Financial Policy Memorandum: Leverage Ratio and Countercyclical Capital Buffer Requirements

MEMORANDUM FOR FSC FROM Financial Policy (Matthew Brunton) MEETING DATE 9 October 2019 SUBJECT Leverage Ratio Requirements FOR YOUR Decision It is recommended that the Committee:

1. Note as part of the fourth Capital Review consultation paper the Reserve Bank consulted on the inclusion of leverage ratio requirements (with a preferred option of both disclosure and minimum requirements)
2. Note that we received submissions in May this year. Of those that commented on the leverage ratio, few were supportive and none of the banks supported including leverage ratio requirements
3. Note that in light of the views put forward in submissions, we now propose to provide three options to FSC for leverage ratio requirements: i. As proposed – disclosure requirements and minimum requirements (4% for IRB banks, and 3% for standardised banks) following APRA methodology, ii. Simplified requirements – disclosure and minimum requirements (3% for all banks) and offering a simplified methodology (when consulting), and iii. No requirements – no leverage ratio requirements in the capital framework
4. Note on balance our (FP’s) preferred option is to have no leverage ratio requirements (option three). This is due to the views and additional information provided by submissions suggesting the benefits of option 1 are not as large as initially believed.
5. Agree to one of the options above for the leverage ratio requirements. Purpose
6. This paper presents the background to the proposed leverage ratio requirements as well as submitters views, and asks for a decision on the inclusion of leverage ratio requirements. Background
7. As part of the Capital Review, the Reserve Bank consulted on the inclusion of leverage ratio disclosure and minimum requirements (in the fourth consultation paper: How much capital is enough?). The stated preferred option was for both disclosure and minimum requirements, with a 3% minimum leverage ratio for standardised banks and a 4% minimum leverage ratio for IRB banks.
8. The preferred option was based on advice provided in the 24 August FSO. This advice noted that both the benefits and costs of leverage ratio requirements were relatively low given other proposed changes in the Capital Review (see #7683677). There was also a more detailed discussion at the 21 August BSG (see #7679593).
9. In May this year we received submissions for the fourth Capital Review paper. With regards to the leverage ratio requirements, banks were generally opposed to adopting leverage ratio requirements citing the additional compliance costs they would impose and low benefits they provide.

2 Arguments for (and against) leverage ratio requirements 5. There were two general aims of the leverage ratio. The first was to mitigate the costs of deleveraging in the system, and the second was as a backstop to the risk-based regime. Given the high average risk-weight of bank lending in New Zealand, generally high Tier 1 Ratios of banks, and other back-stop measures proposed in the Capital Review (namely ‘levelling the playing field’ and dual reporting) these benefits were seen as relatively low. 6. As part of the Capital Review, the Reserve Bank said that it would take the Basel and APRA standards as a starting point and deviating where it made sense to do so.1 Adopting leverage ratio requirements as proposed would align with the APRA standards except in that the requirement for IRB banks is set at 4% rather than 3.5%. Adopting a leverage ratio was initially thought to provide benefits to bank funding by providing a comparable metric that was standard international practice. 7. However, leverage ratio requirements also intersects with other aspects of banking regulation; namely the risk-sensitivity of capital ratios through capital requirements and market discipline. However, again due to the high risk-weights and Tier 1 ratios of banks in New Zealand, as well as the current ease of for market participants in constructing leverage ratios, these costs were seen as quite low. There are also additional compliance costs for banks from calculating leverage ratios (including for internal auditing and attestation). 8. In addition to this, there is an interaction with minimum capital requirements and the Escalating Supervisory Response (ESR) in that a higher minimum leverage ratio effectively can increase the point of bank regulatory failure (if it is reached before the minimum capital ratio) and potentially require ESR actions sooner. As such, part of the discussions around the leverage ratio was ensuring that it would not undermine the capital ratio requirements (and by extension the ESR). 9. A final key consideration by FSO was the concerns raised by the IMF in the past around New Zealand not implementing leverage ratio requirements.2 Given the more conservative treatment elsewhere in the capital framework the IMF still provided a “compliant” status for capital adequacy in the 2017 FSAP, however this was on the backdrop of the upcoming consultation of leverage ratio requirements and Basel not intending to fully implement the leverage ratio until 2018. 1 See ‘The calculation of risk weighted assets. Response to submissions’ paragraph 26. 2 In the 2017 FSAP, the IMF described the lack of leverage ratio requirements as “The main conceptual divergence from the Basel framework” (p20).

3 Submitter views 10. The appendix provides a summary of arguments for leverage ratio requirements with views from detailed submissions added in red. 11. One member of the public was in favour, Martin Lubberink was sceptical (in part from the accounting standards) and Ian Harrison was opposed. 12. Genworth commented that “Ideally, a leverage ratio would not be part of the [international] capital framework.” and that “Comparing the capital strength of individual financial institutions by the leverage ratio will give a misleading picture.” However, they also note the advantages of a leverage ratio in that it would:  “provide valuable information about system-wide leverage”,  strengthen market discipline in that it raises questions over banks which have used risk-weighting methods to lower capital requirements,  provide a safe-guard against the risk-based approach, and  be consistent with APRA and Basel standards. Genworth concluded that it favours disclosure over disclosure and implementation, but also sees merits in implementation in order to align internationally. 13. The domestic banks, ANZ, ASB, and Rabobank did not support the leverage requirements, generally regarding the leverage ratio as redundant given the high Tier 1 capital proposals. In particular:  ASB said the proposals “would unnecessarily create additional administrative burden for banks and the Reserve Bank with little or no prudential benefits.”  Kiwibank noted that the proposed capital ratio requirements would already reduce leverage, that an unrealistic risk-weight is required before a leverage ratio becomes binding, and that they did not believe there was any additional information for investors (noting disclosure statement information).  ANZ stated that the extra value of a leverage ratio did not appear material in New Zealand, and that they “have not found that the lack of a reported leverage ratio negatively affects offshore investor perceptions”.  Rabobank said “We appreciate the need to be consistent with international￾best practice, but in our view the RBNZ already has adequate policy tools to measure the build-up of leverage in the system and at individual bank level.” They did not support new metric unless necessary and noted the concept of the leverage ratio not being binding in “normal times” was not well explained. 14. The external experts did not focus on leverage ratio requirements (Ross Levine noted that the discussion was already strong and had no comments to add on the analysis).

4 Options (Accommodating submission views) 15. We propose three options for leverage ratio requirements: i. As proposed – disclosure requirements and minimum requirements (4% for IRB banks, and 3% for standardised banks) following APRA methodology, ii. Simplified requirements – disclosure and minimum requirements (3% for all banks) and offering a simplified methodology. This would be less complex and reduce compliance costs. However, it would be less aligned with international standards and possibly would not provide much more information than already exists in disclosure statements, and No requirements – no leverage ratio requirements in the capital framework at this point. This would be the simplest option as it would not require any changes to the current framework or any further consultation work on the leverage ratio. That is, we effectively maintain the status quo of not requiring a leverage ratio. 16. In light of the submissions, we now view the benefits of leverage ratio requirements as lower than initially identified. As the benefits outlined in the 24 August FSO were already relatively low (due to the New Zealand context of conservative risk-weights, high Tier 1 ratios and other proposed backstop measures in the Capital Review), our preferred option is to not have any leverage ratio requirements in the capital framework for now (option three). 17. An alternative option of having more simplified requirements (option two) is also provided. However, this approach would reduce the benefits of international comparability, as the methodology would be less aligned with APRA and Basel methodology, and it may provide little to no additional information than what is provided in disclosure statements. On the other hand, it could address points raised from the IMF in the past, although it is not certain that the lack of a leverage ratio would result in non-compliance for an FSAP (it could be seen as ‘regulation tailored to national circumstances’). 18. However, we believe a more pragmatic solution will be not to formally adopt any leverage ratio requirements at this point in time. If this were to develop into a material FSAP non-compliance in the future then the issue could be re-visited. 19. There was also discussion at the Banking Steering Group about only requiring disclosure requirements rather than both minimum and disclosure requirements. However, the underlying issues in submissions focused on leverage ratio requirements in general. Our (FP’s) view is that the marginal difference between minimum and disclosure requirements compared to only disclosure requirements is negligible except that the latter would not fully align with APRA and Basel standards (e.g. we expect compliance costs to be similar in both cases). As such, were the RBNZ to adopt leverage ratio requirements our preference is that it includes both minimum and disclosure requirements. Recommendation 20. It is recommended that the Committee decide on one of the three options stated in paragraph 15. 21. On balance, our preferred option is to have no leverage ratio requirements (option three) as our view is the benefits of a leverage ratio are lower than initially believed.

