PRO-CYCLICALITY IMPLICATIONS OF IFRS9 AND THE RWA FRAMEWORK

Transcription

1 PRO-CYCLICALITY IMPLICATIONS OF IFRS9 AND THE RWA FRAMEWORK Brad Carr, Senior Director, Regulatory Affairs Jonathan Ng, Policy Advisor, Regulatory Affairs Hassan Haddou, Policy Advisor, Regulatory Affairs

2 Slide number Title slide Slide number Title slide 3 Executive summary 13 Bank 2A retail 4-5 This analysis 14 Bank 2B diversified 6 Section 1 IRB banks 15 Detailed impact analysis 7 Bank 1A retail 16 Section 3 observations and conclusions 8 Bank 1B whole Observations and conclusions 9 Bank 1C cyclical wholesale 19 Section 4 appendix 10 Bank 1D diversified Methodology and assumptions 11 Detailed impact analysis 23 Reg EL shortfall sensitivity analysis 12 Section 2 SA banks 24 Other studies 2

3 At the 2009 G20 Pittsburgh Summit 1 pro-cyclicality was identified as a regulatory priority. Some initiatives have attempted to address this (CCyB and stress testing) The introduction of IFRS 9 (based on expected loss (EL)), in conjunction with the current risk-weighted asset (RWA) framework in a downturn period, may have unintended implications, in particular greater capital volatility Based on our IRB and SA bank stress scenario, IFRS9 provisioning can erode CET1 between ~ bps, while RWA inflation can erode CET1 between ~ bps. When aggregating both impacts, this has a doubling effect in CET1 erosion The absence of capital relief for holding a higher level of provisions brings additional pressure to banks CET1 ratios in a stressed environment A key consequence will likely be a tightening in credit availability, which prolongs the downturn and this feedback loop would have procyclical effects In addition, a new statement in the Basel III standards suggests ratings for RWA may become even more procyclical than the current framework, which is beyond the modeling performed in this analysis 2 Policy makers are urged to consider the loss-absorption capacity of the additional provisions in the capital framework In this analysis, a key assumption made is the classification of Stage 2 loans (see slide for more detail). We note that individual banks will have different assumptions, in particular staging of loans, which will produce different impacts 1. Leaders statement of the Pittsburgh Summit September, 25th 29th 2009, page 8 2. Ratings systems should be designed in such a way that idiosyncratic or industry-specific changes are a driver of migrations from one category to another, and business cycle effects may also be a driver Basel III: Finalizing post-crisis reforms, December 2017, page 86 3

4 To illustratively demonstrate the pro-cyclicality implications of the new accounting requirements compounded with the current RWA framework during a downturn, we have modelled the CET1 capital ratio impacts for: IFRS 9 estimates in a benign (e.g. current) environment; and IFRS 9 and RWA inflation estimates in a stressed environment across different bank balance sheet scenarios, under IRB and SA approaches The IIF understands there are other areas in the regulatory framework that may have pro-cyclical implications (e.g. initial margin in derivatives), but have focused on the two major frameworks (i.e. IFRS 9 and RWA) to demonstrate the potential impacts on banks The model is based on several simplified assumptions (e.g. PD doubling stress scenario, staging of loans, calculation of lifetime PDs and constant Reg EL to accounting provisions shortfall) to estimate the Day 1 and stressed CET1 impacts Note that the model is not intended to precisely estimate CET1 impacts of IFRS 9 for banks, but to: provide an illustrative example of potential CET1 impacts on our hypothetical bank balance sheets; and highlight the implications of having higher capital volatility for banks 4

5 The model utilizes IRB credit risk data across a sample of banks Pillar 3 disclosures as a baseline to derive average PDs, LGDs, EAD and RWA density across the major Basel asset classes 1,2 The model assumes a severe downturn scenario for: EL provisioning to be approximately a 100% increase in PDs across all asset classes to estimate PIT PDs the RWA framework to be approximately a 75% increase in PDs across all asset classes to estimate a hybrid (PIT and TTC) approach 3 For more detail on our methodology and assumptions refer to the appendix (slides 20-22) The IIF intends to complete similar analysis on the CECL accounting requirements as a subsequent addendum in due course 1. The sample of banks used are six major banks from Europe, Canada and Australia 2. Major Basel asset classes are defined as residential mortgages, credit cards, other retail, corporates, sovereigns and financial institutions 3. Note that exposures are assumed to migrate commensurately to their scaled PD and therefore may be assigned higher risk weights 5

