## 1. Capital Ratios for Banks in the EU

## Source details

**Canonical URL:** [1. Capital Ratios for Banks in the EU](https://www.imf.org/-/media/files/publications/wp/2017/wp17137.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2017/wp17137.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2017/wp17137.pdf.json)

---

### Introduction and scope
- Holding capital against risk-weighted assets (RWA) rather than total assets aligns with Basel risk sensitivity.
- Internal models rely on historical parameters; concerns exist about resulting capital ratios (Le Leslé and Avramova, 2012; Vallascas and Hagendorff, 2013; Behn, Haselmann, and Vig, 2016).
- Complexity and opacity of internally generated risk weights create risks of gaming, mistake, and monitoring difficulty.
- Paper distinguishes itself by:
  - Evaluating risk weights at the portfolio level and across country of counterparty exposure.
  - Documenting risk weight heterogeneity, assessing sensitivity of corporate risk weights to fundamentals, and analyzing counterfactual capital ratios.
  - Using EBA transparency exercise data covering listed and non-listed European banks.

### Regulatory review of bank risk weights
- Basel methods allowed (subject to supervisory approval):
  - Standardized approach (SA) uses prescribed risk weights.
  - Internal Ratings Based (IRB) approaches use bank internal models (Foundation IRB and Advanced IRB).
- IRB model parameters include PD, LGD, EAD, and effective maturity; calibration and supervisory judgment contribute to heterogeneity.
- Representative CAR estimates from literature and policy:
  - Dagher et al. (2016): CAR of 15–23 percent.
  - FSB (2014): 16–20 percent for global systemic banks.
  - US Federal Reserve proposal: more than 18 percent in total loss-absorbing capacity.
- Leverage ratio introduced by Basel III as backstop: currently set at 3 percent of Tier 1 capital; alternatives and calibrations discussed (examples: 4 to 5 percent, and 8 percent suggested by Bair, 2013).
- BCBS and BIS exercises and actions:
  - 2013 Hypothetical Portfolio Exercise found variation driven by underlying risk and practices.
  - 2016 RCAP: alignment of PD outcomes and IRB estimates on average; dispersion in LGD and loss rates remains.
  - Consultations and proposals in 2015–2016 on constraints, floors, and aggregate floor calibration in the range of 60 to 90 percent of exposures at default.
- European supervisory actions:
  - EBA 2014 review of residential mortgage risk weights.
  - ECB planned review of 7,000 IRB models; ECB TRIM launched Feb 2017 targeting 68 banks across 15 countries through 2019.
  - EBA March 2017 report: RWA variability largely explained by portfolio features; remaining variability due to idiosyncratic features, modeling assumptions, and practices.
- Comparative RWA density examples:
  - Average RWA density for European G-SIBs: 35 percent.
  - Average RWA density for U.S. G-SIBs: 58 percent.
  - As of March 2015, RWA/total assets: U.S. GSIBs varied between 40 and 74 percent; European GSIBs ranged between 21 and 55 percent (SNL Financial).

### Variation in risk weights across bank portfolios (EBA transparency data)
- EBA exercise covers 105 banks across 21 European countries (reference dates: December 2014 and June 2015), representing around 70 percent of EU banking assets.
- EBA data granularity enables investigation along three dimensions:
  - Portfolio type: IRB vs SA.
  - Asset class: corporate, retail, mortgage.
  - Country of counterparty exposure: up to 60 reported countries (minimum of 95 percent of total exposures at default and top 10 countries per bank).
- Key portfolio-level findings:
  - Credit risk accounts for 85 percent of total RWA; operational risk 10 percent of RWA.
  - Median IRB credit risk weight (June 2015): 34 percent.
  - Median SA credit risk weight (June 2015): 75 percent.
  - SA credit risk weights are, on average, twice as high as IRB.
  - Country-level IRB credit risk weights range from 22 percent in Sweden to close to 50 percent in Austria.
  - SA credit portfolios vary between slightly less than 60 percent in Malta to 90 percent in Latvia.

### Risk weights by asset class and cross-country heterogeneity
- Corporate portfolio (IRB):
  - Lowest IRB corporate risk weights: Denmark 33 percent and Sweden 34 percent (on average of EAD).
  - Highest IRB corporate risk weights: Ireland 69 percent and Portugal 80 percent.
  - SA corporate risk weights vary between 76 percent (France) and 103 percent (Hungary).
- Retail portfolio:
  - IRB retail risk weights vary: Luxembourg average 11 percent; Spain 46 percent.
  - SA retail median across the EU: 72 percent.
- Mortgage portfolio:
  - Average gap between IRB and SA: 26 percentage points (IRB lower).
  - IRB mortgage exposures with less than 10 percent of EAD subject to regulatory capital in Finland and Sweden.
  - IRB mortgage averages: Austria 25 percent and Ireland 32 percent.
  - SA mortgage (noted): Latvia 35 percent; highest Polish banks at 85 percent.
- Country supervisory/historical influences:
  - Ireland: higher mortgage risk weights due to housing crisis and arrears.
  - Sweden: financial supervisory authority applies a 25 percent floor to mortgage risk weights.

### Variation by country of counterparty exposure
- EBA reports top 10 counterparty countries per bank; exposures at default broken down by counterparty country.
- Defaulted loans are excluded from IRB risk-weight calculations to improve comparability.
- Illustrative example (IRB corporate exposures to Germany):
  - Austrian banks: 59 percent (average).
  - Danish banks: 9 percent (average).
  - Irish banks: 137 percent (average).
  - German banks: 46 percent (average).
- Example descriptive statistics (IRB corporate exposures to the Netherlands):
  - Average 61 percent; Median 47 percent; Std. Dev. 31 percent; Min 41 percent; Max 137 percent.
- Greatest variability observed:
  - IRB and SA corporate: Luxembourg and Poland counterparty exposures.
  - Retail IRB and SA: Ireland and Spain dispersion.
  - Mortgage IRB and SA: exposures to the U.S.
- Cross-border informational opacity can raise or lower risk weights; significant heterogeneity persists after cleaning.