5 Summary of the key arguments for and against a leverage ratio Supporting arguments Opposing Arguments Avoids and mitigates the costs of deleveraging  A leverage ratio can capture systemic risks that are not reflected in risk-weights.  A leverage ratio places ceiling on leverage which can limit the severity of the deleveraging process during a crisis  The risk-based proposals will already cause leverage to fall in the system (Kiwibank). Acts as a backstop  Can limit how much risk-weights may fall, relative to exposure, from regulatory arbitrage.  Only focuses on the portfolio level, and as such is not as effective for the very risky bank profiles.  The Capital Review will also include backstops (output floor, dual reporting).  Would require unreasonably low risk-weights for it to become effective (Kiwibank). Interaction with capital ratios  Banks seem to be currently in the space of a constraining capital ratio rather than a leverage ratio. This will likely be exacerbated by the Capital Review.  It is possible that a leverage ratio requirement can be calibrated to be only constraining in rare ‘special-case’ occurrences.  Any risk-taking behaviour occurring from a lack of risk-sensitivity would naturally result in a constraining capital ratio and return risk-sensitivity.  If a leverage ratio is constraining it could impact the risk sensitivity of bank portfolio decisions. Interaction with market participants  There already exists market constructed leverage ratios.  Evidence suggests that capital ratios are more associated with stock prices than leverage ratios for simpler banks in more robust capital frameworks.  A regulatory leverage ratio would be internationally aligned and comparable ratio, helping banks raise funding from overseas.  One non-bank submitter (Genworth) argued that a disclosed leverage ratio would provide valuable information.  A regulatory leverage ratio may undermine market discipline associated with capital ratios.  Disclosure statements provide enough information for investors that there is little to gain from reported leverage ratios (Kiwibank).  At least one bank (ANZ) has indicated that they do not perceive a benefit from reported leverage ratios in raising funds from overseas. Compliance costs  Using the APRA methodology would lower compliance costs for IRB and standardised banks.  There may be some synergy with dual reporting in the exposure measure calculation for IRB.  There may be additional costs associated with assurance and auditing.  Would add unnecessary administrative burden for both banks and the Reserve Bank (ASB).

MEMORANDUM FOR FSC FROM Financial Policy (Maisie Prior) MEETING DATE 9 October 2019 SUBJECT The Countercyclical Capital Buffer – feedback from submitters, external experts, and considerations FOR YOUR Information It is recommended that the Committee:

1. Note that as part of the Capital Review we have consulted on setting the Countercyclical Capital Buffer (CCyB) to 1.5% for all banks as part of the 15/16% Tier 1 capital ratio.
2. Note that submitters are mostly in favour of having a CCyB as part of the framework, but question the setting of the CCyB, and raise concerns about how the CCyB will work in practice and with other jurisdictions.
3. Note that the external experts have expressed concerns over the CCyB, particularly in terms of its effective operation and operational complexities. They suggest that there is a case for the CCyB to be rolled into the rest of the prudential capital buffer.
4. Discuss the feedback from submitters and external experts, and the policy considerations raised, so that the forthcoming decision paper on the CCyB can reflect the issues of interest for FSC. Purpose
5. To outline the views raised by submitters and external experts on the 1.5% early set Countercyclical Capital Buffer (CCyB) proposed in the fourth Capital Review consultation paper, and to present the key considerations for decisions on the CCyB. Background
6. In the fourth Capital Review consultation paper, we consulted on having an ‘early set’ CCyB, set at 1.5%, as part of the 16% Tier 1 capital ratio. The proposed CCyB would have a positive value at all times, except after a financial crisis. The early set component of the CCyB would be built up automatically once the banking system has returned to profitability and the economy is recovering from the effects of the financial crisis.
7. The rationale for proposing the 1.5% calibration for the CCyB was an aim to balance the macroprudential objective of supporting lending, against the risk that capital buffers will need to be drawn on to absorb subsequent further losses. In the consultation paper, we noted that a 1.5% CCyB could potentially support up to 10% of lending during a crisis (as an upper limit of the lending that it could support).
8. In the consultation paper, we also noted that taking into account international evidence in addition to our own analysis, a higher calibration than 1.5% for the early CCyB would mean that banks could operate at capital ratios that do not meet our soundness objectives. Banks that fall within the adjusted regulatory buffer during a crisis would be subject to escalating supervisory response (ESR).
9. We committed to publishing our approach to setting the CCyB in more detail alongside the final policy position for the Capital Review. We stated that this would include proposals for how quickly the CCyB would be increased following a financial crisis, supervisory

2 expectations that would be in place once the buffer is cut, and indicators to guide both the release of the buffer and increases beyond the early set CCyB.

1. Introduction
2. This paper outlines the views expressed on the CCyB proposal, both by submitters and external experts. While submitters were generally in favour of having a CCyB as part of the capital framework, there were mixed views on the setting of the CCyB. Submitters also raised questions around how a CCyB would work in practice. The External Experts expressed doubts around operationalising the CCyB, especially as part of the proposed prudential capital buffer, and questioned the CCyB’s role in the capital framework.
3. This has led Financial Policy to open up the possibility of reconsidering the CCyB’s role as part of the capital stack. This paper presents some considerations for the Committee reflecting these views.
4. FSC previously agreed to using the early strategy for setting the CCyB in order to increase the usability of Through-the-Cycle (TTC) capital buffers, and also agreed that the neutral level of the CCyB will be calibrated based on the desired TTC settings (see #7778449 and minutes #7798058). Earlier papers have also outlined the pros and cons of using a CCyB compared to a conservation buffer (see #7304976), and the strategies for setting the CCyB (see #6994986).
5. To refresh the Committee, the difference between the two CCyB strategies are that a late CCyB raises capital buffers when risks are high, while an early CCyB begins building the CCyB towards a prescribed neutral level, even if risks are not elevated. The early (neutral) CCyB setting is effectively part of the TTC capital calibration that can be removed during a severe downturn. At the time, this choice for setting the CCyB at 1.5% reflected a preference for having more usable buffers following a crisis. The hope is that lowering the CCyB could support bank lending, which could reduce the cost of crises.
6. Views of submitters and external experts 2.1. Submitters Submissions on the CCyB were mostly supportive of its inclusion in the framework, but questioned whether an early or late set CCyB was appropriate…
7. Submitters that commented on the CCyB were mostly supportive of its inclusion. Some of the submitters who supported a CCyB wanted it to be set above zero only when risk conditions were elevated (late set), while others support the idea of setting the CCyB above zero in normal conditions (early set).
8. Both ANZ and ASB note the early set CCyB is inconsistent with the original intent of the CCyB, submitting it should be set to zero in normal conditions and only increased when risks were elevated (i.e. during periods of excessive credit growth). ASB noted that raising the CCyB in a period of slow credit demand risks dampening economic growth and limits options for the RBNZ to increase buffers during an expansionary credit period.
9. Some submitters (Kiwibank and Rabobank, among others) asked for a larger amount of the proposed capital buffer to be attributed to the CCyB. Kiwibank noted that moving to higher capital ratios gives the RBNZ the opportunity to put in place a larger CCyB than proposed (they suggest an increase in the CCyB to 2.5% offset by reducing the increase in the conservation buffer to 6.5%). They note that this may encourage credit growth in a recession when banks might otherwise be concerned about the track of future capital ratios. They also suggest that during a severe crisis, stress modelling suggests banks

