Submission to the RBNZ on the Proposal for Changing Capital Adequacy for New Zealand Banks
Harbour[1] has written two short research papers on the RBNZ’s proposal to lift capital held in New Zealand by banks (see here and here). These reports highlighted that higher lending rates, lower deposit rates and potentially lower risk-weighted asset growth could be expected outcomes that might be associated with lower investment and GDP. There are broader implications for our capital market that are also worth considering. We doubt that investors will accept either a lower cost of debt financing or a lower return on equity reflecting increased equity capital and hence we anticipate higher medium costs. This belief generates an optimal level of capital lower than 16%. We would, however, allow greater flexibility in the use of the proposed Countercyclical Capital Buffer. Finally, we believe uncertainty in transition costs warrants a longer period for the introduction of strong bank capital.
At a high level, we support the six principles of the review. Banking systems globally have significantly increased capital requirements. The most obvious example is Australia’s move to a 10.5% unquestionably strong level. We note Australia has yet to introduce a Countercyclical Buffer suggesting 10.5% may just be a stopping point to yet higher bank equity capital.
New Zealand’s banking system is unique in our reliance on Australian bank shareholders to provide capital; in exposure to the systemic risks from housing (like Australia); and exposure to relatively high levels of household debt. We note that stress tests[2] currently signal that individual banks carry a significant capital buffer. The latest 2017 stress test pointed to a potential 6.9% point fall in CET1 capital, with a 6.4% to 7.4% range across the banks. This scenario would eat significantly into the current bank capital buffers. Whilst this scenario may have very significant output and unemployment costs, the “banks could absorb material losses while remaining solvent”. These stress tests provide estimates that come with a wide range of uncertain outcomes, but in this scenario, it is not obvious that the banking system itself would become an independent cause to exacerbate the output impacts or sustain the sort of banking crisis and permanent output costs outlined as a concern in the capital review. In other words, higher capital, while useful for reducing more extreme events and scenarios, wouldn’t itself be useful to mitigate external shocks.
Our perspective
Our judgement is that something closer to 14.0% to 14.5% for the bank capital stack seems a more likely median position through the cycle that optimises the trade-offs. This would still be a substantial increase in bank capital when the move to a 90% IRB floor shift is included. We push for a larger (0-3%) countercyclical capital buffer (CCyB), allowing the RBNZ to vary capital stack between 13.0% and 16.0%. We would want to see equalisation across financial institutions, with flexibility reflecting risk assessments of the economy and sectors.
We think that significant policy change has uncertain impacts and we recommend taking relatively smaller steps over a longer horizon with a definitive date to pause and re-examine the evidence.
Our key points:
- Transition costs could be significant and unpredictable
- Slow down the proposed timetable – implement over 10 years not 5 years
- Interest rate impacts could be larger and remain uncertain
- Loan growth could be constrained and right now credit is somewhat fragile
- Push for a 0-3% CCyB, set at 1.5% now, be flexible
- Treat D-SIBs and other banks with the same capital stack
- Reconsider the role of Tier 2 capital
Harbour is interested in the broad economic impacts of the proposals, the impact on bank shareholders, on bank debt investors, and on the structure and risks in New Zealand’s capital markets. We have conducted an independent analysis and undertaken direct research with Australian banks and shareholders.
In addition to responding to some detailed questions, we also raise specific issues that we consider merit more inquiry and debate.
We agree (support) with more in the proposals than we disagree with, and broadly concur with the direction of the proposals. But as with most submissions, there tends to be a focus on controversies and hence a negative tone can permeate responses. We request that you normalise our response accordingly and we look forward to engaging with you on this very important proposal.
We think that rapidly adding higher capital requirements for New Zealand banks could carry higher than anticipated transition costs. We found the RBNZ proposal relatively silent on the costs and benefits of transition and the case for the relatively fast (five year) timetable for higher capital was not made persuasively. We note that some of the literature[3] advises against using conventional monetary policy (lower interest rates) to ameliorate or offset transition costs. With rates in New Zealand already near intergenerational lows the scope for monetary policy to offset an unexpected significant tightening in monetary conditions as a result of rapid adoption of higher bank capital is limited.
We note that this is a clear area in which the bank is seeking feedback and that in a recent speech the Governor suggested some flexibility around the timing of adding capital. We welcome a stronger consultation on timing, and the potential for a pause to assess the transition costs would be a useful consideration. This is a key recommendation in this submission, and we make our comments on transition costs at the end of this document. Five years should move to ten years, including a pause to assess the evidence.
