Quantitative Impact Study 5 Overview on the Results of the...

Confidential

CEBS 2006 118

16 June 2006

Quantitative Impact Study 5 Overview on the Results of the EU countries

Table of contents

Executive summary ..................................................................................................................2 1. Introduction .....................................................................................................................5 2. Summary of results across approaches..........................................................................9 3. Standardised Approach ................................................................................................15 4. IRB Approaches............................................................................................................18 5. Securitisation ................................................................................................................33 6. Impact of the trading book review and the double default treatment ............................35 7. Changes in Minimum required capital for Operational Risk ..........................................38 8. Impact of the CRD on Investment Firms in the EU .......................................................41 Annex A: Methodology for the aggregation of individual QIS 5 bank data .............................42 1 Introduction ...................................................................................................................42 2 Notation.........................................................................................................................42 3 Calculation of the change in minimum required capital on a portfolio level ..................44 4 Aggregation of results across banks.............................................................................48 Annex B: Risk Parameters .....................................................................................................51 1. PDs ...............................................................................................................................51 2. LGDs.............................................................................................................................53 Annex C: Operational risk ......................................................................................................63

2/65

Executive summary

To evaluate the effects of the Capital Requirements Directive (CRD) on the level of the regulatory minimum capital requirements in Europe, a fifth Quantitative Impact Study (QIS 5) was undertaken in 18 EU countries1, Bulgaria as an accession candidate and Norway as a member of the European Economic Area (EEA)2. This report summarises the preliminary results based on data from 49 Group 1 and 213 Group 2 banks which CEBS received in the first quarter of 2006. These data were also taken into consideration in the aggregations carried out by the OC/QIS Group of the Basel Committee on Banking Supervision for the G10 and non G10-countries. National supervisors reported that data quality has significantly improved since the previous exercise. Nevertheless, several rounds of data quality checks were carried out between February and May 2006.

The QIS 5 is an important milestone in banks’ implementation of the CRD as well as an essential step in the discussion on the proper calibration of the CRD for the regulatory minimum capital requirements. In contrast to previous QIS exercises – the last QIS results at a European level were published in 2003 – aspects like the new requirements for the estimation of the LGD parameter and the new issues related to the Trading Book were included in the data collection. A comparison of the results between QIS 5 and QIS 3 is therefore difficult also taking into account that macroeconomic and credit conditions were relatively favourable in the participating countries at the time of QIS 5. Furthermore, since QIS 3 the risk-weight functions have been revised and calibrated on an Unexpected Losses-only basis. The results in this report include the 1.06 scaling factor for credit risk-weighted exposures which was introduced in 2004.

The results of the exercise show that minimum required capital (MRC) under the Pillar 1 provisions of the CRD (including the 1.06 scaling factor to credit risk-weighted exposures) would, on average and on the basis of the present rather favourable macroeconomic juncture, decrease relative to the current regime (see table 1). For example, for Group 1 banks, minimum required capital under the “most likely approach” (the approach an institution is expected to apply after the implementation of the CRD) would decrease by 7.7%, and for Group 2 banks by 15.4%. These figures represent the total overall change in MRC for credit risk, operational risk and market risk under Pillar 1. Possible additional capital requirements due to the so-called Supervisory Review Process (‘Pillar 2’), which, if applied, would offset at least partly the reported decrease in MRC, have not been taken into account in this study.

1 Belgium, Cyprus, Czech Republic, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Luxembourg, Malta, Netherlands, Norway, Poland, Portugal, Spain and the UK. 2 Data of Norway and Bulgaria are always included whenever this report makes reference to “EU banks” or “EU results”.

3/65

Table 1 Change in minimum required capital relative to current regime [%]

Standardised Approach

FIRB Approach

AIRB Approach

Most likely approach

EU Group 1 -0.9 -3.2 -8.3 -7.7

EU Group 2 -3.0 -16.6 -26.63 -15.4

Data based on the different sample of banks for the various approaches. The figures do not take account of the transitional floors as described in Article 152 (1 to 4) of the CRD.

In general, the results are broadly in line with the figures which were obtained at the G10 level. Compared to the G10 figures the relevant approaches for Group 1 banks (Foundation and Advanced IRB Approach) and for Group 2 banks (Standardised and Foundation IRB Approach) in the EU sample show on average a slightly larger decrease in MRC.

Compared with the current regulation, the largest decrease in the MRC is due to residential mortgage exposures, followed by other retail exposures. Because of their relatively long maturity, residential mortgage exposures are sensitive to interest rate risk, which is subject to specific scrutiny (and possible corrective action) under the Supervisory review Process (Pillar 2). The reported reductions are larger under the IRB Approaches than under the Standardised Approach. Corporate exposures (including SMEs) account for the third largest reduction, again in particular under the IRB approaches. These results might be influenced by the favourable macroeconomic situation in most countries. Unsurprisingly, operational risk capital charges represent the largest increase in MRC compared with the current regulation. Other increases, albeit relatively much smaller, come from capital charges on undrawn credit lines, OTC derivatives and repos, securitisation exposures and bank and sovereign exposures, as well as equity exposures.

In order to analyse the incentives for banks to move to the more advanced approaches, the capital requirements for banks providing data on at least two different regulatory approaches were compared. This analysis shows that the CRD provides on average an incentive both for EU Group 1 banks and Group 2 banks to move to the more sophisticated approaches. MRC under the Advanced IRB Approach is 6.8% lower than under the Foundation IRB Approach for Group 1 banks and MRC under the Foundation IRB Approach is 12.4% lower than under the Standardised Approach (not taking account of the transitional floors according to Article 152 (section 1 to 4) of the CRD).

For operational risk the results are not as comprehensive as for credit risk, because only one Group 2 bank that provided data on the Advanced Measurement Approach (AMA) did also provide data on the Standardised approach (TSA) and vice versa. However, for Group 1 banks the available data indicate that the incentive structure to move from the TSA to the AMA is demonstrated. The same holds true for a move from the Basic Indicator Approach (BIA) to the Standardised Approach both for Group 1 and Group 2 banks.

The CRD has not introduced any changes in the rules for calculating capital requirements for market risk in the trading book apart from the rules for specific market risk. The number of

3 Only four countries have Group 2 banks in their sample which provided information on the Advanced IRB Approach. Therefore, the aggregated figure of -26.6% is statistically not as representative as the results for the other two approaches. In most cases these banks have a specific business profile and their customers are typically involved in specialised branches of industry.

4/65

banks having submitted data for the new approaches for specific risk is very limited and cannot be regarded as representative. The report therefore does not contain a specific chapter on the effects of the CRD on the change in MRC for market risk.

No separate QIS was carried out for investment firms. However, some investment firms were included in the data set as part of the figures provided by the banking groups to which they belong. The only country which has conducted a separate analysis was the UK (see section 8).

On an overall basis the results obtained from QIS 5 (based on the current scaling factor of 1.06) are broadly in line with the expectations of the European supervisors.

The effective impact of the implementation of the CRD will depend on the risk profile of the institutions at the relevant implementation dates and will be limited by the transitional floors until the end of 2009.

CEBS expects that in the course of implementing the CRD, supervisors will ensure that banks will maintain a solid capital base throughout the economic cycle. CEBS believes that mechanisms, including the Supervisory Review Process, are in place to achieve this goal. Effective implementation of stress testing and Pillar 2 requirements as set out in the CRD may play an important role in this regard. Moreover, in the first three years after implementation the transitional floors defined with respect to the minimum required capital according to the current regulatory regime were introduced as an additional safeguard. National authorities will continue to monitor capital requirements during the implementation period of the CRD.

5/65

1. Introduction

The main purpose of QIS 5 at the EU level is to provide an assessment of the impact of the new capital framework – as set out in the Capital Requirement Directive (CRD) – on the institutions to which it applies in, and active in, the EU. The results reflect changes in minimum capital requirements for credit risk, market risk and operational risk under Pillar 1 of the CRD.4

The exercise has been carried out in parallel with the QIS 5 exercise organised by the Basel Committee for G10 and non-G10 countries.

The QIS 5 results reflect the Basel Framework as implemented in the CRD. For QIS 5-purposes, supervisors provided participating banks with data on national implementation (i.e. decisions on how to apply national discretions). Outcomes therefore reflect the ways in which the CRD will be implemented within EU jurisdictions and take account of EU specific issues like the treatment of covered bonds, permanent partial use and intra-group exposures.

National supervisors collected data from their institutions in the fourth quarter of 2005. After a first quality check at the national level, the results were submitted for an aggregated analysis in late February and early March 2006.

Chapter 1 of this report describes the sample of institutions from which the results have been obtained, elaborates on the quality of the data and on the methodology used for aggregating them.

Chapters 2, 3 and 4 summarize the results of the exercise for the different approaches applicable under the CRD for calculating the capital requirements for credit risk (Standardised Approach, Foundation IRB and Advanced IRB approach) and highlight the main drivers (PD, LGD, Exposure Value and regulatory calculation difference) behind the changes in minimum capital required compared to the Basel I regime.

Chapters 5 elaborates further on the exposure class ‘securitisation positions’ and chapter 6 on the impact of the trading book review.

Chapter 7 gives an overview of the results obtained for operational risk. Finally, chapter 8 elaborates briefly on the role of investment firms in the QIS 5 exercise.

In order to maintain the confidentiality of results, all scatterplots showing individual bank results have been aggregated across all countries for Group 1 and Group 2 banks. However, two different size classes in terms of capital were considered, distinguishing banks with more than 10 billion Euro capital from those with up to 10 billion.

1.1 Sample 18 out of the 25 EU countries participated in the QIS 5 exercise5. Two other European countries – Bulgaria and Norway – have also submitted data. As Bulgaria is an EU accession

4 The term ‘pillar 1 requirements’ is derived from the structure of the Revised Framework (Basel II). Minimum capital

requirements for credit risk, market risk and operational risk are specified in the first pillar of the Revised Framework. The corresponding articles in the CRD are laid down in title V, chapter 2. Additional requirements relating to the institutions’ own assessment of their risk management are specified in the second pillar of the Revised Framework. The corresponding articles in the CRD are set out in title V, chapter 3.

5 Belgium, Cyprus, Czech Republic, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Luxembourg, Malta, Netherlands, Norway, Poland, Portugal, Spain and the UK.

6/65

candidate and as Norway is a member of the European Economic Area (EEA), in which the CRD will be implemented as well, the data from these two countries have also been included in the sample.

The participating banks are assigned to the two different categories Group 1 and Group 2 banks. As in QIS 3, Group 1 banks are banks which fulfil all of the following three criteria:

- the bank is internationally active;

- the bank is diversified; and

- the bank has original own funds in excess of 3 bn Euro.

Out of the 20 participating countries 49 Group 1 banks and 213 Group 2 banks provided data for the exercise. Therefore, in total 262 workbooks have been considered.

The supervisors in the participating EU countries were asked to ensure that the QIS 5-sample of banks consists only of banks which are not consolidated in another EU country in order to avoid double counting. Therefore foreign subsidiaries of groups located in another EU country are not part of the QIS 5-sample of banks of the host country. For instance, the data of a Belgian subsidiary of a Germany based banking group have been included in the consolidated data provided by Germany, but not in the data provided by Belgium. However, the data from such subsidiaries were often considered in national country reports as they include useful information on the impact of the CRD on national markets. There are few examples of banks (4 in total in only two countries) which are subsidiaries of head institutions located outside the EU. However, due to the small number of these banks the impact on the overall results is small.

Table 2

QIS 5 participation

Group 1 Group 2

Total RSA FIRB AIRB Total RSA FIRB AIRB 49 34 42 27 213 194 91 16

Since banks may have provided data for more than one approach, the total number of participating banks is in general less than the sum of the banks providing data for one of the approaches. – * The reporting dates in 2004 refer to the German banks which are included on the basis on their QIS 4 results. Most of these banks are Group 2 banks, but there are also 2 Group 1 banks which are included on QIS 4 basis. – ** For submissions dated December 2004 a statement that data earlier than June 2005 is representative was provided.

The participating banks were instructed to carry out all computations on the basis of national implementation. Therefore all important EU-specific rules in the Basel II Framework as set out in the CRD (for example the specific risk-weights for covered bonds and the 0% risk-weight assigned to intra-group exposures) are properly reflected in the QIS 5 workbooks submitted by the banks and have been considered both in the G10/non G10 reports of the Basel Committee and in the EU cross-country aggregation carried out in this report. For this reason in this report no separate analysis is provided focusing on the quantitative impacts of EU specific regulations. However, it has to be taken into account that some portfolio-specific results – in particular decreases in MRC obtained for the interbank portfolio – are certainly driven by the low risk-weights which were assigned to many exposures in these portfolios.

Many banks considered more than one approach in their data collection to enable an assessment of whether the incentive structure to move to a more advanced approach is set appropriately in the CRD. Table 3 shows the number of QIS 5 participants from the EU countries for which the respective approach is the “most likely approach” (i.e. the approach which the institution is expected to adopt after implementation).

7/65

Table 3

Most likely approach of QIS 5 participants

Group 1 Group 2

RSA FIRB AIRB RSA FIRB AIRB

Total 2 21 26 116 83 14

The number of banks using the different approaches to operational risk in their capital calculation is presented in table 4. Compared to QIS 3 much more Group 1 banks provided figures for the Advanced Measurement Approach (AMA), the same holds true for Group 2 banks and the Standardised Approach (TSA). Only 2 Group 2 banks provided AMA figures. National supervisors pointed out that the implementation of the AMA still poses a challenge for many banks.

