Capital review - The role of preference shares in the capital regime

Ref #7885777 v1.0 MEMORANDUM FOR FSO FROM Susan Guthrie DATE 17 Nov 2017 SUBJECT The role of preference shares in the capital regime FOR YOUR Information and Agreement Background

1. The Capital Review was announced in March 2017. The second paper in the Review, released in July, addressed the question “What should qualify as bank capital?” Consultation on this paper closed on 8 September.
2. The second capital review paper outlined a package of preferred reforms. Option 4 included non-redeemable preference shares, with no contingent trigger, as AT1 capital and subordinated term debt (again, no contingent trigger) as Tier 2 capital. Removing the redeemable feature is an important issue.
3. The arguments against allowing perpetual preference shares classified as Tier 1 capital to be redeemable primarily relate to the implications for the permanence of capital. If the shares are redeemable, the capital may only be available to the bank until the first optional call date or, if the shares are not redeemed on the first optional call date, depending on the specific terms and conditions, until any quarter thereafter. In contrast, a well-capitalised bank will have permanent, committed funding. We recommend FSO: a) Note that on 15 September FSO asked for further analysis of five issues relating to the definition of capital. One of these issues was preference shares. b) Note that the second consultation paper had proposed recognising only non-redeemable preference shares. However banks submitted that there is no investor appetite for such instruments and Deutsche Craigs submitted the non-redeeming requirement could be easily circumvented. This feedback has prompted a re-look at the role preference shares might play in the capital regime. c) Note that the arguments against accepting redeemable preference shares as Tier 1 capital remain but that redeemable preference shares could reasonably be included in the regime as Tier 2 capital. d) Note that further public consultation is likely to be required on redeemable preference shares. e) Agree that:  The preferred way forward, given feedback, is for redeemable and non￾redeemable preference shares to be accepted as Tier 2 capital.

2 Ref #7885777 v1.0 4. The adverse implications of having uncommitted capital - i.e. capital that can be redeemed - are easy to see. If a bank gets into difficulty, it may be reluctant to compound its deteriorating reputation by failing to redeem its preference shares on the first available call date. Contrast this to ordinary share capital which remains permanently available to the bank, whatever the bank’s condition. So rather than being a stabilising force for the bank, redeemable shares can be a source of added pressure and instability at a time the bank is already under stress. 5. This fundamental dynamic is not made any less real by the requirement that Reserve Bank approval be obtained before a bank redeems its preference shares. If the bank is weak and approval is not given, the reputation of the issuing bank still deteriorates when the preference shares are not redeemed. So rather than being a stabilising force for the bank, redeemable shares can be a source of added pressure and instability. 6. In contrast to these arguments against allowing redeemable preference shares to qualify as Tier 1 capital, the banks report that they have no interest in, and investors have no appetite for, non-redeemable preference shares. Financial services firm Deutsche Craigs said, pointing to the experience pre-Basel III, that banks will easily find ways to offer redemption at a future date for ‘non-redeemable’ preference shares (for example, a sister company to the bank will agree to buy the shares from investors at a future date, giving effect to ‘redemption’ in all but name). 7. This feedback led FSO to request FP to review again the potential role of preference shares in the capital regime. In particular, should redeemable preference shares be acceptable and, if so, what class of capital should they qualify for? 8. The purpose of this memo is to outline the results of the further analysis that has been done and present our recommendation. 9. It is worth noting at the outset that preference shares have not featured a great deal in the current regime. Just $1.3bn out of a total of AT1 issued capital has been in the form of preference shares. ANZ has issued $300m and ASB $1.0bn (in two separate issues). In all cases, the preference shares have been issued to the parent. 1.3 7.2 AT1 and Tier 2 instruments issued under Basel III by the large four NZ banks, NZ$bn pref shares contingent debt

