Basel III – the view from Europe
Article in the Risk Magazine, July 2018, from Christian Ossig, Uwe Gaumert and Michaela Zattler.
The Basel Committee document “Basel III: Finalising post-crisis reforms”, published on 7 December 2017, is driven largely by criticism of banks’ internal models and the variability in their results under Pillar 1 capital calculations. It is both legitimate and sensible to try to reduce this variability.
Views nevertheless continue to differ on how this objective can best be achieved. We believe the new framework for credit risk capital in particular will need very careful examination to make it suitable for implementation in Europe.
To tackle risk-weight variability, the Basel Committee eventually opted to introduce an output floor of 72.5% based on all standardised approaches. The consequences of this Basel reform in Europe have been hotly debated.
The floors (not just on the outputs, but also on some inputs) were proposed with the intention of limiting the unwarranted variability in model results simply and effectively. They will certainly achieve this goal, but at a very high price.
Capital requirements will actually become less risk-sensitive, since the risk-sensitive measurement results delivered by models will ultimately be adequately reflected only for larger risks. In our view, the Basel Committee has taken the easy and wrong way out. Europe – and especially the European Banking Authority (EBA) and European Central Bank – has done a much better job by focusing on a cause-based approach, which involves identifying the reasons for model variability.
In fact, there is a strong case that such steps reduce the need to deploy the capital floor at all in Europe. The ECB’s Targeted Review of Internal Models is the biggest and most comprehensive effort worldwide to date aimed at reviewing internal models for credit, market and counterparty credit risks, including the extensive use of on-site inspections.
The EBA’s work on standardising internal ratings-based approach (IRB) models for credit risk has created new, harmonised guidelines on the application of the definition of default, on estimating the probability of default (PD) and loss-given-default (LGD), and on the treatment of defaulted exposures. There has also been a banks’ internal backtesting techniques(RTS) and guidelines on LGD estimation appropriate for an economic downturn.
Finally, the EBAs yearly benchmarking exercise in accordance with Article 78 of the fourth capital requirements directive provides an additional model validation approach beyond banks' internal backtesting techniques, by using a sector-wide comparison of estimates delivered by banks' models.
The Basel floors are also less appropriate for Europe because the financial system has a distinctive structure. The most robust estimates of the quantitative impact of Basel III for Europe were issued on a best-efforts basis by the EBA in December 2017. But the impact assessment used out-of-date portfolio data from December 2015. The EBA came to the conclusion that banks will see their Tier 1 capital requirement rise by an average of 12.9%, with 6% of this increase accounted for by the floor alone.
The report also seemed to confirm the often expressed hypothesis that big banks, which in Europe mostly use internal models for Pillar 1 capital, will be much harder hit by the reforms than smaller banks that tend to use the standardised approach for credit risk already.
For a number of reasons, however, the results need to be treated with caution. First, the Basel document contains a number of new and sometimes not totally clear-cut definitions, whose precise interpretation is likely to have a significant effect on the overall results. For example, there is a new due diligence process in the standardised approach to credit risk that has never been used before.
Second, these figures do not show the impact of the Fundamental Review of the Trading Book, and it is unclear whether the reform of the standardised approach for measuring counterparty credit risk was taken into account either. Yet both those reforms will have significant effects on the end result.
We therefore assume that the EBA’s impact assessment to be conducted in the coming months on the basis of the European Commission’s May 2018 call for advice will find banks need much more capital. This is because the final Basel framework is a global compromise, which was unable to take adequate account of circumstances in Europe.
Compared to other major financial centres, the European financial sector is characterised by some important fundamental features, which should be considered when designing an appropriate regulatory regime for banks.
The vast majority of investments in Europe are funded through the intermediation of banks. Various studies show that bank loans account for 75% of all financing in Europe, with only 25% coming from the capital markets and other non-banks.
Since banks are the primary source of finance in Europe, they provide funding for investments of all maturities and in most cases hold these in their books until they mature. This is because the European secondary loan markets, like the primary capital markets, are comparatively underdeveloped. Europe does not have many government-backed programmes like Freddie Mac and Fannie Mae in the US, to which banks can transfer risk in order to reduce their need for capital.