A second observation is that we suspect that the proposal for a 16% capital requirement may induce significant regulatory arbitrage, encouraging rapid growth in New Zealand’s non-bank financial system. We note already an increase in so-called ‘warehouse facilities’ provided for non-banks from Australian and other global banks. Whilst this growth is from a low base, we note that the bank capital review itself is largely silent on the issue of the potential role of non-bank financial institutions funded by non-NZ deposit-taking institutions in increasing credit growth. The extent to which leverage in the broad financial system generally exacerbates economic shocks has implications for the cost-benefit analysis used in the RBNZ’s modelling. New Zealand has a clear track record in using tax-payer funding to bail out non-banks, and over the long-term policies need to evolve to reduce just shifting risk from one sector to another. This may require policy co-ordination between the RBNZ and APRA and further regulation of the non-bank sector.
An underlying theme is that the large Australian owned banks, with circa 88% market share ought to have higher capital reflecting their systemic role. Some global regulatory bodies require systemically important banks to hold higher capital than smaller banks. Many of these banks are Globally Systemically Important Banks. These differ from our D-SIBs. Our view is that the negative impact on society from any bank failure can be significant. The example of Northern Rock comes to mind, where the failure of a relatively small financial institution had a major impact on public confidence.
Whilst the capital positions of the non D-SIBs in New Zealand are potentially more constrained given their ownership structures and restrictive shareholding positions[4], this can be overcome through the capital market and global capital funding. The RBNZ paper seems to make a case that the smaller banks are special and would struggle to gain capital through retained earnings compared to the D-SIBs[5]. We don’t agree that retained earnings is the only path towards stronger equity capital and we think that the output costs and risk appetite for any bank failure are similar enough in a New Zealand context to not differentiate between institutions with the exception of assessing sectoral and specific risk issues.
Cost of capital, dividend policies and shareholders’ reactions
- We doubt the Modigliani-Millar model holds in the context of Australian-owned banks.
The reaction of global and Australian capital markets to this proposal will largely determine the costs of holding significantly higher capital. We think bank shareholders (and debt holders) are unlikely to accept significantly lower potential returns for holding higher capital (and hence less risk). Our view differs from the assumptions in the Reserve Bank’s consultation papers. We have had numerous discussions with the Australian banks and institutional shareholders and the evidence in capital markets in our opinion is that capital structure matters particularly when the cost of debt is largely fixed. We do not think that debt costs will fall materially for NZ D-SIBs as the marginal cost of debt reflects the financial health of the entire bank entity.
In the case of the Australian-owned subsidiary banks in New Zealand, the cost of debt funding is largely set by the credit rating of the overall corporate. Further in terms of dividends and returns to shareholders, we observe already the extent to which shareholders prefer high dividends and stock buybacks in Australia, as opposed to retaining earnings for further investment at lower profitability (return on equity).
We do not think that increasing equity in NZ banks via retained earnings is an option that will be palatable for shareholders if doing so requires dividends to be cut. In the absence of a bank crisis, we would expect significant resistance from Australian shareholders to cut dividends to provide higher capital for the NZ subsidiaries. Instead, it seems more likely that banks would engage in a combination of raising fresh capital from shareholders; selling down non-core assets; repricing net interest margins; or cutting risk-weighted assets.
The evidence that shareholders in Australia like buybacks and dividends has been widely canvassed and reduces the impact of the Modigliani-Miller[6] thesis. We agree that the academic and theoretical literature suggests that investors should be indifferent to various tax effective income streams and capital injections. However, in practice, this is not what we observe. Both retail investors, and increasingly, institutional investors (through funds that focus on yield) are important providers of risk capital. Our most likely scenario is that shareholders push banks to stem the growth of lending in New Zealand in marginal lending sectors while also seeking to re-price interest margins.
A special issue arises for the Australian banks with very large relative businesses in New Zealand. We note that APRA’s proposed APA222 25% subsidiary restriction could have a significant impact on some Australian banks’ capability in terms of increasing capital in New Zealand. This highlights the difficulty of seeing debt-equity substitution across the sector.
We cannot find enough evidence in Australia to provide a reasonable estimate of the offset impact on the cost of capital from debt to equity substitution. We note that the RBNZ Capital Review Paper also largely cites global evidence. Cline[7] , for instance, cites evidence for the US over the 2002-2013 period which suggests that less than half the offset has been observed. We disagree with the RBNZ’s assumption of a 50% offset effect. We do not think that debt costs will fall meaningfully, and we do not think that shareholders will accept a substantially lower return on equity.
We suspect a reasonable range for an offset impact on the cost of capital for New Zealand D-SIB banks, given the Australian ownership and debt rating structures, is less than 50%.
