Comparative analysis of the Regulatory Capital calculation across major European jurisdictions April 2013
CONFIDENTIALITY Our clients industries are extremely competitive, and the maintenance of confidentiality with respect to our clients plans and data is critical. Oliver Wyman rigorously applies internal confidentiality practices to protect the confidentiality of all client information. Similarly, our industry is very competitive. We view our approaches and insights as proprietary and therefore look to our clients to protect our interests in our proposals, presentations, methodologies and analytical techniques. Under no circumstances should this material be shared with any third party without the prior written consent of Oliver Wyman. Oliver Wyman Oliver Wyman MAD-CXV00111-002
Contents Contents Executive Summary 1 1. Context, objectives and scope 4 2. Methodological framework 5 2.1. Sources of information 5 2.2. Methodology 6 3. Main findings 10 3.1. Credit Risk 11 3.2. Market Risk 15 3.3. Operational Risk 15 3.4. Other supervisory practices 16 3.5. Advanced models penetration 16 4. Quantitative impact assessment 18 4.1. Consolidated impact 18 4.2. Santander 20 4.3. BBVA 22 4.4. CaixaBank 24 4.5. Sabadell 26 5. Glossary 28 Oliver Wyman MAD-CXV00111-002
List of Figures List of Figures Figure 1: Spanish entities evolution Transition to less conservative European practices 3 Figure 2: Spanish entities evolution Transition to more conservative European practices 3 Figure 3: Overview of contents covered in the dimensions long list 7 Figure 4: Identification of major regulation differences among geographies 8 Figure 5: Materiality assessment illustration 9 Figure 6: Overview of the main findings: Spain vs. other jurisdictions 10 Figure 7: Detail on Credit Risk findings PD modelling 11 Figure 8: Detail on Credit Risk findings LGD modelling 12 Figure 9: Detail on Credit Risk findings EAD modelling 13 Figure 10: Detail on Credit Risk findings Definitions, asset categorisation and other dimensions 14 Figure 11: Detail on Market Risk findings 15 Figure 12: Detail on Operational Risk findings 16 Figure 13: Detail on Other supervisory practices findings 16 Figure 14: European advanced models adoption consolidated analysis 17 Figure 15: European advanced models penetration main differences 17 Figure 16: Spanish entities evolution Transition to less conservative European practices 19 Figure 17: Spanish entities evolution Transition to more conservative European practices 19 Figure 18: Impact on Santander Transition to less conservative European practices 20 Figure 19: Impact on Santander Transition to more conservative European practices 21 Figure 20: Impact on BBVA Transition to less conservative European practices 22 Figure 21: Impact on BBVA Transition to more conservative European practices 23 Figure 22: Impact on CaixaBank Transition to less conservative European practices 24 Figure 23: Impact on CaixaBank Transition to more conservative European practices 25 Figure 24: Impact on Sabadell Transition to less conservative European practices 26 Figure 25: Impact on Sabadell Transition to more conservative European practices 27 Oliver Wyman MAD-CXV00111-002
Executive Summary Executive Summary The introduction of Basel II regulation and Capital Requirement Directives have aimed to deliver better alignment between entities capital and the actual risks borne. Specifically, this has encouraged a more accurate measurement of the regulatory capital associated with each individual risk-bearing asset allowing for the idiosyncrasy of each individual financial institution: the key metric for regulatory capital calculation - Risk Weighted Assets (RWA) - is estimated through internal models (in the advanced method) that are developed by each individual institution according to the rules and guidelines set by the regulation. Those rules and guidelines deliberately allow for some degree of freedom to the supervised institution to adopt the methodological choices and model design decisions that are considered the most appropriate to correctly measure their risks. Furthermore, supervisory authorities have the leeway to further specify and cascade the rules into the local market. The individual choices adopted by each institution are subject to the approval of the local supervisory authority. As a result, differences in methodology and practices to calculate capital requirements exist both for institutions subject to different supervisory authorities and among those in the same jurisdiction. This evidently raises the question to what extent the capital requirement levels across different institutions are comparable. In order to address this question, this report describes findings from a comparative analysis of RWA calculations across the largest European countries (France, Germany, Italy, Spain and UK). The analysis uses as reference the RWA calculation for four major Spanish institutions and compares this to the calculation derived from the enforcement of practices applied in the other jurisdictions. In addition, as we expect a convergence in capital ratios to be a gradual trend due to the arrival of the Single Supervisory Mechanism - we illustrate the potential impact of potential convergence on capital ratios under two scenarios: a) a composite of conservative European practices for each methodological component and b) composite of