Reserve Bank Capital Review Industry Forum

Reserve Bank Capital Review Industry Forum 21 February 2019

Contents 2 • Opening remarks • Competition protocols • Timeline of Capital Review and next steps • Risk appetite framework and calibration • Afternoon tea • Quality of capital • Changes to IRB framework • Other issues

3 Opening remarks: Adrian Orr

4 Competition protocols

5 Timeline of Capital Review and next steps

6 Objectives and Principles • Promote the maintenance of a sound and efficient financial system by setting the most appropriate capital adequacy framework for New Zealand • According to principles of Capital Review, capital should…

1. Readily absorb losses ahead of creditors and depositors
2. Take account of the relative risk of banks’ exposures
3. Not vary substantially between different methods for determining capital requirements
4. Reflect the risk inherent in NZ financial system and the RBNZ’s regulatory approach, and therefore the outcome should be conservative relative to peers
5. Be practical to administer, minimise unnecessary complexity, and account for home-host regulatory relationships
6. Be transparent to enable effective market discipline

7 What we’re proposing  Tier 1 capital of 16/15 percent of RWA  Recalibrate internal models to around 90 percent of standardised  Capital buffers tied to Escalating Supervisory Response framework • Enhanced role for capital buffers (including countercyclical, DSIB) • Leverage ratio – disclosure and minimum (4/3 percent of exposures) • 5 year transitional period 10.5% 14.9% 13% 18% 17% 0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20 Current minimum Current levels Current levels (with IRB changes) Proposed (systemically important) Proposed (non-systemically important) % Tier 2 Tier 1 Conservation buffer DSIB buffer Countercyclical buffer Voluntary buffer

8 What we’re proposing 11.8% 16% 14.1% 15% $35bn $48bn $5.5bn $5.9bn 0 10 20 30 40 50 60 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 Current outcome (comparable basis) Proposed minimum Current outcome Proposed minimum Current outcome Proposed minimum Current outcome Proposed minimum Tier 1 ratio Tier 1 capital Large banks Small banks

9 Clarity on regulator-regulated relationship • More efficient model approval process • Escalating Supervisory Response (ESR) – greater clarity about supervisory actions with a graduated buffer approach

10 Timeline – near term • Another industry forum penciled in Auckland (March) • Analytical note on Risk Appetite Framework (March) • Consultation period extended (3 May) • Open to further discussions with industry during the consultation period, including bilateral meetings if desired • Release of final decisions, accompanied by Regulatory Impact Statement (Q3)

11 Further work • Consultation on further elements of the framework:  Near term: – Identification framework for systemically important banks (March) – Internal model change process (workshop with affected banks)  Later in 2019 and beyond: – Mutual capital instrument – Leverage ratio design (if decision to proceed) – Escalating Supervisory Response framework and trigger points – Strategy for setting the countercyclical capital buffer – Operational risk framework (pending APRA finalisation) – Tier 2 (subject to current consultation) • Dovetail with changes to Banking Supervision Handbook as Capital Review decisions are implemented

12 Proposed transition Quarter / year Proposal Q3 2019 • Confirm final Capital Ratio decisions • New AT1 instruments need to meet revised standards Q4 2019 • Start of transition to higher ratios • Implement changes to IRB framework (floor / scalar) 2020 • Dual reporting • Revised Standardised Measurement Approach (Op Risk) • Leverage ratio requirements • Transition to higher capital ratios 2021 • Transition to higher capital ratios 2022 2023 2024

13 Proposed transition 8.5% 10% 11.5% 13% 14.5% 16% 8.5% 9% 10.5% 12% 13.5% 15% 13.4% 11.6% 14.2% Q3-2018 end-2019 end-2020 end-2021 end-2022 end-2023 Q3-2018 end-2019 end-2020 end-2021 end-2022 end-2023 Systemically important banks Non systemically important banks Tier 1 requirement including prudential buffers Current Tier 1 capital ratio Tier 1 capital ratio after changes to IRB framework

14 Risk appetite framework and calibration

Contents 15 • The risk appetite framework • The quantitative modelling • Output impacts. What basis for claims of “win-win”?