### Calculation and aggregate composition of risk weights (Appendix A)
- Risk weights computed as RWA ("Risk exposure amount") divided by Exposure value ("exposures at default").
- Credit risk share of total RWA: 85 percent.
- Operational risk share: 10 percent.
- Market risk share: less than 5 percent for banks in all EU countries except Denmark, Germany, Hungary, Sweden, and the U.K.
- Analysis focuses on corporate, retail, and mortgage exposures across IRB and SA portfolios.
- Treatment of nonperforming exposures:
  - Defaulted assets are subtracted from both numerator and denominator for IRB portfolios to compute risk weights from "good quality" assets.
  - This adjustment is not needed for SA portfolios because provisions are already deducted from Exposure Value.

### Illustrative bank-level detail: Nordea Bank (selected June 2015 figures)
- RWA component shares (Jun-15):
  - RWA - Credit Risk: 78.4
  - RWA - Market Risk: 5.8
  - RWA - Operational Risk: 11.4
  - RWA - Other*: 4.4
- IRB share of Nordea's credit portfolio (June 2015, percent):
  - Corporate Portfolio: 98.8
  - Mortgage Portfolio: 96.5
  - Retail Portfolio: 88.6
  - Total Credit Portfolio: 96.7
- SA share of Nordea's credit portfolio (June 2015, percent):
  - Corporate Portfolio: 1.2
  - Mortgage Portfolio: 3.5
  - Retail Portfolio: 11.4
  - Total Credit Portfolio: 3.3
- Nordea average risk weights:
  - IRB portfolios: vary between 8 and 42 percent.
  - SA portfolios: vary between 59 and 100 percent.
- Cross-country Nordea examples (selected, IRB Corporate and SA Corporate where shown):
  - Denmark: Corporate IRB 35.2; Corporate SA 99.9.
  - Sweden: Corporate IRB 35.3; Corporate SA 100.0.
  - United States: Corporate IRB 36.2; Corporate SA 100.0.

### IRB corporate risk weights and firm fundamentals (regression evidence)
- Baseline regression relates bank i’s IRB corporate risk weight in country c at time t to:
  - Fundamentals of counterparties (Leverage, Interest coverage, Liquidity, Stability (z-score), Market value (market-to-book)),
  - Bank controls Z (Corporate portfolio share, Pretax ROA),
  - X (real GDP growth in counterparty country),
  - Country fixed effects.
- Panel A (median firm fundamentals) selected coefficients (Table 3):
  - Leverage 0.601 0.621 [0.113]*** [0.154]***
  - Interest coverage -0.656 -0.605 [0.278]** [0.222]***
  - Liquidity -0.491 -0.314 [0.149]*** [0.158]*
  - Stability -8.085 -7.053 [1.874]*** [3.057]**
  - Market Value -7.142 -4.180 [1.743]*** [2.231]*
  - Pretax ROA 20.792 14.432 9.963 15.220 13.165 [8.732]* *[7.689]*[8.700][7.945]*[9.072]
  - Real GDO growth 0.690 0.425 0.214 0.833 0.511 [0.269]* *[0.279][0.346][0.317]* *[0.283]*
  - Observations 481 316; R-sq uared 0.407 0.413 0.396 0.408 0.405 0.407 0.415 0.418 0.409 0.405
- Panel B (75th percentile) selected coefficients:
  - Leverage 0.129 0.149 [0.118] [0.098]
  - Liquidity -0.272 -0.223 [0.120]** [0.131]*
  - Stability -5.198 -4.743 [0.849]*** [1.478]** *
  - Real GDO growth 0.569 0.483 0.271 0.841 0.531 [0.295]* [0.287]* [0.357] [0.289]** * [0.297]*
  - Observations 481 316; R-sq uared 0.391 0.402 0.391 0.402 0.397 0.405 0.423 0.422 0.412 0.407
- Interpretation:
  - Lower leverage, higher interest coverage, greater liquidity, better valuation, and greater stability correlate with lower corporate risk weights.
  - Stability and market valuation have larger economic significance than leverage and liquidity.
  - IRB models capture median firm risk better than tail (75th percentile) risk.
- Expected Default Frequencies (EDFs) from Moody’s (Table 4) as alternative proxy:
  - EDF_25th perc. 7.321 -9.612 [4.671] [3.692]* *
  - EDF_50th perc. 2.435 -1.029 [1.063]* * [1.436]
  - EDF_75th perc. 0.681 0.331 [0.503] [0.522]
  - EDF_90th perc. -0.136 -0.01 [0.141] [0.139]
  - Corporate portfolio -0.658 -0.649 -0.681 -0.657 [0.203]*** [0.200]*** [0.205]*** [0.212]***
  - Pretax ROA 25.708 26.199 27.526 27.555 [8.812]* * * [8.646]** * [8.752]* * * [10.241]* *
  - Observations 471 308 471 308 471 308 453 292; R-sq uared 0.391 0.406 0.392 0.404 0.391 0.405 0.385 0.398
- EDF results: corporate risk weights do not show consistent sign or significance for EDF thresholds; accounting-based fundamentals drive coefficients more consistently than market-derived EDFs.

