commitments should be captured by financial regulation to the extent possible. Thus, if the through-the-cycle expected loss provisioning model would be implemented we would argue for an inclusion of off-balance sheet items. For simplicity and to ensure consistency with regulatory capital requirements the exposure values should be calculated using the credit conversion factors as defined in Basel II.
Question 6: At this point, the suggestion is not to include the option for competent authorities to allow internal methods to determine expected losses across an economic cycle. As an alternative to the regulatory approach to calculate counter-cyclical factors, would it be desirable to allow firms' internal methodologies (to be validated by supervisors)?
In line with our general remarks it is of great importance that we closely analyse the effect of proposed regulation on markets in terms of level playing field, incentives for regulatory arbitrage and cross boarder activities before any new measures are decided upon.
Besides, deciding upon a more exact model framework and model inputs requires thoughtful consideration, as to take into account not only the risks in individual banks but also of the banking system as a whole. At the same time the model framework should not undermine transparency from an investor perspective.
It should be pointed out that effectiveness of a regulatory approach to determine the counter cyclical factors would to a large extend depend on availability of good quality data. As the calibration of both α and β-factors on a national level depends also on the definition of a business cycle it may imply significant differences across member states
Question 7: Should the exposure class of Article 86 (i.e. for credit institutions subject to the IRB approach) be used irrespective of the fact that the credit institution may be under the Standardised approach? It may be noted that a mapping between exposures class under the Standardised approach and under the IRB is already used in the prudential reporting system of some Member States. As an alternative, should countercyclical parameters be defined for the 16 exposures classes under the Standardised approach? (See ANNEX 1, suggested Article 74a (1).)
Question 8: Please give your views on the following approaches:
- the Spanish model of through-the-cycle expected loss provisioning;
- a 'simplified' Spanish model.
In particular, we would welcome views on the relative merits of both options in terms of the building up of provisions in a graduated manner over time (See ANNEX 1, suggested Annex IXb).
Again, Swedish authorities do not believe that dynamic provisioning is the best way forward to mitigate procyclical effects of financial regulation (see general remarks)
Question 9: Should new risk categories (as suggested above) be introduced along the lines of the Spanish system or, alternatively, should the current risk categories of the CRD (e.g. credit quality steps in Annex VI) be used? (See ANNEX 1, suggested Annex IXb.)
We fail to see the logic in introducing a new set of risk categories in addition to the risk categories defined in Basel II. The most reasonable approach, both in terms of simplicity, administrative burden and cost would be to use existing categories.
Question 10: Is the 'location of the borrower' (as opposed to the booking of the exposure) the right approach, with a view to avoiding regulatory arbitrage? (See ANNEX 1, suggested Annex IXb 2.)
Question 11: Will the data to determine counter-cyclical factors be easily available? Sweden has no central credit register. To determine macro counter cyclical factors Swedish authorities would have to rely on information from individual credit institutions.
Question 12: Please give your views on the methodologies for calculating the through-the-cycle expected loss provisions at consolidated level. (See ANNEX 1, amended Article 73.)
Question 13: Please give your views on the scope of disclosure requirements for through-the-cycle expected loss provisioning. (See ANNEX 1, suggested amendment to Annex XII (17).) We support the transparent disclosure requirements as proposed.
RESIDENTIAL MORTGAGES DENOMINATED IN A FOREIGN CURRENCY
Question 14: Do you consider that the risk weights suggested will be effective in discouraging unsafe practices and irresponsible lending in foreign currency denominated housing loans?
Question 15: Do you consider a loan to value ratio of 50% or less is sufficient objective evidence that the borrower has sufficient private wealth to withstand currency movements and potentially correlated movements in property prices?
Generally, the Swedish authorities support measures to curb bank lending in foreign currency to customers receiving their revenues in another currency. Such lending has been a contributing cause to financial crises in many cases, also in Sweden in the 1990s. It should be noted that Pilar II under the current capital requirement framework already allows supervisory authorities to account for individual risk profiles, such as a large exposures in foreign currency denominated loans, when determining the level of capital required.
The present proposal has some important overarching problems:
A. Hitherto, there has been a strong principle nationally as well as internationally that a bank's credit decision should be based on the borrower's assessed repayment capacity. Collateral should only be seen as a supplement. Deviations from this principle (such as the US sub-prime lending, which put the collateral first) has led to major bank problems. Capital requirements are always primarily based on the risk weight of the borrower but may be adjusted if there is sufficient collateral. The current proposal for calculating the capital requirements on property loans in foreign currency puts the risk weight of the collateral first. The repayment capacity of the borrower will make no difference to the risk weight. This deviation from basic principles will create inconsistencies and is potentially dangerous.
B. The proposal will only "hit" one category of foreign currency loans. Should the proposal be implemented, a borrower having a non-collateralised foreign currency loan will have a much lower risk weight and hence lower loan costs. This will cause perverted incentives, for instance in pricing products.