6 Section 1 IRB banks

7 Bank 1A is predominately retail focused with 65% of its exposures in mortgages and 20% in credit cards and other retail 14% Bank 1A (Retail focused) 14% Bank 1A (Retail focused) - Stressed environment 12% (80 bps) 12% (150 bps) (200 bps) 10% 10% 8% 8% 6% 12.0% 11.2% 6% 12.0% 4% 4% 8.5% 2% 2% 0% CET1 (pre IFRS9 impact) IFRS9 impact CET1 (post IFRS9 impact) 0% CET1 (pre stress impact) RWA inflation IFRS9 impact CET1 (post IFRS9 impact) Day 1 CET1 impact of IFRS 9 is 80 bps because of long tenor mortgages Incremental CET1 impact stressing the IFRS 9 provisioning is 120 bps (i.e. total of 200bps) In conjunction with RWA inflation during a stress, the CET1 impact is 270 bps 7

8 Bank 1B is predominately wholesale focused with 50% of its exposures in corporates, 20% in sovereigns and 15% financial institutions 14% Bank 1B (Wholesale focused) 14% Bank 1B (Wholesale focused) - Stressed environment 12% (30 bps) 12% (90 bps) (100 bps) 10% 10% 8% 8% 6% 12.0% 11.7% 6% 12.0% 10.1% 4% 4% 2% 2% 0% CET1 (pre IFRS9 impact) IFRS9 impact CET1 (post IFRS9 impact) 0% CET1 (pre RWA inflation IFRS9 impact CET1 (post stress impact) IFRS9 impact) Day 1 CET1 impact of IFRS 9 is 30 bps Incremental CET1 impact stressing the IFRS 9 provisioning is 70 bps (i.e. total of 100bps) In conjunction with RWA inflation during a stress the CET1 impact is 160 bps 8

9 Bank 1C has the same portfolio composition as Bank 1B, however all of its corporate exposures are with specialized lending activities (income producing commercial real estate) 1 14% Bank 1C (Cyclical wholesale focused) 14% Bank 1C (Cyclical wholesale focused) - Stressed environment 12% (30 bps) 12% (150 bps) 10% 10% (110 bps) 8% 8% 6% 12.0% 11.7% 6% 12.0% 4% 4% 9.4% 2% 2% 0% CET1 (pre IFRS9 impact) IFRS9 impact CET1 (post IFRS9 impact) Day 1 CET1 impact of IFRS 9 is 30 bps Incremental CET1 impact stressing the IFRS 9 provisioning is 80 bps (i.e. total of 110bps) In conjunction with RWA inflation during a stress the CET1 impact is 230 bps 1. The SL-IPRE calculations are on the assumption that RWA is IRB-modeled and not under the Slotting Criteria 0% CET1 (pre RWA inflation IFRS9 impact CET1 (post stress impact) IFRS9 impact) 9

10 Bank 1D has a diversified portfolio with 48% of exposures in retail and 52% of exposures in non-retail Bank 1D (Diversified) Bank 1D (Diversified) - Stressed environment 14% 14% 12% (50 bps) 12% (90 bps) (150 bps) 10% 10% 8% 8% 6% 12.0% 11.5% 6% 12.0% 4% 4% 9.6% 2% 2% 0% CET1 (pre IFRS9 impact) IFRS9 impact CET1 (post IFRS9 impact) Day 1 CET1 impact of IFRS 9 is 50 bps Incremental CET1 impact stressing the IFRS 9 provisioning is 100 bps (i.e. total of 150bps) In conjunction with RWA inflation during a stress the CET1 impact is 190 bps 0% CET1 (pre RWA inflation IFRS9 impact CET1 (post stress impact) IFRS9 impact) 10