Table 4

Number of CEBS banks using the different approaches to operational risk

Approach Group 1 Group 2

Basic Indicator Approach 0 138

Standardised Approach 30 72

Advanced Measurement Approach 19 2

1.2 Data quality The data quality has significantly improved compared to previous simulation exercises. Banks are in a more advanced stage of implementation and generally speaking more data are available, thus significantly improving the accuracy of the IRB estimates. Compared to previous simulation exercises a higher number of banks were able to calculate the capital requirements using the more advanced approaches for credit risk as well as for operational risk.

Furthermore, in earlier exercises details on the exact CRD text and national implementation were not yet available. Their availability during QIS 5 was crucial in improving data quality and reflecting decisions on national discretions in the final results.

The better data quality is reflected in a higher coverage of institution’s exposures and risk-weighted exposures. QIS 5 instructions suggested calculating the capital requirements using the new approaches for a significant portion of their consolidated exposures (possibly at least 80%): the actual coverage is on average 97.7%. Based on the assumption that the included exposures included are representative of the overall banking and trading book exposures, the risk-weighted exposures and expected losses were scaled up to 100%. If riskier (or less risky) exposures were left out of the calculation inaccuracies could therefore occur.

However, the implementation of the most recent changes to the final version of the CRD are not fully reflected in the data provided by the institutions:

i) only a small number of banks distinguished adequately between economic downturn and default-weighted LGDs;

8/65

ii) only a few institutions reported data using the new internal models for exposure values for counterparty credit risk exposures; and

iii) the implementation of the double default treatment in the IRB Approaches varies from country to country: In some countries no bank adopted this option; in other countries at least one Group 1 bank assigned exposures to the Double Default risk-weight curve.

Some further issues remain as far as data quality is concerned. Accounting regimes may differ between countries, the use of CRM techniques still poses some problems and, LGD and exposure value modelling still need further improvement. Furthermore, in relation to the QIS 5 rule of assigning the guaranteed exposure (or a part of this exposure) to the regulatory portfolio of the guarantor, some banks have experienced problems in moving the guaranteed exposures across portfolios. It should be noted that these issues are mainly IT related. CEBS expects these issues mainly to affect the accuracy of individual banks’ results rather than the overall magnitude of the impact on the results of QIS 5.

1.3 Comparability with QIS 3 data

When comparing QIS 5 data to previous simulations, mainly QIS 3, the results should be carefully interpreted for the following reasons:

i) the sample of banks is different, by country and by approach.

ii) various changes to the CRD framework have occurred since QIS 3 was completed, including different correlation factors for the qualifying revolving exposures (QRE) and the other retail risk curve, the move to a UL-only framework for computing risk-weighted exposures, a change in the treatment of reserves, and a multiplier applied to credit risk-weighted exposures.

iii) improvements in risk management, and especially compliance with the CRD, which not only improve data quality, but also result in an overall reduction of risk and consequently of capital requirements

iv) different macroeconomic conditions and thus different states of the credit cycle. Even though the situation is not the same in all EU countries, on average the overall economic conditions seem to have improved since QIS 3. Therefore – all other things being equal – the estimates of risk parameters (mainly PDs and LGDs) can be different.

1.4 Methodology The QIS 5 results were calculated on the basis of the UL-only approach for computing minimum required capital (MRC). UL risk-weighted exposures were computed for each of the regulatory portfolios like corporate, sovereign, interbank and the retail portfolios and include, unless stated otherwise, the 1.06 scaling factor. The risk-weighted exposures also capture market risk and operational risk. Under the IRB approaches the Expected Loss (EL) was computed and compared with the eligible reserves. If the provisions exceed the Expected Losses the difference can be considered as eligible Additional own funds, up to 0.6% of RWA for credit risk. For purposes of QIS 5 analyses the so-called regulatory calculation difference (EL minus provisions) was allocated to the regulatory portfolios on an EL basis in order to compare changes in MRC (CRD versus current regime) at a portfolio level. The formulae for the changes in minimum required capital on the level of individual portfolios and entire banks are set out in detail in the methodological annex (Annex A).

In order to aggregate results of the various countries in the sample and to determine the overall change of capital in the system, country weights were assigned to each country’s total change in MRC. Results of Group 1 banks were weighted by the proportion of Original own

9/65

funds plus Additional own funds, less deductions, of all Group 1 banks in the banking system of each country, irrespective of whether or not they participated in the data collection exercise. For Group 2 banks, the results were weighted based on the capital of Group 2 banks in each country.

The described method of weighting the individual country results is the same as applied for G10 countries in the report on the QIS 5 results by the Basel Committee.

As already mentioned in Section 1.1 subsidiaries of groups located in another EU country were not considered in the QIS 5-sample of banks of the host country in order to avoid double counting.

The tables in sections 2, 3 and 4 of this report provide aggregated information on changes in minimum required capital (under pillar 1) at the bank level and for individual portfolios. Usually the portfolio specific tables include three types of information: For each of the regulatory portfolios the size of the portfolio is expressed in terms of minimum required capital according to the current regime. The second important piece of information is the change in MRC for the specific portfolio which shows the impact of the CRD on the exposures in this particular portfolio. For the change in MRC the comparison to the current regime is expressed in terms of risk-weighted exposures (and possibly deductions) in the specific portfolio relative to the current regime regardless of the share of these exposures in the overall exposures of the bank. In contrast to this, the contribution defined as the product of size and percentage change in MRC demonstrates the extent to which extent the overall result of a bank is driven by a specific portfolio. The change in the MRC at the bank level can then be obtained by summing up the individual portfolio contributions as well as the additional capital requirements for operational risk. In the context of the impact of individual portfolios on the overall result a “positive contribution” or a “negative contribution” to the overall result means that a specific portfolio either contributes to an increase in capital requirements or to a decrease.

The technical details of these concepts (including the related formulae) can be found in Annex A.

The Basel Committee and CEBS have developed prudential filters especially to mitigate the impact of IFRS on regulatory capital. However, prudential filters have not been investigated within QIS 5 and they have not been incorporated in the QIS workbooks. Raw accounting figures had to be adjusted by using prudential filters in order to make the regulatory capital of IFRS and non-IFRS banks comparable. The QIS 5 did not aim to investigate the impact of the adjustments which had to be made to the accounting figures in order to arrive at a coherent basis for regulatory capital. The QIS 5 started with numbers on actual regulatory capital, i.e. Original own funds, Additional own funds and Ancillary own funds and the supervisory deductions. It was left to the national supervisors to provide their participating banks with guidelines on how to implement the prudential filters in the calculation method for regulatory capital.

Finally, the QIS 5 results only refer to the effects on the minimum capital requirements for credit risk, market risk and operational risk under Pillar 1. Possible additional capital requirements which may be set by national supervisors under pillar 2 were not in the scope of QIS 5.

2. Summary of results across approaches

In this section a summary of the results of all QIS 5 participating banks is provided for the three credit risk approaches available under the CRD (Standardised Approach, Foundation

10/65

IRB and Advanced IRB) as well as for the most likely approach, i.e. the approach that individual banks are most likely to adopt at implementation.

It should be noted that results for the IRB Approaches have been calculated on the basis of a scaling factor of 1.06. This factor, which is applied to the credit risk risk-weighted exposures both in the Basel Accord and in the CRD, was introduced in 2003 to offset the decrease in capital which resulted from the shift to the UL-only calibration.

The transitional floors of 95% in 2007, 90% in 2008 and 80% in 2009 were not considered in the tables of this section. The proper calibration of the risk-weight functions is a target related to a longer time horizon. Introducing the floors would contradict this longer-term perspective by looking only at the transitional years. For this reason the transitional floors have not been part of the calibration discussion at any time.

2.1 Overall results Table 5 presents the overall results at group levels for the different approaches.

For Group 1 banks the results for the Standardised Approach show an average decrease in MRC of 0.9% with respect to the current regime. For banks calculating the capital requirements with the IRB Foundation Approach the MRC decreases, on average, by 3.2%; and the decrease is still higher for the Advanced IRB Approach (8.3%).

Focussing on the most likely approach, the results show an average decrease of 7.7%. This is mainly explained by the fact that 53% of Group 1 banks plan to adopt the AIRB Approach at the time of implementation.

Table 5

Overall results Change in minimum required capital relative to current regime [%]


FIRB Approach

AIRB Approach

Most likely Approach

Group 1 -0.9 -3.2 -8.3 -7.7 Group 2 -3.0 -16.6 -26.6 -15.4

At the country level, the dispersion of country average results increases as the more sophisticated approaches are considered. For the AIRB Approach the minimum and maximum values are -38.7% and +6.2%, respectively, whereas for the Standardised Approach the minimum and maximum values are -11.1% and 12.0%. This is consistent with the risk-sensitive nature of the IRB Approaches as opposed to the Standardised Approach, where capital requirements are constrained by a limited range of fixed risk-weights. Therefore, the different risk profile of single banks’ portfolios are better captured. Furthermore the variety and uncertainties that estimation methodologies present for some banks contribute to the high dispersion of the results.

For Group 2 banks, the results for the Standardised Approach are much more representative than for Group 1 banks, given that about half of the Group 2 banks (116 out of 213) have chosen the Standardised Approach as their most likely approach, whereas only two out of the 49 Group 1 banks have done the same. Table 5 shows an average decrease in MRC of

11/65

3.0%, which can be mainly explained by the larger size of the retail portfolio (including the SMEs classified as retail) compared to Group 1 banks (49.3% versus 35.5%). For some countries this size difference is even more apparent.

IRB results are also more favourable for Group 2 banks, even though they are presumably less representative, in terms of most likely approach and data quality. In the FIRB Approach MRC show a decrease of 16.6%, and in the AIRB Approach a reduction of 26.6%. Considering the most likely approach, which is the Standardised Approach for more than 50% of the banks, the average MRC decreases by 15.4% in comparison with the current regime.

The following histograms (charts 1 to charts 4) show the dispersion in the change of MRC relative to the current regime among the participating Group 1 and Group 2 banks. The dispersion within countries for the different approaches seems to be caused by differences in portfolio characteristics and disparity and uncertainties in methodologies for estimation; the variation among countries is driven, additionally, by the different points of the economic cycle. In general, the variation in the results seems to be somewhat bigger for the Group 2 banks. However, compared with QIS 3 and in particular for the Group 1 banks not many outliers can be found.

12/65

Charts 1

Change in banks’ minimum required capital, Standardised Approach versus current regime [%]

Change Standardised vs Current by size - Group 1

-40%

-20%

0%

20%

40%

60%

80%

Change Standardised vs Current by size - Group 2

-40%

-20%

0%

20%

40%

60%

80%

Charts 2

Change in banks’ minimum required capital, Foundation IRB Approach versus current regime [%]

Change FIRB vs Current by sizeGroup 1

-70%

-60%

-50%

-40%

-30%

-20%

-10%

0%

10%

20%

30%Change FIRB vs Current by size

Group 2

-70%

-60%

-50%

-40%

-30%

-20%

-10%

0%

10%

20%

30%

≤ 10bn >10bn € capital

≤ 10bn >10bn € capital ≤ 10bn >10bn € capital


13/65

Charts 3

Change in banks’ minimum required capital, Advanced IRB Approach versus current regime [%]

Change AIRB vs Current by sizeGroup 1

-50%

-40%

-30%

-20%

-10%

0%

10%

20%

30%Change AIRB vs Current by size

Group 2

-50%

-40%

-30%

-20%

-10%

0%

10%

20%

30%

Charts 4

Change in banks’ minimum required capital, most likely approach versus current regime [%]

Change ML vs Current by sizeGroup 1

-80%

-60%

-40%

-20%

0%

20%

40%

60%Change ML vs Current by size

Group 2

-80%

-60%

-40%

-20%

0%

20%

40%

60%



14/65

2.2 Incentive structure One of the main objectives of the new framework is to provide banks with a range of different methodologies (both for credit and operational risks) which, on average, should give adequate incentives to improve their risk measurement systems and therefore progressively apply the most sophisticated approaches for the calculation of their capital requirements.

In order to evaluate whether the incentive structure is in line with the supervisory intentions, the following table compares the results between the Standardised and FIRB Approach and between the FIRB and AIRB Approach. Only banks that provided data on both approaches, whether Standardised and FIRB, or FIRB and AIRB are included. The comparison of the changes in MRC for two approaches of the same bank directly indicates whether there is a capital relief for the more sophisticated approach or – alternatively – the MRCs increase in spite of the use of the more sophisticated approach. The first alternative would give a negative algebraic sign in the columns of the following table 7, the second alternative a positive one on the level of individual banks.

Table 6 shows that on average the incentive structure works well for all three approaches and for both groups of banks.

It is especially important to consider the incentive structure between the two IRB approaches for Group 1 banks. According to table 6 they can reduce their MRCs considerably by using the Advanced IRB Approach instead of the Foundation IRB Approach.

Group 2 banks (apart from limited exceptions) also have a strong incentive to move to the more sophisticated approaches, i.e. to one of the IRB approaches. Regarding the AIRB, there are only a few observations available for Group 2 banks. Therefore these data are statistically not as representative as the data for the AIRB results for the internationally active Group 1 banks.

Table 6

Incentive structure Minimum required capital for Foundation IRB relative to Standardised, and minimum required capital for Advanced IRB relative to Foundation IRB [%]

Group 1 Group 2 FIRB/

Standardised AIRB/FIRB FIRB/ Standardised AIRB/FIRB

Average -13.5 -6.8 -12.4 -6.7

For “FIRB/Standardised” columns, only those banks which provided data for both the Standardised and the Foundation IRB Approaches are included. For “AIRB/FIRB” columns, only those banks which provided data for both the Foundation and the Advanced IRB Approaches are included. The figures do not take account of the transitional floors as described in Article 152 (1 to 4) of the CRD.