less conservative practices. The analysis considered 120 elements of the RWA calculation. Each element was considered across jurisdictions to assess material differences and to calculate the impact on the for each of the Spanish banks involved in the exercise, applying Oliver Wyman expert judgement. The main conclusions are as follows: Significant differences are observed in the methodology applied, both across and within jurisdictions. This is driven both by choices adopted by individual institutions and supervisory practices No single jurisdiction is an overall outlier in terms of level of conservatism adopted, despite significant differences across the 120 elements considered. Spain tends to be on the more conservative side of the spectrum, but within a tight spread across jurisdictions. Spain s relative position can be described as follows: 1
Executive Summary 1. Observed Spanish RWA calculation practices are relatively more conservative than the ones observed in the other jurisdictions for Credit Risk the key driver of RWA for the four largest Spanish institutions, representing ~85% of total RWA: a. PD (Probability of Default), Spanish practices are broadly aligned with peers. The less conservative Spanish guideline of permitting the use of external data is offset by more conservative practices such as adjustments for client attrition in sample PD calculations. b. Loss Given Default (LGD): Spanish practices are more conservative, driven by the application of conservative buffers in Large Corporate portfolios and more stringent modelling guidelines (e.g. downturn development samples and/or collateral haircuts, unresolved cases treatment and defaulted assets capital charges) c. In addition, the Spanish supervisor and individual institutions apply some other conservative practices on other relevant credit risk RWA calculation parameters, but all have a minor impact - (e.g. Exposure at Default (EAD) downturn add-ons and floors at single transaction level, synthetic securitisations that are not allowed to reduce capital, shortfall calculation, etc.) 2. Differences are in general less material for Market and Operational risk RWA practices: a. In Market Risk, Spanish practices are broadly aligned with peers, primarily due to the application of a less conservative VaR multiplier coefficient (3 vs.3 5.5 in other geographies), which is offset by other more conservative practices (e.g. VaR temporary weighting practices in average VaR calculation and lack of recognition of diversification benefits across geographies) b. In Operational Risk on the AMA approach, Spanish practices are more conservative driven by not allowing for international diversification effects. No material differences have been found under the standard method The degree of adoption of internal models on the different asset classes has been found to be a key driver for the difference across jurisdictions. None of the Spanish banks in the analysis have validated neither internal models method for counterparty credit risk and nor advanced internal models approaches for specialized lending (common practice is slotting criteria), not and only one bank (BBVA) is adopting the AMA approach for operational risk measurement. In addition, the standardized approach is still adopted for RWA calculation of major foreign subsidiaries (namely in USA and Latin America). The quantitative impact of the differences identified was estimated for the Spanish entities included in the analysis. We simulated the transition of the entities portfolios to two scenarios: Under a less conservative scenario (defined as the composite of the least conservative practices in each element observed across jurisdictions): in this 2
Executive Summary scenario, the ratio for the average Spanish bank would increase from 10.6% to 11.4-12% (Figure 1 below) Under a more conservative scenario (defined as the composite of the most conservative practices in each element observed across jurisdictions): in this scenario, the ratio for the average bank would broadly remain unchanged.(figure 2 below) Figure 1: Spanish entities evolution Transition to less conservative European practices Oper. and Count: 15-30bps Credit Risk Spain 5-10bps Credit Risk - Other geographies: 25-35 bps 45 75bps 11.4 12.0% 10.6% 30 40 bps 0 10 bps 5 10 bps 0 5bps 10.9 11.2% Current Core Capital LGD PD Other Credit Risk Other risks without Advanced models with Figure 2: Spanish entities evolution Transition to more conservative European practices Oper. and Count: 5-8bps Credit Risk Spain 5-10 bps Credit Risk - Other geographies: 10-12 bps 10.6% 10.3 10.7% 20 30bps 10.5 11.0% (10) 5 bps (5) 5bps (5) (1)bps (5) 1bps Current Core Capital LGD PD Other Credit Risk Other risks without Advanced models with 3
