Capital review - Next steps for the definition of capital

Ref #7882693 v1.0 MEMORANDUM FOR Financial Oversight Committee FROM Financial Policy (Principal author: Susan Guthrie) DATE 12 September 2017 SUBJECT Next steps for the definition of capital FOR YOUR Decision Recommendations: It is recommended that the Committee: a. Note the submission period for the second Capital Review consultation paper (the “Numerator Paper”) has closed. The Numerator Paper addressed the question “What should qualify as regulatory capital?” b. Note that, based on the information and arguments received to date, a core aspect of the consultation, namely removing contingent capital instruments from the framework, remains a preferred reform. c. Note that, further thinking may be required on the role of preference shares in our regime, given the relatively modest role they currently play, the self-reported lack of appetite for issuing the proposed instruments among banks, and the potential unintended consequences of retaining them in the framework. d. Note that the possibility of including a CET1 instrument suitable for banks structured as mutual societies seems worthy of investigation as part of the Capital Review. e. Note that, if the banks’ arguments against departing from Basel III are accepted, and taking into account the likely response of the big four banks to Option 4 (as articulated in their submissions), there are grounds for adopting a common equity-only regime in contrast to Option 4. f. Agree that Financial Policy should do further work looking at the role of preference shares in the regime and the possibility of including a new CET1 instrument suitable for banks structured as mutual societies. g. Note that a delayed, and extended, transition period may alleviate bank concerns and would be justifiable if it delivered an appropriate, effective, simple and enduring capital definition. h. Agree that steps should be taken to remove a remaining near term uncertainty for banks arising from the Numerator Paper. This uncertainty relates to the early recall provisions of contingent capital instruments issued under the current regime. i. Agree that a statement, articulating an in-principle decision with respect to contingent features in capital instruments in the capital regime going forward, should be made in the near future. j. Agree to the timeframes outlined at the end of the paper.

2 2 Introduction

1. A wide-ranging Capital Review was announced in March 2017. Two consultation papers have been released thus far, the first seeking feedback on the issues to be covered in the Capital Review (the “Issues Paper”) and the second seeking feedback on proposed reforms to the definition of capital (the “Numerator Paper”). The submission period for these two papers has now ended. A third paper, addressing the basis against which to measure capital adequacy, will be released shortly (the “Denominator Paper) and other papers are likely to follow.
2. The consultation period for the Numerator Paper ran for 8 weeks, beginning on 14 July and ending on 8 September.
3. 15 submissions have been received in response to the Numerator Paper. In addition, 22 submissions were provided in response to the Issues Paper, including quite detailed responses that relate to the definition of regulatory capital. Two meetings to discuss the Numerator Paper were held with banks: • There was a roundtable meeting sponsored by the NZBA and attended by Financial Policy, IRD, NZBA and representatives of the banks. The tax implications of contingent debt, and the reference to tax in the Numerator Paper, were the primary focus of the meeting. • A meeting was held with the Treasurer and other members of ANZ senior management. Much of the concerns raised by ANZ were subsequently reflected in bank submissions.
4. The feedback provided in submissions to both consultation papers, and feedback provided in meetings, have informed the preparation of this paper.
5. The purpose of this paper is to identify the areas of our consultation that we believe we should commit to, or, in the light of submissions, give some further thought to. Hence, on the latter point, this paper is a ‘direction of travel’ paper. New Option
6. Submitters have raised some important points that lead us to believe a new option (Option 6) for the numerator framework should be considered. This new option would retain desirable elements of the options we have consulted on, but have added elements. We provide a high-level outline of Option 6 in this paper, and explain why we think this Option is worth exploring further. This paper is not asking FSO for a decision to proceed with Option 6, rather it seeks FSO’s agreement to do further work on Option

3 3 7. Option 6 would introduce three new proposals, and retain some dimensions already consulted on. In particular, Option 6 would continue to have, as a core component, the removal of the contingent features of qualifying capital instruments. 8. It is worth noting that proceeding down the Option 6 path would require a further consultation, both because there are new features to consider, and because some of these features are likely to be contentious. Also note that even without the ‘new option’ further consultation would be required on the detail of, say Options 4 or 5 of the Numerator Paper, before they could be implemented. 9. By and large, most of the issues raised in submissions to the Numerator Paper were anticipated or raised in earlier submissions or meetings. Subsequent sections of this paper summarise the feedback we have received. However, there were three particular themes evident in the submissions that have prompted us to consider developing a further option, Option 6. 10. The first of these themes is a reasonably common view, and it is the preference ‘not to be unusual internationally’. The argument that is made here is that there is a commonly understood international framework available that includes contingent instruments, and departing from that framework will be costly. This view can overstated to a degree, as local tax, accounting, legal and resolution frameworks always imbue subtle and sometimes not-so-subtle differences to capital instruments. Moreover, a departure that is clear cut - such as removing the contingent feature of capital instruments - could be expected to be understood by markets. On the other hand, it is a reasonable proposition that markets will not want to invest the time needed to understand regional differences. 11. We would draw a distinction, however, between a regulatory framework that offers bespoke capital instruments, and therefore departs from international norms (for example, accepting instruments without contingent features), and one that merely offers a subset of the instruments accepted in the international regime. In the latter case, we are not at all persuaded by the argument that (undefined) costs arise simply from being different. 12. The second theme in submissions, relevant to Option 6, was raised by investment advisers. This was a concern about the impact, on the depth of the New Zealand capital markets, of removing contingent debt, and the ability to redeem preferences shares. While the advisers are ‘talking their book’, they do make a fair point. If market depth was adversely affected by reforms to the definition of capital, this would be an unintended consequence. Our focus, however, is on having a coherent and appropriate prudential framework, rather than market development. 13. The third theme relates to the small mutual banks. Co-op, TSB, SBS, and interestingly, Nelson Building Society provided a joint submission. These submitters raised concerns about the capital-raising challenges they face, and pointed to the benefits of including

4 4 common equity capital instruments that are suitable for banks structured as mutual societies. We have effectively introduced instruments of this nature for NBDTs, and overseas bank regulators, such as the APRA, OSFI and the Bank of England, have introduced these instruments, or appear about to. 14. Of these themes, the most important is the first - international comparability. While this may not have been the submitters’ intention, the arguments made against introducing bespoke capital instruments do not lead us to believe our concerns about contingent capital instruments (AT1s, T2s, and preference shares) are misplaced. Rather, these arguments raise questions in our mind about whether we should continue down the path of allowing bespoke instruments (as is envisaged in Option 4). Our hesitancy is reinforced by the large banks’ own arguments that they wouldn’t issue such instruments (the small banks may well wish to issue Option 4’s Tier 2 instruments to the local market). Option 4 does, of course, include common equity which is defined as per Basel III. 15. Moreover, the current composition of capital calls further into question whether it is worthwhile including bespoke capital instruments in the regime. These (non-contingent) instruments would replace contingent debt and contingent preference shares, but these instruments currently make up a relatively small proportion of total capital. 16. While Option 6, unlike Option 4, would remove preference shares and subordinated (non-contingent) debt from the list of qualifying capital instruments, banks could still issue these instruments (or contingent versions of them) for funding purposes if so desired. Option 6 would limit regulatory capital to common equity (defined and measured as per Basel III). 17. On the second issue, financial market development, to the extent we think this is a genuine issue, we could do further work exploring the potential impact of requiring banks to list a minimum volume of nominated instruments, for example (or something similar). There are academic arguments about the benefits of listing, including enhanced monitoring and market-discipline, and self-discipline incentives. This feature could be included as a module of Option 6, to address some stakeholder concerns about the market impact of removing qualifying instruments, or be a stand-alone piece of work to consider in due course. 18. On the third issue of capital instruments suitable for mutual societies, we think there is merit in positively responding to the call for work in this area. And, as with the above point, it would be all the more relevant if we were to resile from offering bespoke instruments into the capital framework. 19. We welcome feedback from FSO on these issues.

5 5 High level summary of the banks’ feedback 20. While the banks’ submissions covered a wide range of topics, at the centre of the big four banks’ concerns, and other submitters such as DeutscheCraigs, seems to be the prospect of Basel III-compliant, and APRA-aligned, capital instruments being removed from the capital framework. The small banks, while valuing the competition-levelling effects of removing contingent debt, would prefer for new instruments to be included in the regime, rather than removing contingent debt: “Rather than restrict capital products, New Zealand would benefit from widening its requirements as have other jurisdictions, such as Canada, where parliament has enacted legislation that enables co-operatives and mutual organisations to issue alternative instruments that qualify as Tier 1 capital.”1 21. Option 4 proposes removing contingent debt from the capital framework, departing from the types of preference shares that are permitted under Basel III and by APRA, and permitting a type of subordinated debt, that is not present in Basel III and in APRA’s regime, as Tier 2 capital. 22. The big four banks say they would not issue the bespoke instruments permitted under Option 4 or, if they did so, would issue them only to their parents. The banks say this response would have adverse flow-on effects, namely: the banks will have less diversified capital funding and so will be more reliant on parent entities; the cost of capital will increase; the Reserve Bank will have less information by which to assess the banks (some of the capital instruments currently permitted are traded on NZDX); New Zealand investors will have fewer options available to them and, related to this last point, the New Zealand financial markets generally will become less developed (i.e. fewer issues, a narrower range of instruments). Some banks also expressed concerns that foreign lenders would become less confident in New Zealand banks if the regime departed from Basel III. 23. The big four banks also submitted that they believe contingent debt provides effective loss-absorbing capital. 24. DeutscheCraigs said if the RBNZ wanted to adopt a more conservative stance it should simply increase the CET1 ratio, rather than cease to recognise internationally compatible instruments potentially issued by New Zealand banks. 25. Each of these issues is looked at in detail later in the paper. But by way of a general response, the banks appear to be over-stating the case against reform: 1 Joint submission from SBS, TSB, Co-operative Bank and Nelson Building Society.