### Harmonization counterfactual analyses and effects on capital ratios
- Three uniform-risk-weight scenarios applied by asset class and counterparty country:
  - Scenario 1: median risk weight by asset class and counterparty country (plausible).
  - Scenario 2: 75th percentile of the risk weight distribution (more severe).
  - Scenario 3: maximum risk weight by asset class and counterparty country (extreme).
- Capital ratios examined: CET1, Tier 1, and Total capital ratios (same denominator: RWA).
- Scenario 1 (median) effects:
  - Banks in the Netherlands: average decline between 10.8 to 12 percentage points in regulatory capital ratios.
  - Banks in Germany and Sweden: reduction around 2 percentage points on average.
  - Banks in Finland, Ireland, and Portugal: boost to capital ratios by more than 1 percentage point.
  - No bank breaches minimum regulatory ratios under Scenario 1: CET1 4.5 percent, Tier1 6 percent, Total 10.5 percent.
- Scenario 2 (75th percentile) effects:
  - Banks in Austria, Ireland, Italy, Norway, and Spain barely affected; Portugal experiences slight boost.
  - Banks in the Netherlands: decline around 14 percentage points.
  - Banks in Belgium, France, Germany, and Sweden: reduction in capital ratios of 3 to more than 4 percentage points.
  - No bank of the 77 in the sample breaches Basel III minimum requirements under Scenario 2.
- Scenario 3 (maximum) effects:
  - Capital ratios in France, Germany, Netherlands, and Sweden would decrease by more than 10 percentage points.
  - Under Scenario 3 breaches:
    - 7 banks would not meet CET1 minimum.
    - 9 banks would not meet Tier1 minimum.
    - 38 banks would not meet Total capital minimum.
    - 8 out of 20 countries where internal models are used would be affected.
- Overall assessment:
  - Under non-extreme but severe harmonization scenarios, banks are unlikely to need to raise capital to meet Basel III minimums.
  - Extreme uniformization (maximum) could cause significant breaches for multiple banks.

### Conclusions and policy implications
- Substantial variation in bank risk weights exists across asset classes and countries of counterparty exposure in EBA data.
- Corporate IRB risk weights are sensitive to accounting-based firm fundamentals but not to market-derived EDFs.
- Harmonization of risk weights may be achievable without large aggregate capital impacts under plausible scenarios.
- Policy recommendations:
  - Encourage harmonization discussions; harmonization might be achieved without large capital impacts.
  - For banks with risk weights markedly below peers, conduct detailed reviews of internal models to determine whether higher capital is justified.
  - Preserve risk sensitivity of the Basel framework rather than reverting to a simple leverage ratio.
  - ECB’s TRIM program aligns with these objectives and could be extended to non-eurozone countries in Europe.

### Appendix I — The 2015 EU-Wide Transparency Exercise (data and classification)
- Exercise discloses detailed data for 105 banks; banks classified in 10 categories.
- Item counts (selected):
  - Capital: 150101 - 150137 (37 items)
  - RWA: 150201 - 150210 (10 items)
  - P&L: 150301 - 150336 (36 items)
  - Credit Risk: 150501 - 150556 (74 items)
  - Leverage: 151001 - 151003 (3 items)
- Portfolios classified into: No breakdown; SA Portfolio; IRB Portfolio.
- EBA item mapping:
  - SA corporate: item 303; SA retail: item 404; SA mortgages: item 501.
  - IRB corporate: item 303; IRB retail: sum of 409 and 410; IRB mortgages: item 406.
- Appendix II contains bank-level average risk weights (IRB portfolio) by exposure type and counterparty country with summary statistics (Count, Average, Median, Std. Dev., Min, Max).

*Source: wp17137 (IMF Working Paper).*

### 1. Capital Ratios for Banks in the EU.................................................................23

### 1. Capital Ratios for Banks in the EU.................................................................23

### Major sections (as listed in the content unit)
- 1. Capital Ratios for Banks in the EU.................................................................23
- 2. Risk Weighted Assets Density......................................................................24
- 3. RWA, IRB/SA Portfolio Decom
position, and Risk Weights...................................25
- 4. IRB/SA Average Risk Weights by Credit Exposure.............................................26
- 5. IRB Average Risk Weights in the Europe.........................................................27
- 6. Changes to Capital Adequacy Ratios...............................................................27

### Tables (as listed)
- 1. Median Risk Weights across IRB/SA Portfolios of European Banks.........................29
- 2. Descriptive Statistics for Bank Risk Weights by Country of Counterparty Exposure......30
- 3. IRB Corporate Risk Weights and Firm
 Fundamentals..........................................33
- 4. IRB Corporate Risk Weights and Expected Default Frequencies..............................35

### Appendices (as listed)
- I. The 2015 EU-Wide Transparency Exercise.......................................................36
- II. Average Risk Weights, IRB Portfolio.............................................................42

*Source: wp17137 - 1. Capital Ratios for Banks in the EU.................................................................23*

### References....................................................................................................20

### wp17137 - References....................................................................................................20