1. Context, objectives and scope The Eurozone is taking the required steps to create the Single Supervisory Mechanism as a core element of banking union. Under the deal agreed on by member states in December 2012 and approved by lawmakers in March 2013, the Supervisor will be an agency within the European Central Bank whose decisions will need final approval by the ECB's governing council. This Single Supervisor will be responsible for the coherent and consistent application of the Single rulebook in the Eurozone; the capital calculation framework being one of its key elements. Oliver Wyman was mandated to perform an international comparison of the RWA calculation across the large European jurisdictions. The work was commissioned by four major Spanish financial institutions: Banco Santander, BBVA, CaixaBank and Sabadell. The objective of the analysis is to identify different practices applied in the actual RWA calculation and to assess what potential impact their application might have on the portfolios of the entities in scope. This was achieved by: Identifying the key elements of the RWA calculation, both from a regulatory and methodological perspective Compare the RWA calculation practices generally adopted in Spain with those adopted in the other large EU jurisdictions (France, Germany Italy, UK) Identify the main differences impacting RWA calculation, and quantitatively estimate their impact on the Spanish institutions ratios All Pillar 1 assets that consume regulatory capital (on and off-balance sheet) have been considered: credit, market, counterparty and operational risk. The rest of the document is structured into four main sections: Description of the methodological framework, Summary of the main findings Quantitative impact assessment Conclusions 4
2. Methodological framework 2.1. Sources of information Multiple data sources were used over the course of the exercise: 1. Public sources o Pillar 3 reports and annual financial statements of institutions across the jurisdictions within the scope of the exercise o Spanish, Italian, French, Germany and UK Regulation on Banking Supervision 2. Non-public data from the Spanish banks: o RWA distribution by type of risk, segment, geography and loan status (performing, restructured and defaulted) o Credit risk parameters drill down: Gross Exposure, EAD, average PDs, historical default rates, average LGDs (both secured and unsecured) and CCFs were provided by segment, loan status (splitting the international activities where relevant) 3. Methodological questionnaires and interviews: a set of approximately 120 questions was submitted to the four participant entities, complemented with methodological interviews held in order to obtain a detailed understanding of the entities methodology and regulatory guidelines 4. Oliver Wyman proprietary information: Oliver Wyman international experts provided input regarding the specific methodological choices and practices adopted across the different jurisdictions 5
2.2. Methodology A three-step approach has been followed in order to perform the Risk Weighted Asset calculation: 1. Definition of the comprehensive list of levers 2. Prioritisation of levers according to materiality criteria 3. Assessment of quantitative impact of prioritised levers 2.2.1. Definition of the comprehensive list of levers A long list of potential levers was defined for the calculation of RWAs across the different jurisdictions. The steps performed to create the list were the following: An initial list of levers was defined by the Oliver Wyman team, following a structured framework (see Figure 3 below) populated using Oliver Wyman expert judgement. The objective was to cover all types of risks that generate Regulatory Capital under the current Basel framework; with a special focus on credit risk and its calculation components, given its importance in the business model of the Spanish institutions The initial list was iterated with each of the participant institutions and completed with additional dimensions that covered the key concerns of the institutions The final list was then consolidated by Oliver Wyman, incorporating all the feedback received 6
Figure 3: Overview of contents covered in the dimensions long list RWA s Credit risk Market risk Standard IRB Base Advanced Basic 1 Internal models parameters PD LGD EAD Maturity Unrated exposures modelling Prudential buffers Guarantees eligibility and others Regulatory capital Operational risk Counterparty risk Standard AMA Current value Standardised EPE 2 Standardised approach Sovereign exposures, POEs and FIs treatment Eligibility of lower RWs Liquid exposures treatment Provisions on defaulted assets Off balance sheet items Others Deductions Pillar 2 add-ons Shortfall and floor Overall validation 3 Asset type coverage, definitions and others Asset class categorisation Default definition Portfolio coverage requirements Treatment of low default portfolios Treatment of equity exposures Legal and validation framework As a result of this step, a list of approximately 120 detailed questions covering potential idiosyncrasies across jurisdictions was defined. 2.2.2. Prioritisation of levers according to materiality criteria Once the initial list of potential levers was identified, the questionnaires for each participant institution were populated. The levers from the initial long list were subsequently prioritized according to their materiality in explaining differences in the level of conservatism adopted by the jurisdictions and entities under consideration. 7