15/65

3. Standardised Approach

3.1 Overall and portfolio-specific results Table 7 provides the average results for the Standardised Approach on a portfolio basis for Group 1 and Group 2 banks. On average there seem to be no significant differences between the two groups of banks both for the percentage change with respect to the current regime and for the relative contribution of single portfolios.

The main contributor to the overall results is the mortgage portfolio (-7.8% of contribution to MRC change for Group 1, -7.2% for Group 2). The remaining retail portfolios also show a negative contribution. However, their magnitude is larger for Group 2 banks, which typically have higher retail activity. Other retail, SME retail and QRE contribute to the overall result of Group 2 banks with -3.1%, -1.7% and -0.3% respectively (-1.0%, -0.9% and -0.2% for Group 1 banks).

These findings are broadly in line with the results of previous Quantitative Impact Studies, and are consistent with the risk-weighting treatment provided in the new framework for retail exposures: i) the lower risk-weight for mortgages (35% instead of 50%), and ii) that for other retail, including SMEs (from 100% to 75%).

The corporate portfolio shows a slight decrease in MRC both for Group 1 and Group 2 banks while the MRCs of the SME corporate portfolio also slightly decrease for Group 1 banks but increase for Group 2 banks. Interpreting the results for both corporate portfolios it has to be taken into account that most counterparties are not externally rated and that a different treatment for defaulted loans has been introduced under the new framework.6

All retail portfolios show a decrease in MRC both for Group 1 and Group 2 banks. This decrease can be observed for almost all participating countries.

Changes in capital requirements for the sovereign and bank portfolio should be interpreted carefully. Under the current regime, capital requirements for these portfolios can be close to (or even equal to) zero. Consequently, even capital requirements which are modest in absolute terms can have a big effect in relative terms, since it is an increase from a very low base. This is reflected in the contribution column in the tables below where, although the change in capital requirements looks large, the contributions remain well under 2%. Both sovereign and bank portfolios show remarkable percentage increase in MRCs. However, in terms of contribution the bank portfolio of Group 1 is the more important driver of the increase in MRC (bank: 1.8%, sovereign: 0.4%) due to its relative size (bank: 6.0%, sovereign: 0.4%).

Another significant driver of the results is the related entities portfolio, with a 2.0% and 0.9% contribution for Group 1 and Group 2 banks respectively. This result is mostly driven by changes in the treatment of insurance entities.

6 The risk-weight of past due loans depends on the amount of provisions which are compared with the outstanding amount of

the loan.

16/65

Table 7

Portfolio specific results for the Standardised Approach, Change in minimum required capital relative to current regime [%]

Group 1 Group 2

Portfolio % of current

MRC

% change in MRC

Contrib. % of current

MRC

% change in MRC

Contrib.

Wholesale

– Corporate 17.7 -1.9 -0.3 10.1 -6.1 -0.6 – Sovereign* 0.4 97.6 0.4 0.2 27.2 0.1 – Bank 6.0 29.0 1.8 6.0 -11.1 -0.7 SME Corporate 8.3 -5.1 -0.4 13.0 1.5 0.2 SL 5.4 -6.4 -0.4 1.7 -0.6 0.0 Retail – Mortgage 27.7 -28.2 -7.8 24.5 -28.5 -7.2 – Other retail 4.4 -23.6 -1.0 15.3 -20.0 -3.1 – Revolving 0.7 -22.9 -0.2 1.4 -22.3 -0.3 SME retail 2.7 -22.2 -0.9 8.1 -20.6 -1.7 Equity 1.2 18.3 0.2 1.8 1.8 0.0 Purchased Receivables 0.1 -19.3 -0.1 0.2 -0.7 0.0 Other exposures 3.0 0.0 0.0 3.3 0.0 0.0 Securitisation 2.6 12.9 0.4 1.2 5.0 0.1 Trading Book CCR 1.6 34.4 0.9 0.1 46.6 0.1 Specific risk 1.3 6.5 0.1 0.4 2.6 0.0 General market risk 2.1 0.9 0.0 1.7 0.0 0.0 Related entities 5.7 19.9 2.0 2.3 37.2 0.9 Large exposures 0.2 0.0 0.0 0.2 0.0 0.0 Deductions 5.0 -0.5 0.0 3.2 -1.2 0.0 Partial Use** 4.0 -3.2 -0.2 5.2 1.2 0.1 Operational risk 5.5 9.0 Overall change 100.0 -0.9 100.0 -3.0 * The large percentage change in capital for the sovereign portfolio arises because some sovereign exposures are allocated a 0% risk-weight under the current regime. Applying any risk-weight under the new approaches gives an infinite percentage increase in capital requirement for those banks with only such exposures, even if in absolute terms the change in capital requirements is small. – ** This row includes Standardised Approach capital requirements for exposures subject to partial use under the IRB approaches for banks also providing data for at least one IRB approach.

17/65

Securitisation and Trading book counterparty risk also provide positive contributions to the average result; a detailed analysis of these portfolios is provided in Sections 5 and 6.

Not surprisingly, in relation to equity exposures, both the percentage change and the relative contribution are relatively small, given the 100% risk-weight both in the new Standardised approach and in the current regime. Only a few countries differ from this pattern (making use of national discretions granted by the CRD in this respect).

3.2 Past Due exposures

The following scattergrams show the percentage of exposures that are past due for the banks that provided data for the Standardised Approach.

Charts 5

Proportion of exposures past due [%]

Percentage of exposures past due - Group 1

0%

2%

4%

6%

8%

10%

12%

14%

16%

18%

20%

Percentage of exposures past due - Group 2

0%

2%

4%

6%

8%

10%

12%

14%

16%

18%

20%

Although the percentage of past due exposures might look quite dispersed, most Group 1 banks (about 70%) show less than 2% of exposures past due, and only a small fraction (less than 10%) of the institutions shows 5% or more exposures past due. For Group 2 banks, similar results can be observed. Most banks (73%) also have less than 2% of exposures in default, only a small fraction (less than 9%) shows 5% or more exposures past due. The Group 2 sample also contains a few outliers. The averages in both groups are 1.4% and 1.7% for Group 1 and Group 2, respectively.

By ordering the portfolios from high to low percentages, past-due exposures on average seem to occur mainly in the SME retail, other retail, SME corporate, retail QRE, retail mortgage and corporate portfolio. As regards the distribution across countries, there is no evidence of a clear relationship between the percentage of past due exposures and the economic condition of individual countries. This is not surprising, considering the time lags between the economic cycle and the credit cycle. Furthermore, this might also depend on the unsecured portion of the loan, the national rules on the definition of past due and on the applicable risk-weight.


18/65

4. IRB Approaches

4.1. Capital drivers under the IRB approaches In general, the favourable results for many banks (in terms of a decrease in MRC) are due to a high portion of retail and residential mortgage exposures. In many countries the corporate and SME portfolios (including SME retail and specialised lending) also significantly contributed to a reduction in MRC. Many supervisors explicitly mentioned favourable macroeconomic conditions as the main reason.

Because of their relatively long maturity, residential mortgage exposures are sensitive to interest rate risk. The reported decrease in minimum requirements for these portfolios might therefore be partially offset by capital requirements under pillar 2 relating to this risk type. Such possible requirements have not been gauged the in QIS 5 exercise, however.

In some countries related entities contributed significantly to an increase in MRC, mainly driven by changes in the treatment of insurance entities (as already mentioned, this observation also holds true for the Standardised Approach).

The capital requirements for securitisation exposures will probably rise under the CRD. However, the QIS results are mainly driven by the risk-weights for liquidity facilities and higher capital requirements for deeply subordinated positions. These facilities are certainly overrepresented and – according to the qualitative analysis provided by some national supervisors – the capital reducing effects for investors might not be properly reflected.

Many banks claimed that the impact of the CRD is not that favourable for banks having large Trading Books with high volumes of OTC derivatives. The removal of the 50% cap on the risk-weighted assets for OTC derivatives which was applied under the current regime often results in an increase in the MRCs even for counterparty credit risk exposures with a good rating.

Operational Risk is obviously a capital driver because there was no capital charge for Operational Risk under the current regime. This remark especially applies to countries where the increased profitability of banks in the recent years has led to particularly high contributions from the Operational Risk capital requirements.

Equally the sovereign portfolio contributes to an increase in MRC because of the 0% risk-weight which is assigned to OECD-countries under the current regime. Furthermore, in some countries the AAA rating is not necessarily used for exposures to public sector entities anymore as it commonly was at the time of QIS 3. In general, the estimated PDs were quite conservative which was often due to the methodological uncertainties which still exist. In some cases increasing capital requirements for sovereign and interbank exposures were not directly observed in the respective exposure classes but contributed indirectly via the “partial use” portfolio as the banks were allowed to report the exposures eligible for a permanent partial use separately.

4.2. Foundation IRB Approach Table 8 presents the contribution of the regulatory exposure classes to the change in MRC for the Foundation IRB approach.

First of all it can be seen that the retail portfolios are the main drivers of the overall reduction in MRC for both groups of banks.

19/65

For Group 1 banks the largest negative contribution results from the residential mortgage sub-portfolio (-9.4% contribution) for which the capital requirements are about 65.7% less than under the current regime. SME retail and the other retail portfolios contribute -2.1% and -0.8% respectively. However, the QRE sub-portfolio provides an exception to the general trend for a reduction in MRC in the retail portfolio. QRE shows an average increase in MRC of 0.9% which is mainly due to the relatively high PD and/or LGD levels for some countries.

The corporate and SME corporate portfolios show a moderate decrease in MRC. In combination with the large sizes (24.0 % and 6.9 %) the contribution to the overall decrease is 2.8% (=2.3%+0.5%).

Sovereign and equity portfolios contribut.e to the increase in MRC. The impact of the former is common to the majority of EU countries, caused by mapping of banks’ estimated PDs with external ratings and the fact that a 0% RW is assigned to OECD-countries in the current regime. The 0% RW is relevant for some countries, where the grandfathering clause for equities (i.e. 10 years exemption for the IRB calculation) has not been implemented or only partially implemented.

Other significant differences for Group 1 banks are the larger contribution from the bank exposures (due to the larger size of the portfolio) and the related entities portfolio.

For Group 2 banks the mortgage and, to a lesser extent, the other retail portfolios are the main drivers of the considerable average reduction under the FIRB Approach; the relative contribution for these sub-portfolios are -16.3% and -5.9%, respectively. The decrease in capital requirements for the mortgage portfolio at least partially reflects the effect of increases in housing prices over the last decade; this effect may be exaggerated by the problems banks still face in estimating economic downturn LGDs. The contribution to the decrease in MRC for the corporate portfolios is 4.0% (=2.3%+1.7%) which is a somewhat bigger decrease compared to Group 1 banks.

The results on the partial use treatment are quite widely dispersed across the different countries. It is not easy to observe from the QIS workbooks to which type of exposure the partial use treatment was applied within QIS 5.7 For Group 2 banks the possibility of a permanent partial use seems to be more crucial as the share of exposures which received the partial use treatment is bigger (Group 1: 2.7%, Group 2: 7.9%). However, as the figures for the contributions of this “portfolio” show, the average contribution is not necessarily big if the share in the overall exposure is big. Size and contribution seem to be uncorrelated. Relatively high contributions indicate that banks have applied the partial use predominately to sovereign exposures which include a high share of exposures to OECD countries and, consequently, produce big increases in minimum required capital.

7 In the QIS 5 instructions banks were asked to assign exposures to the partial use treatment only if these exposures qualify

for the permanent partial use. The input cells were not intended to include exposures to which partial use is applied during transition periods or in cases where internal estimates of the PDs were not available yet but will be in the near future.

20/65

Table 8

Portfolio specific results for the Foundation IRB Approach, Change in minimum required capital relative to current regime [%]

Group 1 Group 2


MRC

% change in MRC


MRC

% change in MRC

Contrib.

Wholesale

– Corporate 24.0 -9.6 -2.3 10.9 -21.2 -2.3 – Sovereign 0.5 276.8 1.5 0.1 314.7 0.5 – Bank 4.9 7.7 0.4 4.0 -16.4 -0.8 SME Corporate 6.9 -6.5 -0.5 14.0 -11.6 -1.7 SL 4.2 4.7 0.2 2.5 20.8 0.6 Retail: (total) – Mortgage 14.2 -65.7 -9.4 25.3 -63.9 -16.3 – Other retail 5.0 -15.4 -0.8 11.6 -42.2 -5.9 – Revolving 1.6 52.5 0.9 1.4 -26.0 -0.5 SME retail 4.2 -50.2 -2.1 6.5 -43.9 -2.9 Equity 1.3 88.8 1.2 1.6 124.5 2.1 Purchased Receivables 0.1 -36.9 -0.1 0.2 3.2 0.0 Other exposures 2.9 0.0 0.0 3.2 0.0 0.0 Securitisation 2.3 3.2 0.1 1.4 12.9 0.2 Trading Book CP 2.3 17.8 0.4 0.2 11.1 0.0 Specific risk 1.5 6.3 0.1 0.5 -1.0 0.0 General market risk 2.9 -0.9 0.0 2.7 0.3 0.0 Related entities 6.0 13.1 0.9 3.0 38.5 1.2 Large exposures 0.2 0.0 0.0 0.1 0.0 0.0 Deductions 12.4 -0.1 0.0 2.9 0.0 0.0 Partial use 2.7 23.2 0.9 7.9 3.1 0.3 Operational risk 5.8 7.9 Overall change 100.0 -3.2 100.0 -16.6 The large percentage change in capital for the sovereign portfolio arises because a 0% risk-weight is assigned to a big share of sovereign exposures under the current regime. Any other risk-weight under the new approaches gives an infinite percentage increase in capital requirement for those banks with only such exposures, even if in absolute terms the change in capital requirements is small.