### I. INTRODUCTION
- Holding capital against risk-weighted assets (RWA) rather than total assets is consistent with the greater risk sensitivity intended by the Basel framework.
- Internal models used by many banks rely on parameters largely based on historical data and previous loss experience; doubts regarding resulting capital ratios have arisen (Le Leslé and Avramova, 2012; Vallascas and Hagendorff, 2013; Behn, Haselmann, and Vig, 2016).
- Recent papers document strategic under-reporting of bank risk (Mariathasan and Merrouche, 2014; Begley, Purnanandam, and Zeng, forthcoming) and regulators have alluded to regulatory arbitrage.
- Complexity and opacity of internally generated risk weights create risks of gaming, mistake, and monitoring difficulty.
- BCBS seeks to address “the issue of excessive variability in risk-weighted assets” to restore “market confidence in risk-based capital ratios” and “promote sound levels of capital and comparability across banks” (BIS, BCBS, 2016a).
- Proposed reforms are expected to increase capital that European banks must hold by a greater amount compared with American peers (The Economist, December 3, 2016).
- Differences contributing to RWA density variation include different asset compositions and wider use of internal models in the EU relative to the U.S.; for example, the stock of mortgages is kept on the balance sheet of European banks whereas US counterparts offload such assets through securitization.
- This paper differs from prior literature by:
  - Evaluating risk weights at the portfolio level and across country of counterparty exposure rather than using RWA density as proxy.
  - Documenting risk weight heterogeneity, assessing sensitivity of corporate risk weights to fundamentals, and analyzing implications of less heterogeneous risk weights under hypothetical counterfactual scenarios.
  - Using a large sample of listed and non-listed European banks leveraging the EBA transparency exercise data.
- EBA data disclose three important dimensions enabling more granular risk-weight calculation:
  - Distribution of bank portfolios across the standardized and internal risk assessment methods.
  - Bank risk weights across major asset classes including corporate, retail, and mortgage portfolios.
  - Largest exposures by country of counterparty for each bank and each asset class (breakdown reported according to the minimum of 95 percent of total exposures at default and the top 10 countries, resulting in exposures to a total of 60 countries).
- The study investigates determinants of corporate risk weights using accounting data and expected default frequencies and derives counterfactual capital ratios using alternative risk weights by asset class and country of counterparty exposure.
- Organization of the paper: Section II (regulatory background), Section III (variations across portfolios), Section IV (variations by SA/IRB, asset class, and country), Section V (sensitivity to fundamentals), Section VI (counterfactual capital ratios), Section VII (conclusion).

### II. REGULATORY REVIEW OF BANK RISK WEIGHTS
- Post-crisis Basel reforms first focused on the numerator of the capital adequacy ratio (CAR) and later shifted attention to the denominator (RWA) and risk weight heterogeneity.
- RWA are derived using risk weights expected to reflect intrinsic risk characteristics of each asset; some variation in risk weights across portfolios is to be expected.
- Basel framework allows a range of methods to measure portfolio risk subject to supervisory approval:
  - Standardized approach (SA) uses prescribed risk weights.
  - Internal Ratings Based (IRB) approaches use bank internal models (Foundation IRB and Advanced IRB under Basel II).
- IRB models require parameters including probability of default (PD), loss given default (LGD), exposure at default (EAD), and effective maturity; calibration and supervisory judgment contribute to heterogeneity.
- To the extent variations in risk weights do not reflect differences in risk exposure, comparability of capital ratios is reduced.
- Reported cross-country CAR variation (example): CAR varying from 12.5 percent for Portuguese banks up to as high as 28 and 25 percent, respectively, for the Netherlands and Sweden (see Figure 1).
- Estimates of sufficient CAR:
  - Dagher et al. (2016) find CAR of 15–23 percent in advanced economies would have avoided creditor losses in the past.
  - Financial Stability Board (FSB, 2014) estimate for global systemic banks: 16–20 percent.
  - US Federal Reserve’s proposal: more than 18 percent in total loss-absorbing capacity (Board of Governors of the Federal Reserve System, 2015).
- Leverage ratio introduced by Basel III as a backstop to risk-sensitive capital; currently set at 3 percent of Tier 1 capital; calibration within the range of 4 to 5 percent could provide a more consistent and effective backstop (Fender and Lewrick, 2015); alternatives argued include a minimum leverage ratio of 8 percent (Bair, 2013).
- BIS RCAP exercises:
  - 2013 Hypothetical Portfolio Exercise: more than 100 major banks and 32 large banking groups in 13 jurisdictions; concluded variations driven by mix of differences in underlying risk and banking and supervisory practices (BIS, 2013).
  - 2016 RCAP report: sample sizes—RWA analysis: 35 major internationally active banks across 13 jurisdictions; EAD analysis: 37 banks across 17 jurisdictions; findings—close alignment of actual PD outcomes and IRB estimates on average, but not for LGD and loss rates; dispersion of A/E outcomes across banks is similar; report describes sound practices for independent model validation.
- BCBS actions:
  - December 2015 review of standardized approaches to credit risk aimed to reduce differences under SA and suggested removing certain exposures (large corporates and financial institutions) from IRB (BIS, 2015).
  - January 2016: agreed to complete work on addressing excessive variability in RWA; March 2016 consultation on additional constraints on IRB models, including floors, to reduce complexity and improve comparability and address excessive variability in capital requirements for credit risk (BIS, 2016c).
  - One option considered: aggregate floor calibrated in the range of 60 to 90 percent of exposures at default to replace Basel I floor.
- European authorities:
  - EBA 2014 review of risk weights for residential mortgages to understand sensitivity to key model parameters (EBA, 2014).
  - Single Supervisory Mechanism planning to review 7,000 IRB bank models over four years to ensure internal models are “solid, credible, and consistent” (Nouy, 2015; ECB, 2016).
  - February 2017: ECB began Targeted Review of Internal Models (TRIM) involving on-site missions to 68 banks in 15 countries stretching over 2019 to reduce inconsistencies and unwarranted variability; TRIM could either raise or lower capital requirements for individual banks (ECB, 2017a; ECB, 2017b).
  - March 2017: EBA report on consistency of RWA for “high default portfolios” covering 114 institutions across 17 EU countries finds RWA variability can be explained largely by portfolio features (proportion of defaulted exposures, country of counterparty, portfolio mix); remaining variability likely due to idiosyncratic features, modeling assumptions, and risk management and supervisory practices.
- Comparative RWA density example:
  - Average RWA density for European G-SIBs: 35 percent.
  - Average RWA density for U.S. G-SIBs: 58 percent.
  - As of March 2015, ratio of RWA to total assets: U.S. GSIBs varied between 40 and 74 percent; European GSIBs ranged between 21 and 55 percent (SNL Financial).