The initial list questionnaire was completed using three sources: 1. Specialised teams within the participant entities, whose day to day work focuses on the analysis and calculation of RWA for the different types of risks covered in the exercise. These teams completed the questionnaire and performed several rounds of iteration in order to clarify responses given 2. Interviews and iteration with experts regarding other geographies practices, with both Oliver Wyman staff and other institution employees with specific knowledge on the topic (e.g. responsible for AIRB Modelling methodology in a major European peer, employees from Operational Risk AMA modelling functions, etc.) 3. Oliver Wyman expertise, obtained in the execution of several engagements on related topics (e.g. building of rating and scoring tools, support on approval processes with banking supervisors, etc.) By using the detailed information on the long list of levers, a prioritisation exercise was performed in order to focus the analysis on the most relevant levers. Relevance was assessed as a combination of material differences present across jurisdictions with the potential impact that the differences would have if applied to current Spanish portfolios. As a result of this prioritisation exercise, approximately 50 questions were identified for further analysis. That allowed us to understand how Spanish RWA calculation practices compare to those of other large EU jurisdictions as a result of a fact based, granular analysis. Figure 4: Identification of major regulation differences among geographies Question Italy France UK Germany How updated must the.... time series be? Are external data series allowed for the Central Tendency calculation?.... Are there any underweight of specific crisis years in the Central Tendency? Is the Probability of Default denominator corrected in order to remove client or operation attrition? How are proceeding from foreclosure for real estate assets estimated?................. Key differences Lever 1 Lever 2 Lever 3 Lever 4 Lever 5 Lever 6 Lever n More conservative European practices average Less conservative 2.2.3. Assessment of quantitative impact of prioritised levers Finally, the quantitative impact of applying the identified differences was estimated for each participant entity based on its current portfolio. The objective was to show 8
the potential quantitative implications of a transition to a more conservative scenario or to a less conservative scenario. In order to properly calculate the impact on each portfolio, a detailed quantitative data request was submitted to the participant entities, both covering exposure distribution and the main parameters for the calculation of RWA at risk class level. Additionally, we completed the quantitative analysis with an international benchmark on the RWA parameters across the other European jurisdictions through public sources (Pillar 3 reports, financial reports, etc.) and Oliver Wyman proprietary information. The impact was estimated in terms of resulting ratio, measured in basis points, and distributed across the different groups of levers within each risk type and within each component of the credit risk calculation, as illustrated in the figure below. Figure 5: Materiality assessment illustration Current Core Capital Lever 1 Lever 2 Lever n Core Capital 9
3. Main findings The analysis reveals that a wide variety of practices applied in the calculation of risk weighted assets. These differences can be observed not only in entities of different geographies, but also in entities in the same jurisdiction. Although some of the differences are driven by methodological choices taken by the entities, some others are directly linked to Supervisors' discretion. Within this range of practices, it can be noted that there are no clear outliers within the different supervisory bodies in terms of conservatism. Spanish practices, when compared with the pool of analysed geographies, are broadly positioned between the European average and the conservative end, with the exception of Market Risk. Figure 6: Overview of the main findings: Spain vs. other jurisdictions Dimensions More conservative Less conservative PD LGD 1. Credit risk EAD Definitions and categorisation Others 2. Market risk 3. Operational risk 4. Others A detailed description for the observed differences in each of the categories is provided in the following sections. 10
3.1. Credit Risk 3.1.1. PD modelling Main differences across jurisdictions identified in PD modelling are related to the cycle definition, allowance for external data for Central Tendency calculation, allowance for continuous rating scales and adjustments for client attrition in the sample used for PD calculation. The overall assessment for PD modelling is that Spanish practices are broadly aligned with other European RWA calculation practices. Figure 7: Detail on Credit Risk findings PD modelling Spanish practices PD denominator is weighted by number of active months (e.g. clients that matures after 6 months are weighted as a half) On the masterscale guidelines, continuous rating scales are not allowed Use of external data is allowed for extrapolating central tendency to longer historical series More conservative approaches All good clients that left the bank are removed from the denominator (GER) Less conservative approaches All the clients are equally weighted in the PD denominator, independently of the number of months (ITA, FRA and UK) Use of continuous rating scales is allowed (UK) External data not allowed (FRA and ITA) External data is allowed, but 5 20% conservativeness add-on in required (UK) Impact on capital levels depends on the entity availability of internal data series and central tendency projection methodology Central Tendency cycle Only recent years of recession is used for the central definition from 1991 to 2012 tendency, enforced by the lack of allowance to use external data on central tendency (FRA and ITA) If the central tendency cycle contains more than one recession, allowed to perform the treatment of crisis years in the central tendency calculation in order to avoiding recession years double counting No specific adjustments for cyclicality in the Central Tendency calibration required yet (e.g. variable scalars method) Not applied in the remaining jurisdictions Adjustments for cyclicality are already considered (UK) Relevance 11