21/65

4.3. Advanced IRB Approach Table 9 shows that, as with the FIRB approach, the main drivers of the overall reduction in MRC are, apart from the corporate portfolio (Group 1: -4.4%, Group 2: -5.3%) mainly the retail portfolios. As far as the corporate portfolio for Group 2 banks is concerned the reductions both in terms of change in MRC and contribution are even bigger than for Group 1 banks.

More specifically, for Group 1 banks the relative contributions of residential mortgage and SME retail are, respectively, -8.5% and -2.0%. For Group 2 banks, which are on average more specialised in this area, the relative contributions of the two portfolios are -5.1% and - 7.3%, respectively. Considerable decreases in MRC can also be observed for the other retail portfolio (Group 1: -23.2%, Group 2: -44.0%). Again, the impact on the bank level is bigger for Group 2 banks as their share of other retail exposures is significantly higher (for example 4.4% for Group 1 banks versus 32.5% for Group 2 in terms of MRC relative to the current regime).

For the partial use the same remarks as for the Foundation IRB Approach apply. However, the contribution of the partial use exposures to the overall change in MRC is much lower for the Advanced IRB Approach relative to the Foundation IRB (Group 1: -0.3%, Group 2: 0.2%). The different figures are mainly driven by the different sample of banks which adopt both approaches.

22/65

Table 9

Portfolio specific results for the Advanced IRB Approach, Change in minimum required capital relative to current regime [%]

Group 1 Group 2


MRC

% change in MRC


MRC

% change in MRC

Contrib.

Wholesale

– Corporate 25.2 -17.6 -4.4 11.8 -44.5 -5.3 – Sovereign 0.4 178.1 0.8 0.1 687.1 0.5 – Bank 3.0 1.8 0.1 3.0 8.6 0.3 SME Corporate 7.2 -23.6 -1.7 11.1 -45.2 -5.3 SL 4.9 -24.6 -1.2 2.6 -17.7 -0.6 Retail: (total) – Mortgage 12.9 -65.4 -8.5 7.9 -64.6 -5.1 – Other retail 4.4 -23.2 -1.0 32.5 -44.0 -15.0 – Revolving 1.5 71.1 1.1 1.2 -62.2 -1.0 SME retail 4.1 -48.5 -2.0 12.2 -57.0 -7.3 Equity 1.3 67.3 0.8 0.9 175.4 1.9 Purchased Receivables 0.0 75.3 0.1 0.0 Other exposures 2.8 0.0 0.0 4.8 0.0 0.0 Securitisation 2.3 19.4 0.4 0.4 20.3 0.2 Trading Book CP 3.5 5.9 0.2 0.2 15.0 0.0 Specific risk 1.3 0.5 0.0 1.4 -0.2 0.0 General market risk 2.3 -0.6 0.0 1.3 -0.2 0.0 Related entities 8.3 12.2 1.0 4.2 29.7 1.3 Large exposures 0.0 0.0 0.0 0.1 0.0 0.0 Deductions 10.9 0.0 0.0 2.5 -0.1 0.0 Partial use 3.8 -7.3 -0.3 1.9 9.1 0.2 Operational risk 6.3 7.5 Overall change 100.0 -8.3 100.0 -26.6 Even though there is only one IRB approach to retail, the results differ from those presented in the previous table due to a different sample of banks.

23/65

4.4 Analysis of the risk parameters PD and LGD

4.4.1 PD One of the main drivers of the results for IRB approaches are the PDs estimated by banks. Since PD is the only risk parameter that has to be estimated in both IRB approaches, no distinction between the two approaches is made here. However, it should be taken into account, as a preliminary caveat, that a comparison of PDs across banks and across countries can provide only rough information about the quality of the portfolios, given that a wide range of factors are likely to affect the PD values:

i. the statistical methodologies adopted by the banks;

ii. the definition of default used to estimate and calibrate the rating systems;

iii. the master scale used to aggregate different rating systems within the same banking group; and

iv. as already mentioned, the economic situation in the different markets.

All average PDs discussed in this report are exposure-weighted averages. Defaulted assets were not considered as the impact of taking into account the 100% PD which is usually assigned to defaulted assets would be too big. The average PDs at a portfolio level are given in table 10.

For Group 1 banks the average PD is 1% for the corporate portfolio (ranging from 0.6% to 1.8%); for Group 2 banks the average value is 0.8%, the dispersion being somewhat wider than for Group 1 banks (from 0.3 to 4.2%). The average PD in the SME corporate portfolio is 2.2% for both Group 1 and Group 2 banks. However there is a quite big dispersion for Group 2 banks (1.3% to 9.5%). Among the participating countries, almost all countries have achieved over 65% of exposures which are below a PD of 0.8% and at most 3% of defaulted exposures in the corporate portfolios of Group 1 and Group 2 banks. This is likely to be due to favourable macroeconomic conditions and a better credit environment in most countries.

In the retail portfolio, PDs are relatively high for QRE, other retail and SME retail exposures (on average, 3.7%, 4.3% and 3.3% respectively for Group 1 banks); and the dispersion is also quite high (for example, ranging from 0.7% to 12.4% for QRE exposures). In contrast to this, for the residential mortgage sub-portfolio (average PD 1.5%) the dispersion is lower. A similar pattern can be observed for Group 2 banks.

24/65

Table 10

Average PDs

average PD [%] Defaulted [%]

Group 1

Corporate 1.04 1.90 Bank 0.22 0.20 Sovereign 0.13 0.10 Retail Mortgage 1.52 1.90 Retail QRE 3.69 4.40 Other Retail 4.33 5.70 SME Corporate 2.20 4.30 SME Retail 3.26 5.00 Group 2 Corporate 0.83 2.50 Bank 0.11 0.10 Sovereign 0.04 0.40 Retail Mortgage 1.39 1.30 Retail QRE 2.33 3.30 Other Retail 2.32 5.10 SME Corporate 2.16 7.70 SME Retail 3.66 5.20

As another possibility to compare the risk-profiles of the different regulatory exposure and sub-exposure classes the internal estimates of the PDs were mapped to four different rating-buckets which broadly reflect the rating grades of A and better, BBB, less than BBB, and defaulted exposures. The corresponding distribution of exposures across these four buckets is provided in table 11.

In the wholesale portfolio (corporate, bank, sovereign) there are no significant differences between Group 1 and Group 2 banks. For corporate exposures about 40% are classified A and better and about 30% are classified BBB; the defaulted exposures represent 1.9% and 2.5% of the portfolio for Group 1 and Group 2 banks respectively. However, as expected, in bank and sovereign most of the exposures are highly rated (PD less than 20bp): 86% and 93% for Group 1 banks, 85% and 98% for Group 2 banks.

25/65

Table 11

PD Quality Distributions

PD bucket < 0.2% 0.2% to 0.8% 0.8% to 99.99% Defaulted

Group 1

Corporate 38.5 31.8 27.8 1.9Bank 86.2 9.1 4.5 0.2Sovereign 93.4 3.3 3.2 0.1Retail Mortgage 30.8 34.6 32.7 1.9Retail QRE 15.6 28.9 51.1 4.4Other Retail 13.3 21.6 59.5 5.7SME Corporate 14.1 31.5 50.2 4.3SME Retail 11.7 24.8 58.4 5.0Group 2 Corporate 41.9 32.6 23.0 2.5Bank 85.3 13.4 1.1 0.1Sovereign 98.1 1.0 0.6 0.4Retail Mortgage 35.0 33.4 29.2 1.3Retail QRE 47.0 23.8 25.8 3.3Other Retail 23.4 29.4 42.1 5.1SME Corporate 16.8 27.7 47.8 7.7SME Retail 10.3 26.2 58.2 5.2

The SME corporate portfolio is on average of lower credit quality for both groups of banks: around 50% of the exposures are rated less than BBB and a significant proportion of exposures is in default (4.3% and 7.7% for Group 1 and Group 2 banks respectively).

The average credit quality is lower for retail portfolios. For Group 1 banks only 13% of the exposures included in the other retail sub-portfolio are rated A and better; about 60% is rated less than BBB; the defaulted exposures represent about 6% of the total. The distribution for Group 2 banks is not significantly different. Also the credit quality for SME retail is lower than for the SME corporate portfolio, both for Group 1 and Group 2 banks.

In the QRE portfolio, for Group 1 banks about half of the exposures are rated less than BBB. The distribution in the mortgage sub-portfolio is on average more skewed towards the higher-rated classes; the portion of defaulted exposures is lower than for the other sub-portfolios (1.9% for Group 1 banks, 1.3% for Group 2 banks). Additional information is contained in Section 1 of Annex B.

26/65

4.4.2 Collateralisation and LGDs under the Foundation IRB approach

Obviously, the LGD estimates banks provided for QIS 5 purposes strongly depend on the ability of the banks to take fully into account the eligible types of collateral and to capture the CRM benefits at the reporting date.

Charts 6 show that there is a wide range in the percentage shares of exposures the banks reported as being secured both for Group 1 and Group 2 banks. This reflects the different ability of institutions to capture the CRM benefits for QIS 5 purposes. As the scatterplots show there are banks which were not able to take collateral into account at all.

Table 12 shows the exposures secured by collateral by country for Group 1 and Group 2 banks. The analysis was carried out on the basis of Foundation IRB banks8 because no separation was possible under Advanced IRB. Over 85% of corporate exposures for Group 1 banks are on average reported as unsecured, which may in part be due to limited reporting of CRM in the QIS 5 data collection exercise. For Group 2 banks, where commercial mortgages are likely to represent a big share of the corporate portfolio, 88% are reported as unsecured.

Table 12

Exposures secured by collateral [%], Foundation IRB Approach

Average Corporate Bank SME Corporate Sovereign

Group 1 banks 14.7 15.7 29.4 1.9 Group 2 banks 11.6 11.6 19.0 3.7

Both for Group 1 and Group 2 banks more than 96% of sovereign exposures are unsecured, but given the relatively small size of this portfolio and the high quality of these exposures (over 93% and 98% of these exposures with a PD less than 0.20% for Group 1 and Group 2 banks, respectively) the contribution to the overall change in MRC is small. There is some dispersion across countries with values ranging from -3.0% to 5.8%.

Similar arguments apply to the bank portfolio where only 15.7% of Group 1 banks’ exposures are secured9 (11.6% for Group 2 banks). Around 86% of these exposures have a PD lower than 0.20% for Group 1 and Group 2 banks.

8 In order to determine the percentage share of uncollateralized exposures all exposures assigned to the LGD values of 75%

and 45% were taken into consideration. 9 Repo-style transactions where banks adjusted the exposure value amount are treated as unsecured in this analysis.

27/65

Charts 6

Proportion of exposures secured by collateral, banks by size10

Percentage of secured exposures Group 1

0%

20%

40%

60%

80%

100%

120%

Percentage of secured exposuresGroup 2

0%

20%

40%

60%

80%

100%

120%

In the wholesale portfolio the average LGD values under the Foundation IRB Approach are as follows: 43% in corporate, 39% in banks, 45% in sovereign. The averages are similar for Group 2 banks (42%, 41% and 43%, respectively). As expected, the distribution of values is very narrow, except for a few banks which show much lower LGD values. This is due to a high portion of exposures with financial collateral (mainly in the sovereign portfolio).

4.5 LGD under the Advanced IRB Approach

LGD is, together with PD and exposure value, the main risk parameter in the IRB framework; its value has a linear impact on capital requirements. As already mentioned above, methodologies and systems for LGD calculations are still under development in most banks. This is reflected in the variety of approaches adopted in QIS 5 and may also explain to a large extent the dispersion observed in the results.

Average LGDs reported in this document are exposure-weighted averages for the respective portfolios in order to be consistent with the method the average PDs were computed. LGDs for defaulted assets were not considered in this calculation. However, the distribution of defaulted exposures across the different LGD buckets does not deviate much from the distribution of non-defaulted assets, i.e. results would be almost the same if defaulted assets were included. As far as LGDs for the retail mortgage portfolio are concerned the calculation in table 13 used the LGDs after the 10% floor has been applied. In contrast, for the LGD distributions in the Annex B.2 LGDs before the 10% floor has been applied were taken into account, as these LGDs better reflect the internal estimates.

10 As already mentioned in the explanatory remarks related to Table 2 some participating banks were included on the basis of

QIS 4-submissions. By technical reasons – as the QIS 4 workbooks were structured slightly different – such banks were not included in those scatterplots which directly refer to risk-parameters, for example PDs and LGDs. However, all banks regardless of whether their QIS 4 or QIS 5 workbooks were used for the analyses, were included in Tables and Charts related to changes in MRC both on entire bank and portfolio level.


28/65

Few banks in the EU sample have fully incorporated downturn considerations in their estimates of LGDs. Difficulties in the assessment of downturn effects have been encountered across all portfolios. This is mostly due to the lack of downturn history. Some banks worked with conservative adjustments (eg. scaling factors or add-ons) instead of sophisticated estimates to account for limitations in the methodologies and/or lack of data.

With respect to the AIRB, the LGD values estimated by banks are obviously more dispersed and, as expected, lower on average than the regulatory values. For corporate exposures the average LGD does not significantly differ from the FIRB regulatory value: 38% for Group 1 banks, 35% for Group 2 banks). In a few cases the estimates provided by single banks are even higher than the regulatory values, given the conservative approach adopted in the estimation as well as the high ratio of unsecured subordinated claims and also a high ratio of undrawn lines.

