Comments on RBNZ capital review

1 Comments on RBNZ capital review Thorsten Beck1 December 9, 2025 This is a report on the review of the capital requirements, undertaken by the RBNZ in 2025. First of all, I’d like to express my gratitude to the staff of the RBNZ that have accompanied my two fellow experts and me throughout the process in a very diligent manner, providing us with all the necessary documentation, making themselves available for any follow-up discussion, and organising a number of very valuable conversations with stakeholders of this process and review. The 2019 review and regulatory risk appetite The 2019 review and subsequent decision on capital requirements went beyond Basel III minimum standards and many other advanced countries in terms of imposing higher capital requirements on banks in New Zealand. This includes both higher levels, but also higher risk weights. On the other hand, the New Zealand system does not include a leverage ratio, i.e., unweighted capital-asset requirement. The higher capital requirements can be justified with a higher reliance of the New Zealand economy on its banks, limited diversification benefits by banks within New Zealand and, more generally, the risks that come with a small open economy. The current review, including the public consultation, argues that risks have declined because New Zealand has introduced several reforms to its financial safety net. It also refers to a higher risk appetite, although one wonders whether this is a politically driven cyclical element. I would first like to make a few remarks about risk appetite. There is a critical difference between monetary and financial stability (both under the purview of RBNZ) in that monetary stability can be clearly defined (e.g., inflation at or around 2%) and observed, while financial stability can be primarily observed when it is absent, i.e., during times of systemic distress in the financial system. A reasonable degree of risk appetite on societal level isto minimise the risk ofsystemic distress given negative repercussions for the economy and society and – ifsuch systemic distress happens – minimise these negative repercussions. This is different from risk appetite on the bank-level, which does not take into account such negative repercussions for society and economy nor for the broader banking and financial system. It also implies that the financial safety should focus on both objectives – minimising the risk of systemic distress (primarily by 1 Florence School of Banking and Finance, European University Institute, and CEPR.

2 forcing sufficiently high capital buffers) and crisis management tools to be used in the case of systemic distress, including going- and gone-concern tools for banks in fragility. At this point, I would like to make a provocative point. The argument has been made that it is ultimately for elected officials to determine the risk appetite for bank regulators and supervisors; however, political economy tells us that risk appetite as revealed preference might be cyclical: given the limited time horizon of elected officials, short- term gains might be preferred even at the cost of longer-term higher risk (materialising in fragility in the medium-term, i.e., beyond the next elections). There is thus a strong case to be made to focus on societal preferences beyond shorter-term political preferences. An independent, ‘technocratic’ institution like the RBNZ is very well positioned to take such a longer-term view. While in the ideal world, there would be an alignment between Treasury and RBNZ, the absence of such an alignment should not prevent the RBNZ from focusing primarily on financial stability. It is also important to note that the 2019 model assumed risk neutrality and focused on the economic cost of a systemic crisis vs. the cost of lower lending volume and higher lending costs, so one cannot really say that the approach was too risk averse. While one can argue that requiring a 0.5% probability of a crisis in a given year (or one in 200 years) might be considered too risk-averse, the RBNZ assessment showed even lower average GDP for a 1% probability of a crisis in a given year (or one in 100 years). This analysis is based on the enormous economic costs that a crisis imposes on a country. Laeven and Valencia (2020) showed that the median fiscal costs of a banking crisis in high-income countries was 7% of GDP and the output costs during the first three years after the start of the crisis amount to 35% of GDP. Both the high fiscal and output costs of systemic banking distress in 2008/9 ultimately contributed to the sovereign debt crisis in several jurisdictions in Europe in the 2010s. It is important to keep these numbers and their societal implications in mind, when discussing the societal risk appetite! This brings me to a broader point, in that capital requirements – and bank regulation more broadly - should be seen as part of the overall financial safety net. In addition to bank regulation, the financial safety net consists of bank supervision, lender of last resort, deposit insurance and bank resolution. The different components interact with each other and can also strengthen each other. In some circumstances, however, a lack in one area has to be compensated with a stronger focus in another area. In the case of New Zealand, there seem to be no Pillar 2 capital requirements or guidance (P2R and P2G, respectively) as is the case, for example, in Australia and the European Union.2 As such Pillar 2 requirements come on top of the statutory Pillar 1 requirements, their absence would call for higher Pillar 1 requirements. More on the risk environment The global macrofinancial landscape is worse than in 2019. Risks arising from changes in the geopolitical and -economic landscape can have severe impact on the New Zealand economy and thus its banking system (to name a few: tariffs with their impact on trade 2 The average P2R in the euro area, for significant institutions, was 2.25% in 2024, based on data from ECB Bank Supervision.