11 Detailed dollar breakdown of: (1) Day 1 CET1 impact of IFRS 9, (2) IFRS 9 provisioning in stress and (3) RWA inflation in stress Banks CET1 (after Reg EL adjustment) 1,2 ($m) IFRS 9 impact 3 ($m) CET1 (post IFRS 9 impact ($m) 4 RWA ($m) Bank 1A Base case 42,622 (2,867) 39, ,180 Bank 1A stressed environment 42,622 (8,221) 34, ,007 Change - (5,354) (5,354) 50,827 Bank 1B Base case 50,748 (1,154) 49, ,899 Bank 1B stressed environment 50,748 (4,521) 46, ,955 Change - (3,367) (3,367) 36,056 Bank 1C Base case 52,093 (1,401) 50, ,109 Bank 1C stressed environment 52,093 (5,280) 46, ,181 Change - (3,879) (3,879) 64,072 Bank 1D Base case stress 47,129 (2,100) 45, ,740 Bank 1D stressed environment 47,129 (6,449) 40, ,991 Change - (4,349) (4,349) 30, All banks are assumed to have a $2,000m regulatory EL shortfall to accounting provisions ( shortfall ). We note the shortfall varies with each individual bank and in some recent examples (i.e. Canadian banks) has absorbed most of the IFRS 9 CET1 impacts. In the appendix (slide 23) we have provided sensitivity analysis as to how the shortfall changes the day 1 IFRS 9 CET1 impact of our hypothetical banks. We further note that the shortfall will also change for individual banks in a stressed period, but have kept this constant for comparability across the benign and stressed scenarios. 2. All banks CET1($) starting point are back solved to achieve a 12% CET1 target ratio 3. IFRS 9 impact is the $ impact on CET1. The IFRS 9 impact is the increase in provisioning from the incurred loss approach and in excess of the regulatory EL to accounting provisions 4. All banks are assumed to already exceed their deferred tax asset (DTA) threshold (10% of CET1). As such, the additional DTA created by the higher provisions will be fully deducted from CET1 (i.e. neutralize any benefits from additional DTA on CET1 ratios). 11

12 Section 2 SA banks

13 Bank 2A is predominately retail focused with 70% of its exposures in mortgages 2, 20% in other retail, 5% in SME corporates and 5% in sovereigns 14% Bank 2A (Retail focused) 14% Bank 2A (Retail focused) - Stressed environment 12% (80 bps) 12% (60 bps) (160 bps) 10% 10% 8% 8% 6% 4% 12.0% 11.2% 6% 4% 12.0% 9.8% 2% 2% 0% CET1 (pre IFRS9 impact) IFRS9 impact CET1 (post IFRS9 impact) Day 1 CET1 impact of IFRS 9 is 80 bps Incremental CET1 impact stressing the IFRS 9 provisioning is 80 bps (i.e. total 160bps) In conjunction with RWA inflation during a stress the CET1 impact is 140 bps 1. Only two bank balance sheet scenarios are presented given that the balance sheets commonly observed in SA banks are akin to 2A and 2B examples 2. Approximately 70% of the mortgage exposures have a LTV ratios less than 80%; under stressed conditions, it is assumed that this proportion drops to 50% 0% CET1 (pre RWA inflation IFRS9 impact CET1 (post stress impact) IFRS9 impact) 13

14 Bank 2B has a diversified portfolio with 30% of its exposures in mortgages 2, 20% in other retail, 20% of exposures in SME corporates, and 20% in rated corporates and 10% of sovereigns. 14% Bank 2B (Diversified) 14% Bank 2B (Diversified) - Stressed environment 12% (70 bps) 12% (40 bps) (150 bps) 10% 10% 8% 8% 6% 4% 12.0% 11.3% 6% 4% 12.0% 10.1% 2% 2% 0% CET1 (pre IFRS9 impact) IFRS9 impact CET1 (post IFRS9 impact) Day 1 CET1 impact of IFRS 9 is 70 bps Incremental CET1 impact stressing the IFRS 9 provisioning is 80 bps (i.e. total 150bps) In conjunction with RWA inflation during a stress the CET1 impact is 120 bps 1. Only two bank balance sheet scenarios are presented, given that the balance sheets commonly observed in SA banks are akin to 2A and 2B examples. For comparison to our other scenarios, the 2B example relatively aligns to 1D 2. Approximately 70% of the mortgage exposures have a LTV ratios less than 80%; under stressed conditions, it is assumed that this proportion drops to 50% 0% CET1 (pre RWA inflation IFRS9 impact stress impact) CET1 (post IFRS9 impact) 14

