EY IFRS 9 impairment banking survey

Transcription

1 EY IFRS 9 impairment banking survey 06

2 IFRS 9 Financial Instruments represents the most fundamental change to a financial institution s accounting methodology, risk management practices and operational processes The adoption of IFRS 9 will require significant enhancements to a financial institution s organizational engagement, data, systems, quantitative models and governance, amongst other areas. The methodology under IFRS 9 will also introduce more judgment, complexity and volatility in reporting, resulting in a need for intensive oversight and enhanced stakeholder scrutiny. Furthermore, the impact on the financial institutions operational processes and financial reporting will not be limited to the transition period and adoption date. It is expected to continue well after the implementation date of the standard ( January 0) due to the scale of change required by the standard and the availability of public information in the postimplementation period. In 06, EY performed a second IFRS 9 global impairment survey and engaged 6 financial institutions to participate, including global systemically important banks (G-SIBs). The survey is intended to assess the state of readiness in the implementation of the IFRS 9 program with a particular focus on impairment. This paper outlines the survey results across the global participants and provides a view on the current status of the overall IFRS 9 program, including how banks are structuring the implementation plan, decisions around critical policies within the expected credit loss methodology and the level of compliance with the Enhanced Disclosure Task Force (EDTF) recommendations. For further insights on this topic, please get in touch with the contacts provided in the appendix to this survey or your local EY contact. We would be glad to assist you if you would like to discuss the results contained in this report, including where your institution falls within the benchmarking study. Yolaine Kermarrec Andre Correia Dos Santos Page EY IFRS 9 impairment banking survey 06

3 Introduction The Survey EY first conducted an IFRS 9 survey in June 05 to provide participants with a state of readiness benchmark, and to identify emerging trends in the implementation of the new standard. This second round conducted in Spring 06 captures the progress made by banks over the past nine months. All results are presented on an anonymous basis Participants Profile Overall state of readiness We surveyed 6 top tier IFRS-reporting financial institutions worldwide 7 have a balance sheet in excess of 600b; 9 have a balance sheet between 00b and 600b banks are global systemically important banks (G-SIBs) 5 banks are under the scope of Sarbanes-Oxley Act (SOX) banks use advanced internal rating based approach (A-IRB) for all of their portfolios Banks have made significant developments in the past nine months on the implementation of their impairment program. Most banks are progressing well into the build phase in this workstream, with a focus on identifying data and system requirements in order to advance towards the deploy phase While certain impairment policy decisions are still under debate (i.e., multiple scenarios), we expect that participants will continue to refine their policies leading up to the parallel run as we observe greater market convergence on these topics There has been some development on classification and measurement (C&M), albeit at a slower pace than impairment. This is consistent with the relative view of complexity associated with C&M. However, banks should not underestimate the IFRS 9 requirements in this area, particularly with regards to the test of solely payment of principal and interest (SPPI) for non-standardized contracts Most banks expect to start a full parallel run in Q 07, although some banks have had to delay this to Q or even Q in the face of slower progress than expected due to scope increases and updates to implementation guidance Most banks expect to disclose a first quantitative impact assessment to the markets during 07 Page EY IFRS 9 impairment banking survey 06

4 Geographic representation of the survey participants United Kingdom Australia Austria Belgium Switzerland Sweden Singapore Italy Ireland Germany France 7 Denmark Canada 6 total number of surveyed banks Page EY IFRS 9 impairment banking survey 06

5 Content IFRS 9 project status and governance 5 5 Impact assessment 6 7 Significant deterioration 5 Definition of default Measurement of expected credit loss 0 6 IFRS 9 disclosures 9 Page EY IFRS 9 impairment banking survey 06

6 . IFRS 9 project status and governance Status of impairment program A shift to the build phase: Policy 7 Most banks have now moved into the build phase, especially on modeling (see pages -6), data and systems, which banks have prioritized Disclosures 7 5 As policy and design decisions are being finalized, with certain topics under development, including multiple scenarios, most banks aim to allow flexibility for refinements Modeling 7 is widely seen as one of the major challenges of IFRS 9: IFRS 9 poses numerous challenges with regards to availability (historic data at origination for stage allocation, specific features for classification, external market data for measurement), granularity (for disclosures and management information), quality and storage Systems Operating model and control framework 0 Most banks have, therefore, started identifying data gaps early on in their programs. Most have focused their disclosure efforts on quantitative disclosures and related data points Banks are at an early stage on designing and implementing the end to end target operating model: Design early Design advanced Build Deploy & test Live Banks are often still focused on the deployment of models to meet the immediate requirements of the standard, as opposed to delivering an end-to-end credit risk reporting solution to support the monthly business as usual (BAU) impairment results production Numerous significant areas of judgment will require robust governance and control process which are transparent to auditors and external stakeholders Page 5 EY IFRS 9 impairment banking survey 06