6 6 • The capital that is relevant to the big four banks’ arguments - and according to those arguments, would not have been issued - is non-common equity issued to third parties. The amount of affected capital is very small, just $1.8bn compared to $26bn of common equity in the big four banks and total risk weighted exposures of $254bn. While total non-common equity issued by the big four banks is $8bn, just 22% of that has been issued to third parties.2 Not issuing the bespoke instruments permitted under Option 4 to third parties is unlikely to have any material impact on the cost of capital, or financial market development in New Zealand - the scale of the issuance has simply been too small to have this sort of effect. • One reason the instruments that would be removed from the regime under Option 4 have played a relatively minor role to date is simply because they only became eligible capital relatively recently (1 January 2013). The banks have not become dependent on these instruments partly because they haven’t had time to become so. This is one reason why it makes sense to act relatively soon to remove the instruments from the regime, given they are a poor fit for New Zealand - the banks are not, as yet, dependent on them to any significant degree. • If the Reserve Bank has an interest in deepening New Zealand’s capital markets, a relevant option is to look at requiring the big four banks to list a portion of their ordinary shares on NZDX. Work exploring this option has not been done, but could be included in the Capital Review. • If the big four banks opt not to take up the option of raising capital through the bespoke instruments included in Option 4, the cost of capital implications, and flow-on economic effects, can be expected to be modest. If, as the banks imply, they will be limited (by choice) to common equity capital only, replacing all currently issued non-common equity capital (including that issued to parent entities) with common equity is equivalent to increasing the CET1 ratio by 300 basis points in total or 60 basis points each year for five years. Based on international estimates, this could be expected to lead to an increase in lending rates of 15 to 24 basis points in total or up to 5 basis points each year for five years. 3 • Since 2014 the Australian parent banks have increased, or will soon have increased, their CET1 ratios by, on average, 250 basis points, so there is a 2 These figures include an estimate of w$1.0bn for WNZL ‘s AT1 issue to its parent, due to occur this week. 3 Refer FSO Paper (2016), Literature Reiew on Optimal Capital Ratios, FSO September 2016. Documentum # 6665392. The paper reported that recent estimates of the interest rate impact of a 100 basis point increase in the CET1 ratio ranged from 5 to 8 basis points.

7 7 precedent for this scale of change not being disruptive. Moreover, given the increase in common equity in the Australian parents, there appears to be plenty of scope for the parent entities to increase their equity investments in the New Zealand subsidiaries (where CET1 ratios have been flat or declining over the same period). • Considerations of how much capital domestic banks will issue domestically (versus offshore) should not be a factor in the design of the capital regime. What matters is certainty about the loss-absorbing effectiveness of reported capital, and the complexity of the capital regime. The depth of local financial markets reflects the needs of both local and foreign issuers, and not just banks, and local investors. There is nothing preventing Australian parent banks from actively raising capital here (as Westpac Australia did in 2016). To date, the big four banks have raised less than 30% of their capital though domestic issues, preferring instead to issue to their parent entities. • There is nothing about Option 4, or an equity-only regime for that matter, preventing New Zealand banks, or their parents, funding themselves with contingent debt. If there is indeed investor appetite for these instruments, this will surely remain the case. The only thing that will change under Option 4 (or an equity-only regime) is that the banks will not be able to include these instruments as reported regulatory capital. • Reducing the reliance on parent entities (the out-sourcing policy) is a separate issue to regulations that govern how banks structure their funding (the capital regime). Irrespective of where their capital is sourced, New Zealand banks must have processes in place that ensure they have the legal and practical ability to control and execute core functions (as required under the Reserve Bank’s outsourcing policy). 26. If giving banks the option of raising capital through bespoke (although long-established) instruments, in addition to common equity, is not likely to be taken up the banks, one obvious alternative to Option 4 is to limit total capital to common equity. This would: • reflect the banks’ indicated action; • take into account our concerns about the loss absorbing effectiveness of going￾concern contingent debt; • recognise the role of bank resolution policies; • deliver on our preference for a simple capital regime;

8 8 • apply a capital definition that complies with Basel III and APRA’s rules; • would be favourable for international lender confidence in New Zealand banks (as the ratio of CET1 capital to risk weighted exposures would be higher than it is today); and • deliver on the “relatively more conservative” principle. 27. If nothing but common equity is recognised in the regime, the capital constraints facing banks structured as mutual societies would need to be acknowledged. Under Option 4 these constraints are addressed by the removal of contingent triggers. However, alternatively, we could include in the capital definition an internationally-comparable CET1 instrument designed for mutual societies (indeed, the mutual banks’ submission requested that such an instrument be included in the regime). Instruments like this are permitted in Europe, the UK and Canada. APRA is currently consulting on such an instrument. 28. Appendix 1 to this paper contains a discussion of issues raised by submitters, but not discussed in detail below. Issue 1: Optionality in funding and market depth The submitters’ view 29. While they have predominantly issued contingent debt to their Australian parents, the big four banks indicated in their submissions that, they value having the option of issuing contingent debt to third parties. Depending on the circumstances, issuing to third parties directly might be more cost-effective for the group than having the parent issue on the bank’s behalf. This hasn’t proven to be the case very often in the past. Since the present regime took effect in 2013, only 27% of the contingent debt issued by the big four banks has been issued to third parties. Note: the WNZL 2015 Tier 2 issue showed how WBC and WNZL can depart. WBC issued AT1 to third parties and used it to purchase Tier 2 issued by WNZL. 30. The big four banks report that if New Zealand’s capital regulations depart from those of APRA (as Option 4 does), they would not issue capital instruments to third parties (the instruments accepted in New Zealand would not be recognised by APRA as group capital). The banks argue that this outcome would be to the detriment of New Zealand investors. While this outcome may be true of the big four banks, there is no obvious disincentive for the small banks to issue Option 4-compliant instruments.

9 9 31. WNZL and others said that Option 4 will make New Zealand banks more dependent on their parents to ensure adequate capitalisation. The banks say this is in contrast to Reserve Bank policy. ASB explained their concern as follows: “this will significantly increase these banks and the broader New Zealand economy to the economic risks profile and investment decisions of their parents. In addition, this outcome is contrary to the Reserve Bank’s broader principle of ensuring banks in New Zealand remain self-reliant, including access to external sources of capital and funding particularly in stressed scenarios.” 32. The banks also argue that contingent debt offers an investment option that is valued by New Zealand investors and, if contingent debt is no longer accepted as capital, this type of investment will disappear from the New Zealand landscape. The NZBA reports that, in contrast to what was reported anecdotally in the Numerator Paper, “participation by wholesale investors has occurred over time”. 33. DeutscheCraigs articulated the concern about the local market as follows: “We see a risk that if the RBNZ implements these proposals, and specifically removing conversion so that capital instruments will attract a tax haircut, or removing trigger events …then we will not see any capital issuance from the major New Zealand banks. At best we will see the Australian banks issue direct into the New Zealand market, as Westpac did in 2016. In our view this would be a significant detrimental change forced on the capital markets resulting in not only a restriction of the investment opportunities available to New Zealand investors but, importantly for the RBNZ, the removal of signalling provided from secondary market pricing of these securities.” Our response 34. There is nothing about Option 4 that would prevent the Australian parents (or their New Zealand branches) issuing contingent debt in the New Zealand market and claiming this as group capital (with the group using this funding to invest in the ordinary shares issued by the New Zealand subsidiary). While this would add marginally to the parent’s risk exposure to the New Zealand subsidiary, shareholders in the parent entity would be fully compensated for the added risk.4 35. In the event New Zealand banks restructure their capital increasingly towards common equity, as a response to Option 4, foreign lenders will have increasing comfort as to the level of creditor risk. Introducing a regime that leads to a greater reliance on common 4 The dividends earned by the parent from the ordinary shares issued by the New Zealand subsidiary will exceed the parent’s interest payments on the contingent debt. This income is sufficient to compensate for the fact that the New Zealand subsidiary may incur a loss without the AT1 debt issued by the parent being triggered to be loss￾absorbing.