3 typically lose 300-400bps of capital, and their own internal modelling supports that assessment. 13. Harbour Asset Management also note that there could be merit in having a flexible CCyB of 0%-3%, set at 1.5%, that could be adjusted up or down as needed. …submitters also questioned whether the CCyB works in practice… 14. Some submitters note that there is a lack of international evidence around the use of CCyBs. That is, while a cut in the CCyB would allow banks additional headroom to expand balance sheets as part of the response to a financial crisis, the efficacy of the CCyB as a broader macro-stabilisation tool is untested internationally. BNZ requested that caution should be exercised in using the tool more generally to manage the business cycle. 15. Genworth also expressed some practical concerns over the operation of the CCyB, considering the two phases of a systemic crisis. a. During a systemic crisis: the most immediate concern of prudential regulators and government is to restore market confidence. Using the CCyB during this phase may not be beneficial, as the challenge typically is to convince markets that bank capital levels are strong. It may be counterproductive to lower the regulatory minimum in such circumstances. b. After the crisis: after the immediate crisis passes and market confidence is restored, the question is whether the CCyB could help to mitigate a credit crunch. The reasons financial institutions may cut back on lending after a crisis are: concern over liquidity and stability of funding sources, re-evaluation of the credit worthiness of loan applicants, and a reduction in demand for loans by households and businesses that typically go into a phase of balance sheet de-leveraging. 16. Genworth notes that to the extent that a bank’s re-evaluation of risk results in higher estimates of risk weighted assets (RWAs), a reduction in the minimum regulatory Tier 1 may encourage more lending at the margin, particularly to higher risk borrowers. However, in a post-crisis period, the financial institutions would likely be more focused on retaining high published capital ratios to maintain market confidence. 17. A couple of submitters also questioned how ratings agencies would respond if the CCyB requirement decreases during a crisis. Some submitters believed that rating agencies may downgrade banks. … and were mixed on the CCyB’s use as part of the RBNZ’s toolkit… 18. Some submitters note that in the past, the RBNZ has used other macroprudential tools and sector-specific RWA overlays to reduce New Zealand bank balance sheet risks. ASB noted that these tools potentially have resulted in a more targeted de-risking of banking sector exposures than otherwise would have been achieved via implementing a CCyB. 19. Other submitters noted that given the relatively low level of the Official Cash Rate (OCR) compared to historical levels, the ability to lower prudential capital requirements by adjusting the CCyB would improve the RBNZ’s ability to stimulate the economy during a recession. …Banks also noted there may be an increase in capital required at group level as a result of jurisdictional reciprocity… 20. Banks with parent banks in other countries note that the reciprocity principle may apply to both an early or late CCyB, resulting in higher capital requirements at group level.

4 21. ASB stated that the introduction of the CCyB will increase the cost of capital for the Australian parent banks (because the parent’s regulatory capital minimums will increase in proportion to their New Zealand exposures), resulting in higher required returns on regulatory capital consumption exposures in New Zealand. They state that this provides motivation to reduce RWA by limiting lending to capital intensive sectors. 22. BNZ also questioned the need for trans-Tasman reciprocity in the CCyB given that business cycles are less than perfectly correlated across between Australia and New Zealand, and that New Zealand-based assets make up a relatively small share of group RWAs. … and one submitter asked for us to publish indicators for increasing/decreasing the CCyB. 23. Kiwibank noted that it would be useful for the banking sector if the RBNZ established a transparent metric that would prompt explicit consideration of increasing or decreasing the CCyB, noting that APRA already communicates on the indicators it assesses around deploying the CCyB. 2.2. External Experts External experts expressed some doubts around having a CCyB in the framework, largely due to operational concerns… 24. Professor Ross Levine is sceptical about the effectiveness of the CCyB. He is unsure whether the RBNZ will be able to accurately assess when to turn the CCyB off. As such, he asks for additional analysis on the practical implementation of the CCyB, noting that the analysis in the fourth consultation paper assumes that policymakers are aware at where in the business and financial cycles the economy is, and can adjust CCyBs with sufficient lead time (at least one year ahead) to navigate through troubled times. He questions if bank regulators will know in advance when to adjust the CCyB, and whether there is any evidence that central bankers implement monetary policy actions at least one year before financial system risks emerge. 25. Similar to Professor Levine, Professor David Miles notes that the CCyB’s effective operation seems to require that the RBNZ can assess where in the cycle the economy is. He posits that it is difficult to tell the difference between an economic shock to banks that is complete and one that is merely the first stage of a crisis that is about to become much worse. As an example, he notes that in the UK the hit to confidence following the demise of the mortgage lender Northern Rock at the end of 2007 proved not to be the eye of the storm – in fact it was a mild precursor to something that was much more serious and which came in late 2008. 26. Professor Levine is also sceptical about the implicit assumption that policymakers will not use CCyBs in other situations, i.e. when the economy is not facing banking system distress. He is concerned that they may not be used at appropriate times, and there is potential that they could be abused for political purposes. … but Professor Miles notes that there are some flexibility benefits from building the CCyB into prudential capital buffers… 27. Professor Miles notes that although the CCyB and ESR applying to banks falling below required capital levels are distinct concepts, they share a common aim, to allow banks to continue operating at closer to normal levels after a temporary hit to their capital than would be the case with a more rigid system of requirements. However, Professor Miles is unsure whether the CCyB is the most useful way to build in this flexibility, noting that it adds complexity to the rules.

5 … however both external experts suggest the CCyB be rolled into the prudential capital buffer. 28. Professor Miles believes that there is a case to be made on simplicity grounds (and in light of the difficulty of working out where in the cycle an economy is) to roll the CCyB into other parts of the prudential capital buffer leaving a single buffer of 10% for larger banks and 9% for smaller banks. He notes that in practice what is far more important than how a prudential capital buffer is comprised is its overall size. 29. Professor Levine does not believe that banks should be able to distribute dividends (or bonuses) when the CCyB has been removed during times of crisis, and thus (implicitly) advocates for having a larger conservation buffer. 3. Considerations on the CCyB moving forward 30. In light of the views of submitters and external experts, there are a number of competing objectives and considerations to take into account when making final decisions on the CCyB. The key issue to address is whether we want the proposed capital buffers to be more usable (and support lending) following a crisis, or whether we want these buffers to be maintained. 31. Considerations include: a. Effective operation/operational complexity: i. The ability of the RBNZ to assess where in the cycle the economy sits and when to take the CCyB off, or build it up (and how indicators may be used). ii. How the CCyB works with the ESR, as well as how dividend restrictions operate with the CCyB. b. Whether the CCyB works in practice: the extent to which lowering the CCyB would support lending following a crisis. c. Early vs late set CCyB considerations: i. Whether it is better to have the flexibility provided by the early CCyB (to be a more usable buffer that could be adjusted up or down); or, ii. Whether it is better to have a late CCyB, in light of operational concerns. It may be worse to lower capital requirements too early rather than raising them too late (asymmetric impact). d. Reciprocity with other countries: if an early CCyB will result in higher capital requirements at the group level, and how branches may be captured. APRA has indicated that there is appetite to explore having an early set CCyB in Australia. e. Consideration of other tools: when the CCyB will be the best tool to use, or whether other tools (e.g. LVRs, sector-specific overlays, ESR) to reduce bank balance sheet risks may be more appropriate. 32. We ask FSC to: a. Discuss the feedback raised by submitters and external experts; and, b. Discuss whether the considerations listed are appropriate, and whether there are any additional considerations to take into account when making decisions on the CCyB.

MEMORANDUM FOR Financial Stability Committee FROM Financial Policy (Author: Charles Lilly) MEETING DATE 9 October 2019 SUBJECT Capital Review: decisions on denominator proposals FOR YOUR Decision It is recommended that the Committee:

1. Note that the feedback we received on ‘denominator’ aspects of the December 2018 Capital Review proposals was largely in line with our expectations. There were a range of views on the appropriate relative calibration of the IRB and Standardised approaches, and submitters provided several alternative options to our proposals (an increased IRB scalar and an output floor).
2. Note that the external experts commented on the difficulties regulators face with operating IRB, and expressed scepticism that IRB should be allowed to generate significantly lower capital requirement outcomes than the Standardised approach.
3. Note that the two primary alternatives to our proposals suggested by submitters (adoption of the new Basel III Standardised approach, or a higher DSIB requirement) would imply that we need to increase our target Tier 1 capital ratio for the IRB banks from 16% to 18% to deliver the same level of soundness, as our 16% target was on the basis of an RWA amount that included the proposed IRB changes.
4. Agree that that the key denominator proposals (an IRB scalar of 1.2 and an output floor of 85% of the current Standardised approach) proceed as consulted on.
5. Agree that in respect of the timing of the changes: a. The increase to the scalar will be considered as part of overall Capital Review transition arrangements, in November; and b. The output floor (and dual reporting) will first apply for reporting on 30 June

Purpose

1. This paper reports back to the Committee the feedback we received on proposed changes to the IRB approach in our December 2018 Capital Review consultation paper. The proposed changes aim to narrow the difference in average risk weighted assets (RWA) outcomes between the IRB approach and the Standardised approach, in line with the Capital Review’s principles. The paper seeks confirmation of the proposals, as a step towards our final calibration decisions for the Capital Review. What we proposed
2. In the July 2018 response to submissions on the third Capital Review consultation paper (denominator issues), we made the following in-principle decisions:  IRB will remain part of the capital adequacy framework.  IRB banks will be required to report credit risk RWA calculated using the Standardised approach (‘dual reporting’).