7 . IFRS 9 project status and governance Status of classification and measurement program Banks are still in the design phase for policy, BMA and SPPI: Policy Business model assessment Bottom-up SPPI testing* Efforts invested seem proportional to the impact banks expect at transition on their balance sheets, with around a quarter of banks having started bottom-up SPPI testing back to the contracts A few banks referred to open questions remaining on technical interpretation of IFRS 9 (e.g., on certain prepayment options) Banks have focused less on data, systems, operating model and control framework: 5 6 These workstreams have a significant dependency on the financial impact assessment (BMA and SPPI) Systems Operating model and control framework Design early Design advanced Build 5 6 Deploy & test 6 Live Most banks are still in the process of performing a data gap analysis new data fields to track term features for BAU governance around SPPI, market data required for the valuation of portfolios transferring to fair value through profit or loss (FVTPL), new data points for statutory and regulatory disclosure requirements Embedding BMA and SPPI in BAU processes will require enhancing trade capture and new product approval processes, as well as the governance around central storage of standardized contracts. Other processes impacted include structuring, impairment, valuation, hedge accounting and reporting *SPPI testing back to individual contracts Page 6 EY IFRS 9 impairment banking survey 06

8 . IFRS 9 project IFRS 9 hedge accounting 0 Apply IFRS 9 hedge accounting in full Apply IFRS 9 hedge accounting for all hedges except for portfolio fair value hedges Most banks will not apply IFRS 9 hedge accounting in 0: As they are awaiting for further clarity and progress on the International Accounting Standards Board (IASB) macrohedge project, and in the context of the wider cost reduction agenda, banks are cautious not to spend too much, too early, on their IFRS 9 hedge accounting program. This may all need to change once the outcome of the macro project is known Banks also see little benefit in applying the standard, and a few stated their hedging strategy is mainly macro hedging New disclosure requirements: Banks are required to apply new disclosure requirements for hedge accounting from January 0 onwards, irrespective of whether they will apply IFRS 9 hedge accounting then. New disclosures relate to risk management strategy for each hedged exposure, the amount, timing and uncertainty of future cash flows, the effects of hedge accounting on financial position and performance, and the option to designate a credit exposure at fair value through profit or loss (FVTPL) - Remain on IAS 9 hedge accounting as long as permitted - Focus current efforts on developing new disclosure requirements of IFRS 7 - Remain on IAS 9 hedge accounting for 0 - May adopt IFRS 9 hedge accounting earlier than when required Page 7 EY IFRS 9 impairment banking survey 06

9 . IFRS 9 project Governance and coordination Governance Coordination Finance - Finance for C&M and hedge accounting - Risk for impairment - Finance for C&M and hedge accounting - Finance and risk for impairment Other Exclusively centralized implementation Implementation governance There is a clear trend for the implementation efforts of C&M and hedge accounting to be predominantly led by Finance Impairment continues to be jointly governed by both Finance and Risk. The joint leadership between Finance and Risk of the impairment project will require banks to clearly articulate the roles and responsibilities across both functions within a detailed target operating model Additional success factors in the project implementation include the integration of interactions with other functional units such as operations, IT and affected business units A number of banks are working to enhance their existing control framework in this area due to the intricacies within the IFRS 9 end-to-end processes 6 Centralized implementation with a few exceptions depending on portfolios and systems - Centralized guidelines, coordination and IFRS 9 ECL models - Decentralized implementation - Centralized guidelines and coordination - Decentralized implementation including IFRS 9 ECL models Centralized implementation efforts Project coordination is generally aligned with the operating model, with half of the participants applying an exclusively centralized implementation, where possible, depending on IT infrastructure, geographical constraints and local regulatory requirements Banks have indicated that they need a robust communication strategy in terms of disseminating information to the relevant department or business units when using a centralized implementation Page EY IFRS 9 impairment banking survey 06

10 . IFRS 9 project Parallel run Calculating IFRS 9 numbers in a tactical platform Live Currently running H 06 H 06 Q 07 Q 07 Q 07 Q Full parallel run Tactical platform Calculating IFRS 9 numbers within a tactical platform will be an imperative first-step in determining the impact of the standard. It will also provide useful information to participants on data issues and system limitations that will require resolution prior to the parallel run Over half of the participants have indicated that they will have the ability to generate impairment figures within a tactical platform by the end of 06, with just under one third of participants reporting that they currently have this platform in place Parallel run Live Q 07 Q 07 Q 07 Q 07 Other Only a third of participants plan to complete at least a 9 month parallel run of IAS 9 and IFRS 9 figures. This figure has declined compared to our previous survey where more than half of participants intended to perform at least months of parallel run The majority of the large European and global banks will have completed at least one reporting period under a parallel run in 07 Page 9 EY IFRS 9 impairment banking survey 06

11 . IFRS 9 project Budget m m < 5m 5m < 5m 5m < 5m 5m < 0m 0m < 60m 60m < 00m 00m < 5m 5m /Undisclosed Expected total budget C&M, impairment, and hedge accounting As expected, there was a very wide range of budgets for IFRS 9, with an element of uncertainty on the actual spend for implementation While larger and more global banks are spending more than smaller locally focused banks, the scale of cost does not appear linearly related to the size of the bank. For example, a larger bank may have budgeted 5x the cost, but the balance sheet is only 0x larger. This strikingly illustrates the cost differences with implementing more complicated solutions While impairment remains the costlier element of IFRS 9 implementation, there is a wide range of cost estimates for C&M. The banks with the larger C&M budgets are more advanced with their programs, suggesting many of the banks with low cost estimates for C&M may be underestimating the cost and complexity m m < 5m 5m < 5m 5m < 5m 5m < 0m 5 6 Classification & Measurement Impairment 0m < 60m Hedge Accounting 60m < 00m /Undisclosed 5 Page 0 EY IFRS 9 impairment banking survey 06