### III. VARIATION IN RISK WEIGHTS ACROSS BANK PORTFOLIOS
- EBA transparency exercise (published November 2015) provides bank-by-bank information using December 2014 and June 2015 reference dates for 105 banks across 21 European countries (representing around 70 percent of EU banking assets).
- The granularity of EBA data allows investigation of bank risk weights along three dimensions:
  - Portfolio type: IRB and SA methods.
  - Asset class: corporate, retail, and mortgage exposures.
  - Country of counterparty exposure: across 60 reported countries (breakdown covers minimum of 95 percent of total exposures at default and the top 10 countries per bank).
- The EBA data permit calculation of bank risk weights for sovereign exposures though sovereigns are outside the paper’s focus.
- Data enable prima facie evidence on variations in risk weights at the portfolio level, by asset class, and by country of counterparty exposure for major banks using internal models in Europe.
- Analysis enabled by EBA data covers both listed and non-listed European banks, addressing prior limitations of cross-country research that relied on aggregate institution-level RWA density or samples concentrated on listed banks.

*Source: wp17137 - References....................................................................................................20*

### Appendix A explains in more detail how risk weights are inferred from the EBA data. They

### wp17137 - Appendix A explains in more detail how risk weights are inferred from the EBA data. They

### Calculation and aggregate composition of risk weights
- Risk weights are calculated as the ratio of “Risk exposure amount” (RWA using the BIS lexicon) to “Exposure value” (“exposures at default” under the BIS lexicon).
- At 85 percent of total RWA, credit risk is the largest bank risk component.
- Operational risk amounts to 10 percent of RWA.
- Except for banks in Denmark, Germany, Hungary, Sweden, and the U.K., the market risk share of RWA is less than 5 percent for banks in all other EU countries.
- Analysis focuses on credit portfolio split by corporate, retail, and mortgage exposures across IRB and SA portfolios.

### Differences by portfolio type (IRB vs SA) and country-level medians
- Some countries rely primarily on SA: Cyprus, Hungary, Latvia, Malta, Poland, and Slovenia.
- Lowest IRB share among EBA banks: Portugal at one-half; highest IRB share: Finland and Sweden at 97 percent.
- Average credit risk weights: SA are, on average, twice as high as IRB.
- Median IRB risk weight for banks in the EU as of June 2015: 34 percent.
- Median SA risk weight as of June 2015: 75 percent.
- Country-level IRB credit risk weights range from 22 percent in Sweden to close to 50 percent in Austria.
- For SA credit portfolios, risk weights vary between slightly less than 60 percent in Malta to 90 percent in Latvia.

### Risk weights by asset class and cross-country heterogeneity
- Corporate portfolio (IRB): significantly more variability than SA.
  - Lowest IRB corporate risk weights: Denmark 33 percent and Sweden 34 percent (on average of EAD).
  - Highest IRB corporate risk weights: Ireland 69 percent and Portugal 80 percent.
  - SA corporate risk weights vary between 76 percent (France) and 103 percent (Hungary).
- Retail portfolio:
  - IRB retail risk weights vary widely: Luxembourg average 11 percent; Spain 46 percent.
  - SA retail median across the EU: 72 percent.
- Mortgage portfolio:
  - Gap between IRB and SA is on average 26 percentage points (IRB lower).
  - IRB mortgage exposures with less than 10 percent of EAD subject to regulatory capital in Finland and Sweden.
  - IRB mortgage risk weights: Austria 25 percent and Ireland 32 percent (averages).
  - SA retail (here noted for mortgage SA portfolio): Latvia 35 percent; highest Polish banks at 85 percent.
- Country-specific supervisory or historical factors:
  - Ireland: deep housing crisis and mortgage arrears contribute to higher portfolio-average mortgage risk weights.
  - Sweden: financial supervisory authority applies a 25 percent floor to mortgage risk weights.

### Variation by country of counterparty and asset class
- EBA sample reports each bank's ten largest counterparty countries; exposures at default are broken down by counterparty country.
- Defaulted loans are excluded from risk-weight calculations to improve comparability.
- Appendix B contains full matrices of average risk weights by portfolio type, asset class, and counterparty country (Tables B1 and B2).
  - Example: IRB corporate exposures to Germany — Austrian banks apply on average 59 percent, Danish banks 9 percent, Irish banks 137 percent, German banks 46 percent.
- Table 2 summarizes descriptive statistics for risk weights by country of counterparty:
  - Example (IRB corporate exposures to the Netherlands): average 61 percent, median 47 percent, standard deviation 31 percent, minimum 41 percent, maximum 137 percent.
- Greatest variability:
  - IRB and SA corporate: counterparty exposures to Luxembourg and Poland.
  - Retail IRB and SA: largest dispersion in Ireland and Spain.
  - Mortgage IRB and SA: greatest differences for exposures to the U.S.
- Cross-border lending introduces informational opacity that can lead to higher or lower risk weights; even after cleaning and disaggregation significant heterogeneity remains.