In the SME corporate portfolio the average LGD is 35% (27% for Group 2 banks). For bank and sovereign portfolios the values are 38% and 28% respectively (39% and 38% for Group 2 banks), even though for these two portfolios the dispersion of values is higher, ranging from 20% to 55% in the bank portfolio, from 1.6% to 43% in the sovereign portfolio.

In the mortgage retail portfolio the average LGD for Group 1 banks is quite low (16%), even though on an aggregated basis above the 10% regulatory floor; for Group 2 banks the average LGD is slightly higher (21%). Certainly, the 10% floor for the LGDs for the retail mortgage portfolio will act as a mitigating factor with regard to economic downturn conditions.

For the other retail portfolios the average LGDs for Group 1 banks are the following: 39% for SME retail, 48% for other retail, 55% for QRE; for Group 2 banks the values are not significantly different (32%, 42%, 52% respectively).

Tables 13 shows banks’ average LGD estimates for the IRB Retail and AIRB portfolios for Group 1 and Group 2 banks.

Table 13

LGD averages for different portfolios

IRB Retail AIRB Average

Other QRE RRE SME Corp. Sov. Bank

SME Corp.

Group 1 banks 47.9 55.0 16.1 38.8 38.1 27.7 37.7 35.1 Group 2 banks 42.2 51.9 21.4 31.7 35.2 38.2 39.4 26.7

For Group 1 banks, the LGD values of the corporate portfolio are quite similar across countries with only few exceptions while the dispersion for the interbank exposures is higher for most countries.

The range of LGD values goes from the 10% floor to 26% in the mortgage portfolio and from 22% to 60% in the other retail portfolio. The highest values for retail can be observed for the QRE exposures (up to 75%) which is plausible because QRE exposures are uncollateralized by definition.

With the LGDs of Group 2 banks there is apparently a large variation across countries in the other retail portfolio (13% up to 63%).

29/65

Another obvious point is the low average LGD value for QRE exposures for some countries. It needs to be carefully checked by the supervisors whether the assignment of exposures to the QRE risk-weight curve is fully compliant with the CRD. In some cases banks might have assigned low LGD values where they have other types of exposures to the same costumer and the LGD reflects a customer specific LGD rather than an LGD assigned to the uncollateralized part of the exposure (e.g. an overdraft facility). Additional information – especially on the variation of internal estimates of the LGDs and the distribution of LGD estimates – is included in Section 2 of Annex B.

For defaulted exposures, an important aspect of the IRB formulae is the fact that the UL charge for these exposures is calculated as the difference between the downturn LGD and the best estimate of EL. Consequently, if the two LGD values are equal the capital requirement on those defaulted exposures is zero.

Given the methodological challenges that the estimation of the downturn LGD poses at the moment for most banks, only a sub-sample of the QIS 5 participating institutions have estimated different values for defaulted exposures.

More specifically, for Group 1 banks in the corporate portfolio 11 out of the 26 AIRB banks (42.3%) provided different values for downturn LGD and best estimate of EL. As expected, for Group 2 banks the percentage is much lower (18.8%, represented by only 3 banks). For both groups of banks the percentage is even lower for the other wholesale portfolios (bank and sovereign).

About 20 banks distinguished between downturn LGD and best estimate of EL for the retail portfolios. For the banks which provided the data, the UL charges (as a percentage of the exposure) are plotted in a scattergram (see charts 7). Focussing only on Group 1 banks, for which data are more significant, the UL charge on the corporate portfolio equals 3.8% of the exposure at default, for the bank portfolio this value equals 6.7%; the UL charge is even higher for the sovereign portfolio (13.3%).

Overall, preliminary discussions with banks showed that the different LGDs to be estimated by banks for the Advanced IRB Approach might cause incentive problems between the Foundation IRB and Advanced IRB Approaches: As already mentioned above, the UL-based risk-weighted assets for defaulted exposures are zero under Foundation IRB because there are no UL risk-weighted assets for defaulted exposures as no distinction is made between downturn and expected LGD. If a bank estimates two different LGD values for defaulted exposures it needs to hold capital for 8% of the non-zero UL risk-weighted assets which would be not the case if the bank applied the Foundation IRB rules for the same exposures. Due to lack of data (see charts 7 for the few available observations) there is no estimation possible at the moment on how material this issue is. However, in both approaches – Foundation and Advanced IRB – provisions must at least be equal to the EL in order to avoid deduction from capital.

Banks claimed to find more evidence on the economic significance of the downturn LGD for the retail portfolios than for corporate exposures. Focussing on the residential mortgage portfolio for the Group 1 banks which provided different values for downturn LGD and best estimate EL the UL charge on this portion of activity equals 3.6% of the exposure at default. The values are even higher for QRE (8.3%) and the other retail portfolios (5.0%).

30/65

Charts 7

UL charge as difference between downturn LGD and best estimate of EL

UL charge by size - Group 1

0%

5%

10%

15%

20%

25%

30%

35%

UL charge by size - Group 2

0%

5%

10%

15%

20%

25%

30%

35%

Many banks used quite conservative exposure value estimates for undrawn lines especially for commitments which attracted a zero capital charge under the current regime such as lines which are not unconditionally cancellable and have maturity below one year. Even if lines are unconditionally cancellable by contract, where they are not in practice the exposure value estimates under the Advanced IRB Approach require the use of internally modelled (positive) CCFs. For banks having high volumes of these lines the incentive to move to the Advanced IRB Approach could be somewhat disturbed.

4.6 Regulatory calculation difference An important part of the IRB framework is the impact of the banks’ provisioning policies on the capital requirement calculations. As already mentioned, the UL-only calibration introduced a higher consistency with the theoretical paradigm under which capital must cover unexpected losses and provisions must cover expected losses. According to this principle, if provisions (general and specific) are not sufficient to cover the overall expected losses (“shortfall”), the regulatory calculation difference (RCD)11 must be deducted from capital; conversely, in the case where provisions exceed the EL, the positive difference can be recognised as a component of additional own funds, although within a specified limit.

In the scatterplots below, as positive RCD yields an increase in MRC, the dots representing those banks are above the horizontal axis. Banks with an excess (i.e. a negative RCD) can be found below the horizontal axis as the excess negatively contributes to the change in MRC compared with the current regime.

11 The regulatory calculation difference is defined by Expected Losses minus Total Amount of Provisions. The Total Amount of

Provisions comprises specific and general provisions. For further details see the methodological Annex A.


31/65

Charts 8

Regulatory calculation difference

Contribution RCD FIRB by sizeGroup 1

-10%

-5%

0%

5%

10%

15%

20%

25%

30%

35%

Contribution RCD FIRB by sizeGroup 2

-10%

-5%

0%

5%

10%

15%

20%

25%

30%

35%

Contribution RCD AIRB by sizeGroup 1

-10%

-5%

0%

5%

10%

15%

20%

25%

30%

Contribution RCD AIRB by sizeGroup 2

-10%

-5%

0%

5%

10%

15%

20%

25%

30%

Table 14 shows that, on average for Group 1 banks, in FIRB the shortfall contributes about 3% (1.6% for Group 2 banks) to the overall change in MRC. In AIRB there is also positive contribution of 2.4% and 1.1% respectively for Group 1 and Group 2 banks.

The different sizes of the shortfall in the Advanced IRB and Foundation IRB Approaches for Group 1 banks can be mainly explained by the different sample of banks under the two approaches as well as by the differences in the LGDs that were used for both approaches.



32/65

Table 14

Regulatory calculation difference [%]

Foundation IRB Advanced IRB

Group 1 Group 2 Group 1 Group 2

Average 2.8 1.6 2.4 1.1

In some countries the shortfall contributes significantly to the overall result; conversely in other countries, the excess of provisions over the EL reduces the overall minimum required capital.

Compared to QIS 3 and to previous analyses of the coverage of the Expected Losses by provisions there seems to be a trend that many banks have changed from a “shortfall-bank” into an “excess-bank” during the last few years or – at least – have reduced the shortfall. Even if the average values of the RCD are positive – i.e. Expected Losses are not 100% covered by provisions – a higher level of coverage has been achieved in QIS 5. This observation is in line with the discussion of the current position of many countries in the economic cycle – macroeconomic and credit conditions were more favourable in the participating countries at the time of QIS 5 – and the related discussion of the PDs which seem to be lower on average than in QIS 3.

4.7 Equity As in previous exercises, the IRB equity rules seem to lead to a clear increase in MRC, independent of the method chosen by the bank. However, the impact could largely be hidden at the aggregate level by the grandfathering option (i.e. 10 years exemption under the IRB calculation for all or some equity exposures in the portfolio at the implementation date) which is an explicit national discretion granted by the CRD. At the time the QIS 5 exercise was conducted not all the EU countries had already decided if and to what extent they want to make use of the grandfathering clause. Therefore, the average data presented below would need to be revised if more countries decided to apply the grandfathering option.

In the countries where the option will be used on a large scale, including almost all equity exposures, only small increases in MRC can be observed.

For the equity exposures not subject to a grandfathering clause and being material (higher than 10% of the sum of Original own funds and Additional own funds), most banks (30 in total) have used the Market Based Approach. The PD/LGD Approach has been adopted by 13 banks but has often been applied only to a small part of the equity portfolio.

For the banks using the Market Based Approach (MBA), the change in the MRC varies from around 120% to 270%. This is consistent with the risk-weight of the Simple Risk-Weight Approach and with the fact that only a few banks have applied the Internal Model Approach (in those cases, the increase of the MRC is less relevant, hence the incentive structure seems to work properly).

Under the PD/LGD Approach, the change in MRC ranges from 100% to 175% and is therefore lower than for banks applying the MBA Approach. However, this result is not robust due to the small size of the sample.

33/65

5. Securitisation

Due to the complexity of securitisation products and the rapid changes in the strategies of institutions the data quality was not equally good for all banks participating in the QIS 5 exercise. However, in comparison to previous studies (QIS 3 and 4) banks made significant progress with respect to the coverage of the transactions and the reliability of reported data.

Even if a certain trend is observable some countries seem to depart from it. This effect could be traced back to differences in the banks' strategies but also to different current regulations. Therefore it has to be kept in mind that the results only show an average and results on a single country basis could diverge significantly.

The incentive structure for Group 1 banks appears to work properly, with the overall change in MRC being much more important for banks using the Standardised Approach than for the ones using the IRB Approach (+21.2% in the Standardised Approach versus +7.9% in the IRB Approach). The same holds true for Group 2 banks (-3.9% versus -14.6%).

The study also shows that the MRC change relative to the current regime for Group 1 banks will be different compared to the impact on Group 2 banks. This could be attributed to different activities in the securitisation market that the groups are undertaking. Group 2 banks are predominantly investors in high-quality (investment grade rated) securitisation positions. These positions will receive a favourable treatment (lower risk-weights) under the new capital requirements which explains the decrease in the overall capital requirements for Group 2 banks under the Standardised as well as under the IRB Approach. In accordance with the activities described above the main contribution to the overall decrease for Group 2 banks is coming from the rated RWA (-49.6% for the Standardised and -52.3% for the IRB Approach).

In contrast, Group 1 banks are in many cases acting as originator or sponsor of securitisation transactions, which usually means that they retain deeply subordinated positions which lead to higher capital requirements than under the current regime.

Therefore the overall increase for Group 1 banks could be attributed to two main issues. As mentioned above Group 1 banks tend to hold more risky positions (including many first loss tranches) which will be deducted more rigidly (see in table 16 the increase in positions to be deducted of 17.5% and 13.7%). The second important driver of the increase in capital charges between the current regime and both approaches is the future treatment of liquidity facilities.

Liquidity facilities, which represent an important part of Group 1 banks’ positions, currently usually have a zero risk-weight (commitment with an original maturity of less than one year). This current practice does not adequately reflect the economic risk the provider of such a facility is bearing. Therefore the risk insensitive current practice will be replaced by a more

Table 15

% Change in minimum required capital for the equity portfolio

Group 1 Group 2

Std FIRB AIRB Std FIRB AIRB

Average 18.3 88.8 67.3 1.8 124.5 175.4

34/65

appropriate treatment resulting in an overall capital increase under both the Standardised and the IRB Approaches.

Table 16

Change in minimum required capital for the securitisation portfolio [%]


IRB Approach

Group 1

Total RWA 3.7 -5.8 thereof

Rated RWA -17.1 -10.1 Unrated RWA 23.4 43.4 Inv. interest early amortisation 0.0 0.0 Correction for Cap -2.6 -29.4 Correction for provisions N/A -9.7 Positions to be deducted 17.5 13.7 Overall change in MRC versus Current 21.2 7.9 Group 2 Total RWA -67.5 -60.3 thereof

Rated RWA -49.6 -52.3 Unrated RWA -16.9 -7.4 Inv. interest early amortisation 0.0 0.0 Correction for Cap -1.0 -0.6 Correction for provisions N/A 0.0 Positions to be deducted 63.6 45.8 Overall change in MRC versus Current -3.9 -14.6

This table only includes banks for which complete QIS 5 workbooks were available.

The decline or lower increase in the capital requirements under the IRB Approach, in comparison with the Standardised Approach can mainly be attributed to the introduction of the Internal Assessment Approach (IAA), which is mainly used in the context of the above mentioned liquidity facilities. Under the Standardised Approach the results are much more conservative particularly if these facilities did not qualify as eligible ones. Even if this is still a new approach in the early stage of development and banks have faced some difficulties in applying it properly and comprehensively, the scope of application and the quality of the results increased in comparison to earlier studies.