12 . IFRS 9 project Budget (cont d) BAU yearly budget post-implementation 0 Over half of our survey respondents have not fully quantified the cost on their BAU budget, with many banks explicitly indicating that it was premature to form such a view BAU costs are expected to be mainly related to additional resources to run models and for licenses (if applicable). Some banks mentioned there will be incremental costs for control functions, reporting, and disclosure, but these are yet to be quantified 0 6 m m < 5m Page EY IFRS 9 impairment banking survey 06

13 . IFRS 9 project Synergies Existing programs leveraged for IFRS 9 aggregation and reporting systems improvement (e.g. BCBS 9) Use Pillar III disclosures as a starting point Roll IRB out for portfolios previously under a standardised approach Preparation for ECB TRIM Others, including regulatory requirements Processes linked for IFRS 9 governance FinRep and CoRep Use stress testing platform to incorporate FLI Capital planning / ICAAP / Forecasting Risk management reporting Asset Liability Management (ALM) Fair Value Measurement processes There are many areas within most banks that are either being leveraged for the benefit of IFRS 9, or subject to the same governance process to ensure consistency and accuracy In terms of leveraging existing or developing processes for IFRS 9, the two areas most cited are: roll out of IRB models, which are generally used as a starting point for IFRS 9 impairment calculations, but then developed and operated as separate models data improvement initiatives, such as BCBS 9 or data warehouse developments For governance, the three areas most cited are: stress testing to ensure consistency (where appropriate) in the forward looking information being used; the IFRS 9 platform will generally be leveraged for stress testing going forward capital planning and ICAAP to ensure some level of integration between IFRS 9, risk management framework and overall capital management FinRep, given the template information that requires reporting has an element of IFRS 9 While governance may cover and promote consistency (where applicable) between IFRS 9, regulatory capital calculations, and stress testing, there was little indication that the processes themselves have a high degree of overlap Total in both graphs is more than 6 as most banks selected more than one option Page EY IFRS 9 impairment banking survey 06

14 . IFRS 9 project Effective interest rate (EIR) approach Plan to revisit current EIR approach 7 Yes Existing short-cuts will be removed EIR impacts both the discounting in IFRS 9 models as well as the measurement of interest revenue The majority of banks noted they will not revisit their current EIR approach as a result of the introduction of IFRS 9 on the grounds that there is no material difference with their current short-cuts, however some are keeping their options open (see slide 5) As part of its implementation of IFRS 9, a bank will need to consider whether approximations used in determining EIRs under IAS 9 remain appropriate given the more significant role that discounting has in measuring impairment under IFRS 9 (e.g. discounting of cash shortfalls that may occur a number of years into the future) Of the banks that have determined they need to remove short cuts (such as straight line approximation), some noted that they will do this due to emphases on the discounting and income that will be magnified under IFRS 9 No No material difference between existing short-cuts and what is required by IAS 9/IFRS 9 N/A EIR approach already fully compliant with IAS 9/IFRS 9 Page EY IFRS 9 impairment banking survey 06

15 . IFRS 9 project IT architecture and governance Number of ECL calculators in IFRS 9 solution One ECL calculator More than one ECL calculator ECL calculator Most participants are setting up a single, central platform as ECL calculator, supporting various IFRS 9 models banks plan to use more than one ECL platform depending on portfolios (wholesale vs. retail, IRB vs. standardized); legal entities, and purpose (ECL calculation for stages and vs. stage ; live production vs. simulation) Frequency of BAU IFRS 9 ECL calculation Frequency of BAU ECL calculation More frequently than monthly Most banks plan a monthly IFRS 9 ECL calculation and management information, which is seen as good practice for credit monitoring purpose banks plan to run an ECL number on a quarterly basis, aligned with their external reporting 6 Monthly Quarterly A couple of banks plan to calculate the ECL numbers more than once per month, essentially in a pre-production environment, with not full governance applied on the number for the intra-month calculation(s) Page EY IFRS 9 impairment banking survey 06

16 . IFRS 9 project IT architecture and governance (cont d) Documented data taxonomy Most banks have identified data gaps early on in their programs (e.g. lifetime PD curves at origination, certain unlikely to pay factors, date of transfer to stage ) in order to develop data remediation solutions. gaps not only include data limitations but also data reconciliation issues (i.e. data mismatch) 6 7 Most banks are in the process of documenting a data taxonomy (e.g. detailed glossary) and establishing sourcing processes in an iterative manner Yes No In progress used for significant deterioration and ECL calculation 9 Current month-end data Older data 6 Mix current/older data, depending on entities and portoflio Use of month-end data for stage allocation and ECL calculation Although participants aim to use data as current as possible, especially for year-end reporting, many recognize that they will have to use lagged data in order to meet financial close timelines Most banks planning to use prior period data referred to a time-lag of one month; a few referred to a time-lag between a couple of days and months In a number of cases it depends on portfolios (corporates data being less timely), legal entities, and type of data Lagging data would be adjusted for significant events occurring prior to closing (e.g. in credit and volumes), and related governance is still in development. This is an area of subjectivity and will likely gain external audit attention Page 5 EY IFRS 9 impairment banking survey 06