3 Ref #7885777 v1.0 A re-look at the role of preference shares in the capital regime. Ordinary shares versus preference shares 10. In addition to considering whether capital is committed - i.e. the principal sum is likely to be available to the bank through bad times as well as good - it is important to consider how else the instrument might absorb unexpected losses incurred by the bank. In this regard, it is helpful to contrast preference shares with ordinary shares. 11. It is widely accepted that common equity (shareholder contributed capital and accumulated retained earnings) is the highest quality capital and as such it receives “Tier 1” status under the Basel III standards. Common equity protects depositors from unexpected bank losses on a “going concern” basis. It does this by acting as a “sink”, in effect imposing unexpected bank losses on shareholders, who absorb the loss, thereby enabling debts to be paid. 12. The mechanism of the loss transfer is two-fold: dividends may be zero (the amount paid is at the full discretion of the bank board) and the value of each share may fall (while many factors determine the value of ordinary shares, one factor is the net asset value of the company and this falls, or eventually turns negative, when there is a loss). 13. In order to pay dividends to ordinary shareholders, the company must be solvent. There is a two-part solvency test that must be met (debts must be able to be paid as they fall due and assets must exceed liabilities). 14. Traditionally ordinary shares remain on issue in perpetuity. However, the Companies Act 1993 made it legal for companies to repurchase their issued ordinary shares. The specific circumstances in which a share repurchase is permitted is dictated by the Constitution of each company. It is envisaged that the circumstances in which ordinary shares can be repurchased will be quite limited in the context of the banks. 15. When a bank is placed in liquidation (or wound up for another purpose) holders of ordinary shares are last in line to receive reimbursement for their investment. By being the most subordinate in liquidation, common equity absorbs losses on a gone￾concern basis as well. 16. In exchange for acting as a sink for bank losses, shareholders have the prospect of “upside” in the value of their investment (eg in wind-up or if sold prior to wind up), and have voting rights in the bank and therefore have the right to influence what the bank does and what risks the bank accepts. 17. “Preference share” is a term that covers a wide range of instruments that provide funding to a bank. Depending on the specific terms and conditions, the preference share could be treated as a liability or equity under accounting standards. As a general principle, it appears to be the case that if the issuer is under no obligation to pay any amount to the holder of the instrument, it counts as equity. 1

1 Based on a quick search through the accounting standards, it appears that the accounting definitions of “liability” and “equity” are given in NZ IAS32. Preference shares that have discretionary distributions appear to be accounted for in New Zealand as equity (NZIAS 32 paragraph AG26). More generally, preference shares that contain “no contractual obligation” to pay an amount to another under “conditions that are potentially unfavourable to the issuer” appear to be equity, as outlined in IAS 32 section 16. Elsewhere NZIAS states that “an equity instrument is any contract that evidences a residual interest in the assets of an entity after deducting all of its liabilities”. If we accept preference shares are not liabilities, the

4 Ref #7885777 v1.0 18. Like debt instruments, distributions paid on preference shares are fixed in advance, and typically consist of a margin above a benchmark interest rate. In other words, the contracted value of the distributions is not sensitive to short term profits. However, in theory, if the bank has little or no profit to distribute, distributions may not be paid to the holders of preference shares (distributions are discretionary) and legally distributions cannot be paid if the bank would be insolvent after making the distribution. If the pre-set distribution is not paid on preference shares, dividends cannot be paid to holders of ordinary shares (this is one reason for the label “preferred”). 19. There is no “upside potential” for preference shares. Unlike ordinary shares, the value of the preference share in wind-up is not related to the net assets of the bank, but is a fixed value equal to the face value of the preference share. 20. Preference shares typically do not confer votes on holders - so holders of preference shares typically have no ability to influence the strategic direction of the bank, or the risks the bank accepts. 21. While preference shares closely resemble debt - there are pre-set distributions based on benchmark interest rates, distributions are payable ahead of dividends on ordinary shares, and the holder’s claim ranks ahead of ordinary shares in wind-up - there are two important differences:  In New Zealand, a bank cannot be put into bankruptcy proceedings because a promised distribution on a preference share that has been accepted as bank regulatory capital has not been paid. In contrast, the non-payment of interest on many debt instruments can be grounds to place a bank in bankruptcy proceedings.  When calculating whether a bank is solvent, only debts are measured against bank assets, not preference shares accounted for as equity. Going concern capital 22. An important question is whether preference shares can be considered loss￾absorbing and, if so, whether they can be considered loss absorbing capital on a going concern basis. 23. The fact that distributions on preference shares are, in legal terms, discretionary, and that debt ranks ahead of preference shares in liquidation, means preference shares are technically capable of being loss absorbing. Moreover, the fact that failing to pay distributions on preference shares cannot be an event of default means the loss can, in theory, be absorbed by preference shares while the bank remains a going concern. Distributions can be nil - imposing a loss on holders of preference shares - yet the bank can continue without the threat of default from preference share-holders. This outcome points to preference shares, in theory, providing going-concern capital. 24. However practical considerations suggest preference shares will typically only provide meaningful loss absorption on a gone concern basis. Unlike ordinary shares, there is no potential upside with preference shares. Hence the sole emphasis of holders of preference shares is on distributions. It will be less damaging to the bank’s reputation to pay no dividends on ordinary shares (holders have the upside to look

repayment of the share in liquidation is dependent on there being positive net assets in the bank (i.e. repayment of the preference shares is funded out of net assets).