Cline says that higher capital requirements impose increases in lending costs, with associated output costs. We note in work cited by the RBNZ, Cline[8] has more recently published work suggesting an 11.7% to 14.1% optimal capital ratio of risk-weighted assets for banks. We note that the optimal capital level is sensitive to this key assumption on the offset impact which is difficult to observe ex-ante. This is another reason why we recommend a longer transition and patience to observe empirical outcomes.
In sum, after consultation we are of the view that Australian and global shareholders will move to push for sustaining return on equity, and hence we expect a combination of higher net interest margins (either higher mortgage rates or lower deposit rates) exceeding the RBNZ’s expectation, and/or a targeted reduction in risk-weighted assets in specific New Zealand lending sectors. Whilst we do not hold the extreme view in the market provided by UBS estimates, we are closer to the CLSA estimates of a 50-70 basis point re-pricing of net interest margins, plus some degree of risk-weighted asset growth constraint in certain sectors.
Consequences of higher bank capital
We agree that the RBNZ needs to factor into a cost-benefit analysis the risk that net interest margins are higher. However, we would push for a higher negative impact on capital formation and growth than assumed in the RBNZ’s paper. We also think that risk-weighted asset growth could be constrained, especially over a transition period if the capital proposals were introduced in five years. Both these views suggest a lower optimal capital ratio than suggested in the RBNZ’s capital review.
In that regard, we consider that the second-round effects from this capital review might have a significant impact on GDP and fiscal risks (from the non-bank financial system). Our view is that the Reserve Bank’s model, which captures the costs of higher capital, could be amended to include an estimate for the impact lower risk-weighted assets may have on investment and GDP growth. We doubt that the local banks or the non-bank financial system have the near-term capacity to either provide sufficient interest rate market pricing tension, or the capital to fill the void to back higher risk-weighted asset growth.
We suggest that the RBNZ adopts a longer transition period with more flexibility and the potential for a pause and reassessment of actual transition outcomes after five years.
Capital allocation across the economy
Many commentators (and potential submitters) have possibly focussed on the RBNZ’s change to a 1 in 200-year risk assessment from possibly an earlier 1 in 50 or a 1 in 100-year risk assessment.
In our view, there are many industries and sectors in New Zealand that face risks where further investment now could substantially improve GDP and reduce the volatility of outcomes, and potentially improve well-being.
Examples include investing in social housing, the electricity network, highway safety, early adoption of 5G, decarbonisation, investing in pre-school education, child health, and more spending on disease prevention, detection and treatment. We are financial market participants, not public policy experts, but the RBNZ request we submit our risk appetite on the move to a 16% capital level. Our view is that this is higher than an optimal long-term level of capital given competing needs.
Calls for $20bn to $30bn more capital in the banking system in New Zealand need to be assessed in a wider context, as capital is scarce and the trade-offs in improving potential GDP (and reducing the volatility of GDP, and outcomes for the most at-risk people in society) are very evident.
We note that INFINZ has called for the Productivity Commission to provide a cost-benefit analysis. That may be a useful exercise, although our reading of any cost-benefit exercise would still be reliant on assessing many assumptions which are hard to gain strong relevant global comparisons.
Again, for Harbour, this points towards caution and flexibility in the speed of adopting significantly higher capital in the banking system.
Countercyclical Capital Buffer needs more flexibility
A final technical point we would like to raise is our view that the proposed 1.5% Countercyclical Capital Buffer (CCyB) could be made much larger. We would propose at least a 3.0% buffer and we would take at least 1.5% from the conservation buffer. Some of this is semantics and signalling but most of all we want to see the RBNZ have more tools for dealing proactively with a dynamic economy.
We doubt that a sustained 16% is the “optimal” capital level for the systemically important banks.
In our view, this doubt is implied in the RBNZ’s own analysis given the 1.5% CCyB (so 14.5% is actually a likely central target expressed by the RBNZ). We think it is more useful to consider 16% as an extreme or maximum level of capital.
Our preference is for central banks to have more tools in a crisis. Interest rates are possibly too low to provide enough monetary stimulus to gain traction in a significant downturn. We are all learning that macro-prudential tools are an important adjacency for monetary policy and for the security of the financial system.
In the context of an end target of 14.0% to 14.5% we would recommend an early set 1.5% CCyB and provide scope to step this up to 3% overtime. We think that in extreme circumstances the CCyB could be set to 0%.
Our judgement is that something closer to 14.0% to 14.5% for the capital stack seems a more likely median position through the cycle that optimises the trade-offs. This view also takes into account that the RBNZ also recommends moving the bar significantly on the IRB capital floor further reducing leverage in the banking system.