17 . Impact assessment Disclosure of impact to the market Date of public disclosure of quantitative impact assessment Most banks are undecided on timing of public disclosures which is surprising due to the recommendations produced by the Enhanced Disclosure Task Force (EDTF) Quarter undecided Q Many banks are planning to disclose a quantitative impact assessment in Q 07 (mainly banks with October yearends) or at beginning of 0 in their 07 annual reports. This is at the latest point recommended by the EDTF Despite the fact that regulators around the world are starting to request a preliminary IFRS 9 impact analysis, banks are still refining their models and policy decisions in key areas (e.g. significant deterioration) in order to improve the reliability of the numbers and full governance in place before a public quantitative impact is disclosed Q Q Q Prior to G- CRAECL 07 - Quarter undecided H 07 H 07 During 0 Page 6 EY IFRS 9 impairment banking survey 06

18 . Impact assessment Proportion of good book expected to be stage 0-5% % % % % % % > 50.% Not disclosed or calculated 0-5% % % % % % % > 50.% Not disclosed or calculated 0-5% % % % % % % > 50.% Not disclosed or calculated Total Wholesale 5 Retail Total Most banks estimate that no more than 0% of their performing exposures will be assessed as stage. However, some expect considerably more. The portfolios will vary in quality, but it is possible that some of the differences arise from differences in approach The majority of banks who cannot calculate this number are banks that have not yet been requested to submit a quantitative impact assessment to their regulator Wholesale As would be expected, it is generally SME loans that show the greatest percentage of exposures in stage, followed by corporate loans The bank with >50% of good book expected to fall into stage plans to recognise a lifetime expected credit loss at initial recognition due to undue cost and effort undertaken in determining significant increase in credit risk Retail Overall, credit cards show the largest proportion of exposures in stage in this portfolio For other types of retail lending there is diversity. For instance, in most cases where the calculation had been made at this level, a relatively low percentage of the mortgage portfolio was assessed to be in stage, but for some of the banks mortgages were one of the larger contributors Page 7 EY IFRS 9 impairment banking survey 06

19 . Significant deterioration Drivers All banks consider using a combination of quantitative and qualitative drivers structured as primary and secondary drivers, plus backstops. The primary driver is meant to be the most early indicator and is generally based on a relative measure while the others cover more obvious (absolute) signs of deterioration such as forbearance or delinquency Primary drivers - the survey reveals two major trends: Use of IFRS 9 lifetime probability of default (PDs), guided by scores and ratings, OR Use of Basel adjusted M PD or scores (often equivalent) for retail exposures or use of ratings for corporate exposures Secondary drivers and backstops - Answers are homogeneous around using forbearance, delinquency (mainly 0 days past due (DPD) as a backstop), watchlists or a combination of those. As expected, watchlists are less frequently mentioned for retail compared to corporates. Also retail watchlists concepts tend to largely overlap with forbearance, delinquency as well as fixed levels of scores or PDs Use of variation of M PD - Banks will have to demonstrate that they are not missing any significant increase in risk of a default beyond months. This may require further adjustments based on macroeconomic forecasts. However, these indicators are still considered very relevant as they are well understood and have been used and tested for a long time Use of variation of lifetime PD - The obvious challenge on transition is to have data available at origination date for existing portfolios (including forward looking information). Some banks mentioned they would have to use proxies on transition (Basel scores, through the cycle (TTC) PDs, latest information available or lending policy cut-offs) Use of variation of ratings Ratings are considered more forward-looking by nature as they involve more expert judgment based on a wider range of information, including more prospective information (borrower s financials, sectorial information etc.) and look beyond a month horizon. Depending on their calibration, they may also require demonstrating the associated PDs reflect current circumstances and reasonable forecasts Transitional vs. strategic approach - The challenges faced on transition are obviously less significant for banks using Basel scores or PD although some issues may still arise depending on when the IRB models were built. It remains to be seen whether in the longer term, the development and increasing use of lifetime PD curve (including forward-looking elements) may result in more convergence towards the use of this more sophisticated quantitative measure Page EY IFRS 9 impairment banking survey 06