On top of that, the European securitisation market was badly hit in the wake of the financial crisis which started in 2007. As a result, a feature of European bank balance sheets is the high proportion of long-term assets. According to ECB statistics, 55% of all loans in the eurozone in 2017 had a term of over five years. The risk-weighted assets (RWAs) associated with such long-term positions therefore have a strong influence on banks’ need for capital and cannot be adjusted at short notice.
SMEs and real estate
A further factor determining the balance sheets of European banks is the funding of small and medium-sized enterprises (SMEs). As many as 99% of all companies in the EU are SMEs, and two thirds of all employees in the private sector work for SMEs. Most of these companies cannot afford, or are unwilling to pay for, an external rating and are consequently dependent on banks to obtain funding. Similarly, a large proportion of banks’ RWAs are made up of loans to unrated SMEs.
Property finance is an equally important business segment for European banks. ECB statistics found that 35% of all loans granted in 2017 were used to buy or build residential property. Residential real estate finance accounts for 62% of long-term eurozone lending (loans of five years or more). What the ECB’s figures do not, however, show is that a substantial proportion of the commercial property sector is also financed through banks. Owing to the special features of this business segment, it is often handled by specialised institutions.
The risks carried on the balance sheets of European banks therefore emanate to a very large extent from property finance. But property markets differ across Europe, sometimes widely so. Differences are especially marked with respect to the definition of property rights, the enforceability of legal claims, loan approval criteria and lending practices, and property valuation methods. All of these have a significant influence on the risk the bank is taking.
In a few European countries, land register entries are not always available, either because they do not exist or because the authorities do not maintain them properly. In other countries, laws designed to protect tenants and homeowners prevent or prolong foreclosures, or the legal system is so inefficient that it takes many years to liquidate collateral.
National standards for approving loans can also differ widely. There are basically two approaches: either to focus on the ability to sell the collateral as a means of reducing the exposure to loss, or to look at the borrower’s ability to repay the loan from their income or other assets. Both approaches are used to a greater or lesser extent depending on the country involved.
Finally, there are different procedures, with differing degrees of conservatism, for determining the value of a property. Many countries use a market-value-driven approach, while a few others like Germany use a much more conservative approach (a so-called mortgage lending value approach). Standards on the frequency of property revaluations also differ.
Until now, risk-sensitive methods of determining the need for regulatory capital have been able to reflect these European specificities very well. Internal models have enabled individual PDs to be estimated for borrowers without external ratings. The special features of national property markets have been reflected in different loss rates and LGD estimates. In consequence, the widespread use of internal models by European banks may be regarded as a sound response to their market environment and explains the basically positive attitude of European supervisors to the internal ratings-based (IRB) approach.
The output floor will now massively disrupt this long-established and well-functioning system which has been in place since the implementation of Basel II in Europe.
The drawbacks of the output floor are closely tied to its reliance on the standardised approaches, such as that for credit risk. In this context, the reform of the standardised approach to credit risk included in the December 2017 package does indeed point in the right direction, by partially enhancing risk sensitivity – for instance, by adopting a more granular structure of sub-asset classes used to assign risk weights.
Nevertheless, the new standardised approach – in the eyes of IRB model users, at any rate – is still not risk-sensitive enough. Capital requirements calculated on the basis of standardised approaches react less strongly or not at all to a change in the risk of a position. Under the IRB approach, ratings have to be adjusted whenever the bank has new material information on a borrower.
As institutions have built up adequate data histories that also cover a financial crisis, internal rating procedures are likely to be particularly well able to capture their risks accurately. When the floor comes into play – specifically, the risk estimates delivered by an internal model are often well below the floor level – the institution could take on risks that its model would perceive as higher without the standardised approach necessarily indicating an increase in risk.
This would have no impact on capital requirements until model outputs reached the floor level. This effect is caused by the typical overstatement of risk under the standardised approaches in conjunction with the much too high floor of 72.5%. The upshot is a perverse prudential incentive of huge proportions.
Nevertheless, we take the view that, in the interests of global compliance, the output floor of 72.5% of the sum of RWAs calculated on the basis of standardised approaches should be introduced in Europe as well. After lengthy negotiations, Europe agreed to the introduction of this tool and should consequently abide by the outcome of these negotiations. Despite this, we believe that ways of limiting the adverse effects of the floor need to be considered in European implementation.