2  The current Standardised approach for credit risk will remain in place for now, though we would review the new Basel III Standardised approach (B3SA) once APRA has finalised its plans.  The new Basel III Standardised Measurement Approach for operational risk will apply to all banks, in due course.  Market risk capital requirements will not change for the time being. 3. Parallel to the proposed changes to capital ratio requirements, in the December 2018 consultation paper we proposed further changes to the credit risk RWA framework:  Exposures in the Sovereign and Bank IRB asset classes will move to the Standardised approach.  The scalar applied to IRB RWA will increase from 1.06 to 1.2.  Credit risk RWA calculated using the IRB approach (including the 1.2 scalar) will be subject to a backstop floor of 85% of the corresponding RWA calculated using the current Standardised approach, calculated at an aggregate level. 4. Combined, the December 2018 proposals are expected to result in RWA calculated using the IRB approach increasing to an average of 90 percent of RWA calculated using the Standardised approach, based on the four IRB banks’ data and portfolios as at March 2018.1 Under the proposed calibration, RWA calculated under IRB would be the binding constraint, meaning IRB would continue to be the determinant of capital allocation for all four banks (the 85% floor to Standardised would only act as a backstop). 5. The indicative transitional arrangements included:  The higher IRB scalar taking effect in the first year of the Capital Review transition, given no systems changes are required of banks to implement it, and  The floor to Standardised taking effect following a 12 month systems build for dual reporting (given IRB banks need to build their Standardised calculation and reporting before they can put a floor into effect). 6. In considering the proposed changes to the IRB approach, it is important to note that the 16% Tier 1 capital ratio proposal assumed that the IRB scalar would increase to 1.2. If we were to soften our IRB recalibration (e.g. retain a scalar of 1.06 instead of 1.2), to achieve the same quantum of capital we considered was needed to meet the proposed 1-in-200 level of soundness, we would need to set Tier 1 capital ratio requirements for the four IRB banks of 17.8%. Objectives and principles 7. The proposed combination of changes (a higher IRB scalar and an output floor at an aggregate level) aims to strike a balance between retaining risk sensitivity in the capital framework and avoiding unjustified differences in outcomes between the two risk weighting methodologies. This follows from two of the principles of the Capital Review:  Capital requirements should be set in relation to the risk of bank exposures (regulatory capital requirements need to reflect the underlying risk of exposures in order to avoid distorting banks’ risk-return decisions); and  Where there are multiple methods for determining capital requirements, outcomes should not vary substantially between methods (there shouldn’t be a structural bias in the relative calibrations of the two approaches). 1 Note that the 90 percent estimate includes the impact of the June 2019 adjustment to ANZ’s residential mortgage and farm lending RWA, which we assume are permanent. Including the value of the expected loss in excess of eligible allowances for impairment deduction under IRB (times 12.5) increases the relative credit risk RWA under IRB to approximately 93 percent of Standardised as at March 2018.

3 What submissions said 8. Submissions and subsequent feedback can be collected into five key areas:  There were differing views on the appropriate relative outcome for IRB RWA relative to Standardised RWA: i. There was debate on either side as to whether a 90% average outcome is the right level. ii. Some banks suggested we should focus on narrowing outcomes even further in key lending areas for Standardised banks (e.g. residential mortgages), rather than aligning the calibrations at an aggregate level. iii. Our proposed calibrations diverge from those of peer regulators, which would weaken the international comparability of IRB bank capital ratios (Basel III reduces the IRB scalar from 1.06 to 1, and has a 72.5% floor).  It was suggested the B3SA should be adopted as an alternative to adjusting the IRB approach. The B3SA would increase risk sensitivity over our current Basel II Standardised approach, but would also likely result in a lower average capital requirement. In this way it could i. ‘narrow the gap’ in average outcomes between the two groups of banks; ii. limit the potential loss of risk sensitivity from having a binding output floor on IRB outcomes; and iii. improve risk sensitivity for Standardised banks.  Some submissions suggested that a ‘level playing field’ could better be created with a higher DSIB buffer, instead of tweaking IRB, given the four IRB banks are likely to be designated as DSIBs. If a 16% Tier 1 requirement was kept for DSIBs, this would mean non-DSIBs would have lower capital ratio requirements.  Some submissions suggested that current IRB outcomes overstate differences to the Standardised approach, as they reflect the current benign state of the credit cycle. The scalar and output floor should be set on a through-the-cycle, average basis (i.e. lower than 1.2/85%). Additionally, consideration should be given to a ‘ceiling’ on IRB outcomes relative to Standardised to prevent an undesirable (in their view) overshooting of IRB outcomes in a downturn.  IRB banks suggested that, if the IRB scalar increases as we have proposed, the transition should be phased in over time (e.g. over two years). 9. We have previously reported to the Committee on feedback from domestic mortgage￾focussed banks, specifically their perception that the proposed IRB changes do not go far enough in levelling outcomes for mortgage lending. Our analysis shows that under our proposals, IRB banks’ capital requirements for residential mortgage lending will be approximately 90% of the Standardised outcome (or 96% if a DSIB buffer of 1% is included in the calculation).2 What the external expert reviewers said 10. James Cummings noted that the proposed changes to IRB are in line with changes to the international framework and reflect industry concerns about inconsistencies in capital outcomes resulting from the use of internal models. He noted that there is limited research literature on whether the use of internal models provide more accurate estimates of unexpected losses than implied by the Standardised approach. He recommended that we should monitor through dual reporting the extent to which IRB models provide more accurate estimates of unexpected credit losses, and revisit the decision to retain IRB accordingly.

2 See #8499570. Note that these figures incorporate the June 2019 overlay on ANZ.

4 11. Ross Levine noted the incentives on IRB banks to use their models to reduce regulatory capital requirements, and the asymmetric information between IRB banks and regulators. He noted that our proposals aim to further limit the risks of continuing to use IRB models (beyond the existing controls, e.g. model approval process) while gaining the potential benefits. The proposed mitigants (floor and scalar) are an ad hoc but not unreasonable compromise given the informational and resourcing constraints that we operate under as a regulator. Though the accuracy-competitive balance issue is “complex, messy, and unlikely to yield a simple answer”, he recommends the Reserve Bank devote additional resources towards ensuring the RWA framework:  Measures RWA as accurately as possible for both classes of banks  Creates as unbiased a competitive environment as possible 12. David Miles noted that challenges of internal models we identified are not unique to New Zealand. While there are theoretical benefits from having banks measure their risks using their internal information, the incentives for capital minimisation and challenges facing a resource-constrained regulator create a tension in retaining IRB. He noted that, to the extent the proposed increase in IRB RWAs is reversed, we would need to set the Tier 1 requirement higher than 16%. This is because the calibration of overall quantum of capital was completed taking the RWA changes as given. He also noted that, if a floor to Standardised RWA became binding, we should question retaining IRB as in that scenario it would not be providing its principal benefit (closer alignment of capital to risk). 13. Overall, the messages we take from the three external experts are:  The Reserve Bank is not unique in its unease with allowing IRB, and the issues and tensions we have identified are shared by other regulators.  While there are theoretical benefits of allowing IRB, there is not yet a strong case for large divergences in overall outcomes/calibrations relative to Standardised.  The proposed changes (higher scalar and floor) are a pragmatic response to the issues identified, in the Reserve Bank’s context.  A tightly binding floor to Standardised would reduce the case for allowing IRB. Other considerations 14. As part of the proposed changes, Financial Policy will review how the IRB framework is operated – for example, what the model approval process looks like, and the degree to which we intervene in the way banks design their models. While our current approach to monitoring IRB outcomes is largely reactive, and in many respects lighter handed than APRA (for example), a relaxation in our approach could see IRB outcomes drift towards the 85% floor. We are not requesting any decisions from the Committee at this stage, but do want to make the Committee aware that this as a question that we will present more of our thinking on in due course. 15. By and large, we see the supervisory overlays we currently have in place (e.g. residential mortgage and farm lending risk weight floors) as being permanent realignments of banks’ models, which would not be unwound in future. The 90% target we are calibrating our proposals to include the impact of these overlays, and so allowing the overlays to be unwound would see an undesirable weakening of the relative calibration of IRB to Standardised. we anticipate these new models will be calibrated to produce similar capital outcomes to the level of the overlays (not the outcomes of the old models). More generally, the points to the need for ongoing monitoring and maintenance to support the credibility of the New Zealand IRB framework. s(18)(c) (i) s(18)(c)(i) s(18)(c)(i)