20 . Significant deterioration Drivers retail IFRS 9 lifetime PD Adjusted IFRS 9 m PD IFRS 9 m PD Basel PD Ratings or scores Fixed level of score or PD Watchlist Forbearance 0 DPD Delinquency other than 0DPD LCR simplification Other 9 9 Retail Primary Secondary Backstop NB. Other risk metrics generally correspond to scores or ratings. Most banks selected this indicator in combination with m PD NB. The graph and commentary refer to the secured retail portfolio as banks displayed similar trends across all retail portfolios NB. bank does not have retail portfolios Primary drivers Around half of the banks converge towards using IFRS 9 lifetime PDs as their main indicator. Although not always specified, this generally means comparing the average marginal PDs over the residual term to what was expected at inception for the same period The other half intends to use Basel M PD or scores (generally equivalent measures). Among these banks, a few banks mentioned they would use fixed levels of score (i.e. an absolute measure). They consider this is a good proxy for assessing deterioration and easier to operationalise A few banks mention using IFRS 9 M PD (with one mentioning adjustments to reflect longer term expectations). Compared to Basel PD, these measures would not be calibrated through a cycle Secondary drivers and backstops Answers are very homogeneous around using forbearance, delinquency (mainly mentioned as a backstop), watchlists or a combination of those Retail watchlists are more mechanical than for corporate exposures and tend to largely overlap with forbearance, delinquency as well as fixed levels of scores or PDs Given the number of entities aligning non-performing forborne assets and stage, forbearance may mostly be relevant in stage during probation periods, when the assets are no longer in default Page 9 EY IFRS 9 impairment banking survey 06

21 . Significant deterioration Drivers wholesale Wholesale Primary drivers Half of the banks intend to use the same primary indicator for corporate compared to retail IFRS 9 lifetime PD Adjusted IFRS 9 m PD IFRS 9 m PD Almost all banks using lifetime PD for retail also use it for corporates (with one bank mentioning it will use watch list instead) Then, rating deterioration is more commonly mentioned Basel PD Ratings or scores Fixed level of score or PD Watchlist Forbearance 0 DPD Delinquency other than 0DPD Ratings are considered more forward looking than retail scores. Rating calibration can be through the cycle (with some stickiness around better ratings) and may therefore require demonstrating they don t miss out any deterioration compared to forward looking lifetime PDs. However, they remain a favoured criteria as they build on a robust process and are aligned with risk management Watchlists are mentioned by a few banks (with one combining it with adjusted PD). Almost all these banks are currently enhancing their watchlist process (see slide ). Those banks tend to use ratings as secondary driver LCR simplification Other Primary Secondary Backstop Secondary and backstop drivers Watchlist is the most common secondary drivers. Most banks using lifetime PD or rating changes as primary indicators use watchlist as secondary or backstop criteria. Interestingly, majority of the banks are not proposing any enhancement to the watchlist process NB. The graph and commentary refer to the corporate portfolio as banks displayed similar trends across all wholesale portfolios Forbearance is mentioned similarly to retail While 0 DPD is generally considered as a lagging indicator for corporate, this is mentioned as a backstop Page 0 EY IFRS 9 impairment banking survey 06

22 . Significant deterioration Drivers debt securities Debt securities Primary drivers IFRS 9 lifetime PD Adjusted IFRS 9 m PD IFRS 9 m PD Basel PD Ratings or scores 7 0 Participants illustrate more variety in their primary approaches to determining significant deterioration for debt securities. This could be due to their being less attention drawn to the application of this concept to this product category The main difference to be noted for debt securities is the use of the low risk simplification, which is mentioned as a primary indicator by 0 banks. The other banks are evenly split between using lifetime PD and ratings Fixed level of score or PD Watchlist Forbearance 0 DPD Secondary and backstop drivers Fewer secondary drivers and backstops were mentioned, reflecting simpler approaches; the main secondary drivers and backstops remain watchlist, forbearance and delinquency Delinquency other than 0DPD Deterioration credit spreads LCR simplification Other Primary Secondary Backstop NB. banks did not respond for debt securities Page EY IFRS 9 impairment banking survey 06

23 . Significant deterioration Watchlist enhancement process Plan to enhance watchlist process 6 0 Yes No The increasing reliance on the watchlist process as an indicator of significant deterioration has resulted in the review of whether there are enhancements required to this process in order to allow for appropriate monitoring of credit The majority of participants have indicated the existing watchlist processes will not be enhanced under IFRS 9. This is largely due to existing watchlist processes being identified as fairly robust and an ongoing review of the adequacy of the process outside of IFRS 9 This point is particularly relevant for a few banks using it as their primary indicator. Almost all of them intend to enhance their watchlist process around the following areas: reinforcing the link with ratings, demonstrating the relative measure, reinforcing the link with governance and processes and making sure it is symmetrical. Things may also have to be removed such as concentration risk For the other banks, the comments show that the issue is still being investigated. Consistency of watchlist definition across the bank is clearly a key point Interestingly one bank mentions that credit analysts will have access to the modeling output which will bring more quantitative expertise into their analysis Page EY IFRS 9 impairment banking survey 06

24 . Significant deterioration Use of probation periods for transfer out of stage Plan to apply probation periods before returning assets from stage to stage Yes Yes for certain triggers No 0 Over a third of the banks declare that they intend to apply probation periods to their stage assets In most cases, the probation period only applies when the transfer to stage has been initially triggered by indicators other than the primary trigger, especially: forbearance (potentially influenced by the European Banking Authority (EBA) definitions when relevant), with a few banks referring to a -year probation; delinquency (minimum cure periods are considered in relation with the 0 DPD criteria), with a few banks referring to a to 6 month probation; and, watchlist, with a few banks referring to a 6-month probation period for instance This means that probation periods would not apply to assets transferred in stage for other reasons (based on PD models for instance) Banks mention that probation periods are currently under review and are part of the calibration exercise which will continue over the coming years Another third of the banks state that they won t apply any probation period, especially as probation periods are already included in ratings and PD models and will therefore naturally apply via the use of ratings or PD indicators Finally another third of the banks is still undecided. This is consistent with the answers collected on the articulation with EBA definitions Page EY IFRS 9 impairment banking survey 06