3 including diversion in trade; geopolitical tension in Taiwan Strait, and, more generally, rivalry between US and China). While I concur that some of the recent reforms in the financial safety net might have reduced the risk environment and strengthened the resilience of banks in New Zealand, this seems more than off-set with the higher risk arising from the global environment. More generally, there seems to be a global trend towards loosening regulatory capital requirements (sometimes mis-represented with valid concepts such as proportionality or simplification). I am not convinced that the current push towards looser regulation is a good one. Most academic economists were not convinced that the Basel III capital requirements were high enough (again New Zealand has gone farther than the Basel III minimum standards) and the current global risk environment (as discussed above) certainly does not call for a loosening of regulation. Capital requirements and growth There is no evidence for a linear relationship between the level of bank capital requirements, on the one hand, and bank lending and economic growth, on the other hand. Increasing capital requirements beyond a certain threshold will certainly result in lower financial intermediation, i.e., lower lending to the real economy. I do not think that the New Zealand banking system is anywhere close to such a threshold. On the other hand, lowering capital requirements below a certain, very low, threshold, might fuel a credit boom, but this might also result in higher medium-term fragility. Between these two thresholds, there is no clear relationship between capital levels and lending or economic growth. In this context, however, it is important to differentiate between effects of transitioning to new capital requirements and effects in steady state. As capital requirements have been increasing over the past years, following the 2019 review, it is not surprising to see negative effects on lending and positive effects on the cost of lending, though these effects are certainly difficult to ascertain and separate from the effects of the pandemic and policy reactions to the pandemic. However, the academic literature has widely agreed that there is no clear and significant relationship between capital requirements and levels of capital, on the one hand, and lending, on the other hand, in the long-run and with capital requirements stable and this relationship might even be positive (e.g., Gambacorta and Shin, 2018). Two evaluations of the Basel III reforms on lending and the cost of lending have not provided evidence of a growth-dampening effect of these reforms. Specifically, BCBS (2022) does not find considerable evidence of negative effects of the Basel III reforms on lending, but rather that banks complying with the Basel III requirements lowered their costs of both debt and equity funding. This decrease was more pronounced for those banks with lower initial capital ratios, while the banks with lower initial CET1 ratios experienced lower loan growth than their peers. FSB (2019) finds heterogeneity across jurisdictions in the effects of Basel III reforms on SME lending. There is some evidence that the more stringent risk- based capital requirements under Basel III slowed the pace and, in some jurisdictions, tightened the conditions of SME lending at the pre-reform least capitalised