The 14.0% to 14.5% optimal capital level reflects our assessment that the costs in terms of either higher net interest rates or lower net lending growth will have higher costs in terms of capital formation and lower GDP. Our submission is also led by our reading of the literature on optimal capital (Cline 2016 for instance). We think the modelling conducted by the RBNZ does not take into account transition costs, and the RBNZ has a higher estimate for the Modigliani-Miller effect than is likely in practice. As a result, while we would have a capability of stretching over time to a 16% capital position with a full CCyB, we would anticipate an over the cycle 14.0% to 14.5% average.
Additionally, we observe that banks generally will hold their own buffer above regulatory minimums. As Cline notes: “The reasonable policy approach, then, would seem to be to set the regulatory requirement at the central estimate of the optimal level, in the expectation that in practice the banks would add a cushion…”.
We think that the central estimate is circa 14.0%-14.5% for NZ banks, still materially above current capital levels.
Do not differentiate between D-SIBs and other banks in the first instance
- And flex the CCyB or other buffers on a case by case basis
We would not differentiate the core capital requirement between the D-SIBs and other banks. This reflects our position that public confidence is as much impacted by the failure of a smaller institution as a large bank and that we often observe less diversification in smaller financial institutions – in geography, in loan concentration - and that funding for smaller institutions can be crowded out in a crisis.
We think that the RBNZ could impose CCyB variations across the bank sector – differentiating for size, sector exposures or other perceived risks.
We clearly think that the RBNZ may wish to apply a larger buffer when identified risk positions are higher (e.g. extended housing prices, rising unemployment, a terms of trade shock, rising interest rates, fiscal risks). We welcome the RBNZ’s introduction of a framework for flexing CCyB and we think it is good policy to pre-publish many of the reasons why you may flex the CCyB.
Another reason for higher CCyB in the capital stack is the potential under Basel III reciprocity for APRA to accept this capital within their Level 1 capital for the Australian banks. This could, therefore, lessen the actual capital that may have to be raised by the banks. Potentially this could significantly reduce the costs of the additional capital, however, it may depend on how APRA views New Zealand held CCyB. We note that the RBNZ is engaging with APRA on this issue.
In addition, we note that APRA’s 10.5% CET1 target is premised on a 0% CCyB. We suspect that changes in the near term with APRA noting that they are analysing “whether a non-zero default level of the countercyclical capital buffer should be implemented” to keep banks in the top quartile of global group 2 banks of a fully phased-in Basel III basis.
Risk weightings are also moving to make the banking sector safer
The proposals to significantly alter risk weightings on various lending categories seems to have received little commentary. We understand moving to a 90% Advanced IRB capital floor itself increases capital in the system by an estimated 16%. The trend of increasing capital intensity rose sharply in Australia through 2016 and 2017. Together with a tightening of lending standards (especially the introduction of detailed expenditure analysis), the higher mortgage risk weights had a significant impact on lending growth. The major banks in Australia have seen mortgage loan growth only in the low single digits in the last 18 months as capital standards for the D-SIBs has been tightened significantly. The adjustment process to move to a small gap between IRB and standardised banks approaches seems reasonable, but we wonder how quickly this should be accomplished, again mindful of transition costs.
Overall the broad thrust of this proposal could encourage lending in the non-bank sector. Whilst these risks may not be systemic, they may nevertheless involve both fiscal and output costs that have not been discussed in the report.
Who fills the potential lending gap if the banks reduce New Zealand risk-weighted assets?
The probability of regulatory arbitrage seems very evident. There is already significant arbitrage in implied capital requirements across various banking regulators, but a potential gap of 5.5% points between Australia and New Zealand in capital requirements is likely to see Australian wholesale bank funding to non-bank New Zealand lenders[9]. Although the proposal may substantially reduce risk in the New Zealand bank sector, it may just translate risk to the non-bank sector. New Zealand has a poor history of non-bank financial risks harming investors and borrowers, and eventually tax-payers. We are concerned that exceptionally strong bank sector capital may just push risk to the non-bank sector.
Our judgement is that the funds management community in New Zealand is not currently well resourced to provide significant capital (say $20bn) to support non-bank financial direct lending. In the long term a deeper local capital market may increase the scope to provide capital for bank and non-bank lending, but today in our opinion there is not sufficient funding from investors to fill a $20bn funding gap. This may change of course as the retirement sector develops through, for instance, KiwiSaver savings pools. As a result, in the near term, we think these proposals increase the probability that foreign banks will step into the funding gap and become more prominent in New Zealand. While this may deepen bank competition, history shows that foreign banks can withdraw support at times of stress, reallocating capital more rapidly between countries. This behaviour could add to stress in a pro-cyclical outcome.