25 . Significant deterioration Use of probability weighted approach in significant deterioration 0 Use of a probability weighted approach to reflect multiple scenarios in the significant deterioration assessment 5 In December the Impairment Transition Resource Group (ITG) agreed that banks should incorporate more than one forward looking scenario when assessing significant increases in credit risk, if there is a non-linear relationship between the different scenarios and the risk of default. However, they need not consider only quantitative approaches There are differences of view as to whether probabilityweighted multiple scenarios approach will be used to determine significant deterioration. Some of those banks who intend not to do so report that they do not expect the effect to be significant Others believe that the effect of non-linear relationships is already reflected in corporate credit gradings or watchlists While over a third of the banks do intend to use multiple scenarios to assess significant deterioration, most have yet to decide how they will do this Yes No Page EY IFRS 9 impairment banking survey 06

26 . Significant deterioration Trigger approach when using a relative criterion 9 Trigger approach when using a relative criterion A B C D E A. PD delta (e.g. + xx bps) B. A PD multiple (e.g. PD*xx), or a number of notches for scores or rating (e.g. x notches out of a XX scale) C. A combination of delta and multiple approaches - with both criteria to be met to trigger a transfer to stage D. A combination of delta or multiple approaches, with at least one criteria to be met to trigger a transfer to stage E. Other Total is more than 6 as banks selected two options as they will adopt a different trigger approach for their wholesale and retail portfolios 7 6 Calibration remains very much work in progress as banks are currently testing different sets of triggers Given the exponential shape of the PD curve relative to ratings, the calibration of a significant deterioration has to take into account the fact that PD multiples in very good ratings only represent very small movements in absolute amounts when the same multiple applied to bad ratings can represent a significant change in the absolute amount of PD Therefore, most banks will use a combination of additive and multiple factors to assess significant deterioration. Only a few of them consider having only one factor as a sufficient trigger and these banks recognize that the triggers would have to be set differently for retail and corporate, but also by portfolio, country and taking into account the level of risk at origination Banks using ratings or scores as their primary drivers will use a number of notches (which is close to using PD multiple approaches). They generally intend to adjust the number of notches depending on the position in the rating scale. Depending on the calibration, this could have similar effects to using the low credit risk simplification and introduce subjectivity Note that those using lifetime PD may also try and calibrate PD changes based on the rating scale to try and map PD variation with ratings changes that are considered meaningful from a risk management standpoint Page 5 EY IFRS 9 impairment banking survey 06

27 . Definition of default IFRS 9 and Basel definition of default Align IFRS 9 and Basel definition of default 6 Fully align definitions of default used for IFRS 9 and for regulatory capital calculation purposes Align definitions except for certain exposures where 0DPD are used for regulatory capital calculation purposes N.B. In Europe, the CRR limits the exception to exposures secured by residential or SME commercial real estate in the retail exposure class, as well as exposures to public sector entities Plan to rebut the 90 days past due presumption 6 Yes No Alignment of regulatory definitions Almost all banks intend to align their IFRS 9 definition of default with the regulatory definition, with only few banks intending to use some exceptions relating to DPD Rebuttable presumptions For those banks using a 0 DPD trigger under Basel, full alignment therefore implies rebutting the IFRS 9 90 DPD presumption which some banks are not willing to do as this could be perceived as a lower quality IFRS 9 implementation Many banks also mentioned that their regulatory definition of default is evolving towards a more systematic use of the 90 DPD trigger decreasing the need to rebut the IFRS 9 presumption. Some banks stated that shifting from 0 to 90 days often has little impact for mortgage loans as contagion would cause default ahead of the 0 DPD trigger, whilst others said on the contrary this impact was significant The few banks that intend to rebut the 90 DPD presumption under IFRS 9 will limit it to very specific portfolios (credit cards in Canada, mortgages in the UK, public sector, sovereigns, institutions or under exceptional circumstances for others) 7 Page 6 EY IFRS 9 impairment banking survey 06

28 . Definition of default IFRS 9 definition of default and EBA definition of non-performing IFRS 9 definition of default and EBA definition of nonperforming Align IFRS 9 definition of default and EBA definition of non-performing Do not align IFRS 9 definition of default and EBA definition of nonperforming This question only applies to banks in the scope of EBA regulations Amongst these European banks, more than half intend to fully align their IFRS 9 definition of default with EBA definitions of non-performing exposures for both forborne and nonforborne exposures. One bank mentioned that alignment would be restricted to non-forborne exposures and another one on the contrary will limit it to forborne exposures Such alignment involves the application of probation periods for forborne exposures as the EBA definition requires that forborne exposures remain non-performing for at least one year following the extension of forbearance measures. Other conditions to cease being non-performing include the discontinuation of the credit-impaired and default classification, full repayment (according to the original or when applicable the modified conditions) likely to be made, and no past-due amount by more than 90 days The Basel consultative guidelines on the definitions of nonperforming and forbearance published earlier this year may result in increased harmonization across countries outside Europe. Therefore the mapping between these concepts and IFRS 9 will remain an area of attention Page 7 EY IFRS 9 impairment banking survey 06