10 10 equity seems consistent with the goal of ensuring New Zealand banks have reliable access to external funding and are thus self-reliant. 36. If the Reserve Bank placed considerable value of the information provided by market trading, it would seem preferable to require registered banks (that can issue ordinary shares) to list a portion of their issued shares on NZX. This would offer potentially relevant information, and help develop a deeper and more diversified local equity market. 37. New Zealand banks can be operationally independent of their parents irrespective of how they are funded. This is true whether one is comparing capital injections by parents or another party, or one is comparing common equity versus other types of capital. Once committed, the capital remains available to the New Zealand bank. Option 4 does not impact on operational independence. 38. We have no method of identifying investors in listed contingent debt instruments. Hence, banks are in the best position to know whether wholesale investors participate in this market. No figures were given by the NZBA so the scale of wholesale investor participation is not clear. New Zealand’s definition of “wholesale investor” includes entities that might nevertheless closely resemble “unsophisticated” investors. Issue 2: The cost of capital 39. Two important issues raised by submitters relate to the impact of the reforms on bank funding costs. The first issue relates to whether, and by how much, a change in the funding structure of a bank, prompted by the removal of contingent debt and reforms to preference shares under Option 4, may impact on the bank’s funding costs (its “cost of capital”), assuming there is no change in foreign investor confidence. The second issue relates to whether a change in the capital regime may impact on foreign investor confidence and thus the risk premium embedded in interest rates paid by New Zealand banks. 40. These two issues are discussed separately here for clarity, but in practice they are inter￾related. For example, foreign investor confidence may in part reflect how banks funding structures respond to the relative cost incentives present in Option 4, and the banks’ funding structures may, in turn, reflect banks’ views about how foreign investor confidence has changed as a result of Option 4 being adopted. 41. We do not attempt a complex analysis of these issues, reflecting the jointly determined nature of bank funding structures and foreign investor confidence. Instead, the paper explores a worst case, cost-of-capital and investor confidence-related, scenario under Option 4 and concludes that the likely impact on lending rates, and economic activity more generally, of the worst case scenario is quite modest. The funding structure issue

11 11 is discussed here, and the foreign lender confidence issue, and worst case scenario, is discussed later in the paper. The submitters’ view 42. The big four banks say that contingent capital is cheaper than common equity and perpetual preference shares, and so is worth keeping in the regime. These submitters report that, if only common equity or perpetual preference shares counts as Tier 1 capital, banks will be forced to rely on these types of funding - which the banks characterise as relatively more expensive than contingent debt - and the overall funding cost (each bank’s “cost of capital”) will increase relative to the status quo. The banks’ say the change in the funding structure will flow through into higher spreads on bank lending that will, in turn, lead to a lower level of investment and thus a lower level of potential economic output. 43. The banks’ articulated this concern in varying ways. For example, WNZL pointed to the higher targeted risk profile required of a bank which holds predominantly CET1 capital (the implicit assumption being that the required return - i.e. the cost of capital - has increased): “A framework based on Common Equity Tier One (CET1) incentives a bank to increase its risk profile in order to achieve an adequate return on a greater level of common equity.” 44. The DeutscheCraigs submission looked at preference shares in some detail, stating that preference shares (compliant with the current regime) and contingent debt have identical capital values - “economically the risks in respect of loss absorption (and hence delivering the same loss absorption outcome) are identical” - but that tax considerations make them more expensive. DeutscheCraigs explains that, from a bank’s perspective, since the risk is the same, the yield paid by the bank should be the same whether contingent debt or preference shares are issued. However, preference shares return less to investors if they cannot use the imputation credits provided with the preference shares. Thus the bank has to pay more to investors to issue preference shares. Foreign investors and charitable trusts, for example, cannot typically use imputation credits. 45. DeutscheCraigs estimate that, because of these tax-related considerations, preference shares cost banks more than contingent debt, with the margin historically ranging from 20 to 40 basis points. Changing the regime to remove contingent debt will therefore lead banks to issue relatively higher-yielding capital.

12 12 Our response 46. Balancing this argument about adverse economic impacts several points can be made. Firstly, we know from the theoretical literature (for example, work related to the theorems of Modigliani and Miller), and empirical studies, that changes to a bank’s funding structure, that see greater weight given to equity, do not typically lead to an increase in the banks overall cost of capital of the magnitude implied by a simple comparison of pre￾restructure relative yields applied to the new funding structure. Typically the weighted average cost of capital will increase with the shift towards relatively more equity, but not by as much as the simple comparison of pre-restructure yields (or returns required by investors) would suggest. 47. There are several reasons for this. Increasing common equity at the expense of contingent debt, for example, can be expected to reduce the volatility of the bank’s equity price (or value) and thus reduce the risk premium associated with that equity. This would lead to a lower required return from common equity, an outcome that would moderate the increase in the bank’s overall funding costs (compared to what otherwise would have occurred). 5 48. It is not just the equity risk premium that might benefit from a switch towards common equity at the expense of contingent debt. Borrowing costs may decline as well.6 Many studies have found evidence that increases in the ratio of common equity to aggregate risk exposures, for example, leads to a decline in borrowing costs and the decline may be sufficient to prevent the overall cost of capital increasing. An example is a BIS working paper from 2016: “In particular, we find that a 1 percentage point increase in the equity-to-total￾assets ratio is associated with a reduction of approximately 4 basis points in the overall cost of debt funding (deposits, bonds, interbank borrowing, etc.)….A back of the envelope calculation indicates that the greater retention of net income by the bank as retained earnings would almost pay for itself through lower cost of debt, even if the cost of equity, typically approximated by the Return on Equity, is presumed to be quite high.”7 49. In the literature review completed by Financial Policy in 2016, the empirical studies reviewed indicated that, on average, the pass through of a change in the funding structure of a bank to the bank’s cost of capital is just under 50% - i.e. the change in the 5 Bank of England (2015). Measuring the macroeconomic costs and benefits of higher UK bank capital requirements. BoE Financial Stability Paper no. 35 December 2015. Page 9. 6 Bank of England (2015). Measuring the macroeconomic costs and benefits of higher UK bank capital requirements. BoE Financial Stability Paper no. 35 December 2015. Page 21. 7 BIS working paper #558 (2016)

13 13 cost of capital is, on average, just under half of what would be implied by applying pre￾restructure relative yields or required returns to the new funding structure. 8 50. Secondly, it appears that a modest increase in the cost of capital has only a negligible impact on economic activity. Hence, accepting that a shift towards common equity and away from continent debt is likely to increase the cost of capital, but not by as much as a simple comparison of pre-restructure relative yields would suggest, the economic impact of a modest increase in the cost of capital seems likely to be relatively minor. 51. Financial Policy conducted a literature review in 2016 that looked at the impact of higher levels of equity on lending rates. 9 10 An increase in equity, expected to give rise to some increase in the cost of capital, had a relatively minor impact on lending rates: “…recent estimates suggest that a one percentage point increase in banks’ tier 1 capital ratios leads to around a 5-8 basis point increase in lending rates.” 11 52. On a related note, a 2015 Bank of England study reported that, for every 1 percentage point increase in a risk-weighted capital ratio, lending rates in the UK would increase by 5-10 basis points and that would lead to permanent annual output losses in the UK of just 0.01% to 0.05% of GDP.12 13 APRA is of the view that an “instantaneous 100 basis point increase in the CET1 capital ratio of the four major banks would, if entirely passed on through repricing of loans and deposits, require an increase in margins of approximately 10 basis points.”14 53. Thirdly, there are aspects of Option 4 that directly reduce bank capital costs. Under Option 4 banks can issue Tier 2 subordinated debt without contractual write off (unlike the present rules) and this means the instruments will potentially be less expensive to banks than current Tier 2 instruments (contractual write off arguably adds to the cost of debt, as it is disadvantageous for investors). Moreover, it is unlikely such instruments will 8 Refer FSO Paper (2016), Literature Review on Optimal Capital Ratios, FSO September 2016. Documentum # 6665392 9 Refer FSO Paper (2016), Literature Review on Optimal Capital Ratios, FSO September 2016. Documentum # 6665392 10 Bank of England (2015) also included a literature review. Measuring the macroeconomic costs and benefits of higher UK bank capital requirements. BoE Financial Stability Paper no. 35 December 2015. 11 Refer FSO Paper (2016), Literature Review on Optimal Capital Ratios, FSO September 2016. Documentum # 6665392 12 see Bank of England (2015). Measuring the macroeconomic costs and benefits of higher UK bank capital requirements. BoE Financial Stability Paper no. 35 December 2015. Page 21 13 Refer FSO Paper (2016), Literature Review on Optimal Capital Ratios, FSO September 2016. Documentum # 6665392 14 APRA (2017) Information Paper: strengthening banking system resilience - establishing unquestionably strong capital ratios. 19 July 2017. Page 36.

14 14 require a tax haircut, unlike the situation for some Tier 2 instruments at present. Similarly, removing the contractual write off/conversion of preference shares would potentially reduce the costs of issuing these instruments (vis a vis instruments with these features). 54. From the small banks perspective Option 4 certainly offers the prospect of cheaper Tier 2 capital (relative to the cost of Tier 2 to these banks under the present regime). Because they have no offshore parent to issue to, these banks have typically accepted a tax haircut on Tier 2 issues under the current regime. 55. Fourthly, a particular aspect of trans-Tasman tax policy has been a significant factor contributing to the low cost of contingent debt relative to common equity for the big four banks. This tax efficiency opportunity is due to be removed in 2018. This follows a recent decision by the Australian Tax Office (ATO) and the prospect of a coordinated policy response from the ATO and IRD. Issue 3: Foreign lender confidence The submitters’ view 56. Foreign investor confidence impacts on banks’ funding costs. Associate Professor Martin Lubberink believes international lender confidence would be adversely affected by a departure from Basel III. His view is that New Zealand banks would be disadvantaged if New Zealand deviates from the international Basel standards by removing contingent debt from the capital framework. The banks are of a similar view, stating that deviating from international standards will make it hard for international lenders to assess New Zealand banks. 57. It is presumably this concern that lies behind the NZBA’s, and individual banks’, claims that Option 4 would lead to a “capital constraint” on New Zealand: “The New Zealand banking system will always be highly reliant on offshore wholesale funding to meet its operational and regulatory capital requirements. The adoption of RBNZ’s proposals has the potential to introduce a capital constraint to the New Zealand economy in addition to significant macroeconomic cost impacts through higher funding costs.”15 58. The NZBA also raised the issue of foreign lender confidence directly: “RBNZ’s proposals would bring significant uncertainty and undue complexity for international investors and may impact stability of international funding for New 15 NZBA submission to the Capital Review Paper 2: what should qualify as bank capital? Issues and Options. Page 3