3.1.2. LGD modelling LGD modelling shows material differences across the different jurisdictions, mainly related to conservative buffers in Large Corporate segments, defaulted assets downturn LGD, development sample requirements, haircuts on collateral values and treatment of unresolved cases. The overall assessment for LGD modelling is that Spanish practices are more conservative. Figure 8: Detail on Credit Risk findings LGD modelling Spanish practices Conservative buffers applied in Large Corporate portfolio, keeping the LGD levels close to the FIRB approach for a significant part of the portfolio Defaulted assets absorb capital LGD development sample for RWAs calculation only considers recession years All the unresolved cases are included in the LGD development sample Unresolved cases are projected according to certain restrictions Typically discount rate applied in the LGD cash flows is risk free + 400 bps; regulator requires it for mortgages portfolio More conservative approaches Less conservative approaches No conservative buffers identified in the Large Corporate portfolios (ITA, GER and FRA) No anchoring to foundation values in low default portfolios if more than 20 defaults available in the development sample (UK) Defaulted assets do not absorb capital but generate a higher shortfall ( Italy) Rest of jurisdictions: performing average LGD development sample is not generally only based on a crisis period Unresolved cases only included in the development sample if these cases have more than 7-10 years (ITA) and more than 0-2 years (FRA) Cases included in the development sample are not explicitly regulated but are an entity decision (UK) No projections allowed for the estimation of the unresolved cases (ITA) Discount rate applied of +9% (UK) For mortgages and Real Typically more Estate portfolios, property conservative haircuts are haircuts are applied in order applied: to update collateral values to 20 40% peak to value foreclosure time and to by the foreclosure reflect the decrease in value moment + ~20% sale at sale time haircut (UK) Unresolved case projections are allowed and not strictly regulated (UK and France) In other jurisdictions, typically lower discount rates are applied (entities in France and Italy) Haircuts are applied, but not on top of the use of a downturn sample (Italy and Germany) Relevance 12
Mortgages portfolio LGD is floored at 10% at portfolio level (aligned with most of European peers) Occasionally, conservative adjustments on cure rates are required, involving also requirements in the LGD development sample (e.g. removal of restructured cured operations, etc.) 10% floor is applied at single position level (Italy) The stress on cure rates are usually entity specific across the different jurisdictions; not identified explicit constraints in the cure development sample 3.1.3. EAD modelling Although the impact of the observed differences in CCFs modelling is less material, significant variability has been identified in the EAD downturn add-on requirement and in the introduction of particular floors at single operation level. Figure 9: Detail on Credit Risk findings EAD modelling Spanish practices More conservative approaches EAD downturn add-on A downturn add-on generally required, but estimated by the entity possibility to not include it is applied (UK and in case of lack of sensitivity Germany) to the cycle is proved Negatives CCFs generally not allowed at portfolio level, but in some particular buckets / products this floor can be imposed at operation level Less conservative approaches EAD AIRB modelling not required yet FIRB modelling is the common practice (ITA) Downturn add-on is not required (FRA) Positive CCF typically required only at portfolio level in the rest of the jurisdictions Relevance 3.1.4. Definitions, asset categorisation and other dimensions In addition to the IRB credit risk parameters modelling, differences have been identified in the SMEs retail definition, default definition (including recognition of restructured debt as defaulted), PPUs allowance, shortfall calculation guidelines, and deductions for synthetic securitizations and insurance participations. The overall assessment is that Spanish practices are broadly aligned with other European RWA calculation practices. In addition to the dimensions described below, no other material differences have been identified regarding the Credit Risk Standard approach or capital deductions for the entities covered in the exercise. 13