4 banks, although these effects were temporary. Both studies consider transition effects rather than long-term effects, however. It is also important to consider that the relationship between capital and lending/growth might vary over the business and credit cycle. During downturns (including recessions), better capitalised banks are more likely to continue lending (Popov and Udell, 2012). During good times, on the other hand, capital might not be a constraint for lending (also in light of lower risk weights under the standardised approach) and therefore lower capital requirement might foster a credit boom, which would then have to be countered either by counter-cyclical capital buffers or by borrower-based macroprudential measures to avoid that it results in fragility in the medium-term (Dell’Arriccia et al., 2016). Capital requirements and competition The academic literature has not come to a clear conclusion whether or not competition increases or decreases stability, as empirical studies have shown different results. Importantly, Beck, De Jonghe and Schepens (2013) show an important interaction between competition and a country’s regulatory framework in explaining variation in stability across banks and countries. While I am not aware of recent empirical evidence on the relationship between capital requirements and competition in the banking system, theory would not suggest a linear relationship, either negative or positive. Higher capital requirements can result in a higher franchise value and more careful lending policies of banks, while lower capital requirements can result in lower franchise value and thus more aggressive risk taking and lending, which ultimately will result in more fragility. Critically, the effect of capital requirements on competition and risk￾taking might importantly depend on the ownership structure of banks (Laeven and Levine, 2009).3 Given the unclear relationship between capital requirements and competition (and the possible trade-off between fostering more competition today and more fragility tomorrow), other competition-enhancing policies might be more appropriate, including open banking and improving the switch guarantee. Importantly, enhancing competition does not require necessarily new entry (although it can be helpful); Claessens and Laeven (2004) show that there is no significant relationship between market structure and competition across countries. Comparison with other countries The Oliver Wyman study, commissioned as background for the current review shows that banks in New Zealand have somewhat higher capital ratios than banks in other jurisdictions. However, and as also pointed out by Martien Lubberink in a recent blog entry, the choice of countries is rather eclectic and does not take into account the nature 3 Behn and Reghezza (2025) find no statistically significant relationship between capital requirements and profit efficiency (as indicator of banks’ competitiveness) for a sample of large euro area banks. They also show that higher capital ratios reduce both bank funding costs and the volatility of earnings for less capitalised banks.

5 of New Zealand as small open economy, the heavy reliance on foreign-owned banks, the smaller size of the banks and the potentially higher sectoral concentration of the New Zealand economy. The comparison of capital requirements in New Zealand and Australia is especially problematic. Countries in Central and Eastern Europe that are host countries to Western European parent banks typically require higher capital ratios than home country supervisors in Western Europe.4 A similar rationale could be applied to the comparison between Australia and New Zealand. So, while the comparison suggests that capital requirements in New Zealand might be higher than in other countries, there are important reasons for this, including the character of New Zealand as host country. Composition of capital buffer One change proposed in the current review is to eliminate AT1 capital and thus simplify the capital stack. There are increasing doubts globally on the usefulness of AT1 capital, for either going- or gone-concern, so I do not think that eliminating this as part of the capital stack has much downside. I concur with the idea to replace AT1 capital with CET1 capital, although I would prefer to replace all of it rather than reducing total Tier 1 capital and partly replacing AT1 with Tier 2 capital. Granularity of risk weights The risk-weighted capital-asset ratios require risk weights that properly reflect the product of (i) probability of default and (ii) expected loss given default. The 2019 review has set the risk weights higher than in Australia and higher than foreseen under Basel III and in other countries. In general, I agree with the current approach to redefine risk weights and thus align them to historical loss data for New Zealand, with one important caveat. While loss data for individual loans across different categories might be useful for risk weights, it is important to (i) calculate such loss rates across the cycle and (ii) use longer time-series to properly take account of tail risks (events of low probability but with high negative impact). This becomes even more important as New Zealand has not been using the counter-cyclical capital buffer in an active way. While more granular risk weights can help better capture the risks of specific asset classes, one important challenge to keep in mind, is regulatory arbitrage, i.e., banks shifting loans (or assets more generally) across different categories to minimise risk weights. However, this can be addressed through supervisory actions targeting misclassification; but this would require more intrusive supervision, which currently 4 Calculating the average sum of CCoB, CCyB, SyRB, and the higher of G-SII and O-SII buffers, across seven home countries within the euro area yields 5.1% and across six host countries 6%, thus a difference of 90 bp. Calculations based on data from ECB Bank Supervision.