15 15 Zealand banks….NZBA queries whether RBNZ has sought comment from the major rating agencies on the potential implications of Option 4 on ratings of New Zealand banks. We understand that each rating agency has its own method of determining capital adequacy and the change proposed by the RBNZ could have significant impacts on those calculations, and therefore on the New Zealand banking industry’s reputation internationally.” Our response 59. It seems unlikely that Option 4 would lead to a significant rise in the risk premium and thus the cost of borrowing for New Zealand banks. There are many reasons to hold this view. For example, international lenders may place considerably more weight on common equity capital (CET1) than AT1 and total capital, when assessing the strength of the New Zealand banks, and the definition of CET1 is unchanged under Option 4. Indeed, CET1 capital certainly seems relatively more important to foreign investors than other types of capital: • APRA’s recent enquiry into what constitutes “unquestionably strong” capital ratios focused on CET1, for example, stating that the CET1 ratio is “…most likely to engender confidence”. 16 • Related to the above point, APRA reports that the capital adequacy assessments done by ratings agency S & P, for example focuses on capital that is “roughly equivalent to Tier 1 capital” (not total capital).17 60. International ratings agencies have expressed a preference for less complex capital regimes. The value of simplicity in capital regulations formed the basis of S&P’s 2010 submission to the Basel Committee on the capital standard proposals for example.18 61. In addition to capital, other factors such as bank governance, internal risk processes, concentration risk in the lending book and so on can be expected to impact on the lending decisions by foreign investors. These considerations are not affected by Option 4 (and may, in fact, evolve in a way that supports lower borrowing costs, as a result of other changes delivered by the Capital Review). The financial strength of the Australian parent may also be a factor influencing the borrowing costs of the big four banks. 16 APRA (2017) Information Paper: strengthening banking system resilience - establishing unquestionably strong capital ratios. 19 July 2017. 17 APRA (2017) Information Paper: strengthening banking system resilience - establishing unquestionably strong capital ratios. 19 July 2017. 18 Stand and Poor’s submission to the 2009 consultation document published by the Basel Committee. “Standard & Poor’s Response to the Basel Committee’s Proposals On Bank Capital and Liquidity. April 15 2010. Other submissions, from other organisations and S&P, are publicly available as well.

16 16 62. We should also not overlook the possibility that, under Option 4, foreign lenders may develop a level of comfort with the unique features of New Zealand’s regime and thus accept the level of reported total capital under Option 4 as being a meaningful measure of total capital. Option 4 does not propose novel instruments, after all, it merely proposes a move to funding instruments similar to those accepted under the previous Basel regimes (Basel I and II). 63. New Zealand has departed from international norms before, when introducing the statutory management regime for example, with no loss of investor confidence. Foreign investors, and ratings agencies and others that inform them, have been found in the past to be willing to invest the resources required to understand the particular nuances of the New Zealand regime. 64. It is not the case that, in practice, complying with Basel III leads to common instruments being issued in all jurisdictions. This is because, while the high level requirements included in the Basel III standards are adopted by compliant regimes, the legal and other regulatory frameworks, and accounting norms, differ meaning that, in practice, the contractual features may differ. These local circumstances are taken into account by ratings agencies and foreign investors. 65. If we assume the worst case scenario under Option 4, namely that total capital is the only factor driving the risk premium applied to New Zealand banks (not CET1 capital), and only CET1 capital held in New Zealand banks is recognised by international lenders, the departure from Basel III would mean that New Zealand banks would have to increase their level of CET1 capital. 19 Only in this way could they maintain international investor confidence (and thus the risk premium) and achieve their targeted borrowing at interest rates that would have prevailed had there been no reform. 66. This would also constitute the “worst case scenario” in terms of the cost of capital implications of banks restructuring their funding as a result of Option 4 - in this scenario we assume the banks replace all non-compliant capital with common equity. 67. If Tier 1 capital, not total capital, is the relevant factor, then the estimates of the new common equity required under Option 4’s worst case scenario, and related costs, reported in paragraphs 35 to 43 below, need to be reduced by a third. If CET1 capital is the relevant factor, no new CET1 capital would be required at all. 68. In aggregate, in this worst case scenario, assuming no grandfathering of existing instruments, the big four banks would immediately have to raise $8bn in new common equity capital (rather than newly-compliant types of preference shares and subordinated debt), using the proceeds to recall $8bn raised from AT1 and Tier 2 instruments that 19 The banks would have to hold sufficient common equity to meet whatever Tier 1 or total capital benchmarks the international lenders apply, since any other capital instruments recognised in the New Zealand regime would not be recognised by international lenders.

17 17 would be non-compliant under Option 4.20 $8bn equates to 3% of the total risk weighted exposures of the big four banks. 69. In practice, it is likely the non-compliant instruments would be grandfathered, for example with 20% of the face value ceasing to be recognised as capital each year for the first five years after the new capital definition came into effect. This means, each year, for the first five years of the new regime, common equity held in the big four banks would have to increase by $1.6bn. This equates to 0.6% of aggregate risk weighted exposures. In effect, under Option 4’s worst case scenario, the big four banks would face the equivalent of an increase in the CET1 ratio of 60 basis points. 70. To put the $1.6bn annual requirement in context, the big four banks paid out $2.6bn in dividends on ordinary shares in the past financial year. Bank data relevant for the worst case scenario 21 Total contingent debt/pref shares22 $bn Risk weighted exposures $bn Ratio of current contingent instruments/RWE % Annual implied new common equity issuance $bn Most recent paid out annual share dividend $bn ANZ 2.7 87.1 3.1% 0.5 1.4 ASB 1.8 52.9 3.4% 0.4 0.2 BNZ 1.4 60.7 2.3% 0.3 0.3 WNZL 2.0 53.8 3.7% 0.4 0.7 Total big four 8.0 254.5 3.1% 1.6 2.6 71. Relevant context is also provided by APRA’s recently released assessment of what constitutes “unquestionably strong” capital ratios.23 APRA reported that, in the case of the four largest Australian banks, achieving “unquestionably strong” status requires an increase in CET1 capital ratios of, on average, 100 basis points above the December 2016 levels. This increase will come on top of increases that have been occurring since the publication of the first Financial System Inquiry report in mid-2014. APRA says that, since mid-2014, “in meeting this new benchmark …the four major banks will have, on average, increased their CET1 ratios by the equivalent of more than 250 basis points”.24 72. APRA has indicated that banks will be expected to deliver the additional 100 basis points to the CET1 ratio over the next two and a half years, and ideally sooner. The increase in New Zealand bank CET1 ratios, implied by the worst case scenario under Option 4, is 20 This value will be an over-estimate if any of the banks subsequently voluntarily take new tax haircuts on existing capital instruments. 21 Data sourced from disclosure statements prepared by each bank for their financial year end. 22 This figure reflects all contingent debt and AT1 preference shares issued by the big four banks up until the date of this report (i.e. it includes those instruments issued after the banks’ most recent annual reports). The reported WNZL figure has been increased by $1.0bn to reflect the new AT1 instrument issued by WNZL on September 7th . 23 APRA (2017) Information Paper: strengthening banking system resilience - establishing unquestionably strong capital ratios. 19 July 2017. 24 News release announcing the APRA (2017) Information Paper: strengthening banking system resilience - establishing unquestionably strong capital ratios. 19 July 2017.

18 18 similar in scale, and potentially deliverable over a similar time frame, as the increases that have occurred, or are due to occur, in Australia since mid-2014. The increases that have occurred in Australia do not appear to have been disruptive to the banks or to the wider economy. 73. Also relevant context is the fact that, in the recent past, in contrast to the upward trend in the CET1 ratio of Australian parents and their international peers25, with one exception, the reported CET1 ratios of the New Zealand big four banks have declined. APRA reported that between June 2014 and December 2016 the reported CET1 ratios of the four large Australian banks rose by 100 basis points (or 250 basis points using internationally comparable measures). 26 This compares to an average decline of 23 basis points in the reported ratios of New Zealand’s big four banks over the same period, and an essentially flat trend overall. Given this relative “slippage” in CET1 ratios compared to overseas banks, an increase in the CET1 ratio for New Zealand banks seems desirable, irrespective of whether it is delivered as an outcome of the worst case scenario under Option 4 or not. New Zealand banking group, change in reported CET1 ratio June 2014 to Dec 2016 Dec 2016 to June 2017 Total period since June 2014 ANZ -80 20 -60 ASB -60 50 -10 BNZ 110 2 112 WNZL -60 40 -20 Simple average of the above -23 6 74. The increase in the large Australian banks’ CET1 capital ratio is also relevant for another reason. Martin Lubberink believes it would be “prudentially unsound” for the Australian parent banks to fund an increase in their common equity investment in New Zealand subsidiaries with anything other than an increase in their own common equity. This arguably over-states the case given that shareholders in the parent would be fully compensated for any additional risk the parent was exposed to by funding their investment in other ways. 75. However, even assuming funding an equity investment with anything other than equity is “prudentially unsound”, the increase in the common equity held in the Australian parent banks since 2014 means there is no obvious constraint on the parents’ ability to acquire larger equity exposures to the New Zealand subsidiaries, of the scale required in the worst case scenario (and, as shown above, in most cases, the additional common equity 25 APRA (2017) Information Paper: strengthening banking system resilience - establishing unquestionably strong capital ratios. 19 July 2017. Page 20 26 APRA (2017) Information Paper: strengthening banking system resilience - establishing unquestionably strong capital ratios. 19 July 2017. Page 17