16 Context for the policy • The Basel standards are a minimum, local context matters • Financial crises have significant economic and social impacts • Established conventions in the academic literature about the relationship between capital, crises and output • RBNZ’s soundness and efficiency mandate • Risk appetite is central to calibrating financial regulation (Basel III, Solvency II in insurance)

17 Four lenses on capital adequacy Stress testing International financial crisis data Risk modelling of NZ banks ‘Optimal’ capital modelling (RBNZ and others) Risk Appetite Framework: • Soundness objective • Capital sufficient to retain the confidence of creditors when subject to an extreme (notional 1 in X) shock • Efficiency objective • Subject to meeting soundness objective, does the capital requirement maximise expected economic output?

18 Conventional expression of the policy problem Source: Firestone, Lorenc and Ranish (2017). Finance and Economics Discussion Series 2017-034, Federal Reserve Board. Relationship of Benefits, Costs and Optimal Capital Level (K*)

19 Marginal costs and benefits of capital Capital Ratio % GDP Marginal Benefit Marginal Cost

20 Conventional relationships deliver an output peak K ratio Output relative to capital low high Output Stability is increasing

21 The RBNZ’s illustration takes one further step - maps output against stability Less stable More stable Expected economic output (GDP) Financial stability Stability and output combination implied by current minimum requirements Capital requirements that maximise expected output (but the level of stability may still be too low) Trading lower expected output for more stability (though expected output still higher than current settings)

22 Risk Appetite Framework • Soundness objective  Capital sufficient to retain the confidence of creditors when subject to an extreme (notional 1 in X) shock • Efficiency objective • Subject to meeting soundness objective, does the capital requirement maximise expected economic output?

23 Four lenses on capital adequacy Stress testing International financial crisis data Risk modelling of NZ banks ‘Optimal’ capital modelling (RBNZ and others)

24 Quantitative modelling

25 Quantitative modelling - introduction

1. Optimal capital literature
2. Loss modelling (Value at Risk model) • Focus was on the relationship between capital and the probability of a crisis • Aim was to produce a range of capital ratios that would deliver market confidence in the face of large shocks, after taking allowance of provisions • Required a quantitative value for the risk appetite, we used 1 in 200 years

26 International financial crisis data Study Capital needed to cap the probability of a crisis at 0.5% Ratio measurement Required amount BCBS (2010) CET1 (Equity) / RWA 10% to 13% (Bank of England restated as 16%+ Tier 1 Ratio) Brooke et al. (2015) (Bank of England) Tier 1 Capital / RWA 14% to 16% Firestone et al. (2017) (Federal Reserve) Tier 1 Capital / RWA 17%+ Dagher et al. (2016) (IMF) Equity / RWA 15% to 23% required to avoid 85% of the banking crises during the GFC

27 Loss modelling approach • Model NZ system as a single bank (precedent in RBNZ modelling, going back to Basel III model in 2012) • Asymptotic Single Risk Factor (ASRF) model (x2 streams) • Some of the decisions required: – What loss indicators? – What banks to include in the historical sample? – How, if at all, to incorporate overseas info? – How, if at all, to incorporate IRB model inputs?

28 Risk modelling approaches Stream A Stream B Historical NPL Historical and model data Simple average all NZ banks Weighted average NZ 99.5% confidence 99.5% to 99.7% confidence Strict solvency Failure level of capital Reference to overseas Reference to NZ IRB Stress test results for LGD Stress test results for LGD R value 0.16 to 0.4 R value 0.24 to 0.32 Tier 1 = 14.5% to 16% Tier 1 = 15.5% base case

29 Stream A output illustration Monte Carlo analysis Confidence level = 99.5% 1.5% < PD < 3% 35% < LGD < 50% 0.20 < R < 0.40 Failure threshold = 0% Median capital ratio = 15.2% Mean capital ratio = 15.5%

30 Stream A output illustration R 16% 24% 30% 35% 40% LGD = 40%, Confidence = 99.5% PD 1.5% 8.1 11.4 14.0 16.3 18.7 2.0% 9.7 13.6 16.6 19.3 22.1 2.5% 11.1 15.5 18.9 21.9 25.1 2.8% 11.9 16.6 20.2 23.3 26.6 3.0% 12.4 17.2 21.0 24.2 27.6