15 Detailed dollar breakdown of: (1) Day 1 CET1 impact of IFRS 9, (2) IFRS 9 provisioning in stress and (3) RWA inflation in stress Banks CET1 1 ($m) IFRS 9 impact ($m) CET1 (post IFRS 9 impact ($m) RWA ($m) Bank 2A Base case 73,934 (5,042) 68, ,119 Bank 2A stressed environment 73,934 (10,585) 63, ,901 Change - (5,543) (5,543) 31,782 Bank 2B Base case 92,425 (5,693) 86, ,211 Bank 2B stressed environment 92,425 (11,776) 80, ,368 Change - (6,083) (6,083) 25, IFRS 9 impact is the $ impact on CET1. The IFRS 9 impact is the increase in provisioning from the incurred loss approach and in excess of the regulatory EL to accounting provisions 15

16 Section 3 Observations and conclusions

17 IRB Banks In summary, the estimated results show that: Day 1 CET1 impacts: ranges between 30-80bps Incremental CET1 impact of stressing the provisioning: ranges between bps Combined with RWA inflation, the CET1 erosion can range between bps under severe conditions SA Banks In summary, the estimated results show that: Day 1 CET1 impacts: ranges between 70-80bps Incremental CET1 impact of stressing the provisioning: ~80 bps Combined with the RWA inflation, the CET1 erosion can range between bps under severe conditions CET1 impacts for IRB and SA banks are more severe for retail focused banks in comparison to wholesale focused banks due to the longer maturities of retail products If rating systems in RWA become fully subject to business cycles (per the final Basel III standards 1 ), then we would have to revise our PD assumption up from 75% to 100%, further compounding these impacts 1. Ratings systems should be designed in such a way that idiosyncratic or industry-specific changes are a driver of migrations from one category to another, and business cycle effects may also be a driver Basel III: Finalizing post-crisis reforms, December 2017, page 86 17

18 In the event of a downturn, regulators may relax capital buffers (for e.g. CCyB) and/or regulatory minima, however this may give limited confidence to the market and its CET1 expectations for banks Impacts of this magnitude on CET1 ratios (even if still complying with regulatory minima) include potential downgrades, increasing banks funding cost and tightening credit availability in order to reduce RWA For example, if Bank 1A s CET1 minimum is 12.0% and during the downturn its CET1 ratio falls to 8.5%, regulators may relax the minimum to 8%. However, the market and its investors would be more concerned that its CET1 ratios eroded by ~350 bps, despite being compliant with regulatory requirements What considerations need to be made by policy makers? Review the treatment of the new provisioning requirements under the capital framework o o neutralization and/or addback to CET1 for the additional provisioning amounts in times of stress; and review the appropriateness of the current DTA threshold limit Further review the stability of the RWA framework continuation of TTC versus PIT nature of the RWA framework, which builds on the work the EBA and the IIF RWA Task Force has conducted 18

19 Section 4 - Appendix

20 DATA The model uses IRB credit risk data from a sample of banks Pillar 3 disclosures to derive a PD band matrix across the major Basel asset classes 1,2 Within each PD band of each Basel asset class, an average PD, LGD, EAD and RWA density is derived using the bank sample data PD bands are categorized using the standardized EBA PD bands Estimated maturity profile is derived from available bank sample disclosures Effective interest rates used for discounting EL in each asset class are based on approximated market rates KEY ASSUMPTIONS IRB All banks are assumed to have a constant regulatory EL shortfall to accounting provisions All banks starting point is assumed to already exceed their deferred tax asset (DTA) threshold (10% of CET1) All banks are assumed to have the same and constant level of other risk RWA (i.e. operational and market risk) All banks are assumed to have a target CET1 capital ratio of 12% Maturity profile is consistent across the bank balance sheet scenarios 1. The sample of banks used are six major banks from Europe, Canada and Australia 2. Major Basel asset classes are defined as residential mortgages, credit cards, other retail, corporates, sovereigns and financial institutions 20