19 19 required of New Zealand banks, under the worst case scenario arising from Option 4, could be achieved simply by suspending dividends). 76. The smaller banks have tended to be less reliant on contingent debt. Of the small banks, Cooperative Bank and Kiwibank have the highest ratio of contingent to debt to risk weighted exposures (2.8% and 2.7% respectively). Unlike the big four, the new issuance of common equity, required under Option 4’s worst case scenario, would exceed the amount most recently paid out as dividends on ordinary shares and these banks, under the worst case scenario, would need to grow more slowly, retaining more profits, in order to retain their risk premium. 27 77. In summary, at worst under Option 4, banks would have to replace existing capital instruments valued at 0.6% of risk weighted exposures with common equity capital for each of the first five years in order to maintain international lender confidence. There would be no requirement for the parents to issue large volumes of new equity to fund this increase because suspending dividends would typically be sufficient, and in any event, common equity in the parents (which provides a suitable basis for equity investment in the New Zealand subsidiaries) has been increasing in recent times. 78. Against the possibility that banks may have to increase common equity by very modest amounts, in order to maintain international lender confidence, we have the prospect of, under every scenario, significant benefits arising from having more certainty as to the loss-absorbing effectiveness of bank capital and a less complex regulatory regime. The capital regime will be easier to comply with, easier to administer and better able to handle shocks.28 79. It is important to note that the above analysis is based on the assumption that there are no changes to the minimum required CET1 ratio, or to the calculation of the risk weighted exposures that form the denominator of the CET1 ratio, occurring at the same time as changes are introduced to the capital definition. The combination of these factors 27 Bank data relevant for the worst case scenario27 Total contingent debt/pref shares $m Risk weighted exposures $m Ratio of contingent instruments/RWE % Annual implied new common equity issuance $m Most recent paid out annual share dividend $m Co-op 32 1,124 2.8% 6.4 2.1 (rebates) Heartland 16 3,742 0.4% 3.2 37.7 Kiwibank 255 9,431 2.7% 51.0 29.0 SBS 22 1,701 1.3% 4.4 0.6 28 Andrew Haldane, in a speech made at the Federal Reserve’s economic symposium Jackson Hole, Wyoming 31 August 2012, made the point that complex regulation is not only “costly and cumbersome, but sub-optimal for crisis control”.

20 20 is what ultimately impacts on banks. These issues are being addressed separately in forthcoming consultation papers, released as part of the Capital Review. Issue 4: contingent debt as going-concern capital The submitters’ view 80. Several submitters expressed the view that AT1 contingent debt will trigger in time to be useful on a going concern basis: 81. Some banks said continuous disclosure requirements “make it highly unlikely that a deterioration in a bank’s financial position would not be recognised.”29 82. The NZBA (incorrectly) says that the triggers involve the Reserve Bank and this means the trigger will be timely. 83. ASB favours the retention of a going concern trigger and potentially increasing it consistent with the trend in Europe (the UK, for example, has increased the AT1 trigger from a CET1 ratio of 5.125% to 7%). 84. The NZBA and others questioned the Numerator paper’s reference to the recent performance of contingent instruments issued by European banks. The NZBA said the economic, political and legal contexts may matter and without this context the relevance of the European examples is unknowable. The NZBA said the Reserve Bank should not base its view on the timeliness of the AT1 trigger based on one case study (Banco Popular). 85. WNZL and the NZBA said non-viability triggers were valuable because of the incentives they provide to management to effectively manage risk. Our response 86. The banks’ claim about the value of disclosure misses the point that accounting information, that forms the basis of disclosure, is a lagging indicator of a bank’s financial position. 29 NZBA submission to the Issues Paper

21 21 87. The NZBA’s claim that the RBNZ is involved in activating the contingent trigger is incorrect - the statement is not true in the case of AT1 capital. 88. The core argument against contingent debt’s effectiveness as going concern capital is timeliness of the trigger. A widely held view in the theoretical literature is that timeliness is an inherent problem with accounting-based triggers (and no other realistic triggers have been developed at this time). The view expressed in the Numerator Paper reflects this literature. 89. The empirical examples included in the Numerator Paper are the only examples available globally from the recent, post-GFC period. Despite NZBA’s claims to the contrary, the Banco Popular case has a relevant lesson for New Zealand - namely, as the literature would lead one to expect, the reported value of the bank, based on internationally comparable accounting standards, was overstated to such a degree the going-concern contingent debt did not trigger until the bank was non-viable. 90. As a general point, several submitters (Martin Lubberink, the NZBA) made no acknowledgement of the fact that Tier 1 capital is more valuable to a bank, from a ratings perspective, than Tier 2 (only Tier 1 capital is considered by S&P when calculating the S&P risk adjusted capital ratio for example). When it came to discussing the loss absorbing effectiveness of contingent debt, Martin Lubberink and the NZBA did not address the most important contingent debt issue - giving effective contingent debt Tier 1 status. 91. In relation to the incentive effects of non-viability triggers, in contrast to WNZL’s and the NZBA’s view, only going-concern triggers (if effective) provide incentives to management to reduce risks. This is because only when a bank has some value (i.e. remains viable) do shareholders have any value to protect. Non-viability triggers only occur once a bank has no value - at this point there is no value remaining in the business and ordinary shareholders, who still own the bank at this point, can be assumed to be indifferent to who owns the bank going forward. The prospect of losing control of the bank when it ceases to have any value does not impact on management incentives throughout the life of the instrument. 92. The issue of delayed triggering of going-concern contingent debt was extensively covered in the Numerator Paper and none of the information or points of view contributed in the submissions was sufficient to counter the view expressed in the Numerator Paper. Issue 5: Preference shares The submitters’ view

22 22 93. Option 4 proposes to recognise only non-redeemable preference shares as bank capital. Submitters were unanimously of the view that preference shares, accepted as capital, should be redeemable. The banks report there is “no investor appetite” for non￾redeemable preference shares. Hence these banks do not support removing the optional call from eligible preference shares. 94. The banks also say the fact that, under the current rules, preference shares cannot be redeemed without Reserve Bank agreement, guarantees its capital qualities. 95. WNZL and others said the contingent feature of preference shares is essential because, removing it will “inhibit the ability of a bank to be recapitalised”. WNZL appears to view preference shares as debt, and recapitalisation as entirely made up ordinary share capital: 96. “The liability under a preference share, without the elements of conversion or write off, remains outstanding. This can be an impediment to bank recapitalisation…” 97. DeutscheCraigs said that, while under Basel II eligible preference shares did not have a redemption feature, banks got around this by simple structuring strategies. Another entity (typically related to the bank) was given the right to buy the preference shares from investors. The submission says such strategies would simply emerge again if redeemable preference shares are not permitted in the regime. Our response 98. Preference shares play a very small role in the current regime. ANZ issued $300m in 2013, and ASB in 2015 and 2016 ($400m each time). In all cases, the issues were to parent entities. Of the $8bn of non-common equity issued by the big four banks (of which $1.8bn was to third parties), only $1.3bn has taken the form of preference shares (none to third parties). 99. We have no way of assessing the factual accuracy of the claim that there is no investor demand for non-redeemable preference shares, but it does beg the question whether there would, in fact, be investor demand if an appropriate rate of return was offered. Hence, what the banks appear to be actually saying is that they are not prepared to pay the rate of return required by investors in non-redeemable preference shares. 100. It is also difficult to reconcile the view that there is no investor demand for non￾redeemable preference shares with the weight the banks attach to Reserve Bank approval for early redemptions - if there is no certainty that the preference shares will redeem, then this feature should be heavily discounted by investors. 101. WNZL’s appears to have a narrow view of “recapitalisation”, namely that recapitalisation is limited to an increase in the capital attributable to ordinary shares. Only in this context does WNZL’s comment about recapitalisation make sense.