As the floor achieves its effect through being linked to the standardised approaches, its adverse effects could be reduced by adding more risk-sensitive and better-calibrated standardised approaches to credit risk for a few selected portfolios. Application of these additional approaches could be optional, so that smaller banks that have so far used the standardised approaches to calculate their capital requirements would not be forced to implement more complex and more risk-sensitive procedures. Because of European specificities, we see a need to include more risk-sensitive options at least for two asset classes: lending to unrated corporates and commercial real estate lending.
Under the new Basel rules, jurisdictions that do not allow any external ratings in the standardised approach (such as the US) may assign corporates identified as “investment grade” a preferential risk weight of 65%. In jurisdictions that do allow the use of external ratings, however, unrated companies with annual consolidated sales of more than €50 million attract a 100% risk weight.
The 65% option should be allowed in Europe as well, not only because of the region’s lower actual loss rates, but also in the interests of an international level playing field.
An exchange listing is one of the criteria Basel suggests for assessing if an unrated corporate should be considered investment grade. Since the affected corporates in Europe are not listed, we need an alternative method of determining what constitutes investment grade.
There are at least two conceivable solutions. First, banks could base this “yes or no” decision on internal ratings which have been approved by supervisors. Investment-grade companies usually have a probability of default of less than 1.5% per year. So all unrated companies with an internal PD of under 1.5% could receive the 65% risk weight. Alternatively, statistics could be analysed to identify a combination of two to three quantitative rating criteria, such as profitability, own funds ratio, or turnover. The thresholds could be set at a level that normally corresponds to investment grade for rated corporates.
An approach of this kind was considered in the Basel Committee’s first consultation document on the new standardised approach to credit risk in 2014, though subsequently rejected as not able to correctly differentiate between different levels of credit risk. But it ought to be possible to find appropriate thresholds for classifying a company as investment grade or not.
Another issue which needs to be considered in depth is the implementation in Europe of the Basel Committee’s proposals for real-estate portfolios. The standardised approach to credit risk sets out two approaches for determining RWAs for residential and commercial property: the loan-splitting approach (which allows banks to classify separately those parts of the loan that are or are not covered by the underlying collateral value) and the loan-to-value (LTV) approach, where risk is determined by the total LTV.
The Basel Committee suggests leaving the choice to jurisdictions. This seems logical given that property markets vary from one country to another. But the question then arises as to how this option should be exercised in Europe. As we see it, this question can only be answered on the basis of the findings of the EC’s current call for advice.
The Basel solution is liable to distort competition in general, since it remains totally unclear what criteria should be used to calculate property value under both approaches.
It would be dangerous, however, to let national supervisors decide which approach to apply. Owing to the existence of the ECB single supervisory mechanism, a situation could easily arise where one country had two separate regimes for valuing property loans, depending on whether banks were directly or indirectly supervised by the ECB. The resulting competitive distortion is surely not something supervisors want.
In fact, the Basel solution is liable to distort competition in general, since it remains totally unclear what criteria should be used to calculate property value under both approaches. There is a lack of standardisation and comparability in Europe with respect to valuation approaches and the availability of data. The obvious solution – namely to mandate the EBA to develop RTS on the matter – has failed to produce satisfactory results in the past. For this reason, a level playing field should be established in the level one text of the capital requirements regulation itself.
On top of that, the current proposals are not risk-sensitive enough for countries with low loss rates. In particular, the low number of categories for commercial real estate lending will lead to a huge overstatement of risk compared with internal models, so additional LTV bands are needed in the suggested LTV approaches. There is enough statistical material available for a sensible calibration.
As far as specialised commercial real estate lending and other specialised lending are concerned, a risk-sensitive slotting system could be developed for the standardised approach, taking the current IRB slotting approach as its basis.
With the high degree of detail in its call for advice, the EC has laid sound foundations for further debate on the European implementation of Basel III. We are sure to see more ideas emerge over the next two years, and we hope this article can be a useful initial contribution to the discussion.
Christian Ossig is the chief executive, Uwe Gaumert is a director of banking supervision, and Michaela Zattler is a divisional manager of banking supervision, at the Association of German Banks (Bundesverband deutscher Banken, BdB). The positions expressed in this article are those of the authors and do not necessarily reflect the position of the BdB.