6 does not seem to take place in New Zealand. In addition, the (mostly downward) adjustment of risk weights reduces overall capital level even more than the proposed changes in capital ratios in the review, under either option, which is of concern. This might have several (possibly unintended) consequences. First, one reason for not introducing a leverage ratio in 2019 was that it would not be binding, given high risk weights (for banks using standardised approach) and high output floor (85%) and a scalar of 1.2 (for banks using internal risk models).5 As risk weights are being reduced, and capital levels under both approaches will thus decline, a properly designed leverage ratio might become binding for some banks at some point. Introducing such a leverage ratio should be considered or – at a minimum – balance sheet analysis of banks undertaken to see whether a hypothetical leverage ratio might become binding. Put differently, one should consider how much head space would remain after the hypothetical introduction of a leverage ratio. It would be important to consider this not on an average across all banks, but rather bank by bank. Second, proportionality – the absolute (but not relative) difference between capital requirements for banks using standardised and for banks using internal risk-based models (IRB) is shrinking with lower risk weights, thus somewhat reducing the benefit for Group 1 banks using IRB. If the objective is to increase proportionality across the three bank groups in New Zealand, it would be important to take into account possible effects of lower risk weights. One important challenge is that risk weight differentiation addresses microprudential risks but does not take into account macroprudential risks stemming from credit cycles; to be more precise, the riskiness of individual (mortgage) loans might be well captured by such risk weights, but not the systemic component. This would require a countercyclical capital buffer, as discussed further below. Proportionality I concur that it is adequate to set different levels of capital requirements for the different groups within the banking system; the failure of a large bank has stronger negative repercussions for economy and society than the failure of a smaller bank, so that higher capital buffers for larger banks are justified both to reduce their failure and to minimise the effects of such failure on the rest of the financial system and the broader economy and society. This is also the reason why Basel 3 has introduced additional capital buffers for G￾SIBs and in the EU, there are additional capital buffers for O-SIIs. Under option 2, Group 1 banks (currently all subsidiaries of Australian banks) have to have internal LAC. In this context, it seems that one very specific concern is the possible transition of a domestic Group 2 institution to Group 1, as – unlike the four current members of Group 1 – such a bank would not be able to rely on internal LAC 5 One strong argument for the introduction of a leverage ratio (i.e., unweighted capital-asset requirement) was the risk that banks would reduce the ratio of risk-weighted to unweighted assets so much that they would be left with a very thin absolute capital buffer. It might be worthwhile to explore the impact of more granular and lower risk weights on the unweighted capital-asset ratio.

7 from the parent bank but only external LAC, i.e., debt issued on markets, which in turn might be perceived as a barrier to further growth (or for RBNZ to reclassify a Group 2 bank as Group 1 bank). Counter-cyclical capital buffer The other issue isthat Isee a limited use of the countercyclical capital buffer. The countercyclical capital buffer was introduced under Basel 3 to increase resilience of financial institutions to credit cycles. It forms a critical element not only in the upturn (to increase resilience while credit is growing rapidly, thus preparing for a subsequent downturn) but also during the downturn phase when these capital buffers can be released. The experience of many euro area countries during the Covid-19 pandemic when a release of countercyclical buffers was not possible (asit had not been built up) resulted after the pandemic in a move towards a positive neutral countercyclical capital buffer that foresees a positive level well before a possible turn in the credit cycle. While the New Zealand framework foresees a countercyclical capital buffer, it is part of the overall capital buffer and does not seem to be actively used. An additional complication is that only Group 1 and 2 banks are subject to the counter-cyclical capital buffer but not Group 3 banks; an activation or reduction of the buffer would thus affect proportionality between Group 1 and 2, on the one hand, and Group 3 banks, on the other hand, with possible competitive repercussions. As this might make the RBNZ more reluctant to use the counter-cyclical capital buffer I would welcome its application across banks of all three groups. I would also advocate a positive neutral rate of 2%. The two options under review As certainly has become clear by now, I am very critical of the general direction toward lowering capital requirements and either of the two options does so. A few weeks ago, however, we were presented with a proposal that would have increased CET1 for Option 2 by one percentage point. Unfortunately, this option seems no longer on the table. The difference between Option 1 vs. Option 2 is only for Group 1 banks, replacing capital with loss-absorbing capacity (LAC). Maybe not surprising, Group 1 banks mostly favour option 2 with internal LAC provided by their parent banks, arguing that this would lower funding costs for them, while Group 2 banks are opposed as it would go against increasing proportionality. However, it is important to assess this option in combination with a realistic (‘hard-nosed’) assessment of resolvability of Group 1 institutions. Also, there is a strong argument (also made by some Group 2 institutions) that shifting away from Tier 2 capital instruments towards internal LAC instruments might reduce the depth and liquidity of markets for banks’ Tier 2 capital instruments. Overall, the choice between Options 1 and 2 has clear implications for competition between Group 1 and 2 banks. Comparing the two options (higher capital under Option 1 and lower capital but LAC