23 23 However, including a contingent term in a preference share does nothing to increase the value of the bank once it has become non-viable, it merely protects the claims of existing shareholders in the (unlikely) event the bank is resurrected and/or there are surplus assets. It does this through writing off the claims of preference shareholders or by creating a new group of ordinary shareholders who now rank equally with existing shareholders, rather than ahead of them as before (conversion). 102. The DeutscheCraigs feedback is relevant and important. It illustrates that not allowing redemption for preference shares may be difficult to enforce in practice. Including non-redeemable preference shares may undermine the goal of having a relatively simple capital regime. 103. If we were to depart from Option 4, and continue to accept contingent triggers in preference shares, but remove contingent debt from the capital definition, we would create incentives for regulatory arbitrage. Banks would have an incentive to structure debt to look like preference shares, in order to capture the contingent trigger feature. Moreover, the legal uncertainty issue remains in relation to contingency and preference shares (there is some uncertainty about the legal effectiveness of means to deliver write off, given New Zealand corporate law, which is an essential part of contingency). 104. Given the arguments against contingency in preference shares, we would strongly advise not recognising Basel III-compliant preference shares in the regime (Basel III requires contingency). Thus even if we agree that permitting redeemable preference shares makes practical sense, Basel III-compliant preference shares are not a good fit with the regime. This would be particularly so if, as we recommend, contingent debt is removed from the regime. 105. We thus have two realistic options with respect to preference shares, assuming contingent debt ceases to be eligible capital: • including redeemable preference shares, or • not including preference shares at all. In this case Tier 1 would consist only of common equity. 106. Of these two options, given the apparent limited appetite from banks and investors, and in the interests of having a relatively simple regime, removing preference shares from the regime seems the better option. However, further analysis may be required on this issue. 107. If preference shares are not included in the regime, the limited capital options for banks structured as mutual societies could be addressed through including an appropriate CET1 instrument for these banks, rather than preference shares. Issue 6: Tax considerations

24 24 The submitters’ view 108. The big four banks seemed to be of the view that the tax-related complexity of contingent debt, and Reserve Bank concerns about the tax treatment of contingent debt, were primary factors in the Reserve Bank arriving at Option 4 as the preferred reform option. These banks, in meetings and in their submissions, argued that contingent debt is taxed appropriately, that the tax effects of contingent debt are not difficult to work out, and that where conversion is envisaged no tax haircut is warranted (in fact, we have no view on the first point and disagree on the last two points) 30. 109. In their submission, the NZBA included a report by PwC which showed that, among other things, by imposing tax haircuts when an instrument is issued, the Reserve Bank is an outlier internationally. The banks therefore propose that, if we are concerned about level playing field arguments, we should keep contingent debt instruments in the regime and simply remove the tax haircut provision. 110. More generally, the big four banks said that the tax-related aspects of contingent debt feature too prominently in the Numerator Paper and they strongly disputed that any “regulatory arbitrage” has gone on. All of these are rather side issues in terms of the proposed reforms. Our response 111. It is not the case that the Reserve Bank has concerns about the tax treatment of contingent debt (it is not part of our mandate), nor do we have concerns about the 30 On the issue of whether the tax outcome of contingent debt absorbing losses is clear and well-established in tax rulings, we disagree with the banks. We disagree because there are circumstances where there is a clear contractual intention to not issue new shares, with loss absorption being achieved by write off. This is an outcome clearly envisaged and intended in many contingent debt contracts. Until recently, this intention has not been addressed in the tax rulings provided by banks (since 1 April 2017 this omission is relevant only for debt issues to third parties, not intra-group issues. Prior to then, it was potentially relevant for all contingent debt issues that included conversion terms). We also disagree with the banks when it comes to the reasonableness of imposing tax haircuts at all. The banks claim that the likelihood tax would be payable by a bank when a contingent debt instrument is triggered is negligible. They say that, at the time an instrument is triggered it is likely that conversion will occur (i.e. shares will be issued) and, even if not, substantial tax assets would be available to the bank meaning no payment would have to be made to IRD. In contrast, our view is that it is that it is quite possible conversion may not occur and for a number of reasons a bank may not be able to utilise any accumulated tax assets (if any exist). The issue of probability was carefully considered and dismissed during the process leading to the introduction of Basel III in New Zealand. However an element of “slippage” emerged in practice with the Reserve Bank accepting narrow tax rulings from banks.

25 25 complexity of the tax treatment of contingent debt (although this has made oversight of capital instruments difficult). What we are concerned about is the added complexity of the financial arrangements used to issue capital, with complex features put in place in response to the tax treatment of contingent debt. 112. What seemed to be motivating the big four banks’ comments about tax was frustration and uncertainty about the recently notified, new interpretation of the current policy (namely that comprehensive tax rulings are required, not rulings that simply focus on shares being issued), rather than the tax implications of Option 4. 113. In fact, under Option 4, tax only arises as a potential issue in the context of Tier 2, and it is not clear that any tax haircuts will be required of long term subordinated Tier 2 debt under Option 4. Hence, based on tax considerations, Option 4 is an improvement for banks relative to correctly applying the current tax-related requirements for Tier 2 capital. The benefit of moving to Option 4 is especially obvious for the small banks that cannot issue ordinary shares - to date these banks, unlike the big four banks, have taken tax haircuts on their Tier 2 debt instruments. Issue 7: Grandfathering The submitters’ view 114. Grandfathering is of potential interest to all banks. The NZBA has asked for full value recognition for instruments up until the first optional call date, from which point the instruments would cease to qualify. “…if RBNZ pursues Option 4, NZBA considers that it should continue to recognise existing AT1 and T2 instruments until their first call date (at which point they would no longer qualify as capital). NZBA considers that approach is preferable (compared to the RBNZ proposal of a five-year phase out) as the existing AT1/T2 instruments have been issued in good faith and at considerable cost under Basel III rules that are still relatively new. Additionally, the existing AT1/T2 instruments continue to provide strong loss absorbency and we see no reason to disqualify these instruments prior to their first call date.

NZBA considers that banks should be able to rely on regulations being effective throughout the expected life of each transaction, unless there are exceptional reasons to change. In our view, no such exceptional reasons exist here.” 115. Our response RBNZ s105

26 26 116. An important general principle is that reported regulatory capital levels must reflect what the Reserve Bank recognises as being effectively loss absorbing. A deviation from this general rule can be accepted for a short time as banks transition to new requirements, but not for an extended period. 117. However, there are concessions that can be made here. On the one hand, given the relatively modest implications for bank capital costs, arising as a result of removing contingent debt and contingent preference shares from the capital regime, it seems reasonable to begin the grandfathering process immediately the new rules take effect and to complete the transition over 5 years. This seems preferable to permitting banks to report the full value of non-compliant instruments for some (potentially considerable) period after the rules change. 118. However, on the other hand, it is arguably more important to have a definition of capital that is a good fit with New Zealand, delivers effective capital, and is likely to be enduring, than to take a hard line on capital reporting. Allowing the grandfathering period to have a delayed start, or to transition over a period beyond five years, would alleviate the (albeit limited in our view) cost pressures on banks and may address some of their immediate concerns. Issue 8: The Capital Review process The submitters’ view 119. The NZBA and several banks signalled that they think the timeframe signalled for the Capital Review is to short. This reflects two concerns: • insufficient time for adequate consultation • The Capital Review will be finished before the current Basel and APRA reviews conclude 120. At a minimum, the NZBA and the banks sought a detailed timetable for the Capital Review, through to completion. 121. WNZL said it “would strongly discourage the Reserve Bank’s approach to implementing changes on a ‘rolling basis’. We consider that it is prudent to work through all aspects of the review before fixed policy decisions are made so the RBNZ, industry and the broader NZ economy can understand the potential impacts of what is being proposed; the potential increases in costs to both banks and the wider economy including any constraints on the availability of capital”.

27 27 122. A number of other banks said a comprehensive cost benefit analysis, incorporating all reforms that emerge from the Capital Review, including changes to the capital definition and the minimum required capital level, should be done. Our response 123. The scope of the Basel and APRA reviews is limited to the denominator, and they do not include the definition of capital. These reviews are not grounds to delay considering the definition of capital. Recommended reforms to the definition of capital 124. Arguably the most important feedback received from the banks is their unwillingness to issue bespoke capital instruments that do not comply with Basel III or align with APRA’s requirements. For reasons explained earlier, this suggests it may be better to adopt a CET1-only capital definition, with inclusion of a suitable CET1 instrument for banks structured as mutual societies, rather than Option 4. In other words, having a common-equity only regime may be more desirable than Option 4. 125. Irrespective of what reform is adopted, it is important to consider the cost of capital implications of removing contingent debt from the regime, and the implications of the reforms for international lender confidence: • If the reforms deliver a change in the funding structure, towards a higher CET1 ratio, all else being constant, the cost of capital implications seem likely to be relatively modest. In contrast, the gains in terms of capital quality, and less complexity in the regime, are significant. The pass￾through of any pre-reform cost difference (between common equity and the debt or preference shares it replaces) is likely to be less than 50%, because of beneficial effects of the reforms on the return required of equity and borrowing rates. Moreover, the economic effects, transmitted through a subsequent modest increase in lending rates, are expected to be small. • It seems possible that foreign investor concerns might reinforce the banks’ commitment to focus on increasing common equity under Option 4, rather than issuing bespoke capital instruments permitted under Option 4. Under the worst case scenario, banks would have to replace all non￾common equity with common equity, over a period of years. This would be equivalent to an increase in the CET1 ratio of 300 basis points in total, or 60 basis points each year for five years.