Figure 10: Detail on Credit Risk findings Definitions, asset categorisation and other dimensions Spanish practices Typically the firm revenue, size or number of employees are considered as explicit filters Restriction related to the fact that companies have to be managed as an SMEs 90 days default definition across all the portfolios (in line with most of the jurisdictions) Most deteriorated restructured debt is considered default European practices are aligned on a 15% PPU allowance Some specific trade finance products cannot be modelled with a maturity lower than 1 year (e.g. export letters of credit, pre-shipment) Under Standard approach, non EU Sovereign exposures in local currency can be weighted at risk-weight established by the non EU regulators only if home regulation recognizes it as equivalent Under Standard approach, no specific lower risk for sovereign exposure <3 months prescribed by local regulator Synthetic securitisations cannot mitigate capital Shortfalls from different portfolios can be offset, so that negative shortfall can be compensated with More conservative approaches Less conservative approaches SMEs retail is entity-specific, leaving room for optimisation (common trend across jurisdictions) Common turnover threshold is 7.5 MM (entities in France) and 2.5 5 MM (Italy) 180 days default definition across all the portfolios, lowering PD but significantly rising the LGD (ITA) 180 days definition for mortgages portfolio. In this case impact is offset by the mortgages LGD floor (UK) However, both UK and Italian practices are moving towards a common scenario in which all the definitions will converge into a 90 days definition Consider restructured debt as defaulted if any loss is recognised (GER and ITA) Not allowed to keep a 10% maximum PPU (ITA) These trade finance products can be modelled with a maturity lower than 1 year (UK) No specific prescription on the equivalency recognition of local regulation outside EU from Home Regulator (ITA) Not material differences of treatment assessed with respect to other geographies In Italy shortfalls from different portfolios cannot be offset Synthetic securitisations mitigate capital (GER and UK) In France shortfall for defaulted assets is not considered (provisions are the best estimate of the Relevance 14
positive shortfalls across portfolios aligned with most of the jurisdictions Insurance equity participations are deducted 50% from the Core Tier 1 and 50% from Tier 2 Occasionally, specific floors are applied for the Real Estate lending segment expected loss) Before, insurance participations acquired before 2007 could be deducted 100% from Total Capital (some French entities). However, this allowance was eliminated in 2013 This practice has not been observed in the rest of the jurisdictions 3.2. Market Risk Some differences have been identified in relation to the coefficient applied to the average VaR calculation and to the recognition of international diversification benefits. Figure 11: Detail on Market Risk findings Spanish practices More conservative approaches Less conservative approaches Relevance Average last 60 days VaR calculation: must be the maximum of a simple average and a exponentially weighted average Rest of jurisdictions: average VaR is calculated as a simple average Entities apply a multiplier of 3 in the average VaR calculation Market Risk diversification benefits are not allowed across international units In other jurisdictions, typically higher observed values (ranging from 3 to 5.5) Generally, entities who calculate VaR at group level account for these diversification benefits The main conclusion relating to market risk is that regulatory guidelines are not the most significant contributors to the differences in capital intensity. In fact, certain lack of explicit guidelines in the context of the recently released Basel 2.5 framework provides significant room for different methodological choices across many relevant levers. 3.3. Operational Risk Main differences in the Operational Risk framework are the recognition of international diversification benefits and the application of insurance policies deductions within the AMA models. No material differences were identified under the Standardised approach. 15
Figure 12: Detail on Operational Risk findings Spanish practices More conservative approaches Less conservative approaches Relevance Once AMA is validated, diversification benefits across different countries are not accountable Spanish entities do not apply the insurance policies capital mitigation associated to insurance policies Other jurisdictions: diversification risk benefits across countries are accountable for capital reduction Some European peers apply the insurance policies mitigation; some UK entities applied and some Italy entities are in process of request (in all cases embedded within the model) 3.4. Other supervisory practices Some specific regulators are more conservative in the Pillar 2 add-ons and floor requirements. Figure 13: Detail on Other supervisory practices findings Spanish practices More conservative approaches Less conservative approaches Relevance Pillar II add-ons No Pillar 2 add-ons incorporated to capital ratios yet aligned with most of the jurisdictions UK is the only country in scope applying Pillar 2 add-ons (risk management quality, capital planning buffers, concentration risk ) However, this dimension does not generate a direct impact on the ratio Floor requirement 80% (aligned with most of the jurisdictions) ITA: 85% 3.5. Advanced models penetration One of the key levers identified is the adoption of advanced models. When analysing the distribution of Risk Weighted Assets between Standard and Advanced approach (see figure below), we observe that Spanish entities have lower adoption of advanced models. 16