However, even if the overall result for Group 1 banks shows an increase for the Standardised and for the IRB Approaches there are some countries experiencing a significant reduction. Overall, the average change in MRC from the securitisation portfolio for Group 1 banks

35/65

varies between -42% and +60% for the IRB Approach. This seems to be the result of different types of positions and differences in current national regulations (some countries already apply a strict treatment under their current national regulation).

One aspect that has to be highlighted is the possibility to apply a cap also under the Standardised Approach (Basel exclude the Standardised Approach from the cap). This will tend to reduce the overall capital requirement. However spot tests showed that not all banks used this possibility (e.g. some banks did not apply the cap because they did not want or were technically not able to calculate the capital requirement before securitisation). The possibility of applying the cap also under the Standardised Approach reduced the change in MRC by 2.6% for Group 1 and 1.0% for Group 2 banks.

6. Impact of the trading book review and the double default treatment

This section discusses the capital impact of the new rules introduced in the final version of the CRD on trading activities and double default (“Trading Book Review”) on the capital requirements for credit risk.

6.1 Double Default The number of banks which were able to submit information on double default is very limited. Thus, any figures derived from the limited data on the potential capital relief due to the lower risk-weights cannot be considered as being statistically significant. Only nine Group 1 banks from four countries applied the DD risk-weight curve to some parts of their exposures.

A majority of banks is likely to benefit from double default treatment. Implementing this treatment is mainly an IT, rather than a modelling issue, and due to other priorities only a very limited number of banks was able to submit data on the effect of double default treatment. However, it is evident that the exposures eligible for this treatment are likely to show a substantial decrease in capital requirements - as increases are ruled out - since banks have the freedom to apply the existing substitution treatment in such cases. The amount of exposure eligible for double default treatment, however, is limited and consequently CEBS does not consider the lack of data on this issue a material shortcoming of the QIS results.

The average capital charge for the exposures treated according to the double default rules is very low (less than 4% in any situation), which is quite natural since the eligibility criteria lead to the application of this treatment only to the obligor and guarantor with the best PDs. The average RW is low in comparison to the RW of other corporate exposures, but this gap may be explained by the fact that all the exposures considered here are by definition unsecured (no guarantees at least, since the exposures to which the double default treatment or the substitution approach are applied are treated separately). Furthermore, this very low average RW is largely compensated by the fact that this treatment only applies to a very small part of the corporate portfolio (less than 1%). Hence, the impact of the double default treatment on the corporate portfolio is minor (the reduction in the risk-weighted exposures of the corporate portfolio ranges from 0.1% to 0.3% on average).

At this stage, it is difficult to conclude whether this is only due to the limited scope of application of the double default treatment, or if this is the consequence of the inability of the banks to apply this treatment consistently on a large scale. Nevertheless, the individual data available show that no bank has included more than 6% of its corporate portfolio in the

36/65

double default treatment. This seems to validate the first option (i.e. only a very small portion of corporate exposures may indeed benefit from this treatment).

Summarizing this subsection CEBS cannot provide definite answers to the question how many banks can be expected to use the double default option in the future. Preliminary figures provide some indication that there should be sufficient incentives for banks to apply this treatment, however.

6.2 Counterparty credit risk The changes that the Trading Book Review made to counterparty risk treatment are quite substantial as more sophisticated methods (compared to the Mark-to-Market Method) are available to determine the counterparty credit risk exposure (exposure value). QIS 5 represents the first opportunity to make a preliminary assessment of the actual impact of the new rules on the banks’ balance-sheets. According to Article 78 and Annex III and IV of the CRD (2000/12/EC) there are several methods available to calculate the Counterparty credit risk exposure value, some of which only apply to OTC derivatives.

Table 17

Number of banks using the various methods for measuring counterparty credit risk

ethod used under CRD Group 1 Group 2

OTC derivatives (Annex III methods)

Mark-to-Market Method12 45 47

Standardised Method 5 7

EPE 11 0

Repo-style transactions (Annex IV methods except for EPE)

Simple method 3 10

Standardised haircuts 33 12

Value at risk 3 0

EPE 2 0

6.2.1 OTC derivatives In QIS 5 most of the participating banks have calculated the capital requirement for OTC derivatives using the Mark-to-Market Method (replacement cost + add-on). Only a small number of banks have made use of the Standardised Method. Due to the small size of the

12 No bank provided information on the Original Exposure Method which is applied by some banks to OTC derivatives under the current regime.

37/65

sample, and also the fact that the banks that made use of this approach only applied it to a small part of their OTC derivatives exposures, the data here are not significant and no conclusions could be drawn.

A slightly larger group of banks (11) have provided data on the EPE approach. This approach is primarily used to measure the exposure value of OTC derivatives (only two banks have also been able to apply this approach also to their repo-style transactions). Furthermore, no bank has used its own estimate for the alpha parameter or provided data indicating that they were planning to do so in the near future. All the EPE results are thus based on the pre-defined regulatory alpha value of 1.4.

For 7 of the 11 banks, the use of the EPE approach results in a decrease in the exposure value, of up to 80%. On average, the reduction in the exposure value for the 11 banks is around 9.3%. Hence the incentive structure is working properly and will not be excessive when banks will eventually use their own estimate of the alpha parameter.

As mentioned above, the lack of data makes it difficult to evaluate whether the use of the Standardised Approach for measuring counterparty credit risk exposures yields lower exposure values and RWAs. The only bank in the QIS 5 sample which provided data of sufficient quality for the Standardised Approach achieved a reduction of 27% for the respective share of OTC derivative exposures.13 Therefore a reliable conclusion whether or not the beta parameter is consistent with the desired incentive structure is not possible.

6.2.2 Repo-style transactions A very limited number of banks have calculated the capital requirements for repo-style transactions with the simple method. Therefore the results are probably not representative.

The comprehensive approach was the most widely used approach for both Group 1 and Group 2 banks (more than 80% of the exposures on average). This results in a strong increase in the exposure value and thus in the corresponding RWA. However, this increase, in terms of percentage change, may be explained by the relatively low exposure value of repo-exposures in the current regime, and the impact of this increase, at the bank level, remains rather small in terms of its contribution to the changes in the MRC. At the individual bank level, the situation is very heterogeneous, with banks having a significant decrease (up to 90%), while others are presenting a huge increase. This result certainly reflects the diverse nature of the portfolios of the banks, with some banks engaged mostly in repo-styles transactions that may benefit from zero haircuts, while others with a more risky portfolio must apply haircuts (to both the collateral and the exposure). This situation is also apparently the result of the fact that several banks have had difficulties in correctly incorporating and treating collateral. Hence, the large overall increase is probably significantly overestimated. In terms of the incentive structure compared with the simple approach, the increase is not problematic for the IRB banks, since the simple approach is not open to them.

The two internal approaches have been adopted in QIS 5 by a very limited number of Group 1 banks. The data are thus too limited to draw any final conclusion, but the first impression is that using VaR might produce a significant decrease of the exposure value (more than 90% for the 3 banks). For the two banks that have used the EPE approach one shows a relatively important decrease of its exposure value (-57%), whereas the other shows a strong increase (81%).

13 This bank applied the Standardised Approach to only 1.4% of its entire OTC derivative exposures.

38/65

6.2.3 Capital charge for Default and Event Risk The CRD has not introduced any changes in the rules for calculating capital requirements for market risk in the trading book apart from the rules for specific market risk. For example, institutions were allowed to replace the 4xVar multiplier for the specific risk surcharge by a 3xVar figure, if they were able to model default and event risk. In this case banks were asked to report the incremental capital charge for default risk.

Only 3 banks have calculated an incremental capital charge for default risk. A sub-sample of 3 banks can certainly not be considered as representative and their results should therefore be interpreted carefully. However, the results of the 3 banks show that the impact from this change on the overall result can be quite important as contributions of up to 5% were obtained. If more banks become able to model default and event risk in the future the incremental capital charge could potentially influence the aggregate figures and could increase them materially.

7. Changes in Minimum required capital for Operational Risk

Table 18 shows the number of EU banks in QIS 5 which have calculated the capital requirements for operational risk according to the different approaches available in the CRD.

As expected, Group 1 banks have filled in the figures for the more advanced approaches, i.e. the Standardised Approach (TSA) and the Advanced Measurement Approach (AMA) whereas Group 2 banks have mostly adopted the Basic Indicator Approach (BIA) and, to a lesser extent, the Standardised Approach; only two Group 2 banks have used AMA models.

Table 18

Number of banks providing data for operational risk approaches

Approach Number of Group 1 banks

Number of Group 2 banks

Basic Indicator Approach 0 138 Standardised Approach 30 72 Advanced Measurement Approach 19 2

Table 19 presents the relative contribution of operational risk to the overall change of MRC.14 It is remarkable that all the figures fall below 10% which was the original target value for operational risk.

14 The results of this table are based on all available information the banks provided. If a bank – for example – provided results

for the BIA and SA, the results of both approaches were considered in the respective columns.

39/65

Table 19

Contribution from the Operational Risk capital charge by approach

Approach Contribution Group 1

Contribution Group 2

Basic Indicator Approach N/A 8.9 Standardised Approach 5.5 7.9 Advanced Measurement Approach 5.9 5.4

Scatterplots related to table 19 can be found in Annex C (charts 32 and 33).

In order to evaluate the incentive structure between the different approaches for operational risk table 20 describes the capital relief which would be obtained on average if a bank decided to implement the TSA instead of the BIA (same for AMA instead of TSA).15 As the table shows, the average risk-weighed assets for operational risk decrease if banks adopt a more advanced method. Thus, for both groups of banks there seems to be an incentive to move to more advanced approaches (from BIA to TSA and from TSA to AMA).

Table 20

Incentive structure Capital Charge for Standardised Approach compared to Basic

Indicator Approach, and Advanced Measurement Approach compared to Standardised Approach [%]

SA/BIA AMA/SA

Group 1 -5.6 -17.0 Group 2 -12.3 N/A*

(*) The sample of Group 2 banks using the AMA is too small, hence no comparison is possible.

The difference between the approaches is particularly significant for Group 2 banks aiming to move from the simplest approach to the Standardised. This might be explained by the large size of the retail activity, to which a 12% beta is applied in the Standardised Approach (as opposed to the 15% alpha in the BIA).

For the TSA and the AMA, the incentive structure seems to be largely fulfilled for Group 1 banks. However, since there is a considerable variation in the results between different banks (see also the charts 35 in Annex C), the AMA figures should be interpreted carefully for the following reasons:

i) some national supervisors report that banks in their national sample are not yet at a stage were they can submit fully reliable AMA data although they intend to implement AMA in due

15 Only those banks are included which provided data both for the BIA and the TSA or the TSA and the AMA.

40/65

course. Expectations are that once these banks have actually implemented AMA, it will improve their MRC results;

ii) banks are at different stages of their AMA implementation projects, partly given the different availability of sufficient event data. It is assumed that these differences in implementation stage account for the larger part of the dispersion in the results; and

iii) another factor that might drive the operational risk impact in the near future is the outcome of current discussions regarding the extent to which banks may take into account diversification effects in their AMA models.

41/65

8. Impact of the CRD on Investment Firms in the EU

The CRD applies both to credit institutions and investment firms. In most countries that participated in the QIS 5 exercise there have not been separate exercises for credit institutions and investment firms, an approach that also holds true for the aggregated EU results presented in this report. This does not mean that investment firms in general are expected to play only a minor role. Some investment firms were included in the data set as part of the figures provided by banking groups, where they belong to such groups. The only country that has conducted a separate analysis of investment firms is the UK. The results of this preliminary study are presented below. Although these results should not be seen as representative for the rest of the participating countries, they give a first indication of how the investment sector might be affected by the CRD. The CRD will apply to some 1300 investment firms in the UK – about 600 securities and futures firms and around 700 investment management firms. The FSA decided to undertake a much more simplified exercise compared to the QIS 5 exercise in order to assess the likely impact of the CRD on UK investment firms and to inform the cost benefit analysis for the FSA’s formal, public consultation on the implementations of the CRD. Only 17 of the 50 representative firms approached submitted data of reliable quality. There are several reasons for this:

- despite progress in investment firms’ CRD awareness and planning over the past year, the vast majority remain relatively unprepared for CRD implementation and are expected to remain on the current rules until 1 January 2008;

- capital may have a significantly smaller role in investment firms than in credit institutions – instead Pillar 2 and conduct of business requirements are the primary means by which firms and regulators ensure that the business of these firms is soundly managed.

The business, size and customer profile of UK investment firms varies considerably. The sample of 6 full-scope (€730k) firms and 11 limited licence (€125k or €50k) firms can therefore be described as only barely representative. The tentative conclusions of the survey are as follows:

• capital requirements for investment firms as a group are estimated to decline by 8% on average under the CRD;

• this outturn is driven by the relatively high proportion of limited licence firms in the sample – the switch from Base Capital + CR + MR + EBR16 to the ‘higher-of’ Base Capital, CR + MR or FOR17 rule dictates that less capital will typically be required for these firms; and

• capital requirements for full-scope investment firms are estimated, on the basis of data from a very small sample, to increase, depending on various assumptions, by between 16% and 27%.