28 28 • If the reforms lead to a higher level of CET1 capital in the banking sector, this is likely to assuage any foreign investor concerns about novel aspects of the New Zealand regime. 126. Baring new information, at this point there appear to be no compelling arguments to deviate from a key feature of the preferred reform package presented in the Numerator Paper as “Option 4” - namely removing contingent debt from the regime. 127. In contrast to what was originally proposed as Option 4, it may be preferable to include in the regime a new CET1 instrument, suitable for banks structured as mutual societies, rather than including non-redeemable preference shares (without a trigger) and Tier 2 capital subordinated debt. Both of these latter instruments would also be suitable for banks structured as mutual societies, but may bring more complexity to the regime and no added benefit beyond what a new CET1 instrument might deliver. This aspect of the reforms requires further analysis. Recapping the arguments against contingent debt 128. The main arguments for removing contingent debt from the capital framework remain compelling. The most important of these arguments are as follows: • Loss-absorbing effectiveness: We have considerable doubt as to the loss-absorbing effectiveness of contingent debt, not necessarily from a legal perspective (unlike contingent preference shares, contingent debt contracts seem to be largely legally effective), but from an economic perspective. The theoretical and empirical literature suggests that the sort of contingent debt that has been required under Basel III, has emerged in the market place, and is likely to emerge, will not be triggered in time to provide going concern capital. In other words, contingent debt is unlikely to absorb losses in such a way as to prevent bank failure (in this way, contingent debt is inferior to common equity as the latter requires no trigger in order to absorb losses). Since contingent debt is likely to only ever trigger when a bank is non-viable, contingent debt has no regulatory value given, when a bank is non-viable, the issue of accumulated losses will be addressed by bank resolution policies (meaning there is no requirement for contractual loss absorption). • Fiscal risks As well as being ineffective at absorbing losses on a going concern basis, and unnecessary on a gone concern basis (given bank resolution policies), contingent debt may be ineffective at preventing fiscal bailouts. This is because when contingent debt has been sold to retail investors there is evidence that, under some circumstances, governments may be unwilling to impose losses on the holders of these instruments. Because the possibility of a fiscal bailout remains

29 29 when contingent debt is sold to retail investors, such instruments entail an element of moral hazard which is disadvantageous for financial sector stability. • Regime complexity Capital regulations do not exist in a vacuum. Local legal frameworks, accounting standards, economic structure, other regulations and the philosophy, framework and resourcing of bank oversight are all relevant factors that shape capital regulation. An especially important context for New Zealand’s bank capital regulations is the regulatory philosophy and framework applied by the Reserve Bank. The Reserve Bank’s regulatory philosophy and framework has led to a regulatory regime with a number of unique features relative to other jurisdictions, including a narrower range of regulatory requirements; an emphasis on ‘self-discipline’ that imparts ultimate responsibility for risk management to directors and senior management, and; relatively high importance attached to the role of ‘market discipline’ from depositors, policyholders, investors, rating agencies and other market participants in contributing to the soundness of a financial institution. In contrast, other regimes may place considerable emphasis on intensive bank supervision, including on-site inspection. Reflecting this context, we have a preference for avoiding unnecessary complexity in the capital regime (and in other areas of bank regulation). In fact, having a less complex regulatory regime is arguably justified on other grounds as well. Less complex regimes are considered by the Bank of England’s Chief Economist Andrew Haldane and others, for example, as being better able to handle shocks. 31 Contingent debt instruments are inherently more contractually complex than ordinary shares, subordinated debt that is not contingent and (non-contingent) preference shares. This inherent contractual complexity reflects the contingent nature of these instruments. This inherent complexity requires an approach to oversight and enforcement that is, if it is to be effective, at odds with the Reserve Bank’s supervisory framework. The poor fit of contingent debt with the supervisory framework was illustrated recently by the non-compliance of Kiwibank’s capital instruments, a situation that was not detected by either Kiwibank or the Reserve Bank at the time the instruments were issued. 31 Andrew Haldane, in a speech made at the Federal Reserve’s economic symposium Jackson Hole, Wyoming 31 August 2012, made the point that complex regulation is not only “costly and cumbersome, but sub-optimal for crisis control”.

30 30 We recommend that, based on the information and arguments received to date, a core aspect of Option 4 - namely removing contingent capital instruments from the framework - remains a preferred reform. We recommend that Financial Policy should do further work looking at the role of preference shares in the regime, and the possibility of including a new CET1 instrument suitable for banks structured as mutual societies. We draw the Committee’s attention to the conclusion that, if the banks’ arguments against departing from Basel III are accepted, and taking into account the likely response of the big four banks to Option 4 (as articulated in their submissions), there are grounds for adopting a common equity-only regime, in contrast to Option 4. We draw the Committee’s attention to the conclusion that having a delayed, and extended, transition period may alleviate bank concerns and would be justifiable if it delivered an appropriate, effective, simple and enduring capital definition. Near term uncertainty introduced by the Numerator Paper 129. Given the significance of the changes envisaged in Option 4, the Numerator Paper has introduced near term uncertainty into the sector. Banks no longer have certainty about the regulatory capital value of the AT1 and Tier 2 instruments they have already issued, or intended to issue in the near future. This uncertainty is hindering the banks’ capital planning. 130. We have attempted to alleviate the near term uncertainty introduced by the Numerator Paper (and the earlier Issues Paper) as much as is practicable. On 8 August 2017 a letter was sent to all the banks explaining that they can continue to issue capital instruments that comply with the current requirements of the Handbook until such time as a new version of BS2A and BS2B takes effect (bearing in mind that, a week or two prior to a new version taking effect, BS16 capital applications would not be accepted for assessment). 32 131. The August letter also explained that any instruments issued under the current rules will be eligible for grandfathering, a transition process that is likely to see the recognised capital value of existing instruments gradually reduced over a five year period beginning immediately upon a new version of BS2A and BS2B taking effect (note, however, that no transition policy has yet been consulted on). 132. An important practical matter relates to a contract provision that is included in all contingent debt instruments and preference shares that currently qualify as capital. This provision gives banks the right to recall the instrument before the first optional call date 32 Refer Documentum 7119771

31 31 when a “regulatory event” occurs that was not anticipated at the time the instrument was issued. Banks will clearly be able to use this provision (if they choose to) for instruments already on issue prior to the Capital Review being announced in March 2017. 133. However it is not clear that banks will be able to use this provision for instruments issued during the Capital Review period (i.e. for instruments issued between March 2017 and the time a new version of BS2A and BS2B applies). This is because reform to the capital definition became a clear possibility from the time the Capital Review was announced. The availability of this early recall provision on instruments issued during the Capital Review period is a remaining near term uncertainty for the banks. We recommend that steps be taken to remove, as soon as possible, a remaining near term uncertainty for banks arising from the Numerator Paper. This uncertainty relates to the early recall provisions of contingent capital instruments issued under the current regime. Removing remaining near term uncertainty 134. The letter of 8 August told banks that they cannot avail themselves of the early recall provision for instruments issued after submissions on the Numerator Paper have closed and the Reserve Bank has made a policy announcement in response to submissions: “In the Reserve Bank’s view, the adoption of a new capital definition is something a bank could be reasonably said to have anticipated once the Reserve Bank has announced the result of the consultation and the policy to be adopted. In the current context, this would mean that the Consultation Paper entitled “Capital Review Paper 2: What should qualify as bank capital?” has closed for consultation, submissions have been considered, and an announcement has been made on the broad option that is intended to be implemented. It may be the case that further consultations are needed on the finer details of the option, but as long as the broad details of the desired option and the intention to implement the desired option have been announced, then, from this date, the reform would constitute an anticipated event. “ 135. In order to clarify the availability of the early recall provision, the policy announcement referred to above does not have to address all aspects of BS2A or BS2B, or even all the reform items envisaged in Option 4. It can merely announce the likely role of contingent debt and contingent preference shares in the regime going forward. 136. Making such a statement does not restrict what can be done with respect to reforming the definition of capital. Making such a statement does not rule out

32 32 consideration of new information or arguments provided in relation to Option 4 or the Numerator Paper more generally, nor does the announcement imply that the Handbook has to be hastily redrafted. But what it does do is eliminate any remaining uncertainty for banks with respect to the availability of the early recall provision on capital instruments issued during the Capital Review period. We recommend that a statement, articulating an in-principle decision with respect to contingent features in capital instruments in the capital regime going forward, should be made in the near future. Redrafting BS2A and BS2B 137. We have the option of quickly following the policy announcement referred to above with specific redrafting of the relevant sections of BS2A and BS2B, to give quick effect to the policy statement. We have been scoping the redrafting required should this option be adopted. However finalising specific redrafting of the relevant parts of BS2A and BS2B, soon after the policy announcement is made, is only one option and arguably not the best way forward. 138. It would seem better, from a quality assurance perspective, to arrive at policy decisions on all aspects of the capital definition before finalising the redrafting of BS2A and BS2B. This could mean the redrafting, while beginning now, would not conclude until early 2018, or mid-2018 at the latest. On the one hand, this would mean a longer period during which banks could continue to issue contingent debt and contingent preference shares and still have these recognised as regulatory capital (albeit they will subsequently be subject to grandfathering). On the other hand, a coordinated approach would offer the prospect of greater internal consistency within the newly worded capital definition-related aspects of BS2A and BS2B and a reduced risk of having to redraft earlier redrafts. On balance the latter option seems preferable. We recommend that that the redrafting of BS2A and BS2B should begin now, with a view to a final version of the Handbook (as relates to the capital definition) taking effect before mid-2018. Recommended time frames We propose the following next steps: • By the end of September 2017 - make a policy statement as to the likely future role in the capital regime of contingent debt and contingent preference shares. • October 2017 to December 2017 - continue to assess new information and feedback on the Numerator Paper in relation to the capital definition; and begin redrafting the capital definition sections of BS2A and BS2B, via a consultative process

33 33 • January 2018 to June 2018 - finalise the redrafting of the capital definition sections of BS2A and BS2B

34 34 Appendix 1: a summary of other feedback provided in relation to the definition of capital