Figure 14: European advanced models adoption consolidated analysis 21% Standard 56% 46% 45% 40% 79% Advanced 44% 54% 55% 60% SPAIN ITALY FRANCE UK GERMANY Even though the four Spanish entities have validated some advanced models, they mainly cover the domestic retail and corporate credit risk. When compared to their peers in other geographies, a significant room for development can be seen in some specific credit risk portfolios and in other types of risk as shown in the figure below. Figure 15: European advanced models penetration main differences Spanish Italian French UK German Entities entities entities entities Entities LatAm local portfolios Credit Risk US local portfolios Sovereign & Public Entities Institutions Operational Risk Counterparty Risk 17
4. Quantitative impact assessment This section details the results of the ratio impact analysis for the four institutions. The impact is decomposed into two groups of levers The impact due to differences identified in those levers applicable to the existing methodological framework, which can be further decomposed into: LGD PD Other credit risk Other risks Impact under the to advanced methodologies In addition, a total impact assessment is also provided. 4.1. Consolidated impact Under the less conservative scenario (defined as the composite of the least conservative practices in each element observed across jurisdictions) we estimate: The impact of levers applicable to existing methodology is 35-65 bps. The main contribution is driven by the LGD component. Additionally, PD, Other Credit Risk and Other Risks also have a material impact The impact of of existing models to advanced methodologies for Spanish and international activities is 45-75bps The total impact under the less conservative scenario is 80-140bps in aggregate Under the more conservative scenario (defined as the composite of the most conservative practices in each element observed across jurisdictions) we estimate: The impact of levers applicable to existing methodology is (25)-10 bps The impact of of existing models to advanced methodologies for Spanish and International activities is 20-30 bps The total impact under the more conservative scenario is (5)-40 bps Spanish entities results under both scenarios are presented in the figures below. 18
Figure 16: Spanish entities evolution Transition to less conservative European practices Oper. and Count: 15-30bps Credit Risk Spain 5-10bps Credit Risk - Other geographies: 25-35 bps 45 75bps 11.4 12.0% 10.6% 30 40 bps 0 10 bps 5 10 bps 0 5bps 10.9 11.2% Current Core Capital LGD PD Other Credit Risk Other risks without Advanced models with Figure 17: Spanish entities evolution Transition to more conservative European practices Oper. and Count: 5-8bps Credit Risk Spain 5-10 bps Credit Risk - Other geographies: 10-12 bps 10.6% 10.3 10.7% 20 30bps 10.5 11.0% (10) 5 bps (5) 5bps (5) (1)bps (5) 1bps Current Core Capital LGD PD Other Credit Risk Other risks without Advanced models with 19
4.2. Santander Under the less conservative scenario we estimate: The impact of levers applicable to existing methodology is 25-50bps The impact of of existing models to advanced methodologies for Spanish and international activities is 60-95bps The total impact under the less conservative scenario is estimated is 85-145bps Under the more conservative scenario we estimate: The impact of levers applicable to existing methodology is (20) - 5 bps The impact of of existing models to advanced methodologies for Spanish and International activities is 20-30 bps The total impact under the more conservative scenario is estimated is 0-35 bps Santander results under both scenarios are presented in the figures below. Figure 18: Impact on Santander Transition to less conservative European practices Oper. and Count: 25-40 bps Credit Risk Spain 4-6 bps Credit Risk - Other geographies: 31-49 bps 60 95 bps 11.2 11.8% 10.3% 20 25 bps 0 10bps 5 10bps 0 5bps 10.6 10.8% Current Core Capital LGD PD Other Credit Risk Other risks without Advanced models with 20
Figure 19: Impact on Santander Transition to more conservative European practices Oper. and Count: 5-9 bps Credit Risk Spain 4-6 bps Credit Risk Other geographies: 11-15 bps 10.3% 10.1 10.4% (5) 0 bps (7) 4bps (3) (0)bps (5) 1bps 20 30 bps 10.3 10.7% Current Core Capital LGD PD Other Credit Risk Other risks without Advanced models with 21
4.3. BBVA Under the less conservative scenario we estimate: The impact of levers applicable to existing methodology is 55bps and 85bps The impact of of existing models to advanced methodologies for Spanish and international activities is estimated between 55bps and 85bps The total impact under the less conservative scenario between 110bps and 170bps Under the more conservative scenario we estimate: The impact of levers applicable to existing methodology between (25) bps and 25 bps The impact of of existing models to advanced methodologies for Spanish and International activities between 20 bps and 30 bps The total impact under the more conservative scenario is estimated between (5) bps and 55 bps BBVA results under both scenarios are presented in the figures below. Figure 20: Impact on BBVA Transition to less conservative European practices Oper. and Count: 10-20bps Credit Risk Spain 10-15bps Credit Risk - Other geographies: 35-50 bps 10.8% 40 50bps 0 8bps 10 17 bps 5 10bps 11.3 11.6 % 55 85 bps 11.9 12.5% Current Core Capital LGD PD Other Credit Risk Other risks without Advanced models with 22