29 . Definition of default Forbearance in stage and definition of stage Forbearance in stage Definition of stage All performing forborne exposures based on the EBA definition (including year probation period) Non-performing forborne assets based on the EBA definition Forborne exposures that are not covered by the EBA definition All forborne exposures in Stage No link between forbearance and staging All defaulted assets (based on the bank's interpretation of default for IFRS 9 purposes) Only defaulted assets for which a loss is expected (LGD > 0) Assets which have not defaulted yet (PD < ) Other Forbearance This question only applies to banks in the scope of EBA regulations Amongst the banks affected by this question, banks intend to fully align stage with EBA definition of performing forborne exposures. These banks generally declare that they will also align EBA non-performing status and IFRS 9 default and stage Such alignment involves the application of quite long probation periods in stage as the EBA definition requires that forborne exposures remain flagged as performing forborne for at least years (after being at least year flagged as non-performing forborne) Only banks said they were not linking forbearance and stage allocation. This mainly reflects that forbearance is an indicator already embedded in ratings or other risk categories and is not considered as a distinct indicator Definition of stage Almost all banks intend to align stage with their interpretation of default for IFRS 9 purposes. Only one bank intends to exclude from stage defaulted assets which are fully guaranteed banks intend to classify exposures which are not yet defaulted in stage. This generally relates to certain types of forbearance or the application of probation periods Total is more than 6 as banks selected two options Page EY IFRS 9 impairment banking survey 06

30 . Definition of default Cross asset contagion Retail Cross-asset contagion across facilities with the same borrower Corporate Crossasset contagion across facilities with the same borrower Corporate Crossasset contagion across counterparties within the same group Cross-asset contagion for default Even though there is general alignment between IFRS 9 and regulatory definitions, banks reveal diversity in the application of the regulatory definition of default across countries Under the Basel definition, default shall reflect unlikeliness to pay or delinquency of a particular obligor observed at the level of the banking group in full. For retail exposures however, the definition of default can be applied at the level of a particular facility, rather than at the level of the obligor. Around half of respondents intend to apply cross-asset contagion for retail exposures while two thirds intend to do so for corporate exposures with the same borrower. limitation is often mentioned as the reason for not applying cross-asset contagion overall The number for corporates drops when it comes to applying cross-asset contagion to borrowers within the same group, with more banks being undecided and more going for contagion only in certain circumstances Finally, many banks do not see applying contagion as an automatic process, but may involve expert judgment and management review in their decision Yes to all facilities/counterparties Yes for some facilities/counterparties No Total for retail is less than 6 as bank does not have a retail portfolio Page 9 EY IFRS 9 impairment banking survey 06

31 5. Measurement of expected credit loss (ECL) Probability of default Use of Basel PD as a starting point Corporate Retail 5 Yes No Majority of banks leveraging Basel PDs The majority of the banks planning to leverage on Basel models, will be re-calibrating the PD value, aiming at converting the value to a point-in-time (PiT) estimate and including forward looking expectations Although almost all of the banks state that they will leverage existing Basel PDs, most banks will also have to develop a new range of models to get full coverage of their books 0 Totals are less than 6 as some banks do not have a corporate or retail portfolio Calculation of PiT PD for IFRS 9 when Basel PD not available Point-in-time estimates already in place in the organisation New point-in-time models Expert judgement approaches Historical loss rates or roll rates Not applicable full Basel PD coverage Other Calculation of PiT PD for IFRS 9 For the development of new PiT PD models where Basel PDs are not available as a starting point, the approaches will vary between portfolios within the respective banks A number of banks are developing new models based on roll rate, historical loss rates and, or expert judgment. The use of expert judgment will require enhanced documentation and support for the methodology A number of banks that already have PiT PD estimates in their current suite of credit risk models intend to leverage these existing models Page 0 EY IFRS 9 impairment banking survey 06

32 5. Measurement of expected credit loss Probability of default lifetime Estimating lifetime PD for IFRS 9 purposes 6 5 Corporate 5 Retail To get to a lifetime PD, transition matrices is the single most popular approach but most banks will use different approaches (e.g. transition matrices and extrapolations and, or lifetime default observations) for different portfolios or products A number of the banks stating that they will use external data will do so in the context of applying external transition matrices or filling gap in their internally built transition matrices. Some will also use external data to build lifetime PD curves that are then used to extrapolate internal data Given the high degree of judgment in choosing the model approach and the incorporation of different forward looking elements, we expect to see granular and detailed disclosures allowing comparability across banks Transition matrices Models based on default observations over the life of the instrument Extrapolations External data Totals are more than 6 as some banks have responded with multiple approaches Page EY IFRS 9 impairment banking survey 06