1. Feedback that is pertinent to the issue of what constitutes capital, but is not outlined in the main paper, is summarised and responded to below, by topic. Common equity instrument specifically for mutual societies The submitters’ view
2. Banks structured as mutual societies, would like to be able to issue common equity capital instruments that reflect their unique nature. This would require the regime to include a definition of common equity that went beyond ordinary shares. Our response
3. With Europe, Canada and the UK including CET1 instruments that relate to mutual societies, and Australia contemplating the same, this does not seem an unreasonable request. Coupon suspension provided by AT1 instruments The submitter’s view
4. Associate Professor Martin Lubberink provided feedback that AT1 contingent debt gives banks the option of cancelling coupon payments and this is a valuable feature as it helps recapitalise a going-concern bank. Our response
5. Against this argument is the fact that flexibility as to whether or not an issuer pays coupons is not a feature unique to the AT1 instruments currently accepted under Basel III. Perpetual preference shares issued by New Zealand banks in the past, and accepted as regulatory capital under pre-Basel III regimes, had non-cumulative, cancellable distributions, and retaining this feature is envisaged in Option 4.33
6. Dividend suspension is, of course, always possible with ordinary shares. 33 Pre-Basel III requirements as to what constitute eligible regulatory capital in New Zealand are outlined in the RBNZ Bulletin Vol. 59 No. 2 June 1996, article by Ken Matthews. By 2010 perpetual preference shares with cancellable distributions were accepted as AT1 capital (refer the version of BS2A that was in effect from October 2010 to December 2013). ASB Capital’s Series 1 and Series 2 Perpetual Preference Shares, issued in 2006 and described in Note 34 in recent ASB disclosure statements, are examples of perpetual preference shares with cancellable distributions.

35 35 Information value in market-traded contingent debt The submitter’s view 7. Another point made by Martin Lubberink is that contingent debt issued to third parties may be listed on NZDX and the trading performance of these instruments provides useful, forward-looking information about the financial health of New Zealand banks. This point was also made by Deutsche Craigs: “We are confident that the pricing of capital instruments will portend deterioration in financial performance much more quickly than the publicly released disclosure statements and in this sense can be a useful tool for regulators. The historical performance of the capital instruments issued by the likes of Credit Agricole and South Canterbury Finance provide evidence of this.” Our response 8. Several responses can be made here. Firstly, if the Reserve Bank placed considerable value of the information provided by market trading, it would seem preferable to require registered banks (that can issue ordinary shares) to list a portion of their issued shares on NZX. This would offer potentially relevant information, and help develop a deeper and more diversified local equity market. Contingent debt and fiscal risk The submitters’ view 9. Reflecting a failure to distinguish between going-concern capital and gone concern capital, or a concession that realistically contingent debt will not act as going-concern capital, the NZBA said contingent debt brings no fiscal risk. The logic of the argument was that, because OBR captures contingent debt and OBR does not compensate investors, there is no fiscal risk associated with contingent debt capital sold to retail investors. Our response 10. This argument is only valid for contingent debt triggers when a bank has become non￾viable (i.e. the NZBA is assuming the loss imposed on creditors always occurs in the context of OBR).

36 36 11. It is also important to note that, if, under OBR, there is a surplus left after depositors and senior creditors are repaid in full, the issue of imposing losses on retail investors in contingent debt instruments becomes real and political considerations could impact on the outcome. Regime complexity The submitters’ view 12. The big four banks do not view the current regime as complex and do not appear to value the prospect of having a less complex regime to comply with, preferring (for reasons outlined in the main paper) to have a regime aligned with Basel III (due to potential impacts on international lender confidence) and APRA (due to the optionality this provides New Zealand subsidiaries of Australian banks). Our response 13. The issue of investor confidence, and optionality around funding, is addressed in the main paper. The issue of regime complexity was extensively covered in the Numerator Paper and none of the information or points of view contributed in the submissions was sufficient to counter the view expressed in the Numerator Paper. Contingent debt global issuance The submitters’ view 14. The NZBA challenged the Numerator Paper’s reporting of Australasian dominance in the issuance of contingent debt with write-off, saying the instruments are popular in the UK and Europe, and pointing to 300 AT1 instruments issued in the Moody’s CoCo monitor. Our response 15. The dominant form of AT1 instruments in the Moody’s register are not contingent debt, but contingent equity securities (akin to preference shares). This point seems to have been over-looked by the NZBA. Resolution and contingent debt The submitters’ view 16. The NZBA said contingent debt that triggers when a bank is non-viable is valuable in that it removes a class of subordinated creditors from having to be dealt with by the resolution authority.

37 37 Our response 17. Including non-viability triggers in the capital regime adds to complexity, with little obvious added value in the context of resolution. Common Equity The submitters’ view 18. Martin Lubberink is of the view that banks will try to exploit weaknesses in the definition of common equity, in order to have low cost instruments classified as equity for prudential purposes. He believes there is a risk, when common equity is not carefully defined, that instruments that do not absorb losses effectively may nevertheless end up being included in common equity. He suggests New Zealand could learn from the European Bank Authority’s work defining CET1 instruments as part of the Capital Requirements Regulation (CRR). Our response 19. This point is potentially valid, but difficult to act on in practice. While the risk of exploiting the definition of common equity is also present in the current regime, removing contingent debt from the framework may create a greater incentive for banks to exploit weaknesses in the definition of common equity. 20. The Numerator Paper signalled that further work might be done on the definition of common equity, if problems were identified. However, after relatively superficial analysis, it appears that significant resources might be needed to identify whether the definition of common equity is particularly vulnerable to exploitation, compared to overseas regimes. At this stage it looks as if the likely gains from this work may not be commensurate with the resources spent. 21. The analysis that has already been done, comparing common equity definitions across capital regimes, suggests that the high level definition of common equity largely aligns across countries. If there are significant differences in what counts as common equity across jurisdictions, the source of these differences would appear to lie in commercial law, accounting frameworks and so on, which will take considerable resources to compare and analyse. SPVs The submitters’ view

38 38 22. Kiwibank’s submission said “appropriately designed SPVs are conduits that do not give rise to any residual risk exposure to the banking group”. Kiwibank rejects the idea that SPVs should be required to issue instruments that the bank would otherwise have issued itself and that the focus, instead, should be on whether the transaction utilising the SPV “results in a genuine external contribution of financial strength to the banking group”. Our response 23. Kiwibank does not acknowledge a key risk for the bank arising from the activities of the SPV, namely reputational risk. This is a valid basis for requiring the SPV to be part of the banking group for capital regulatory purposes. 24. There are international precedents for Option 4’s requirements of SPVs. Under Option 4, if capital issued under an arrangement involving a SPV is to be recognised as banking group capital, the SPV must be part of the banking group for financial reporting purposes. This is consistent with APRA’s regime and the EU bank capital regulations: • APRA will only allow a bank to report as capital an instrument issued with the involvement of a structured entity if the structured entity would be part of the consolidated banking group under the application of Australian accounting standards.34 • EU Regulation 575/2013 (“CRR”) says that, in order to qualify as AT1 capital, an instrument that is not directly issued by the bank must (in addition to other requirements), be “issued through an entity within the consolidation pursuant to Chapter 2 of Title II of Part One” (i.e. the entity must sit within the consolidated group that is subject to capital regulations).35 This suggests the bank must have an ownership interest in the SPV. 25. The Basel III requirements in relation to SPVs are clear: no instrument issued by an SPV will be recognised as group capital unless the instrument would constitute capital had it been issued by the bank itself. This is a requirement of our current framework, and would continue to be a requirement under Option 4. This requirement reflects concerns 34 Australian Prudential Standard 111 (APS), Attachment I “The following requirements must be met for capital instruments issued through an SPV to qualify as Regulatory Capital: (a) the SPV issuing the instrument is a single purpose non-operating entity established for the sole purpose of raising capital for the ADI and the SPV would, in accordance with Australian Accounting Standards, be fully consolidated in the Level 2 group; ….” 35 EU Regulation 575/2013 Article 52 (1)p

39 39 that, if the instruments could diverge, the quality of capital available to the banking group might be less than expected. BS16 processes The submitters’ view 26. ACC and NZSF submitted that the non-objection regime requires reform. The NZBA said the Reserve Bank should weigh up the costs and benefits of increasing its’ resourcing in the area of bank supervision and regulation, and that this should be part of the Capital Review. 27. ASB proposed that the Reserve Bank introduces user pays for non-objections: “The definition of capital should not be determined by the Reserve Bank’s resourcing considerations. To assist the Reserve Bank in its BS16 non￾objections process, issuers provide the Reserve Bank with legal and tax opinions and accompanying Director attestations. As such, overall, we do not consider the level of bank capital issuance is unnecessarily onerous. In addition, the Reserve Bank could consider resourcing the oversight of these issuances on a users pays basis.” 28. SBS, TSB, Co-op and NBS submitted that the non-objection regime should be replaced by an outright approvals process, to create more certainty. Our response 29. The resourcing of bank oversight activities is a separate stream of work, prompted by the FSAP report. Transparency The submitters’ view 30. The NZBA does not support the proposals regarding enhanced transparency. The NZBA’s view is that the documents would not be of any value to the public (the NZBA does not mention the financial media) and the legal opinions may mislead investors. 31. ASB does not object to the BS16 self-assessments and Director attestations being publicly disclosed. Our response

40 40 32. The public interest is served by well-informed financial journalists, and they could be expected to not only understand but critique the capital offers better if they had access to the legal opinions and so on.