16 Expenditure based requirement 17 Fixed overheads requirement

42/65

Annex A: Methodology for the aggregation of individual QIS 5 bank data

1 Introduction

The purpose of this annex is to outline the calculation of minimum required capital (MRC) at the level of individual portfolios and to describe the weighting scheme for the aggregation of QIS 5 data that is applied for two types of analysis:

1. when considering the aggregation of changes in minimum required capital at a portfolio level across banks for a particular portfolio. “Portfolio” in this document refers to the different exposure classes set out in the CRD as well as to several other items with an impact on overall minimum required capital, such as capital requirements for market and operational risk and for the requirements arising from the partial use of the Standardised Approach under the IRB Approaches; and

2. when considering the aggregation of changes in minimum required capital for the whole banking system of individual countries. This includes in particular an assessment of how changes in the individual portfolios contribute to the total change for the banking system.

The formulae which are set out in this annex were the basis for all portfolio and bank-specific results on changes in minimum required capital in this report. The methodology for bank-specific results is completely in line with the cross-country analysis carried out by the OC/QIS Working Group on the G10 and non-G10 level. The numbers in this report were extracted from an SQL database maintained in Frankfurt in which all QIS data of EU countries are stored. The analysis team of the Deutsche Bundesbank provided an Excel-based tool (QIS-Portfolioanalysis.xls) for supervisors to reproduce the numbers for the banks from their own countries.

The annex is structured as follows: First the necessary notation is provided in Section 2. Then the formulae on the changes in minimum required capital and contribution on a portfolio level are set out (Section 3). An additional subsection of Section 3 describes how partial use exposures are treated in the analysis (3.2.1). Subsection 3.2.2 is devoted to the different risk categories of the trading book. Section 4 describes how aggregation across banks is carried out.

2 Notation

Throughout this annex we rely on the following notation:

RWAi, RWAiCurr Risk-weighted assets (Risk-weighted exposures in the terminology

of the CRD) of bank i according to the CRD and the current regime

cRWAi, cRWAiCurr Credit risk-weighted assets of bank i (i.e. risk-weighted assets

without market risk and operational risk)

RWAiPF , RWAi

PF,Curr Risk-weighted assets according to the CRD for a specific portfolio PF of bank i

43/65

RWAiPU , RWAi

PU,Curr Risk-weighted assets of bank i treated under the partial use of the Standardised Approach and the risk-weighted assets for these exposures under the current regime

ELi, ELiPF Aggregated expected losses for the whole bank i and for a specific

portfolio PF of bank i

Di Regulatory calculation difference, i.e. the difference between expected losses and loan loss provisions

Dedi, DediCurr Total deductions under the CRD and the current regime

respectively, including deductions for securitisation, related entities, and other supervisory deductions as described below.

DediSec, Dedi

Sec,Curr Deductions for the securitised portfolio under the CRD and the current regime, respectively

DediRelEnt, Dedi

RelEnt,Curr Deductions for the related entities under the CRD and the current regime, respectively

Dediother, Dedi

other,Curr Other supervisory deductions under the CRD and the current regime, respectively. Note that these amounts do not include the deductions for securitisation and related entities, and deductions due to a possibly negative regulatory calculation difference.

GPiincl,Curr, GPi

incl General provisions included in capital under the current regime and under the Standardised Approach of the CRD

GPiincl,PU General provisions which are eligible for Additional own funds

under the Standardised Approach partial use

MRCi, MRCiPF Minimum required capital for a bank and for the claims of a specific

portfolio PF of that bank under CRD

MRCiCurr, MRCi

PF,Curri Minimum required capital for a bank and for the claims of a specific

portfolio PF of that bank under the current regime

iMRC∆ , PFiMRC∆ Absolute change in minimum required capital for bank i and

portfolio PF for the CRD relative to the current regime, i.e. the differences Curr

i iMRC MRC− , ,PF PF Curri iMRC MRC−

% , % PFi iMRC MRC∆ ∆ Percentage change in minimum required capital for bank i and

portfolio PF for the CRD relative to the current regime, i.e. the expressions ( )Curr Curr

i i iMRC MRC MRC− ,

( ), ,PF PF Curr PF Curri i iMRC MRC MRC−

SizeiPF Size of a portfolio PF in terms of the share of minimum required

capital of the respective portfolio with respect to the current regime compared to the total minimum required capital of the bank with respect to the current regime

44/65

ContriPF Contribution of the portfolio PF to the change in the total minimum

required capital

In the aggregation across all banks of a country we use PF as an index for the portfolios and i as an index for the banks.

Throughout the rest of this annex we rely on the following formulae for the minimum required capital according to the current regime

,= 8% incl CurrCurr Curr Curri i i iMRC RWA Ded GP⋅ + − , (1)

on

{ } ,= 8% max , 0.6% incl PUi i i i i iMRC RWA D cRWA Ded GP⋅ + − ⋅ + − (2)

for the minimum required capital under the IRB Approaches of CRD, and on

= 8% incli i i iMRC RWA Ded GP⋅ + − (3)

for the minimum required capital under the Standardised Approach of CRD.

These formulae take into account that the recognition of general provisions in Additional own funds is different under the Standardised and IRB Approaches and depends on national discretion. For example, there is no recognition of general provisions in CRD under the IRB Approaches. Under the Standardised Approach general provisions are eligible in Additional own funds up to 1.25% of risk-weighted assets.

3 Calculation of the change in minimum required capital on a portfolio level

3.1 Standardised Approach The important factors for the change in minimum required capital are the risk-weighted assets both for the Standardised Approach and the current regime and the amount of general provisions which are eligible under the Standardised Approach and the current regime. We propose to allocate the general provisions to each of the specific portfolios according to the share of risk-weighted assets relative to the overall risk-weighted assets for the entire bank. Thus we obtain:

,, ,

8%% 1.

8%

PFPF inclii i

PF ii PF Curr

PF Curr incl Currii iCurr

i

RWARWA GPcRWA

MRCRWARWA GPcRWA

⋅ − ⋅

∆ = −

⋅ − ⋅

(4)

45/65

3.2 IRB Approaches Under the IRB Approaches, the switch to UL-based risk-weights in the CRD results in risk-weights which are fundamentally different from (EL+UL)-based risk-weights used prior to the Madrid compromise. In the UL-based framework, a shortfall or excess in provisioning affects the numerator of the capital ratio, i.e. the capital figure. In order to take account of the fact that both numerator and denominator of the capital ratio changed and to ensure the comparability of the data, percentage changes in minimum required capital at the level of individual banks were calculated by

{ } ,

,8% max , 0.6%

% 1.8%

incl PUi i i i i

i Curr Curr incl Curri i i

RWA D cRWA Ded GPMRC

RWA Ded GP

⋅ + − ⋅ + −∆ = −

⋅ + − (5)

3.3 Allocation of the regulatory calculation difference for the IRB Approaches The problem we are facing now is that we need to allocate this difference to the portfolios of the bank such that the change in minimum required capital at the level of the entire bank can be computed as the sum of the contributions arising from the different portfolios. This is more difficult than in previous impact studies because some of the elements, e.g. general provisions appearing in formula (5), are available only at an aggregate level. This means that a method is needed to distribute these elements in some reasonable way across the different portfolios. We propose the following formula as a definition of the percentage change in minimum required capital for a single bank portfolio:

{ }( )

,, ,

8% max , 0,6%% 1.

8%

PFPF ii i i

PF ii PF Curr

PF Curr incl Currii iCurr

i

ELRWA D cRWAEL

MRCRWARWA GPcRWA

⋅ + ⋅ − ⋅∆ = −

⋅ − ⋅

(6)

The denominator of formula (6) describes the minimum required capital at a portfolio level according to the current regime. Note that in the definition of the minimum required capital at a portfolio level a certain share of the general provisions which are currently eligible as Additional own funds are deducted from 8% of the risk-weighted assets for this particular portfolio. This share depends on the risk-weighted assets of the portfolio relative to the credit risk-weighted assets of the entire bank.

Since the regulatory calculation difference depends on the expected losses, the most natural scheme for allocating it to individual bank portfolios is by relating it to the expected losses associated with the respective portfolios. Note that the contributions to the change in minimum required capital that result from the change in other supervisory deductions have not been incorporated at the portfolio level and will be reintroduced for analysis only at the level of the whole bank.

The portfolios equity, securitisation and related entities are not considered in the calculation of the regulatory difference Di. In addition, deductions for securitisation and related entities have to be considered in the formulae. Thus, we have to define the minimum required capital of these portfolios separately, eg for securitisation and related entities by

{ }8% , , ,PF PF PFi i iMRC RWA Ded PF Sec RelEnt= ⋅ + ∈

and therefore a modification of formula (6) is necessary for these cases:

46/65

,,

8%% 1

8%

Equity EquityEquity ii Equity Curr

Equity Curr inclii iCurr

i

RWA ELMRC

RWARWA GPcRWA

⋅ +∆ = −

⋅ − ⋅

(7)

,, , ,

8%% 1

8%

Sec SecSec i ii Sec Curr

Sec Curr Sec Curr incl Currii i iCurr

i

RWA DedMRC

RWARWA Ded GPcRWA

⋅ +∆ = −

⋅ + − ⋅

(8)

, ,8%

% 18%

RelEnt RelEntRelEnt i ii RelEnt Curr RelEnt Curr

i i

RWA DedMRC

RWA Ded⋅ +

∆ = −⋅ +

(9)

3.4 Portfolio size and contribution We have to weight portfolios by their size in order to get the overall change in the minimum required capital of the bank from the changes of its portfolios. The size of the portfolio PF is expressed in terms of current minimum required capital:

,

.PF Curr

PF ii Curr

i

MRCSize

MRC= (10)

Accounting for the size of the portfolio the contribution of the change in required minimum capital of the portfolio PF is defined as follows

%PF PF PFi i iContr MRC Size= ∆ ⋅ . (11)

Furthermore, we have to incorporate the contribution of the capital charge for operational risk. Since there is no capital charge for operational risk under the current regime, formula (2) cannot be applied to operational risk. However, since the contribution can be understood as the absolute change in minimum required capital compared to minimum required capital under the current regime, we define

8% OpRisk

OpRisk ii Curr

i

RWAContr

MRC

⋅= .

Now, a simple calculation shows that

,

%Other Other Curr

PF OpRisk i ii i iCurr

PF i

Ded DedContr Contr MRCMRC−

+ + = ∆∑ . (12)

One advantage of this approach is its similarity to the approach taken in the analysis of QIS 3. Based on formulas (5), (6), (7) and (8) we can easily identify the portfolio of a bank that has the greatest influence on the overall result. In the QIS 5 exercise PF is an index running over the following portfolios: corporate, specialised lending, sovereign, interbank, residential mortgage, other retail, QRRE, purchased receivables, related entities, securitisation and the different risk categories of the trading book.

The implementation of the formulae described above is even more complex in the databases and in the portfolio analysis tool. In the Excel workbooks all risk-weighted asset amounts are scaled up by the factor 1/coverage, where coverage describes the amount of risk-weighted

47/65

assets under the current regime a bank was able to include in the QIS survey divided by the total amount of risk-weighted assets under the current regime. Additional care is needed as in the allocation of the regulatory calculation difference the necessary information on the expected losses at a portfolio level needs to be scaled up by 1/coverage. The issue of scaling up is captured in more detail in Section 3.2.4 where exposure categories related to the trading book are considered.

3.5 Treatment of partial use exposures With the portfolio analysis we want to make sure that the overall change in minimum required capital at the bank level is obtained as the sum of the contributions of the specific portfolios. Therefore we also need a formula which captures the partial use of the Standardised Approach for banks adopting an IRB Approach. Clearly the risk-weighted assets arising from the rules of the Standardised Approach compared to those arising from the current regime need to be taken into account. Secondly, we need to consider that a certain sub-amount of the general provisions which were eligible elements of Additional own funds under the current regime can be still recognised as eligible Additional own funds under the CRD. Consequently, we obtain

,

,, ,

8%% 1.

8%

PU incl PUiPU i

i PU CurrPU Curr incl Currii iCurr

i

RWA GPMRC

RWARWA GPcRWA

⋅ −∆ = −

⋅ − ⋅

(13)

This number has to be multiplied by ,PU Curr

Curri

MRCMRC

in order to compute the contribution of the

partial use of the Standardised Approach to the overall bank change in minimum required capital.

3.6 Treatment of trading book and market risk From a computational point of view trading book exposures receive a different treatment in the QIS 5 workbooks and in the Portfolio Analysis Tool to banking book exposures. The reason is not only that these positions are excluded from the allocation mechanism of the regulatory calculation difference but also that the “scaling-up mechanism”18 and the scaling factor of 1.06 do not apply. The following table summarises the treatment of the categories counterparty credit risk, specific risk, general market risk, and large exposures. Note that the scaling mechanisms mentioned in columns 2 and 3 of the following table apply to the risk-weighted assets and should not be understood as factors to be applied to the exposure.

18 Here, “scaling-up mechanism” refers to the procedure of multiplying risk-weighted assets by the factor 1/coverage, as the

banks were allowed to consider the CRD risk-weights only for a certain subset of exposures. Coverage describes the percentage share of exposures which were included in the QIS 5 data collection exercise. According to the QIS instruction the banks had to achieve at least 80% coverage relative to their world-wide exposures.

48/65

Scaling mechanism to capture the coverage

Scaling Factor of 1.06 (under IRB)

Counterparty credit risk Scaling Multiplier

Specific risk No scaling No multiplier

General market risk + settlement risk + Annex V of Directive 93/6/EEC

No scaling No multiplier

Large exposures No scaling No multiplier

The third row of the table includes the trading book exposures which are treated according to IRB rules (Annex V of Directive 93/6/EEC).

4 Aggregation of results across banks

It is agreed that drivers of the overall results on portfolios should be identified and analysed. The question arises as to how the results of the individual banks should be weighted. Applying weights based on capital numbers may have the undesirable effect that a non-zero weight is applied to an empty portfolio.