5 Summary and recommendations 16. The feedback we received on the December 2018 denominator proposals was largely in line with our expectations. Submitters and the external experts identified the trade-offs we are balancing – wanting to promote risk sensitivity, but also wanting to reduce unjustified differences between the two capital approaches – and came to a range of potential solutions. The external experts broadly agreed with the theoretical benefits of retaining the IRB approach but were sceptical of the degree to which it should confer a regulatory capital benefit. 17. The appendix provides more detail on how we have considered the substantive points raised through the submissions process. While there are some merits to the main alternative suggested by submitters (adoption of the B3SA), as currently calibrated this alternative would need to be offset by an increase to the proposed Tier 1 capital ratio requirements for the four IRB banks to approximately 18%, if we want to deliver the same level of soundness as originally proposed. 18. As discussed in our Committee papers ahead of the December 2018 consultation paper, the level of alignment (IRB RWA at 90% of Standardised RWA) is largely a matter of judgement: depending on their perspective, submitters made cases for why it should be lower or higher than what we have proposed, using similar arguments to those we have traversed in our own internal discussions. 19. We recommend that the Committee agree that the December 2018 IRB proposals proceed as consulted on:  The scalar applying to IRB RWAs increases from 1.06 to 1.2; and  A floor of 85% of the comparable Standardised RWA amount is applied to IRB RWA, calculated at an aggregate level. 20. In respect of timing, we recommend that the Committee  Agree that the timing of the increase in the IRB scalar occurs relatively early in the Capital Review transition period, to support the ability of Standardised banks to act as a competitive constraint on IRB banks. We will bring specific timing recommendations to the November Committee meetings, so that these can be considered alongside other aspects of the final calibration.  Agree that the output floor (and associated dual reporting requirement) comes into force from 30 June 2021. 21. Financial Policy intends to report back to the Committee on options around adopting the B3SA (and APRA’s variant of it) in 2020, including any calibrations we think would be necessary for New Zealand. Given we view the B3SA as an enhancement, rather than replacement, of the current Standardised approach, on balance we recommend not deferring the imposition of the output floor until details of our potential adoption of the B3SA are more concrete. Financial Policy will also report back on APRA’s decisions on implementing the Basel III changes in APRA’s IRB approach.

6 Appendix: responses to submissions 22. In the following section we provide a response to the main comments received on the proposed IRB changes. Should we adopt the Basel III Standardised approach (B3SA) instead of recalibrating IRB? 23. The B3SA was finalised in late 2017, and APRA recently concluded consultation on its implementation in Australia. The two most relevant changes compared to the Basel II Standardised approach are:  More graduated risk-weight buckets for residential mortgages by LVR, and differentiation between owner-occupiers and other borrowers.3  Additional risk weight differentiation for commercial property, SME lending, and other retail lending. 24. In addition to increasing risk sensitivity, the B3SA has recalibrated some asset classes, generally downward. For residential mortgages we estimate that the B3SA would result in a 15-20% decline in average RWAs, compared to our current Standardised approach. Banks (both IRB and Standardised) therefore see adoption of the B3SA as an alternative means of ‘levelling the playing field’ for mortgage lending. Instead of lowering Standardised RWAs, our proposals would increase IRB RWAs by about 13%. 25. The position we previously reached is that our current Standardised approach remains fit for purpose at present, as it is already more risk sensitive than the Basel II Standardised approach. We will consider adopting the B3SA once APRA has finalised the Australian implementation. The primary motivation for adopting B3SA would be to further enhance the framework’s risk sensitivity, especially for non-mortgage asset classes. Importantly, we see the calibration of the B3SA as a separable issue to enhanced risk sensitivity; increased sensitivity need not be coupled with a 15-20% average decline in residential mortgage capital, and it’s not clear that moving from our current Standardised approach to the B3SA should be coupled with capital relief. 26. Moreover, as explained above, narrowing differences in IRB and Standardised outcomes by reducing Standardised RWAs would mean that we would need to set higher Tier 1 capital ratio requirements for both IRB and Standardised banks, to achieve the same level of resilience at each bank, and for the system as a whole. Adoption of B3SA should not therefore be seen as a ‘solution’ to the capital adjustments and strategic decisions each bank will need to make to meet the proposed requirements. 27. Aspects of the B3SA mean adopting it would be consistent with our principle of promoting a risk sensitive framework, and we have stated our willingness to consider these improvements in due course. But, unless the Committee would be amenable to make offsetting adjustments to the target Tier 1 ratio (i.e. ~18% instead of 16% for DSIBs), in its current calibration we do not see the B3SA as a viable alternative to meeting the present objective of narrowing the difference in capital outcomes between IRB and Standardised. 28. A final comment on the B3SA relates to the output floor. To comply with the proposed dual reporting requirement and output floor, IRB banks will need to invest in building calculation systems to produce Standardised RWAs.

3 Note that the Reserve Bank’s BS2A Standardised approach already differentiates risk weights for owner￾occupiers and investors, and for loans above 80% LVR. The Basel III Standardised approach increases differentiation of risk weights for LVRs below 80%.

7 29. On one hand, if we intend to adopt the B3SA in the next one to two years a reasonable case can be made that we should defer dual reporting and the output floor until B3SA has replaced our current Standardised approach. This would provide a cleaner transition, and avoid duplication of work for IRB banks, reducing regulatory compliance costs. Based on the 2018 QIS data, the proposed calibration of the output floor would not be a binding constraint on IRB banks, and so deferring dual reporting and the output floor until such a time as the B3SA has been adopted is unlikely to materially affect capital outcomes over the transition period. 30. On the other hand, the B3SA can be seen as an enhancement rather than replacement of the current Standardised approach. It improves risk sensitivity by creating new sub￾asset classes with more targeted risk weights, rather than completely replacing the current framework (e.g. residential mortgages are split into owner-occupier and investor, and new categories are created for commercial property and specialised lending). The information IRB banks would need to assign borrowers to the B3SA asset classes is likely to already exist. While less clean a transition, requiring dual reporting and the output floor to initially be on the basis of the current Standardised approach, with adoption of the B3SA in due course, is not expected to be difficult. Should we set a higher DSIB buffer requirement instead? 31. Another alternative to our proposed approach to narrowing outcomes between IRB and Standardised is to adjust the size of the DSIB buffer, since the four IRB banks coincide with the banks likely to be designated as DSIBs. 32. The objective of the DSIB buffer is to reduce the likelihood and cost of the failure of a systemically important bank, and reduce competitive distortions associated with banks being ‘too-big-to-fail’. These are different to the objective of the proposed IRB recalibration (reducing structural biases in the relative calibrations of the two RWA frameworks). As there is the potential for a bank that uses the Standardised approach to be designated a DSIB, we do not think these two sets of objectives and tools should be conflated. 33. Moreover, as in the discussion of the B3SA, using the size of the DSIB buffer to narrow the differences in outcomes between IRB and Standardised, instead of the IRB scalar, would oblige us to reconsider the 16% target Tier 1 capital ratio for the four IRB banks. To achieve the equivalent level of resilience at IRB and Standardised banks as we have proposed would require a DSIB buffer of 3% (i.e. target Tier 1 ratios for the four IRB banks of 18%, and 15% for other banks). Do the proposals create perverse incentives or outcomes for IRB banks? 34. At the margin, recalibrating the IRB approach via the scalar will increase the incentives for banks to be more selective in the portfolios they choose to develop internal models for. This could lead to ‘cherry picking’, where banks optimise RWA by selectively modelling portfolios depending on the relative capital outcome of IRB and Standardised. 35. Currently, only a handful of low materiality portfolios at each IRB bank use the Standardised approach (e.g. Retail SME at ASB). Despite the supervisory constraints we have put into aspects of the framework and individual models (which raise IRB RWA), by and large the four IRB banks operate as ‘full service’ IRB banks, consistent with APRA’s expectations for their respective parents. In some cases, we observe banks using IRB models even where the current Standardised approach would have resulted in a lower capital requirements, e.g. for some credit card and personal loan portfolios. In part this