31 Stream B output illustration Figure 3: Monte Carlo analysis (PD 1-2%, LGD 30-40%, R 24-32%)

32 Output impacts – what basis for “win-win”?

33 Marginal costs and benefits of equity Capital Ratio % GDP Marginal Benefit Marginal Cost

34 Output assessment • Estimated impact of policy on lending margins 20 bps to 40 bps • Net benefit of policy proposal = 𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝐶𝑜𝑠𝑡𝑠 𝑜𝑓 𝐶𝑟𝑖𝑠𝑖𝑠 × 𝑅𝑒𝑑𝑢𝑐𝑡𝑖𝑜𝑛 𝑖𝑛 𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑜𝑓 𝑐𝑟𝑖𝑠𝑖𝑠 − 𝐿𝑜𝑤𝑒𝑟 𝑠𝑡𝑒𝑎𝑑𝑦 𝑠𝑡𝑎𝑡𝑒 𝑜𝑢𝑡𝑝𝑢𝑡 𝑑𝑢𝑒 𝑡𝑜 ℎ𝑖𝑔ℎ𝑒𝑟 𝑙𝑒𝑛𝑑𝑖𝑛𝑔 𝑚𝑎𝑟𝑔𝑖𝑛𝑠 • Overseas research suggests the PV impact of this change in lending rates on long run GDP could be -0.16% to -0.33% of GDP • Will factor in bank cost estimates to final cost benefit analysis / Regulatory Impact Statement (RIS)

35 Capital and output Source Effect on lending rates (basis points) Effect on GDP (basis points) Federal Reserve Board (2017), Full Pass Through 6.9 -7.4 Federal Reserve Board (2017), Half Pass Through 3.4 -3.7 BCBS (2010) 13 -9 Bank of England (2015) 5 to 10 -1 to -5 Federal Reserve Bank of Minneapolis (2016) 5.7 -5.7 RBNZ meta-study (2016) 5 to 8 -1 to -5

36 The meaning of “win-win” Less stable More stable Expected economic output (GDP) Financial stability Stability and output combination implied by current minimum requirements Capital requirements that maximise expected output (but the level of stability may still be too low) Trading lower expected output for more stability (though expected output still higher than current settings)

37 International comparisons

38 International comparisons can be a misleading basis for assessing capital policy • Multiple sources are available that report relative capital levels of banks globally • BCBS, S&P, EBA • But comparative results are an unreliable guide to capital adequacy • Our policy aim is to calibrate to absolute NZ risk and an NZ risk tolerance, not peer relative outcomes • Comparative results can reveal little about relative regulatory policy settings in different countries • Actual outcomes reflect different Pillar 1/Pillar 2 philosophies, “side letters”, voluntary capital choices etc., and not just nominal regulatory minima

39 Basel III Monitoring Report (BIS) – Dec 2017 data • Limitations: NZ application of Basel framework on average more conservative than other jurisdictions, dataset includes banks with less comparable business models to NZ

40 International comparisons – S&P RAC • Limitation: Standard and Poor’s Risk-Adjusted Capital methodology relies on S&P’s economic risk assumptions (peer group: 4 largest NZ banks, large retail and commercial banks in each country; NZ (p) = pro forma at 17% Tier 1) 0 2 4 6 8 10 12 14 16 18 0 2 4 6 8 10 12 14 16 18 FI NO CZ NZ (p) SE HK DK PL IE NL MY IL AU AT SG NZ S&P Risk￾Adjusted Capital ratio (%) Range Median

41 International comparisons - Basel 0 5 10 15 20 25 30 35 40 45 0 5 10 15 20 25 30 35 40 45 SE FI DK CZ NO PL NZ (p) IE HK NL SG MY NZ AU IL AT Basel Tier 1 capital ratio (%) Range Median • Limitation: Basel framework applied differently across jurisdictions (peer group: 4 largest NZ banks, large retail and commercial banks in each country; NZ (p) = pro forma at 17% Tier 1)