### IRB corporate risk weights and firm fundamentals (regression analysis)
- Baseline regression (equation (1)) relates bank i's risk weight for its corporate portfolio in country c at time t to:
  - Fundamentals of counterparties (vector Fundamentals),
  - Bank controls Z (Corporate portfolio share, Pretax ROA),
  - X (growth rate in real GDP in country of counterparty),
  - Country fixed effects C, and error term.
- Firm indicators (from IMF’s Corporate Vulnerability Utility, 2013–2014, lagged):
  - Leverage (ratio of debt to assets),
  - Interest coverage (EBIT / interest expense),
  - Liquidity (current assets / total assets),
  - Stability (z-score),
  - Market value (market-to-book).
- Findings (Table 3; Panel A median firm):
  - Lower leverage, higher interest coverage, greater liquidity, better valuation, and greater stability correlate with lower corporate risk weights.
  - Stability and market valuation have greater economic significance than leverage and liquidity.
  - Banks with larger corporate portfolio shares and more profitable banks are likely to apply higher corporate risk weights (not uniform across specifications).
  - Higher GDP growth in country of counterparty associates with greater corporate risk weights.
  - Country of origin matters: banks from Denmark, Netherlands, and Sweden assign lowest corporate risk weights on average.
- Findings for 75th percentile (Table 3; Panel B):
  - Results largely maintained except firm leverage and interest coverage lose statistical significance.
  - Economic significance of coefficients is considerably reduced compared with Panel A.
  - Suggests IRB models capture average (median) firm risk better than tail (75th percentile) risk.
- Expected Default Frequencies (EDFs) from Moody’s (averaged over six months ending December 2014 and June 2015) used as alternative proxy (Table 4):
  - Results do not show consistent sign or statistical significance for EDF thresholds.
  - Corporate risk weights reflect accounting-based firm fundamentals but not market-derived probability of default.

### Harmonization counterfactual analyses and effects on capital ratios
- Three hypothetical scenarios apply uniform IRB risk weights by asset class and counterparty country:
  - Scenario 1: use the median risk weight of each asset class per counterparty country (plausible).
  - Scenario 2: use the 75th percentile of the risk weight distribution (more severe but perhaps plausible).
  - Scenario 3: use the maximum risk weight by asset class and counterparty country (extreme).
- Capital ratios examined: Common Equity Tier 1 (CET1), Tier 1, and Total capital ratios (same denominator: RWA).
- Scenario 1 (median):
  - Banks in the Netherlands would, on average, experience between 10.8 to 12 percentage point decline in regulatory capital ratios.
  - Banks in Germany and Sweden: reduction around 2 percentage points on average.
  - Banks in Finland, Ireland, and Portugal: boost to capital ratios by more than 1 percentage point.
  - Under Scenario 1, no bank breaches minimum regulatory ratios: CET1 4.5 percent, Tier1 6 percent, Total 10.5 percent.
- Scenario 2 (75th percentile):
  - Banks in Austria, Ireland, Italy, Norway, and Spain barely affected; Portugal experiences slight boost.
  - Banks in the Netherlands: decline by around 14 percentage points.
  - Banks in Belgium, France, Germany, and Sweden: reduction in capital ratios of 3 to more than 4 percentage points.
  - No bank of the 77 in the sample breaches Basel III minimum requirements under Scenario 2.
- Scenario 3 (maximum):
  - Capital ratios for banks in France, Germany, Netherlands, and Sweden would decrease by more than 10 percentage points.
  - Under Scenario 3, 7 banks would not meet CET1 minimum, 9 banks would not meet Tier1 minimum, and 38 banks would not meet Total capital minimum; 8 out of the 20 countries where internal models are used would be affected.
- Overall: under non-extreme but severe scenarios, harmonization is unlikely to force banks to raise capital to meet Basel III minimums; extreme uniformization could cause significant breaches.

### Conclusions and policy implications
- Substantial variation in bank risk weights exists across asset classes and countries of counterparty exposure in the EBA data.
- Corporate IRB risk weights are sensitive to accounting-based firm fundamentals but not to market-derived EDFs.
- Counterfactual harmonization exercises suggest harmonizing risk weights may be possible without significant aggregate impact on bank capital under plausible scenarios.
- Policy suggestions:
  - Harmonization discussions are encouraged, as harmonization might be achieved without large capital impacts.
  - For banks with risk weights markedly below peers, detailed review of internal models is warranted to determine whether higher capital is justified.
  - Risk-based regulation that preserves risk sensitivity of the Basel framework is preferable to reverting to a simple leverage ratio.
  - The ECB’s TRIM program is aligned with these objectives and could be extended to non-eurozone countries in Europe.

*Source: wp17137 - Appendix A (IMF working paper).*

### References

### References

### Key empirical findings from figures and tables
- Median IRB risk weight is 34 percent across the EU; median SA risk weight is 75 percent for SA portfolios.  
- IRB corporate risk weights vary between 32 percent in Denmark and 84 percent in Ireland.  
- IRB retail risk weights vary between 11 percent in Luxembourg and 45 percent in Ireland; the median SA retail weight is 72 percent across the EU.  
- IRB mortgage risk weights vary between 6 percent in Finland and 31 percent in Ireland; the median SA mortgage weight is 42 percent across the EU.  
- Table 1: Median Risk Weights across IRB/SA Portfolios of EU Banks, June 2015  
  - IRBSA Corporate Exposures51.894.9  
  - IRBSA Retail Exposures27.071.6  
  - IRBSA Mortgage Exposures15.542.3  
  - IRBSA Overall32.975.1

### Distributional and country-level statistics (selected)
- Figures show wide cross-country variation in RWA/TA (Risk Weighted Assets Density), with panel comparisons for EBA banks (June 2015) and US/European G-SIBs (March 2015).  
- RWA composition: credit risk is the main component of RWA; banks in most of advanced Europe use the IRB method.  
- IRB portfolio composition differs across EU banks, with greater variability across asset classes in SA portfolios.  
- Figures and tables present country-by-country descriptive statistics (Panels A–C) for corporate, retail, and mortgage credit risk weights (Percent, June 2015), including Count, Average, Median, Std. Dev., Min, Max for exposures to many counterparties.