8 will reflect that there is still an RWA benefit from using IRB models in the Group’s Level 2 calculation, as APRA’s rules sometimes do not flow through our supervisory constraints. 36. Based on the QIS data, we estimate that the proposals will lead to IRB RWA being higher than Standardised for Commercial Property (an average ratio of 109% of Standardised RWA), Retail SME (103%), and Other Retail (126%). For major portfolios (mortgages, farm lending, general corporate), we expect ratios between 75% and 95%. For individual IRB banks, there is a wider range of relative outcomes. 37. Presented with a menu of options, banks have an incentive to use the approach that results in the lowest regulatory capital requirement (indeed, the observation that IRB currently gives a capital benefit for almost all asset classes is consistent with the ‘full service’ approach of the four IRB banks). Given our policy objective of a closer alignment of capital outcomes between IRB and Standardised necessitates us reducing one of the key benefits of being an IRB bank, we should not be surprised that taking away a current advantage will result in some IRB banks reassessing their choices around their capital calculation. 38. Our expectation is that, while in specific cases there may be an incentive to opt for the Standardised approach,4 the relatively modest scale of the proposed change (a 13% increase in IRB RWA), and other factors (Group considerations, better risk management etc.) will not create large enough potential benefits for IRB banks to want to move a large number of and/or their most material portfolios onto the Standardised approach. 39. Our analysis is that the proposed 85% floor would not be binding at present, as there is a combination of ‘unders and overs’ across asset classes under IRB to produce an average outcome of 90% of Standardised. Even if IRB banks were to opt for the Standardised approach for portfolios where it results in a materially lower outcome than IRB, the output floor at 85% would become increasingly binding on the portfolios that continue to be modelled (since the 85% floor is calculated only on portfolios using IRB). Thus a mechanism would already exists in the proposals to limit the degree to which ‘cherry picking’ can be used to minimise capital requirements. 40. If we are concerned about IRB banks’ potential ‘cherry picking’ resulting in suboptimal capital allocation decisions, provided we stick to the policy objective of achieving a close alignment of IRB and Standardised outcomes, our attention is perhaps best spent focussing on improving the risk sensitivity of the Standardised approach. Does there need to be a ceiling on IRB RWA? 41. The observation that IRB RWAs will be more sensitive to the state of the credit cycle, and therefore we should calibrate the relativity between IRB and Standardised on a ‘cycle average’ basis has some merits. 42. In theory, well-built IRB models will produce RWAs that do not respond to cyclical changes in borrowers’ financial circumstances. We have encouraged (and in some cases, such as residential mortgages, mandated) banks to build IRB models that take such a through-the-cycle view of credit risk. However, the limited evidence we have (see Annex) suggests that New Zealand banks’ IRB RWA will still be somewhat procyclical in some portfolios.

4 ASB’s submission, for example, constructed an example where a bank with low provisioning of defaulted exposures would face a significantly higher capital burden under the proposed IRB changes.

9 43. As calibrated, there is still scope for IRB risk weight inflation in response to a downturn event before we would be at the point where IRB RWAs would be materially above Standardised RWAs. Based on our limited evidence, we expect a moderate-to-severe downturn event could result in IRB RWAs increasing from around 90% of Standardised RWA to around 105%. 44. Having no ceiling on IRB RWA reinforces the incentives for IRB banks to develop better quality regulatory capital models that focus on through-the-cycle risk identification, instead of indicators of borrowers’ short-term credit risk (e.g. payment behaviour) which often drive procyclical RWA outcomes.

1 Ref #8677867 v1.2 PAPER FOR FSC FROM Lucia Frei (Financial Policy) DATE 23 October 2019 SUBJECT Role of incentives regarding the Capital Review - response to Prof. Levine’s External Expert Report FOR YOUR Information It is recommended that the Group: Note: The purpose of this paper is to respond to Prof. Levine’s assessment of the Capital Review and his recommendation to give more attention to the role of incentives on bank decision makers. Financial Policy has considered the incentive effects of different types of capital, governance and ownership variables, as well as incentives in regard to the CBA. Note: Considering the incentive effects of different capital types provides some additional comfort regarding the proposed changes to the forms of capital accepted as regulatory capital. Note: A stylized metric was developed as an intuitive approach to gain an understanding of the configuration of Levine’s moderating ownership- and governance-related factors at NZ’s banks. Overall, the results indicate that the impact of these variables on the incentive effect of capital requirements will be limited. Note: In sum, the cost- and benefit-side incentives effects seem to be relatively balanced, with a slight tendency towards a positive welfare-impact. Even though we do not propose attempting to quantify the incentive effects for the CBA, we will include the role of incentives as a box in the CBA. Note: In sum, considering the issue of incentives in more depth has provided some additional comfort for the Capital Review and the in-principle decisions made. It does not change our positions, but has been a very useful exercise to have completed. Note: This paper may provide useful insights for future bank regulatory reforms; for instance regarding the governance of banks or the design of remuneration schemes. Note: Incentive effects should be monitored during the implementation of final Capital Review decisions. Background

1. This paper considers the relationship between incentives and bank capital, following on from the recommendations in Prof. Levin’s External Expert Report. 1 According to Prof. Levine, bank crises can not only arise from large shocks to banks’ assets, but also from excessive risk-taking incentives for banks arising from moral hazard problems.
2. However, Prof. Levine does not necessarily expect a greater emphasis on incentives to have any impact on the RBNZ’s recommendations regarding capital ratios, since he regards the RBNZ analyses as well-balanced. In discussions with Financial Policy Prof. Levine noted that though bringing the concept of incentives more fully into the analyses would be beneficial, he was not sure how we could embed it formally into our risk appetite framework, for example. Summary of Prof. Levine’s Reflections on Incentives
3. Prof. Levine depicts two main aspects of incentives in his External Expert Report – the incentive-stability connection and the incentive-efficiency connection. In regard to both, the impact of capital regulation depends on how regulation influences the incentives of owners and those of executives, as well as the comparative power of the two parties. 1 This paper does not encompass a discussion on the dynamic effects of capital regulation or on the role of deposit insurance. The topic raised by Prof. Levine regarding the dynamic effects of capital regulation will be considered in another note on the financial ecosystem. The RBNZ’s views on the relationship between capital requirements and deposit insurance were represented in a note to the Minister of Finance on Sept. 24th 2019.

2 4. The following graphic is a simple illustration of the incentive-stability and the incentive￾efficiency connections (for further explanation see Annex, Table 1). The incentives on executive risk-taking behaviour stemming from capital requirements (solid arrows) are moderated by ownership and governance variables (blue boxes). The executives’ behaviour in turn influences the stability and efficiency of the financial system (dashed arrows) through elements, such as the PD level or the credit allocation. Figure 1: Graphic of the incentive-stability and incentive-efficiency connections 5. In addition to these two central aspects, Prof. Levine discusses the incentive effects of contingent debt and of Tier 2 capital. a. The impact of contingent debt on risk-taking incentives depends on the design elements, such as where the trigger is set or the convertibility to equity. Since convertible contingent debt can harm bank decision makers by diluting their ownership stake, it can be an incentive for bankers to manage their bank more prudently. b. Prof. Levine sees Tier 2 capital as a potentially effective mechanism for constraining risk-taking incentives. If uninsured debtholders do not expect government bail-out, they have incentives to dampen risk-taking within banks. Analyses of the role of incentives in regard to the Capital Review 6. The importance that Prof. Levine attributes to incentives for bank executives arises from the US-context, in which bank failures in part stemmed from mismanagement and poor decision making on behalf of bank executives.2 The specific features of the NZ financial system are considered in the following evaluation of the possible incentive effects of bank capital requirements in regard to the Capital Review. 7. Although Prof. Levine has published conceptual as well as empirical studies on bank governance, risk-taking and competition,3 there is currently no developed set of descriptive statistics or metrics for evaluating incentive effects of bank capital regulations on bank decision makers. Notably, Prof. Levine states in his assessment that researchers, such as himself, are just starting to build models of banking systems in which incentives play a central role, meaning that it is still early days in this area of research, so there are limited theoretical or empirical studies to draw on for examining these incentive effects in NZ. Incentive-moderating variables for NZ’s banks 8. The developed stylized metric (see Annex – table 2) presents an intuitive approach to gain an understanding of the configuration of the moderating ownership and governance variables at nine NZ banks, based on a preliminary discussion with bank supervisors 2 It is also more common in the US to have small closely-held entities where the incentive effects would naturally be in sharper contrast to less closely-held entities. 3 For example, Corbae and Levine (2018); Laeven and Levine (2009); Levine (2005) and Levine (2004).