42 International comparisons - Leverage • Limitation: Leverage ratio doesn’t control for different risk profiles (peer group: 4 largest NZ banks, large retail and commercial banks in each country; NZ (p) = pro forma at 17% Tier 1) 0 2 4 6 8 10 12 0 2 4 6 8 10 12 NZ (p) HK PL NO IE SG MY IL CZ NZ FI AT DK AU NL SE Leverage ratio (Tier 1/Exposure) (%) Range Median

43 Afternoon tea

44 Quality of capital

Contents 45 • Recent global policy focus on gone-concern • Why the Reserve Bank has a focus on going-concern • Why we don’t want CoCos in the capital framework • Ordinary share capital for banks structured as mutual societies

46 Recent global policy – gone concern • FSB’s TLAC – Principles and Term sheet released Nov 2015 • EBA MREL – Final Report Dec 2016 • APRA’s Tier 2 proposal – consultation paper released Nov 2018

47 Rationale for Gone Concern capital • Banking crises do great harm • Bailing-in creditors once a bank is non-viable helps contain a crisis, and reduces fiscal risk • Gone concern capital instruments operate to deliver bail-in

48 Why we prefer going-concern • Banking crises do great harm – preventing them makes sense • The case for increasing going-concern requirements is sound • Bailing-in creditors can be problematic: • Potentially lengthy, costly, uncertain outcome • Particularly difficult when hosting a systemic bank • We have options to increase capital that other countries may not have (a reasonable flow of earnings that can be retained)

49 Why we reject CoCos • Our focus is prevention, not resolution. History shows that CoCos may not be Tier 1 material: • Suspending dividends makes a bad situation worse • May not trigger in time (i.e. when the bank is viable) • In NZ CoCos have been primarily sold to parents (fill-in for equity) • History shows they may have fiscal risk (not reliable for resolution) • Uncertainty generated by ‘circuit breakers’ for NZ-issued CoCos sold domestically (not reliable for resolution)

50 Ordinary share capital for mutuals • “Full voting rights” = one vote per member, not share • BS2A’s requirements for distributions and allocation of net surplus assets – dividend policies and society rules the solution ? • Building Societies Act 1965 – Section 11 raises a potential question about the permanence of issued share capital • We are committed to working with the sector to facilitate the issuance of ordinary shares.

51 Changes to IRB framework

52 Objectives • Preserve the risk differentiation and capital allocation benefits of internal models, where we think these exist • Where internal modelling doesn’t offer net tangible benefits, more efficient to use standardised approaches • For a given underlying level of risk, internal models and standardised approaches should produce broadly comparable capital levels • Disparity of outcomes we see in key areas (e.g. mortgages) hard to justify on the basis of different underlying risks • Put risk mitigants in place to allow for more efficient processes • Streamline the model approval process • Reduce reliance on model interventions, overlays

53 How to balance competing objectives? Reduce gaps between IRB and standardised Preserve a risk-sensitive capital framework Reduce gaps across IRB banks Output floor tied to standardised Adjust IRB calibration (scalar) Combination of output floor and scalar More intensive monitoring, enforcement of IRB (e.g. regular benchmarking)

54 Quantitative Impact Study • QIS provided information on how current IRB outcomes compare to standardised approach, used to calibrate output floor and IRB scalar • Currently, IRB approach produces average of 76 percent of standardised 65 75 69 103 76 0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140 Sovereign and Bank Corporate Mortgage Other retail Total IRB RWA as % of Standardised Range of IRB banks Total

55 Proposals • Reduce scope of internal models • Standardise sovereign and banks, operational risk modelling • Recalibrate IRB approach closer to standardised outcomes • Increase scalar so that average IRB outcome is around 90 percent of standardised • No loss of risk differentiation or capital allocation benefits • Output floor of 85 percent of standardised, supported by robust dual reporting • Acts as a backstop, though not expected to be binding • More level playing field for comparable risks • Consulting on calibration (~90 percent outcome, compared to current 76)

56 A more level playing field • Current Tier 1 capital per $100 of mortgage lending, Tier 1 capital at proposed minimum ratios (estimate using public data only) 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 ANZ ASB BNZ Westpac Kiwibank Other banks $ $ Current outcome Proposed minimum (estimate)

57 Other issues