Any proposed weighting scheme should satisfy the following two requirements:

1. equation (11) should hold for the same data aggregated over all banks, i.e. the product of the average size of a certain portfolio times the average change in minimum required capital should be equal to the average contribution of that particular portfolio to the change in minimum required capital for all Group 1 or Group 2 banks in the whole banking system; and

2. The contributions of all the portfolios of the banks plus the contributions arising from general provisions, deductions and operational risk should sum to the total change in minimum required capital (see equation (12)) also in the aggregated results.

The second condition is obviously satisfied by any arbitrary weighting specification, whereas the first requires more attention because the product of averages is not equal to the average of the product. (referring to the product of size and change in minimum required capital ).

We have the choice of applying the bank-weights to either % PFiMRC∆ or PF

iContr . Applying

them to % PFiMRC∆ is definitely not an appropriate definition because this might result in

misleading averages in cases where selected banks do not have exposures in certain portfolios. In such cases we would apply non-zero weights to empty portfolios. The result is misleading because one could not know if the change in minimum required capital is due to the fact that the rules changed or because a bank has no exposure in this portfolio. This effect is avoided if the weights are applied to the contributions, because the contributions include the size of the portfolio as a factor. Consequently, empty or small portfolios do not have a big effect on the aggregate number.

Let us summarise this preliminary discussion:

49/65

4.1 Group 1 banks In order to aggregate the results on a portfolio level as well as to obtain the total change in minimum required capital across the banks we suggest applying the following weight to each bank:

Curri

i Curri

i

MRCwMRC

=∑

, (14)

where the index i runs over all Group 1 banks. Applying these weights we would obtain

PF PFavg i i

iContr w Contr= ⋅∑ (15)

as an average contribution to the change in minimum required capital for the portfolios and for operational risk.

As described above applying the weights wi to determine the average % PFavgMRC∆

analogously would be misleading because of the empty or small portfolios of certain banks. However, there is another possibility for determining % PF

avgMRC∆ . Taking into account that PFiContr was defined as a product which includes the size of the portfolio as a factor (see

formula (12)), expression (15) can be reinterpreted in the following way - the right side of (11) could be understood as a weighting of the % PF

iMRC∆ , where the index i runs over all Group 1 or Group 2 banks of a certain country. The weights are given by ( )PF PF

i i i ii

w Size w Size⋅ ⋅∑ .19 Consequently we obtain:

%

%PF PF

i i iPF iavg PF

i ii

w Size MRCMRC

w Size⋅ ⋅ ∆

∆ =⋅

∑∑

. (16)

For analysis purposes (16) will be very useful to analyse effects on portfolio level. However, it should be noticed that the weights used in (16) are different for each portfolio.

4.2 Group 2 banks Consistent aggregation across banks requires a weighting scheme which reflects the differences in the relative contributions of individual banks to the changes in minimum required capital for the whole banking system. In choosing a weighting scheme, it is necessary to distinguish between Group 1 and Group 2 banks. The reason is that the majority of Group 1 banks participate in the QIS exercise and, therefore, the data collected for these banks is broadly representative for the Group 1 banks of an entire country. This is not the case for Group 2 banks. Since relatively few of the smaller Group 2 banks are participating in QIS 5, they would be underrepresented in the analysis if the same weighting scheme is applied as for Group 1 banks. Therefore, a different weighting scheme should be used for Group 2 banks:

19 We have to include the expression PF

i iiw Size⋅∑ in the denominator of each weight to make sure that the weights sum up

to 1.

50/65

wi = 1 / ( number of Group 2 banks )

The same formulae as in Section 4.1 can be used for the averages of the changes in total minimum required capital, the contributions at a the portfolio level and the changes in minimum required capital on a portfolio level for Group 2 banks. The even weighting of banks without regard to their risk-weighted assets could be considered to be disadvantageous, however it should be reiterated that the sample of Group 2 banks will not be representative for all Group 2 banks in the banking system. This becomes clear when one considers that a QIS 5 Group 2 bank with €10 million in Original own funds may represent a large number of banks of similar size not participating in the study. QIS Group 2 banks with €1 billion Original own funds are representative of significantly fewer Group 2 banks in the total banking system. Applying the same weighting scheme as proposed for the Group 1 banks would have the undesirable effect that data coming from small banks would have little impact on the overall results.

Example: Let us consider the residential mortgage portfolios of three Group 2 banks, or for three Group 1 banks having the same minimum required capital relative to the current regime, i.e. the same weights.

Bank 1 Bank 2 Bank 3

Re . .% s MortgiMRC∆ -40% -60% no exposures

Re . .s MortgiSize 20% 40% 0%

Re . .s MortgiContr -8% –24% 0%

Weights 1/3 1/3 1/3

The contribution to the overall change would be 1/3 (-8% – 24%) = -10.7%, which should be interpreted as follows: If the minimum required capital for all the other portfolios of this bank remains unchanged, the rules of the CRD would, on average, result in a change in minimum required capital of -10.7%. Now we compute Re . .% s Mortg

avgMRC∆ . Note that the simple average would be 1/3 (-40% – 60%) = -33.3%, which is misleading since Bank 3 has no residential mortgage exposures and the average should lie somewhere between -40% and -60%, which are the values of those two banks having residential mortgage exposures. In contrast, formula (11) yields a plausible result. We have

( )( ) ( )( ) ( )

( )1/ 3 20% 40% 1/ 3 40% 60% 1/ 3 0%

53%1/ 3 20% 40% 0%

⋅ ⋅ − + ⋅ ⋅ − + ⋅= −

⋅ + +

according to (15). The -53% is closer to the -60% of Bank 2 than to the -40% of Bank 1 because the residential mortgage portfolio is more important for Bank 2 than Bank 1 in terms of size.

51/65

Annex B: Risk Parameters

1. PDs20

Charts 9

Average PDs for the Corporate portfolio, banks by size

Average PDs by size - Group 1

0.0%

0.5%

1.0%

1.5%

2.0%

2.5%

3.0%

3.5%

4.0%

4.5%


0.0%

0.5%

1.0%

1.5%

2.0%

2.5%

3.0%

3.5%

4.0%

4.5%

20 The scattergrams for the risk parameter PD do not require a distinction between Foundation and Advanced IRB

Approaches.


52/65

Charts 10

Average PDs for the Sovereign portfolio, banks by size


0.0%

0.1%

0.2%

0.3%

0.4%

0.5%


0.0%

0.1%

0.2%

0.3%

0.4%

0.5%

Charts 11

Average PDs for the Bank portfolio, banks by size


0.0%

0.1%

0.2%

0.3%

0.4%

0.5%

0.6%

0.7%


0.0%

0.1%

0.2%

0.3%

0.4%

0.5%

0.6%

0.7%



53/65

Charts 12

Average PDs for the SME Corporate portfolio


0%

1%

2%

3%

4%

5%

6%

7%

8%

9%

10%


0%

1%

2%

3%

4%

5%

6%

7%

8%

9%

10%

2. LGDs

2.1 LGD distribution for main portfolios

Chart 13

LGD distribution of the Corporate portfolio

0

5

10

15

20

25

30

35

40

(0;10) (10;20) (20;30) (30;40) (40;50) (50;60) (60;70) (70;80) (80;90) (90;100)

Sha

re (i

n %

)

Group 1 Group 2


54/65

Chart 14

LGD distribution of the Bank portfolio

0

5

10

15

20

25

30

35

40

45

50

(0;10) (10;20) (20;30) (30;40) (40;50) (50;60) (60;70) (70;80) (80;90) (90;100)

Shar

e (in

%)

Group 1 Group 2

Chart 15

LGD distribution of the Sovereign portfolio

0

10

20

30

40

50

60

70

80

(0;10) (10;20) (20;30) (30;40) (40;50) (50;60) (60;70) (70;80) (80;90) (90;100)

Shar

e (in

%)

Group 1 Group 2

55/65

Chart 16

LGD distribution of the SME Corporate portfolio

0

5

10

15

20

25

30

35

40

45

(0;10) (10;20) (20;30) (30;40) (40;50) (50;60) (60;70) (70;80) (80;90) (90;100)

Sha

re (i

n %

)

Group 1 Group 2

Chart 17

LGD distribution of the Other Retail portfolio

0

2

4

6

8

10

12

14

16

18

20

(0;10) (10;20) (20;30) (30;40) (40;50) (50;60) (60;70) (70;80) (80;90) (90;100)

Shar

e (in

%)

Group 1 Group 2

56/65

Chart 18

LGD distribution of the QRE portfolio

0

5

10

15

20

25

30

35

40

45

(0;10) (10;20) (20;30) (30;40) (40;50) (50;60) (60;70) (70;80) (80;90) (90;100)

Shar

e (in

%)

Group 1 Group 2

Chart 19

LGD distribution of the Retail Mortgage portfolio

0

10

20

30

40

50

60

(0;10) (10;20) (20;30) (30;40) (40;50) (50;60) (60;70) (70;80) (80;90) (90;100)

Sha

re (i

n %

)

Group 1 Group 2

57/65

Chart 20

LGD distribution of the SME Retail portfolio

0

5

10

15

20

25

30

35

40

(0;10) (10;20) (20;30) (30;40) (40;50) (50;60) (60;70) (70;80) (80;90) (90;100)

Shar

e (in

%)

Group 1 Group 2

Chart 21

LGD distribution of the SL HVCRE portfolio

0

10

20

30

40

50

60

(0;10) (10;20) (20;30) (30;40) (40;50) (50;60) (60;70) (70;80) (80;90) (90;100)

Shar

e (in

%)

Group 1 Group 2

58/65

Chart 22

LGD distribution of the SL Other portfolio

0

5

10

15

20

25

30

35

40

45

(0;10) (10;20) (20;30) (30;40) (40;50) (50;60) (60;70) (70;80) (80;90) (90;100)

Sha

re (i

n %

)

Group 1 Group 2

Chart 23

LGD distribution for counterparty credit risk exposures in the Trading Book

0

10

20

30

40

50

60

70

80

(0;10) (10;20) (20;30) (30;40) (40;50) (50;60) (60;70) (70;80) (80;90) (90;100)

Shar

e (in

%)

Group 1 Group 2

59/65

2.2 Variation in average LGDs across countries

Charts 24

Average LGDs for the Corporate portfolio

Average LGDs by size - Group 1

0%

10%

20%

30%

40%

50%

60%


0%

10%

20%

30%

40%

50%

60%

Charts 25

Average LGDs for the Sovereign portfolio


0%

10%

20%

30%

40%

50%

60%

70%


0%

10%

20%

30%

40%

50%

60%

70%



60/65

Charts 26

Average LGDs for the Bank portfolio


0%

10%

20%

30%

40%

50%

60%

70%

80%


0%

10%

20%

30%

40%

50%

60%

70%

80%

Charts 27

Average LGDs for the SME Corporate portfolio


0%

10%

20%

30%

40%

50%

60%


0%

10%

20%

30%

40%

50%

60%



61/65

Charts 28

Average LGDs for the Other Retail portfolio


0%

10%

20%

30%

40%

50%

60%

70%

80%

90%


0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

Charts 29

Average LGDs for the Retail Mortgage portfolio


0%

10%

20%

30%

40%

50%

60%


0%

10%

20%

30%

40%

50%

60%



62/65

Charts 30

Average LGDs for the Retail QRE portfolio


0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%


0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Charts 31

Average LGDs for the SME Retail portfolio


0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

Average LGDs by size- Group 2

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%



63/65

Annex C: Operational risk

Chart 32 shows the contribution of the Standardised Approach (TSA) and the Advanced Measurement Approach (AMA) to the overall change in MRC for Group 1 banks. The Basic Indicator Approach (BIA) was not used by any Group 1 bank. For both approaches, most of the banks show contributions of between 3% and 8%.

Chart 32

Contribution of the Operational Risk Capital Charge for banks adopting Standardised Approach and Advanced Measurement Approach (Group 1 banks), [%]

Contribution TSA by size

0%

2%

4%

6%

8%

10%

12%

Contribution AMA by size

0%

2%

4%

6%

8%

10%

12%

The results of Group 2 banks are included in chart 33. Only three Group 2 banks used the AMA.


64/65

Chart 33

Contribution of the Operational Risk Capital Charge for banks adopting Basis Indicator Approach and Standardised Approach (Group 2), [%]

Contribution BIA by size

0%

10%

20%

30%

40%

50%

60%

70%

Contribution TSA by size

0%

10%

20%

30%

40%

50%

60%

70%

Charts 34 and 35 compare the change in risk-weighted assets at bank-level between different approaches both for Group 1 and Group 2 banks. Even if some Group 1 banks produced less favourable results for the Standardised Approach compared to the Basis Indicator Approach the incentive structure seems to be largely given between both approaches for the majority of banks.

Chart 34

Incentive structure for Operational Risk Capital Charge for Standardised Approach compared to Basic Indicator Approach [%]

Change TSA vs BIA by sizeGroup 1

-25%

-20%

-15%

-10%

-5%

0%

5%

10%

15%

Change TSA vs BIA by sizeGroup 2

-25%

-20%

-15%

-10%

-5%

0%

5%

10%

15%



65/65

Chart 35

Incentive structure for Operational Risk, Capital Charge for Advanced Measurement Approach compared to Standardised Approach [%]

Change AMA vs TSA by sizeGroup 1

-80%

-60%

-40%

-20%

0%

20%

40%

60%


Quantitative Impact Study 5 Overview on the Results of the...

Documents

Transcript of Quantitative Impact Study 5 Overview on the Results of the...