3 and their expert judgement. An in-depth assessment would require the filling of information gaps, for example regarding the calculation of variable remuneration. 9. According to Prof. Levine, bank capital reforms have a greater risk-reducing effect at strong-governance banks with influential owners providing the additional equity. Of the considered NZ owned banks, some banks were regarded as having strong governance but only two other banks have large, influential owners. In addition, no bank was estimated to have owners with a high degree of personal wealth invested in the bank. Hence, a very strong risk-dampening effect may not be present in NZ. 10. Correspondingly, the design of remuneration schemes4 and the level of personal wealth senior executives have invested in their bank are likely to be of greater relevance for the risk-reducing incentive effect. a. Due to stricter remuneration regulation from APRA, the four Australian-owned banks5 (ANZ, BNZ, ASB and Westpac) have remuneration schemes that reduce risk-taking by, for example, aligning executives’ and shareholders’ interests through deferred variable remuneration and malus provisions6 . Furthermore, their senior executives were estimated to have a moderate level of personal wealth invested in the banks. Yet, some negative incentive effects could stem from basing variable remuneration, at least partly, on factors such as the relative total shareholder return7 and/or ROE. b. Except for , the CEOs of the NZ-owned banks receive or more of their remuneration in form of a fixed salary. Prof. Levine regards fixed salaries positively in comparison to variable remuneration based on the ROE. However, at the NZ-owned banks (except for 8 ) the executives do not have significant personal wealth invested in their bank. 11. Overall, the results of the stylized metric indicate that the impact of the moderating variables on the incentive effect of capital requirements will be limited. However, further information would be necessary to assess their impact with greater certainty. Incentive effects of different types of capital 12. Table 1 provides an overview of the different capital types and the incentives these provide for shareholders and debtholders to monitor their bank, where green symbolizes a strong monitoring incentive. Table 1: Capital types and incentives to monitor 13. Ordinary shareholders have an incentive to monitor bank management, since they do not only participate in the profits, but also absorb loss. However, in the event that a bank 4 Please note that, following the FMA/RBNZ Conduct and Culture Review, several banks are in the process of revising their remuneration schemes. 5 Seeing as the NZ subsidiaries of the big four banks are important parts of the parent financial institution, the ultimate shareholders and the ownership structures of the parent banks were considered. 6 Malus refers to the ability to reduce or withhold variable remuneration parts that have been awarded, but not yet paid out in exceptional circumstances such as e.g. in case of misconduct. 7 A relative total shareholder return (TSR) measures an institute’s TSR over the performance period (usually multiyear) relative to the TSRs of selected financial institutes in the peer group. 8 While the level of personal wealth invested in the bank by executives was regarded as moderate at , their remuneration scheme was seen as relatively less risk-reducing. Capital type Monitoring incentive Ordinary shares Redeemable perpetual preference shares Subordinate debt (contingent and non-contingent) Unsubordinated debt Uninsured deposits Secured debt (e.g. covered bonds) Most loss absorbing Least loss absorbing s 18(c)(i) s18(c)(i) s18(c)(i)

4 faces significant losses, shareholders do not have an incentive to invest more capital in the bank, since creditors would benefit foremost. As the four IRB banks in NZ are wholly-owned subsidiaries of Australian banks, the ordinary shareholders are once removed. Thus, a better understanding of the involvement of the parent’s shareholders in monitoring the NZ subsidiaries would be beneficial. 14. Including redeemable perpetual preference shares (RPPS) as a capital instrument has the advantage that additional, new investors, who have an incentive to closely control management actions, are likely to come on board. Furthermore, as these shares would likely be listed in NZ, their pricing would be a signal of the market evaluation of the bank. This new information being available in the market could incentivize managers to behave more prudently. However, RPPS may have the disadvantage of motivating bankers to also pay dividends or redeem the share on the scheduled call date at low bank capital levels, in order to avoid any negative signalling effects. Yet, measures that mitigate against such signalling risks could be implemented by the RBNZ.9 15. Subordinate debt: a. The in-principle decision to not accept contingent convertible capital instruments (Cocos) as regulatory capital was based on complexity, redundancy and reliability concerns. Relative to ordinary shares, Cocos have less monitoring incentive impact, which gives some support to the in-principle decision. On the other hand, theoretically, Cocos have a strong incentive effect, since bank shareholders and managers are motivated to reduce risk-taking in order to avoid the trigger event and a potential loss of control. Yet, as Prof. Levine points out, economic theory shows that the incentive effect depends on the design of Cocos.10 Furthermore, for NZ’s big four banks, the disciplinary effect of Cocos also depends on whether or not the parent holds them. Moreover, the positive incentive effects of Cocos, stemming from the possibility of loss of control, could likewise be achieved by an escalating supervisory response framework and the Open Bank Resolution (OBR) policy. b. Non-contingent, subordinated debt: Shareholders have an incentive to aim for an increase in bank risk-taking after funds from debtholders have been obtained. However, this downside can be counterbalanced by the incentive of uninsured debtholders to limit bank risk-taking and leverage via the price they require for debt. An additional beneficial incentive effect may stem from new investors holding such debt, which would likely be listed locally (analogous to RPPS). 16. Holders of unsubordinated debt have less incentive to monitor the bank than subordinate debtholders due to their preferential claim. Yet, they probably have a greater capacity to monitor than uninsured depositors. In principle, the latter have an incentive to monitor bank risk and, if necessary, demand higher interest rates. However, higher capital requirements could also reduce the motivation of the depositors to closely monitor banks, as the banks would generally be perceived as safer. Lastly, secured debtholders have a relatively low monitoring incentive, because they have preferential claims in case of a bank failure. 17. All in all, considering the incentive effects of different capital types provides some additional comfort regarding the changes to the forms of capital accepted as regulatory capital proposed in the Capital Review. Incentive effects in regard to the Cost-Benefit Assessment (CBA) Analyses 18. Incentive effects from increased bank capital requirements on bank risk-taking behaviour could increase the costs as well as the benefits. As the incentive effects 9 Such measures pertain to the required capital positions after redemption and were described in more detail in a FSC paper regarding AT1 Capital (Ref #8524944). 10 Important design elements are the height of the trigger and the conversion mechanism.

5 depend on a multitude of factors and their interactions, a numerical expression of their influence on the costs or benefits is not attempted. However, several channels of influence regarding the CBA are outlined below in a qualitative way. 19. Possible cost-side incentive effects encompass the following: a. Executives could be under pressure from shareholders, and/or motivated through their remuneration schemes, to maintain their bank’s ROE at increased levels of regulatory capital and, therefore, be willing to take greater risks in credit allocation. Resulting elevated PD or LGD levels could heighten the probability of bank failure. Such a decrease financial stability could incur output losses and social costs. b. As higher levels of equity funding makes banking in NZ more expensive, bank decision makers will be incentivized to cut costs and/or to pass on costs to customers, e.g. through higher interest rates. The latter would involve transferring wealth out of NZ for the foreign-owned banks. c. Similarly, bankers may be motivated to reduce the quantity of credit allocation as a means to attain a higher capital ratio. A reduction of lending activity could curtail investment and consumption and, thus, impact NZ’s GDP negatively. 20. Possible benefit-side incentive effects include the following: a. Requiring higher levels of equity would motivate bank decision makers to reduce debt financing. As this would decrease the debt-related interest payments and banks in NZ are largely foreign-owned, this would transfer wealth into NZ in the form of higher taxable profits. b. Since the regulatory reforms ensure that banks have enough equity to absorb significant losses and signal that government will not bail failing banks out, perceived implicit bank subsidies should be reduced. This could motivate non-bank financial service providers and new market entrants to compete with banks, and thereby raise the overall efficiency of NZ’s financial system. c. Correspondingly, as the higher capital requirements reinforces NZ’s “non-zero failure” regime and emphasis on market-discipline, it could incentivize share- and debtholders to more closely monitor bank management and, if necessary, demand reduced risk-taking. Investors holding preference shares would likewise have an incentive to function as an additional controlling entity. d. Furthermore, higher capital requirements would incentivize bank decision makers to reduce RWAs by decreasing risks in credit allocation. This would lead to lower PD and LGD values and, thus, a reduced likelihood of a financial crises in NZ with its associated output losses and social costs. A more stable financial system can in turn increase investment and GDP levels. 21. In sum, the cost- and benefit-side incentives effects seem to be relatively balanced, with a slight tendency towards the positive welfare-impact. Conclusion 22. Considering the issue of incentives in more depth has provided some additional comfort for the Capital Review and the in-principle decisions made. Yet, the results should be regarded with some caution, due to the limited information available and the lack of applicable metrics or models from economic research. 23. Even though we do not propose attempting to quantify the incentive effects for the CBA, we will include the role of incentives as a box in the CBA. 24. Furthermore, this paper may provide useful insights for future bank regulatory reforms; for instance regarding the governance of banks or the design of remuneration schemes.

I Bibliography Auckland Centre for Financial Research (2016). New Zealand Corporate Governance Index – 2016 financial year end. Barth, J. R., Caprio Jr, G. and Levine, R. (2012). Guardians of finance: Making regulators work for us. MIT Press. Corbae, D. and Levine, R. (2018). Competition, stability, and efficiency in financial markets. In 2018 Jackson Hole Symposium: Changing Market Structure and Implications for Monetary Policy, Kansas City Federal Reserve. Institutional Shareholder Services (2019). Governance Quality Score – Overview and Updates (March 29th 2019). Laeven, L. and Levine, R. (2009). Bank governance, regulation and risk taking. Journal of financial economics, 93(2), 259-275. Levine, R. (2004). The corporate governance of banks: A concise discussion of concepts and evidence. The World Bank. Levine, R. (2005). Finance and growth: theory and evidence. Handbook of economic growth, 1, 865-934. MSCI (2015). MSCI Governance-Quality Indexes Methodology (June 2015).