33 5. Measurement of expected credit loss Loss given default Use of Basel LGD as a starting point Leveraging Basel LGDs Corporate Retail 5 Yes 5 9 No Totals are less than 6 as some banks do not have a corporate or retail portfolio The majority of banks plan to leverage their Basel LGDs and adjust them to make them IFRS 9 compliant The most common adjustments are: Removal of downturn calibrations and floors to make LGDs unbiased Discounting Application of lifetime and forward looking estimates Recovery costs Consideration of derivatives Calculation of LGD for IFRS 9 when not using Basel LGD Corporate 9 7 Retail Historical loss rates Historical loss rates calibrated to PIT and forward-looking expectations Expert judgment Not applicable full Basel LGD coverage Other Developing IFRS 9 LGDs For those who do not plan to leverage Basel LGDs or for portfolios where Basel LGDs are not available, most banks intend to derive PiT LGDs from historical loss rates, commonly combined with macro economic variables and, or expert judgment 5 Totals are less than 6 as some banks do not have a corporate or retail portfolio Page EY IFRS 9 impairment banking survey 06

34 5. Measurement of expected credit loss Loss given default (cont d) Rate to discount LGDs Rate to discount LGDs Corporate Retail Effective interest rate (EIR) The majority of banks intend to use the EIR or a proxy thereof to discount LGDs, the proxy often being the contractual rates Contractual rate Other Contractual rates are used on the basis of the effect not being materially different from the EIR and often in line with the banks current practices in interest recognition, however some banks will perform further assessments before deciding (see slide ) Reflecting the evolution of collateral value in the LGD Totals are less than 6 as some banks do not have a corporate or retail portfolio Use of time-dependent LGDs When it comes to the use of time-dependent LGDs roughly a third of the banks are still undecided Secured Corporate Secured SMEs Yes No Other Some banks suggested they will only use time dependent LGDs where there is an established link between LTV and LGD e.g. mortgages The most common approach for modeled portfolios is to adjust the LGD against the indexed collateral value, with most referring to housing price index for their retail exposures, while the LGDs of non-modeled exposures will be adjusted through expert judgment Secured Retail Totals are less than 6 as some banks do not have a corporate, SME or retail portfolio Page EY IFRS 9 impairment banking survey 06

35 5. Measurement of expected credit loss Forward looking information 7 Number of discrete forward looking scenarios Probability-weighted approach to incorporate forward looking information One scenario with an overlay approach Run multiple scenarios, calculate one ECL based on the weighted average of credit risk factors Run multiple scenarios, calculate an ECL for each, and probability-weight them Modelled approach Sources of forward looking scenarios Total is more than 6 as most banks have selected multiple sources of information Forward looking scenarios The number of discrete forward looking scenarios required to calculate ECLs is still up for debate. Most banks are considering a fixed number of scenarios, with participants either: intending to use three scenarios, representing a best, worst and base case; or remaining undecided on the number of scenarios, although most banks are choosing between using, or scenarios of the banks intending to use only one scenario plan to reflect non-linearities in the loss distribution by means of a management overlay. Another considers that the effect of alternative scenarios is already reflected in their base scenario. This approach will need to be properly documented Nearly all banks are planning to use centralized internal scenarios for consistency with business planning, stress testing and IFRS 9. Centralized scenarios often are built up using external information Probability weighting of forward looking scenarios The most common proposed approach is to derive separate ECL numbers for each scenario and apply a probability weighting to derive the final number Some banks are using only one scenario, either the most likely scenario or a pre-weighted scenario. Some will add an overlay while others did not mention an overlay or remain undecided. Their approach will need to be properly documented Page EY IFRS 9 impairment banking survey 06

36 5. Measurement of expected credit loss Forward looking information (cont d) Forward looking perspective - estimation of IFRS 9 PD 7 Corporate 5 Retail 9 Pre-calibration process Post-calibration process Combination of postcalibration process and management overlay Combination of precalibration process and management overlay Total for corporate is less than 6 and equal to 6 for retail; some banks do not have a corporate or retail portfolio, while some banks selected approaches in combination Forward looking perspective - estimation of IFRS 9 LGD Corporate Retail 9 Keep the LGD static Use stress testing processes Management overlay using expert judgement Other Totals are equal to 6 however some banks do not have a corporate or retail portfolio, while some banks selected two approaches in combination Probability of default The majority of the banks are incorporating the forward looking information at the calibration stage, adjusting the PD calibration target to reflect the macro/customer specific variations. We also note that a minority of banks are looking at incorporating macro-economic factors directly into the PD models All banks are planning to use some form of management overlay in fact, a small portion are planning to rely exclusively on management overlays to adjust the model outcomes Banks do not seem to have finalized their governance process around the management overlays, which is a process that has proven to be time consuming e.g. the bank wide stress test exercises Loss given default Forward looking LGD continues to be an area where banks are less advanced on their thought process compared to PDs, evidenced by almost a third of the banks still being undecided on the approach The majority of banks that have selected Other state that they will use some form of macro economic variable based forecast, for examle within the individual components in the LGD model, such as collateral values for secured loans Banks that state that they will use the existing stress testing processes will only stress the LTV values, affecting secured LGD and leaving the unsecured LGD flat Page 5 EY IFRS 9 impairment banking survey 06