21 The maximum residual maturity is 5 years for all exposures with the exception of mortgages, which assumes the majority of exposures are between 5 years and 10 years The PD band matrix is constant across the case studies All exposures with PDs less than 2.5% is assumed to be Stage 1, all exposures with PDs greater than 2.5% and less than 100% is assumed to be Stage 2 and exposures with PDs equal to 100% is assumed to be Stage 3 Lifetime PDs are assumed to have a marginal linear increase of 20% from 1yr PDs Cyclical wholesale bank balance sheet scenarios assumes corporate PDs are that of corporate IPREs PDs The model uses regulatory LGDs for calculating EL Discount factors (based on the asset class effective interest rate) are used to calculate EL for stage 1, stage 2 exposures (lifetime exposures) The model assumes a severe downturn scenario for: EL provisioning to be approximately a 100% increase in PDs across all asset classes to estimate PIT PDs. Thus a scaling factor of 2 is applied to the base case PDs RWA framework to be approximately a 75% increase in PDs across all asset classes to estimate a hybrid PD (PIT and TTC). Thus a scaling factor of 1.75 is applied to the base case PDs 1 1. Note that exposures are assumed to migrate down the credit curve in line with their assigned average PD 21

22 KEY ASSUMPTIONS SA For each asset class, SA and IRB have similar profiles in terms of credit quality and maturity The model applies for each exposure the appropriate SA risk-weight (current Basel rules) For a given exposure, IFRS 9 calculations are assumed to be identical for SA and IRB banks Regulatory EL shortfall does not apply to SA banks All banks starting point is assumed to already exceed their deferred tax asset (DTA) threshold (10% of CET1) The model assumes a severe downturn scenario consistent with the one used for IRB banks The model assumes that under stressed conditions, the proportion of mortgages with a LTV less than 80% drops from 70% to 50% of the overall portfolio 22

23 Each of the hypothetical banks shown in our analysis have a constant regulatory EL shortfall ( shortfall ). However, in reality the shortfall varies with each individual bank and in some recent examples (i.e. Canadian banks reporting in November 2017) this has partially absorbed the day 1 CET1 impacts for IFRS 9 provisions The below table provides sensitivity analysis as to how different shortfall levels change the day 1 IFRS 9 CET1 impact for our hypothetical bank balance sheets Banks CET1 (pre Reg EL adjustment) ($) Reg EL shortfall ($) IFRS 9 impact ($) CET1 (post Reg EL adjustment and IFRS 9 impact ($) RWA ($) CET1 (%) CET1 change from base case (bps) Bank 1A Base case 44,622 (2,000) (2,867) 39, , % - Bank 1A no Reg EL shortfall 42,622 0 (4,867) 37, , % (60) Bank 1A higher Reg EL shortfall 46,622 (4,000) (867) 41, , % 60 Bank 1B Base case 52,748 (2,000) (1,154) 49, , % - Bank 1B no Reg EL shortfall 50,748 0 (3,154) 47, , % (40) Bank 1B higher Reg EL shortfall 54,748 (4,000) (846) 49, , % 10 Bank 1C Base case 54,093 (2,000) (1,401) 50, , % - Bank 1C no Reg EL shortfall 52,093 0 (3,401) 48, , % (50) Bank 1C higher Reg EL shortfall 56,093 (4,000) (599) 51, , % 20 Bank 1D Base case 49,129 (2,000) (2,100) 45, , % - Bank 1D no Reg EL shortfall 47,129 0 (4,100) 43, , % (50) Bank 1D higher Reg EL shortfall 51,129 (4,000) (100) 47, , % (0) 23

24 While there are still a couple of years until CECL will be effective for GAAP compliant banks around the world, IFRS 9 was effective from 1 January 2018 There have been studies completed by Deloitte, EY and the rating agencies projecting the day 1 CET1 impacts of IFRS 9 to be relatively low (ranging between bps) 1 which agrees with our estimates. This can be partially due to the regulatory EL shortfall to accounting provisions cushioning 2 However, the pro-cyclicality nature of IFRS 9 is an area of concern for the market as highlighted in the European Systemic Risk Board paper 3 and Barclays study 4 The Barclays study estimates a bps CET1 erosion due to additional provisioning in a typical downturn Note that Barclays assumes a typical downturn would experience 5x increase in PDs to normalized levels, which is more severe compared to our model assumptions 1. Sixth Global IFRS Banking Survey, Deloitte, May 2016 Moody s Survey: Capital impact of IFRS 9 adoption will be modest for most banks, Moody s, March 2017 EY IFRS 9 Impairment Banking Survey, EY, August IFRS 9 impact cushioned by regulatory buffer, Fitch Ratings, March Financial stability implications of IFRS 9, European Systemic Risk Board, July IFRS 9 Bigger than Basel IV, Barclays, January