Figure 21: Impact on BBVA Transition to more conservative European practices Oper. and Count: 0-2bps Credit Risk Spain 10-14bps Credit Risk - Other geographies: 10-14 bps 10.8% 0 20 bps (5) 5 bps 10.5 11.0% 20 30 bps 10.7 11.3% (15) (0)bps (5) (0)bps Current Core Capital LGD PD Other Credit Risk Other risks without Advanced models with 23
4.4. CaixaBank Under the less conservative scenario we estimate: The impact of levers applicable to existing methodology is 40-95bps The impact of of existing models to advanced methodologies for Spanish and international activities is 20-35bps The total impact under the less conservative scenario is 60-130bps Under the more conservative scenario we estimate: The impact of levers applicable to existing methodology is (40) - 0 bps The impact of of existing models to advanced methodologies for Spanish and International activities is 15-25 bps The total impact under the more conservative scenario is (25) - 25 bps CaixaBank results under both scenarios are presented in the figures below. Figure 22: Impact on CaixaBank Transition to less conservative European practices 11.0% 30 60bps 10 30bps 0 4bps 0 1bps 11.4 11.9% 20-35 bps Oper. and Count: 15-25 bps Credit Risk Spain 5-10 bps 11.6 12.3% Current Core Capital LGD PD Other Credit Risk Other risks without Advanced models with 24
Figure 23: Impact on CaixaBank Transition to more conservative European practices 11.0% 10.6 11.0% 15-25 bps 10.7 11.2% (30) 0 bps (2) 0 bps (6) 0bps (2) 0bps Oper. and Count: 10-15 bps Credit Risk Spain 5-10bps Current Core Capital LGD PD Other Credit Risk Other risks without Advanced models with 25
4.5. Sabadell Under the less conservative scenario we estimate: The impact of levers applicable to existing methodology is 60-105bps The impact of of existing models to advanced methodologies for Spanish and international activities is 10-25bps The total impact under the less conservative scenario is 70-130bps Under the more conservative scenario we estimate: The impact of levers applicable to existing methodology is (35) - (15) bps The impact of of existing models to advanced methodologies for Spanish and International activities is 10-15 bps The total impact under the more conservative scenario is (25) - 0 bps Sabadell results under both scenarios are presented in the figures below. Figure 24: Impact on Sabadell Transition to less conservative European practices 50 80bps 5 15 bps 5 10 bps n.a. 11.0 11.5% 10 25bps 11.1 11.7% 10.4% Oper. and Count: 10-24 bps Credit Risk Spain 0-1bps Current Core Capital LGD PD Other Credit Risk Other risks without Advanced models with 26
Figure 25: Impact on Sabadell Transition to more conservative European practices Oper. and Count: 10-14 bps Credit Risk Spain 0-1bps 10.4% (30) (15) bps (4) (0)bps (1) 0bps n.a. 10 15bps 10.1 10.3% 10.2 10.4% Current Core Capital LGD PD Other Credit Risk Other risks without Advanced models with 27
5. Glossary AIRB AMA Bps CCF EaD ECB EPE FI FX FIRB IRB IMM LGD PD POE PPU RW RWA SME VaR Advanced Internal Ratings Based Advanced Measurement Approach Basis Points Credit Conversion Factor Exposure at Default European Central Bank Expected Positive Exposure Financial Institution Foreign Exchange Foundation Internal Ratings Based Internal Ratings Based Internal Models Method Loss Given Default Probability of Default Publicly Owned Enterprises Permanently Partially Use Risk Weight Risk Weighted Assets Small and Medium Sized Enterprises Value at Risk 28
Report qualifications/assumptions and limiting conditions This report is for the exclusive use of the Oliver Wyman client named herein. This report is not intended for general circulation or publication, nor is it to be reproduced, quoted or distributed for any purpose without the prior written permission of Oliver Wyman. There are no third party beneficiaries with respect to this report, and Oliver Wyman does not accept any liability to any third party. Information furnished by others, upon which all or portions of this report are based, is believed to be reliable but has not been independently verified, unless otherwise expressly indicated. Public information and industry and statistical data are from sources we deem to be reliable; however, we make no representation as to the accuracy or completeness of such information. The findings contained in this report may contain predictions based on current data and historical trends. Any such predictions are subject to inherent risks and uncertainties. Oliver Wyman accepts no responsibility for actual results or future events. The opinions expressed in this report are valid only for the purpose stated herein and as of the date of this report. No obligation is assumed to revise this report to reflect changes, events or conditions, which occur subsequent to the date hereof. All decisions in connection with the implementation or use of advice or recommendations contained in this report are the sole responsibility of the client. This report does not represent investment advice nor does it provide an opinion regarding the fairness of any transaction to any and all parties. 29
Oliver Wyman Paseo de la Castellana, 13-2 piso 28046 Madrid Tel: 34 91 432 8400 Fax: 34 91 432 8401 www.oliverwyman.com