5 Ref #7885777 v1.0 forward to), than cease paying distributions on preference shares. Hence it seems likely that banks have an incentive to pay distributions on preference shares for as long as possible, even in the face of unexpected losses. 25. If the only binding constraint on paying distributions to preference shares is the insolvency test, it seems reasonable to conclude that only when the bank is close to insolvency, will distributions on preference shares not be paid (and hence will preference shares be loss absorbing). This argument points to preference shares being accepted as gone concern capital only. 26. The GFC experience is consistent with this argument. In the initial stages of the GFC distributions were not deferred, as issuers and regulators were worried about adverse signalling effects. Only when the crisis was well advanced did regulators impose losses on holders of hybrid capital instruments that resemble preference shares. 2 3 The current regime and going concern capital 27. The current regime implicitly acknowledges that, without intervention, preference shares are unlikely to be loss-absorbing on a going concern basis. The current regime imposes limits on what distributions can be made when common equity is low. The distributions that can be made in aggregate on Tier 1 capital (including AT1 preference shares) are subject to an aggregate cap when the bank’s measured common equity ratio (“CET1 ratio”) falls below 7%, this being the sum of the minimum required CET1 ratio (4.5%) and a conservation buffer (currently set at 2.5%). 28. A sliding scale applies. For example, if the bank’s CET1 ratio lies between 4.5% and 5.125%, none of the bank’s post-tax earnings (measured as accounting profits) can be distributed. If the measured ratio lies between 5.125% and 5.75%, 20% of post￾tax profits can be distributed. If the measured ratio lies between 6.375% and 7%, 60% of bank post-tax profits can be distributed. 29. The large four banks appear to operate quite different pay-out policies, so in reality, some (ANZ, WNZL) might find the conservation buffer a more meaningful constraint than others. The ratio of total dividends paid (on ordinary shares and preference shares) to net profit Financial year 2017 2016 2015 2014 2013 ANZ^ n.a. 89% (89%) 99% (61%) 137% (81%) 78% (56%) ASB 47% 27% 136% 51% 15% BNZ n.a. 58% 36% 55% 24% WNZL n.a. 78% 67% 40% 1% ^ the value in brackets is the dividends paid less new issues of ordinary shares

2 Forbes magazine, 15 June 2016 3 IMF (2011). Pazarbasioglu, C., Zhou J., Le Lesle, V. and Moore, M., “Contingent Capital: Economic Rationale and Design Features”. IMF staff discussion note January 25, 2011. SDN/11/01

6 Ref #7885777 v1.0 30. Moreover, if relatively few preference shares have been issued relative to ordinary shares, even if the conservation buffer is breached, providing the bank’s CET1 ratio sits somewhere above 5.125%, preference share dividends may still be able to be paid. 31. Our view is that, given it is reliant on accounting items, the CET1 ratio is an unreliable measure of financial strength and a close-to-insolvent bank may nevertheless report a ratio of 5.125% or more. Thus, despite the conservation buffer, we believe it is possible that distributions may continue to be paid on preference shares even though the bank may be close to non-viable. Given the incentive faced by banks to pay distributions on preference shares at all costs, this suggests it is unlikely preference shares will be loss absorbing on a going concern basis. 32. The arguments about redeemable preference shares failing to provide committed funds, and acknowledging that preference shares are unlikely to absorb losses on a going concern basis via distributions, together suggest redeemable preference shares should not be accepted as Tier 1 capital. Gone concern capital 33. The current regime goes beyond requiring preference shares to be subordinate to debt, in order to qualify as capital. Preference shares must also “write off” or convert into ordinary shares. This seems unnecessary given preference shares are senior only to ordinary shares. Moreover there is some uncertainty as to the legal effectiveness of terms that impose “write off” of preference shares. We thus propose removing this write-off or conversion requirement. 34. Whilst we do not believe preference shares realistically provide meaningful loss absorbency on a going concern basis, we acknowledge they are valuable in wind up. Hence, we have concluded that preference shares are acceptable as capital, but only in the context of Tier 2 capital. 35. Nothing in the above analysis negates the concerns we have about the redeemable aspect of preference shares. The redeemable feature means the capital is not committed and the pressure to repay may add to the woes of a struggling bank. However, we accept that debt instruments accepted as Tier 2 capital have fixed terms, which raises comparable “commitment” issues. Do we allow redeemable preference shares? 36. On balance, it seems that the most significant contribution preference shares can make is in the context of windup - the value contributed by investors is subordinate to all debt and thus acts as a meaningful buffer for depositors and other creditors. This suggests we should accept preference shares as Tier 2 capital, not Tier 1. 37. Given that we conclude preference shares are likely to be at most gone concern capital, and thus eligible for Tier 2 status only, it seems reasonable to accept redeemable preference shares as Tier 2 capital (Tier 2 debt capital is permitted to have a fixed term). Options 38. The above analysis raises a number of issues in the context of preference shares. Given our view is preference shares (whether redeemable or not) are unlikely to be loss absorbing on a going concern basis via distributions, should non-redeemable