Tier 2 capital potentially has a role despite the banks no “gone bank” policy stance
We have reservations about excluding Tier 2 capital from the Conservation buffer capital stack. We disagree that accepting small elements (1-2%) of “restricted” Tier 2 capital would set up a financial system for a “gone banks” scenario. In a related issue, we consider the capital requirements for the mutually owned banks may need a specific solution.
Uncertainty argues for small steps and flexibility in timing
In summary, we think there are significant risks around introducing large policy changes with uncertain outcomes. We would favour a more gradual introduction of higher bank capital with clear timelines for pausing and more flexibility for the RBNZ to exercise a larger range for the CCyB.
When intervening with a patient, doctors are taught to “firstly, do no harm.” The idea is that a desire to do good may fix one problem, but intervention may create other problems or even fail to fix the original one. This idea seems relevant to the implementation of the bank capital proposals. It is widely accepted that there may be anticipated or unanticipated consequences arising from a large shift in capital requirements. Quantifying this or even identifying everything in advance is very challenging. The range of estimates for net interest rate margin changes is only one example of the uncertainty. A broader set of forecasts for the impact on risk-weighted asset growth most likely sets the scene for greater uncertainty, especially in transition costs.
Accordingly, from a risk management perspective, we would suggest a slower, staggered approach to reaching the desired capital levels.
We propose a time schedule that might, for instance, seek to reach 13% to 13.5% Tier 1 capital within 5 years and includes, say, a 1.5% CCyB. We then recommend the potential for the RBNZ to wait for circa 2 years to observe empirical outcomes. The pause approach may substantially lessen risks of higher than expected transition costs or even reveal longer-term flaws in the proposed policy. The pause may also give time for the RBNZ (and APRA) to interpret further Basel rule changes and to also assess the growing role of the non-bank financial sector as these changes are implemented.
At some more distant stage, a further judgement in lifting the target to 14% or 14.5% could be made (either adding a further 1% CcyCB or more to the conservation buffer).
Our overall approach would see a ten year plus timetable and create greater optionality around policy, enable real-time observation of the impact of the policy change and probably be inherently less destabilising. It could also be more politically appealing and therefore reduce the risk of regulatory intervention through a public backlash should interest rate changes prove sharp, or lending drop off substantially. Good policy needs to be sustainable, as the benefit of these proposed changes needs to be measured over perhaps 50 -100 years.
The only negative aspects would be if there was an event that stressed the financial system in the near term, or if personnel changes removed the appetite for a longer time frame for implementing the policy.
Finally, in the near term, given the softer economic confidence, we are experiencing it does not seem like the best time to seek to rapidly increase bank capital, rather a more measured approach could be considered.
We welcome further engagement.
Andrew Bascand
Managing Director
Mark Brown
Head of Fixed Income, Director
Simon Pannett
Senior Credit Analyst, Director
April 2019
[1] Harbour Asset Management manages funds on behalf large wholesale clients and retail advisers, spanning KiwiSaver, superannuation, charitable trusts, IWI, and financial independent advisers. Today our Funds Under Management are a little over $4.7bn. We are a significant investor in both bank debt and bank equity. We also manage money for many banks and bank clients. We, therefore, have deep relationships across the bank sector.
[2] RBNZ 2017 Stress test included a downturn in the Chinese economy, a collapse in the demand for commodities a fall of 35% in house prices, and a dairy payout ratio below $5kgMS.
[3] Cline 2016, “The problem with this line of thinking is that monetary policy should stick to its central mandate—maintaining price stability along with high employment—and not be burdened with extraneous obligations.”
[4] We are looking forward to further work on tier 1 capital for mutually owned New Zealand banks.
[5] RBNZ Capital Review Paper 4 (See paragraph 114 and 115).
[6] The M&M theorem holds that the average cost of capital to the firm is independent of capital structure, because any reduction in capital cost from switching to higher leverage using lower cost debt is exactly offset by an induced increase in the unit cost of higher-cost equity capital as a consequence of the associated rise in risk.
[7] Cline, 2015. Testing the Modigliani-Miller Theorem of Capital Structure Irrelevance for Banks
[8] Cline, 2016. Benefits and Costs of Higher Capital Requirements for Banks
[9] It is our guess however that APRA may introduce a CCyB overtime that would narrow the gap and the adoption of a 14-14.5% central NZ optimal level would reduce further the scope for regulatory arbitrage.