II Annex Table 1: Explanation of the incentive-stability connection and the incentive-efficiency connection Incentive￾stability connection The incentive-stability connection describes the linkage between bank regulation as a factor influencing bank decision makers, who in turn shape the stability of the banking system. Thus, bank regulation can reduce moral hazard problems11, if it induces major decision makers of the bank (influential owners or executives) to have more of their personal wealth invested in their bank. Having more personal wealth at risk incentivizes these decision makers to manage their bank more prudently. In regard to the incentive-stability connection, Prof. Levine discusses three factors that moderate the influence of capital regulation on the risk-taking incentives of bankers: ownership structures, executive remuneration schemes and the governance of banks. a. According to Prof. Levine, increasing capital requirements will reduce risk-taking incentives more, if the additional equity comes from influential owners, rather than from small, diffuse shareholders. b. In addition, the stabilizing effect of increased capital requirements will be greater, if executive remuneration contracts do not reward risk-taking. Well-designed remuneration schemes can help align the interests of executives and shareholders.12 c. Furthermore, the response to capital reforms will depend more on the ownership structure for banks with strong governance systems, while it will depend more on the executive remuneration packages for banks with weaker governance systems. Incentive￾efficiency connection The incentive-efficiency connection describes the linkage between capital regulation as a factor influencing the efficiency of the banking system by altering risk-taking incentives of bankers. Prof. Levine focuses on incentives in regard to bankers’ credit allocation decisions and government subsidies for banks. a. Capital requirements might alter risk-taking incentives for bankers with respect to decisions on the allocation of credit. If decision makers are induced to invest more of their personal wealth in their bank, credit might be allocated to more prudent endeavours. If, however, remuneration components of powerful executives are based on the bank’s ROE, regulation increasing bank equity might induce executives to allocate credit to riskier endeavours. b. According to Prof. Levine, banks can borrow inexpensively, fund themselves with less equity and take greater risks due to explicit and implicit government subsidies. Since depositors and debtholders expect that government will bail them out in case of bank failure, they have weaker incentives to constrain risk-taking and leverage. Hence, he highlights that the expectations of the bank’s debtholders determine the moral hazard problem. If higher capital requirements reduce this implicit subsidy for banks, other financial service providers might emerge and increase the overall efficiency of NZ’s financial system. 11 Results of a study by Levine and Corbae (2018) indicate that regulatory reforms to improve bank governance enhance bank stability and efficiency. By incentivizing bankers to focus more on the long-term value creation, such governance-ameliorating policies lessen traditional principal-agent problems between owners and managers. Furthermore, Levine and Corbae (2018) found that capital requirements and governance are closely linked and that the combination of both shapes bank lending and risk. 12 According to Prof. Levine, considerable evidence shows that changes in remuneration incentivized executives in financial conglomerates to take greater risks prior to the GFC (cf. Barth, Caprio and Levine, 2012).

MEMORANDUM FOR FSC FROM Financial Policy (Matthew Brunton) MEETING DATE 23 October 2019 SUBJECT D-SIB buffer calibration and framework FOR YOUR Information It is recommended that the Committee:

1. Note that as part of the Capital Review we have consulted on the inclusion of a capital buffer for Domestic Systemically Important Banks (D-SIBs), proposed to be set at 1% (so a total Tier 1 requirement would be 16% for D-SIBs).
2. Note that the 16% Tier 1 capital ratio was calibrated with systemic banks in mind. As such any change in the D-SIB buffer calibration should maintain the Tier 1 ratio for D￾SIBs and subsequently lower the Tier 1 requirements for non-D-SIBs.
3. Note that this paper will present submitters’ views and provide a framework to incorporate them into the final decision-making process. It does not provide any final calibration for the D-SIB buffer (that will be for the next FSC paper on the D-SIB).
4. Note that for the New Zealand context, we see the larger impact of D-SIB failure (compared to non-DSIBs) as the key argument for the calibration of the buffer. Other factors will also be analysed to the extent that they are relevant. Purpose
5. As part of the Capital Review consultation document ‘How much Capital is enough?’, the Reserve Bank proposed that 1% of the 16% Tier 1 ratio be attributed to a D-SIB buffer. This paper discusses submitters’ views to the D-SIB buffers and provides further analysis in response. It does not propose a calibration for the D-SIB buffers at this stage. For more detail on analysis and estimates provided see #8559445. Background
6. Following the GFC the Basel Committee on Banking Supervision (BCBS) began developing a framework for identifying, and calibrating additional loss absorbency requirements for, Domestic Systemically Important Banks (D-SIB).
7. Regulators in other jurisdictions have used a range of analysis, both quantitative and qualitative, to make a judgement about the appropriate D-SIB buffer for their banking system. For example, the Hong Kong Monetary Authority (HKMA) leverages the BCBS work for calibrating G-SIBs. In contrast, APRA used an IMF report on the appropriate HLA amount as well as comparisons with peer jurisdictions, but ultimately opted for a 1% D-SIB buffer due to conservative treatments elsewhere in their capital framework.
8. In the New Zealand context, there was discussion over whether 16% Tier 1 requirements on small banks when their failure does not have a systemic impact and whether the CCyB should only apply to large banks. Ultimately though it was decided treating the CCyB and D-SIB as two separate capital components was more appropriate, and that 1% of the proposed 16% Tier 1 ratio would be made-up from the D-SIB buffer. Submitters’ views
9. ANZ, ASB, BNZ and WNZL generally agreed with the proposed D-SIB buffer of 1%, noting that this aligns with APRA. BNZ argues that it could be met with AT1 capital.

4 19. However, we do think there may be grounds for thinking it is less relevant for New Zealand due to the existence of the OBR policy and statutory manager powers. This is also reflected in the S&P assessments of the large New Zealand bank: “… in our view, it is uncertain whether the New Zealand government would provide timely financial support to the country's private-sector banks …. The country's bank resolution framework allows the senior creditors of a bank to absorb losses to help the bank continue operations if the bank were to experience financial distress.” 20. The key question is then how much emphasis we place on potentially lower moral hazards in New Zealand compared to the higher market concentration. Lower moral hazards could, to an extent, reduce the need to have relatively higher risk-tolerance for D-SIBs. Capital and funding gap between DSIBs and non-DSIBs 21. Concerns raised by standardised banks regarding the comparative capital levels held have been addressed through proposed RWA changes. These result in IRB capital requirements that are 96 percent of Standardised requirements (when also accounting for a 1% D-SIB buffer). Refer to #8499570 for more details. 22. However, another lens on the competitive advantage of D-SIBs is the lower funding cost that systemic institutions enjoy from a perceived government support, which is often reflected in their credit ratings. While we do not believe that an implicit government guarantee is as credible in New Zealand as in other jurisdictions (due to OBR), D-SIBs receive credit rating upgrades through parent banks who in turn have perceived Australian government support. As such, it could be argued that there is a distortion in the funding costs across the banking sector via this channel. 23. It is a non-trivial task to determine the appropriate pre-externality funding cost for D-SIBs. For one, D-SIBs naturally have a larger scale, more access to international capital markets, and more diversified portfolios. Some research has attempted to separate out the legitimate funding cost differences from implicit support impacts. For example, Cummings and Guo (2019) found there was a 27-32 bps cost advantage for Australian D￾SIBs issuing senior debt. Following the implementation of Basel III (including a D-SIB buffer), this fell to about 13-20 bps. 24. For New Zealand, we can first look at recent anecdotal evidence. Westpac and ASB (both D-SIBs) and Kiwibank (non-D-SIB) all recently issued senior debt notes with a Moody’s rating A1. Westpac and ASB issued at 85bps above the swap rate, whereas Kiwibank issued at 107 bps above. This difference could be indicative of legitimate funding advantages D-SIBs enjoy (although such anecdotes should be read with caution). 25. We also assess the nominal funding cost differences between D-SIBs and non-D-SIBs in New Zealand (excluding operating costs). This does not differentiate parental/sovereign￾support from legitimate D-SIB funding advantages, but does provide some useful context. For a 1% D-SIB buffer, the equity required for non-DSIBs is reduced and difference in average funding costs falls by 4 bps for mortgage lending. For other sectors non-D-SIBs are active in, the change is around 11 bps. However, for mortgages and rural lending the funding cost difference will already be relatively narrow from RWA changes and for consumer lending non-D-SIB banks would have a funding advantage. Furthermore, deposit insurance may equalise deposit rates across banks (depending on the final regime) and non-D-SIBs could get rating upgrades, further reducing funding cost differences.