C. The steeply rising scale for capital requirements, according to the formula, is not logical. Below 50% LTV ratio, the risk weight is equal to the one for corresponding loans in domestic currency, but already at 60% LTV ratio the risk weight is 290%, increasing to 625% at 75% LTV and 1250% at 100% LTV. This implies that the risk to the bank for low LTV ratios is deemed as equal for domestic and foreign currency loans, but that there is a very large difference in risks at higher LTVs. In fact, the loan costs will probably become prohibitive. In fact, it could be argued that the high risk weights will provide an incentive for borrowers to look for loans only when the interest rate in the domestic currency is significantly above that of the foreign currency, in other words, when the presumed risk of devaluation/depreciation is high. Furthermore, such a steep escalation of capital charge would create strong incentives for banks to postpone collateral re-valuation, make use of non-conservative valuation methods and non-conservative experts, and make other similar efforts to avoid being hit by those prohibitive capital charges, which in turn would be very detrimental to financial stability.
D. The proposal makes no distiction between two floating currencies and a currency pair in which only one currency is floating and the other is fixed. The risk of building up large currency rate tensions in the latter case is, of course, much greater.
In conclusion, since the proposal is fraught with so many weaknesses, we propose that it is reassessed and that also other alternatives are sought to achieve the desired results. For instance, loan to income (LTI) ratios might be discussed. Although it is not suitable to prescribe specific LTI ratios in the CRD, the CRD could provide general directions to national authorities, e.g. stating that "when a bank provides loans in a currency different than the domestic currency to a borrower whose main revenues are denominated in another currency than the loan currency, the national supervisory should ensure that the bank's credit granting process includes assessing that the borrower has a sufficient margin to be able to service the loan also in a situation when value of the domestic currency depreciates in relation to the loan currency. The presumed magnitude of the depreciation should be realistically assessed, e.g. based on economic and market criteria, such as the prevailing interest rate differential."
REMOVAL OF NATIONAL OPTIONS AND DISCRETIONS
Question 16: Is this suggested scope of maximum harmonisation in 2006/48/EC and 2006/49/EC appropriate?
The Swedish authorities are generally positive to maximum harmonisation. A single rule book will contribute further to the achievement of the internal market. Moreover, since it is likely to assume that a harmonized framework would contribute to improved communication and collaboration across jurisdictions, we believe that maximum harmonization would be beneficial also from a financial stability perspective. The scope of harmonisation seems appropriate. However, further work needs to be done. Maximum harmonisation may also require a flexible legislative process to be able to respond to development on financial markets.
Question 17: Is the suggested prudential treatment for both residential and commercial real estate is sufficiently sound?
Swedish authorities generally support the measures aiming at adjusting the legislative framework taking into account the experiences from the current crisis. However the effect of the amendments should thoroughly be analysed (credit institution/stability, borrower, real economy). This said, we have the following comments;
The present crisis has involved residential mortgages in some countries. It is however overly simple and harsh to change the risk weights for all residential mortgages. In the view of the Swedish authorities, the problems that emanate from residential mortgages in this financial crisis have two characteristics.
First, in some cases banks have deviated from the principle of repayment capacity and increased focus on underlying collateral. This has led to increased defaults and credit losses that have propagated through the financial system. It is however not a problem with mortgage lending as such but rather with the risk assessment of the borrowers possibilities to repay.
Second, in some cases mortgages have been granted to individuals who have made their business to buy-and-let. In this context these borrowers could better be described as individuals running private real estate “companies”.
It is therefore premature to introduce higher risk weights on residential mortgages in general as there is very little evidence that “normal” residential mortgages have played any major part in this crisis.
Furthermore, the proposal is not in line with the Swedish experience of real estate lending. (According to a report from Finansinspektionen concerning development on the mortgage market (May 2009) the mortgage market does not constitute a significant risk for the banks or for overall financial stability. However, there are certain risks that individual borrowers will not be able to pay their mortgages in the future due to increased unemployment. According to this report the average LTV is 70 % for new loans and the average for the total portfolio is 58 % .Furthermore, in a report from July 2009 concerning banks lending for commercial real estate in Sweden, Finansinspektionen finds that the current financial crisis has yet to make any obvious impact in the form of insolvency and credit losses in lending for commercial real estate. Compared with the banking crisis in the beginning of the 1990s, banks now stress the importance of a underlying cash flows in real estate companies, and banks have considerable internal valuation expertise which reduces the risk of overly optimistic valuations.)
We note also that the proposed changes of risk weight for mortgages in real estate is not directly applicable on credit institutions applying the IRB-approach (Articles 87.7 and 91 compared to Annex VII and VIII). This may distort competition between credit institutions using the standardised approach and those using an advanced IRB approach. The latter probably have the largest market share in mortgage lending.
The problems identified can – at least partly – be taken care of by strengthening the rules on responsible lending, thereby forcing the banks to focus on the borrowers repayment ability.
Question 18: Is the suggested timeline (2012) for a single definition of default (i.e. 90 days) is appropriate. Yes, the suggested timeline seems reasonable.
FINANSINSPEKTIONEN SVERIGES RIKSBANK
Uldis Cerps Mattias Persson
Head of Department Head of Department