### Changes to capital adequacy ratios — scenario analysis
- Scenario 1: RWA evaluated at the median risk weight by asset class exposure and country of counterparty (chart axis includes -14 to 4 percentage points for CET1/Tier1/Total Capital changes).  
- Scenario 2: RWA evaluated at the 75th percentile of the risk weight distribution by asset class exposure and country of counterparty.  
- Scenario 3: RWA evaluated at the maximum risk weight distribution by asset class exposure and country of counterparty – Worst case scenario (chart axis includes -22, -17, -12, -7, -2 percentage points for CET1/Tier1/Total Capital changes).

### Regression and econometric results (selected coefficients preserved exactly)
- Table 3. IRB Corporate Risk Weights and Firm Fundamentals (Panel A, firm fundamentals evaluated at the median):  
  - Leverage0.6010.621 [0.113]***  [0.154]***  
  - Interest coverage-0.656-0.605 [0.278]**   [0.222]***  
  - Liquidity-0.491-0.314 [0.149]***[0.158]*  
  - Stability-8.085-7.053 [1.874]***   [3.057]**  
  - Market Value-7.142-4.180 [1.743]***   [2.231]*  
  - Pretax ROA20.79214.4329.96315.22013.165 [8.732]* *[7.689]*[8.700][7.945]*[9.072]  
  - Real GDO growth0.6900.4250.2140.8330.511 [0.269]* *[0.279][0.346][0.317]* *[0.283]*
  - Observations481316; R-sq uared0.4070.4130.3960.4080.4050.4070.4150.4180.4090.405
- Table 3. IRB Corporate Risk Weights and Firm Fundamentals (Panel B, 75th percentile): selected coefficients include:  
  - Leverage0.1290.149 [0.118][0.098]  
  - Liquidity-0.272-0.223 [0.120]**[0.131]*  
  - Stability-5.198-4.743 [0.849]***  [1.478]** *  
  - Real GDO growth0.5690.4830.2710.8410.531 [0.295]*[0.287]*[0.357][0.289]** *[0.297]*  
  - Observations481316; R-sq uared0.3910.4020.3910.4020.3970.4050.4230.4220.4120.407
- Table 4. IRB Corporate Risk Weights and Expected Default Frequencies (selected):  
  - EDF_25th perc.7.321-9.612 [4.671][3.692]* *  
  - EDF_50th perc.2.435-1.029 [1.063]* *[1.436]  
  - EDF_75th perc.0.6810.331 [0.503][0.522]  
  - EDF_90th perc.-0.136-0.01 [0.141][0.139]  
  - Coporate portfolio-0.658-0.649-0.681-0.657 [0.203]***[0.200]***[0.205]***[0.212]***  
  - Pretax ROA25.70826.19927.52627.555 [8.812]* * *[8.646]** *[8.752]* * *[10.241]* *
  - Observations471308471308471308453292; R-sq uared0.3910.4060.3920.4040.3910.4050.3850.398

### Data sources and documentation
- Primary data and calculations referenced: European Banking Authority (EBA), European Central Bank (ECB), Bank for International Settlements (BIS), SNL, Bankscope, Fund staff/author’s calculations.  
- Figures and tables pertain to June 2015 (EBA banks) and March 2015 (G-SIBs) snapshots; several analytical exercises rely on EBA data and Fund staff calculations.

### Thematic literature and regulatory context (selection from references)
- Regulatory and supervisory analyses: BIS Basel Committee RCAP and capital floors consultations (2013, 2014, 2015, 2016a, 2016b, 2016c).  
- European authorities: European Banking Authority reports on RWAs and HDP exercise (2014, 2017), ECB TRIM guidance (2017a, 2017b), ECB SSM priorities (2016).  
- Academic and policy studies on RWAs, internal models, and capital: Le Leslé and Avramova (2012), Mariathasan and Merrouche (2014), Plosser and Santos (2014), Barakova and Palvia (2014), Beltratti and Paladino (2016), Behn, Haselmann, and Vig (2016), Fender and Lewrick (2015), Dagher et al. (2016), Fraisse, Le, and Thesmar (2015).

*References and figures/tables as presented in the source document.*

### Appendix I. The 2015 EU-Wide Transparency Exercise

### Appendix I. The 2015 EU-Wide Transparency Exercise

### Overview of the exercise
- The 2015 EU-wide transparency exercise discloses detailed and comparable data for 105 individual banks in Europe.
- Banks are classified in 10 categories.
- Each of capital, RWA, Profit and Loss (P&L), market risk, credit risk, nonperforming exposures (NPE), forborne exposures, collateral, sovereign exposures, and leverage are uniquely identified by an item number.
- Credit risk is the largest risk category and is covered by 74 related items.