7 Ref #7885777 v1.0 preference shares be accepted as Tier 1 capital (as was proposed in Option 4 in the second consultation paper)? The above analysis suggests not. 39. The banks’ clear preference is to be allowed to report redeemable preference shares as capital (versus non-redeemable). The above analysis suggests redeemable preference shares are acceptable as Tier 2 capital. 40. These considerations point to one obvious response to the banks’ feedback: allow banks to continue to report redeemable preference shares as capital, but classify the instruments as Tier 2 capital. 41. This would not be incompatible with having non-redeemable preference shares as Tier 1 capital, as was proposed in Option 4. However, upon reflection, we have concluded that including non-redeemable preference shares in Tier 1 is not ideal (it is unlikely distributions will cease in time for the shares to absorb losses on a going concern basis). Moreover, the Deutsche Craigs feedback suggests it will be difficult to prevent banks taking steps to in fact redeem ‘non-redeemable’ preference shares. 42. If preference shares are accepted as Tier 2 capital it would seem reasonable to align the first optional redemption date requirements of preference shares with those applying to Tier 2 debt instruments. Comparison with Basel III 43. If preference shares, with the features we recommend, are accepted as Tier 2 capital in New Zealand, on the face of it they will not align with the Basel III standards. This is because, reflecting recommendations we have made elsewhere, the instruments will not have a non-viability contingent trigger. 44. Non-compliance with Basel III may mean New Zealand subsidiaries of Australian banks may choose not to issue preference shares to third parties. However, the pattern to date under Basel III has been that New Zealand subsidiaries of Australian banks only issue preference shares to parent banks anyway - not third parties. Moreover it appears that the parents are able to purchase whatever instruments they like from their New Zealand subsidiaries as the instruments are not taken into account when group capital is being measured (the instruments are set aside upon consolidation).4 Thus even if New Zealand’s requirements of preference shares do end up departing from those of APRA, it appears that this will not prevent New Zealand subsidiaries issuing New Zealand-compliant capital to their parents. 45. It is also worth noting that the Basel III standards allow instruments that do not include a non-viability trigger to be Tier 2 capital if legislation and/or regulations impose mandatory losses on subordinated debt when a bank becomes non-viable.5 It seems possible, given this, that elements of OBR might be sufficient for preference

4 However, the nature of the instruments may have an impact on the standalone calculation of RWA and/or capital for the parent. 5 Mexico’s compliance against Basel III was assessed by the BIS in 2015. The RCAP report included the following statement: Point of non-viability The Basel capital requirements provide that all additional Tier 1 and Tier 2 capital instruments issued must include a contractual principal loss absorption mechanism unless the governing jurisdiction of the bank has in place laws that (i) require such Tier 1 and Tier 2 instruments to be written off upon such event, or (ii) otherwise require such instruments to fully absorb losses before taxpayers are exposed to loss.

8 Ref #7885777 v1.0 shares and subordinated term debt instruments that omit a non-viability contingent trigger, to be considered compliant with the international standards for Tier 2 capital. However, this aspect of the Basel III standards has yet to be investigated in any depth. 6

6 Basel technical note that relates to this issue link