8 under Option 2), the first option would be simpler and increase resilience, while Option 2 might make resolution easier if it comes to this point (with the caveat below). While the argument has been made that Option 2 would result in lower lending cost (because of banks’ lower funding costs), this would really depend on the pass-through from funding to lending costs and the competitive pressure that Group 1 banks might face. The use of LAC instruments looks attractive, as they are cheaper for banks while still providing a capital buffer in case of fragility. The critical challenge is to define the trigger point, which – in case of going-concern resolution would have to be relatively high. As already indicated above, this option has to be assessed together with the overall resolution and crisis management framework. While APRA seems to rely on a non￾quantitative trigger6 , at the full discretion of APRA, I would suggest a two-pronged approach: a quantitative trigger that forces RBNZ to intervene and discretionary power to intervene even before a Group 1 bank reaches the trigger threshold. Having a sufficient level of LAC (Option 2) could be a necessary condition for “resolvability”, but is certainly not sufficient, as also acknowledged by the RBNZ. As far as I understand, there is the intention to introduce a new crisis preparedness framework in 2028/29. This new framework would set out other resolvability factors, in addition to LAC. I consider it somewhat premature to transition towards a system that relies heavily on LAC, while the overall framework for bank resolution for Group 1 banks is not in place yet. Finally, while LAC instruments have been implemented across several jurisdictions, to my best knowledge there has been no experience with actually using it during resolution. Conclusions The theme for this capital review is “simple, strong and proportional.” I agree with this focus, but would like to point out that (i) simple would certainly imply option 1 rather than option 2, (ii) strong would certainly imply option 1, but with higher CET 1 (more comparable to the 2019 requirements) and (iii) proportional is important but should include a CCyB for banks of all groups, so that proportionality does not change across the cycle. References Basel Committee on Banking Supervision (2022). Evaluation of the impact and efficacy of the Basel III reforms. Basel, Switzerland Beck, Thorsten, Olivier De Jonghe and Glenn Schepens (2013). Bank competition and stability: cross-country heterogeneity, Journal of Financial Intermediation 22, 218-44. 6 According to information provided to me by the RBNZ staff, a Non-Viability Trigger Event occurs when APRA notifies a bank in writing that it believe that conversion of its bank hybrids, or conversion, write-off or write down of other capital instruments of the bank, or a public sector injection of capital or equivalent support, is necessary to prevent the bank becoming non-viable.

9 Behn, Markus and Alessio Reghezza (2025). Capital Requirements: A Pillar or a Burden for Bank Competitiveness? ECB Occasional Paper No. 2025/376, Frankfurt a.M., Germany. Claessens, Stijn and Luc Laeven (2004). What drives bank competition? Some international evidence. Journal of Money, Credit and Banking 36, 563-84. Dell’Ariccia, Giovanni, Deniz Igan, Luc Laeven and Hui Tong (2016). Credit booms and macrofinancial stability. Economic Policy 299-355. Financial Stability Board (2019). Evaluation of the effects of financial regulatory reforms on small and medium-sized enterprise (SME) financing. Basel, Switzerland. Gambacorta, Leonardo and Hyun Song Shin (2018) Why bank capital matters for monetary policy. Journal of Financial Intermediation 35 Part B, 17-29. Laeven, Luc and Ross Levine (2009). Bank governance, regulation and risk-taking. Journal of Financial Economics 93, 259-75 Laeven, Luc and Valencia, Fabian (2020). Systemic banking crises database II, IMF Economic Review 68, 307-61. Popov, Alexander and Gregory Udell (2012), Cross-border banking, credit access, and the financial crisis. Journal of International Economics 87, 147-61.