### Data classification and itemization
- Category items and counts (excerpt):
  - Capital: 150101 - 150137 (37 items)
  - RWA: 150201 - 150210 (10 items)
  - P&L: 150301 - 150336 (36 items)
  - Market Risk: 150401 - 150491 (25 items)
  - Credit Risk: 150501 - 150556 (74 items)
  - NPE: 150601 - 150626 (18 items)
  - Forborne exposures: 150701 - 150726 (18 items)
  - Collateral: 150801 - 150812 (4 items)
  - Sovereign exposures: 150901 - 150914 (14 items)
  - Leverage: 151001 - 151003 (3 items)

- Portfolios are classified into three types:
  - No breakdown
  - SA Portfolio (standardized approach)
  - IRB Portfolio (internal ratings-based approach)

- For SA portfolios, credit risk information is under exposures numbered 303 (corporate), 404 (retail), and 501 (mortgages).
- For IRB portfolios, credit risk information is under exposures numbered 303 (corporate), the sum of 409 and 410 (retail), and 406 (mortgages).
- In EBA terminology:
  - Risk exposure amount = RWA
  - Exposure value = reported exposure after credit conversion factors or credit risk mitigation techniques

### Treatment of nonperforming exposures and adjusted risk weights
- Defaulted assets are subtracted from both numerator and denominator of the RWA density to compute risk weights from "good quality" assets (applies to IRB portfolios).
- Note: This adjustment is needed for IRB portfolios only, since provisions are already deducted from Exposure Value for SA portfolios.
- The exercise records items that enable this adjustment (listed in Table A4 in the source).

### Illustrative bank: Nordea Bank — data coverage and portfolio composition
- Nordea Bank is provided at consolidated group level and for the largest 10 country exposures for two periods: December 2014 and June 2015.
- Information on credit risk includes 567 entries for each period, totaling 1,134 observations.
- Data distribution by portfolio type (per period):
  - Obs. at the group level: Total 87
  - Obs. for each of largest 10 country exposures: Total 480
  - Portfolio breakdown (per period):
    - 0 No breakdown: 35 (group level), n/a (per country)
    - 1 SA: 21 (group level), 28 (per country)
    - 2 IRB: 31 (group level), 20 (per country)

- Largest 10 country exposures reported: Denmark, Finland, Germany, Luxembourg, Norway, Poland, Russian Federation, Sweden, United Kingdom, and the United States.

### Nordea Bank — RWA component shares (in percent)
- Dec-14:
  - RWA - Credit Risk: 81.8
  - RWA - Market Risk: 5.0
  - RWA - Operational Risk: 11.6
  - RWA - Other*: 1.6
- Jun-15:
  - RWA - Credit Risk: 78.4
  - RWA - Market Risk: 5.8
  - RWA - Operational Risk: 11.4
  - RWA - Other*: 4.4
- *Other risk exposure amounts include counterparty value adjustments (CVA).

### Nordea Bank — credit portfolio shares (June 2015) (in percent)
- IRB share:
  - Corporate Portfolio: 98.8
  - Mortgage Portfolio: 96.5
  - Retail Portfolio: 88.6
  - Total Credit Portfolio: 96.7
- SA share:
  - Corporate Portfolio: 1.2
  - Mortgage Portfolio: 3.5
  - Retail Portfolio: 11.4
  - Total Credit Portfolio: 3.3
- Note: The credit portfolio considered includes corporate, retail, and mortgage loans; it excludes exposures to sovereigns, governments, financial institutions, equity, in default, and securitization.

### Findings on risk-weight differences: IRB vs SA and cross-country dispersion
- Average risk weights at Nordea:
  - IRB portfolios: vary between 8 and 42 percent
  - SA portfolios: vary between 59 and 100 percent
- Cross-country dispersion for Nordea (selected findings):
  - IRB corporate risk weight varies between 35 percent for exposures in Sweden and about 50 percent in Luxembourg.
  - IRB retail portfolio risk weights range between close to 22 percent in Norway and 53 percent in the Russian Federation.
  - IRB mortgage exposures show pronounced dispersion: between 4 percent for exposures in Sweden and 100 percent in Poland.
  - For SA portfolios, risk weights are more homogenous across country of counterparty and credit exposures.

- Excerpts from Nordea Bank risk weights by country (IRB and SA; excludes sovereigns, governments, financial institutions, equity, in default, and securitization):
  - Denmark: Corporate IRB 35.2; Corporate SA 99.9; Mortgage IRB 10.8-31.0; Retail SA 74.5
  - Finland: Corporate IRB 35.5; Corporate SA 100.0; Mortgage IRB 7.3-27.2; Retail SA 72.7
  - Germany: Corporate IRB 44.9; Corporate SA 100.0-48.4; Retail SA 75.0
  - Luxembourg: Corporate IRB 49.8; Corporate SA 99.8; Mortgage IRB 33.7-29.6; Retail SA 73.6
  - Norway: Corporate IRB 40.2; Corporate SA 100.0; Mortgage IRB 11.9-21.6; Retail SA 73.8
  - Poland: Corporate IRB 33.8; Corporate SA 100.0; Mortgage SA 100.0-42.4; Retail SA 75.0
  - Russian Federation: Corporate IRB 45.0; Corporate SA 100.0; Mortgage IRB 30.0-35.2; Retail IRB 53.3; Retail SA 75.0
  - Sweden: Corporate IRB 35.3; Corporate SA 100.0; Mortgage IRB 3.8-35.0; Retail IRB 24.2; Retail SA 73.4
  - United Kingdom: Corporate IRB 38.9; Mortgage IRB 46.8; Retail SA 75.0
  - United States: Corporate IRB 36.2; Corporate SA 100.0; Mortgage IRB 47.2; Retail SA 75.0

### Cross-country averages and dispersion (selected summary statistics from Appendix II tables)
- Appendix II presents bank-level average risk weights (IRB portfolio) by exposure type (Corporate, Retail, Mortgage) and by country of counterparty, with summary statistics including Count, Average, Median, Std. Dev., Min, and Max for many country-exposure combinations (tables B1 and B2).
- Panel excerpts indicate wide variation across countries and exposure types in both IRB and SA portfolios (see full tables in the source).

*Source: Appendix I. The 2015 EU-Wide Transparency Exercise (EBA transparency data and author’s calculations).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2017/wp17137.pdf_