Allowing companies to fail

I suspect (but can’t prove) that creative destruction is one of the under appreciated factors that underpin the health of the economy. There is quite a lot of evidence however that creative destruction has been suppressed since the 2008 Global Financial Crisis. The rights and wrongs of the extent to which bail-outs were and continue to be necessary is too big a topic to cover in this post.

For the record, I do believe that the bail-outs of the banks were necessary at the time but that “bail-in” gives bank supervisors a very real option to avoid having to do this in the future. The increase in capital requirements are also likely to reduce the risk of a bail-in being required. Others may disagree and my views chiefly relate to the Australian banking system which is where my professional expertise is based. The issues associated with COVID-19 raise a whole lot of related but, in many ways, different issues. At the risk of stating the obvious, it’s complicated.

Against that background, I found this short article published on the VoxEU website worth reading as another reminder of the value of allowing companies to fail and/or be restructured. The conclusion of the article (copied below) gives you the key points the authors derive from their research

We investigate a large number of stakeholders that could be negatively affected by a fire sale but find little evidence for negative externalities. The main effect of fire sales is a wealth transfer from the seller to the buyer. Thus, from a welfare perspective, the costs associated with fire sales of corporate assets are much lower than previously thought based on an analysis of seller costs only. From a policy perspective, these findings indicate that the merits of bailouts as a response to the potential losses associated with fire sales are limited, especially given the moral hazard and the other distortions caused by these bailouts. 

We recognise that the economic shock caused by the COVID-19 pandemic is unparalleled since the WWII and the Great Depression, and hence, some emergency measures and bailouts were likely necessary to prevent a meltdown of economic activity. However, one difference between the current crisis and the Global Crisis is the apparent lack of fire sales of struggling companies or investments into such companies at fire-sale prices. Warren Buffett’s Berkshire Hathaway, for instance, invested $5 billion in Goldman Sachs in September 2008 and $3 billion in General Electric in October 2008, while Warren Buffett’s firm has not undertaken any major investments during the COVID-19 crisis (Financial Times 2020). Our results therefore suggest that, at least at the margin, fire sales would have been an effective alternative to bailouts, especially for large bailouts such as for the airlines in the US.

“The merits of fire sales and bailouts in light of the COVID-19 pandemic”, Jean-Marie Meier and Henri Servaes, 18 January 2021.

Dee Hock, the Father of Fintech

Marc Rubinstein writing in his “Net Interest” newsletter has a fascinating story about the history of Visa. The article is interesting on a number of levels.

It is partly a story of the battle currently being played out in the “payments” area of financial services but it also introduced me to the story of Dee Hock who convinced Bank of America to give up ownership of the credit card licensing business that it had built up around the BankAmericard it had launched in 1958. His efforts led to the formation of a new company, jointly owned by the banks participating in the credit card program, that was the foundation of Visa.

The interesting part was that Visa was designed from its inception to operate in a decentralised manner that balanced cooperation and competition. The tension between cooperation (aka “order”) and competition (sometimes leading to “disorder”) is pervasive in the world of money and finance. Rubinstein explores some of the lessons that the current crop of decentralised finance visionaries might take away from this earlier iteration of Fintech. Rubinstein’s post encouraged me to do a bit more digging on Hock himself (see this article from FastCompany for example) and I have also bought Hock’s book (“One from Many: VISA and the Rise of Chaordic Organization“) to read.

There is a much longer post to write on the issues discussed in Rubinstein’s post but that is for another day (i.e. when I think I understand them so I am not planning to do this any time soon). At this stage I will just call out one of the issues that I think need to be covered in any complete discussion of the potential for Fintech to replace banks – the role “elasticity of credit” plays in monetary systems.

“Elasticity of credit”

It seems pretty clear that the Fintech companies offer a viable (maybe compelling) alternative to banks in the payment part of the monetary system but economies also seem to need some “elasticity” in the supply of credit. It is not obvious how Fintech companies might meet this need so maybe there remains an area where properly regulated and supervised banks continue to have a role to play. That is my hypothesis at any rate which I freely admit might be wrong. This paper by Claudio Borio offers a good discussion of this issue (for the short version see here for a post I did on Borio’s paper).

Recommended

Tony – From the Outside

The potential for computer code to supplant the traditional operating framework of the economy and society

I am very far from expert on the issues discussed in the podcast this post links to, I am trying however to “up-skill”. The subject matter is a touch wonky so this is not a must listen recommendation. That said, the questions of DeFi and cryptocurrency are ones that I believe any serious student of banking and finance needs to understand.

In the podcast Demetri Kofinas (Host of the Hidden Forces podcast) is interviewed by two strong advocates of DeFi and crypto debating the potential of computer code to supplant legal structures as an operating framework for society. Demetri supports the idea that smart contracts can automate agreements but argues against the belief that self-executing software can or should supplant our legal systems. Computer code has huge potential in these applications but he maintains that you will still rely on some traditional legal and government framework to protect property rights and enforce property rights. He also argues that it is naïve and dangerous to synonymize open-source software with liberal democracy.

I am trying to keep an open mind on these questions but (thus far) broadly support the positions Demetri argues. There is a lot of ground to cover but Demetri is (based on my non-expert understanding of the topic) one of the better sources of insight I have come across.

Tony – From the Outside

First-time buyers: how do they finance their purchases and what’s changed? – Bank Underground

Some interesting research out of the UK examining the impact of a variety of factors associated with first time house purchases.

Our results support claim that average FTBs are increasingly higher up the income distribution for their age. And slower than average income growth for younger workers have worked against FTBs. But our results challenge the view that average FTB ages have got much higher. And while FTBs on average are borrowing more in nominal terms, they aren’t spending more of their income on mortgage repayments than before: cheaper credit has roughly cancelled out the effect of bigger mortgages.

— Read on bankunderground.co.uk/2020/09/16/first-time-buyers-how-do-they-finance-their-purchases-and-whats-changed/

The dark art of measuring residential mortgage risk

Residential mortgages are one of the seemingly more plain vanilla forms of bank lending. Notwithstanding, comparing capital requirements applied to this category of lending across different types of banks can be surprisingly complicated and is much misunderstood. I have touched on different aspects of this challenge in a number of mortgage risk weight “fact check” posts (see here and here), focussing for the most part on the comparison of “standardised” capital requirements compared to those applied to banks operating under the “internal rating based” (IRB) approach.

A discussion paper (“A more flexible and resilient capital framework for ADIs”, 8 December 2020) released by the Australian Prudential Regulation Authority (APRA) offers a good summary (see p27 “Box 2”) of the differences in capital requirements not captured by simplistic comparisons of risk weights. However, one of the surprises in the discussion paper was that APRA chose not to address one of these differences by aligning the credit conversion factors applied to off-balance sheet (non-revolving) residential mortgage exposures.

Understanding why APRA chose to maintain a different treatment of CCFs across the two approaches offers some insights into differences in the way that the two approaches recognise and measure the underlying risks.

Before proceeding we need to include a short primer on “off-balance exposures” and “CCFs”. Feel free to skip ahead if you already understand these concepts.

  • Off-balance sheet exposures are the difference between the maximum amount a bank has agreed to lend and the actual amount borrowed at any point in time.
  • The CCF is the bank’s estimate of how much of these undrawn limits will in fact have been called on (converted to an on balance sheet exposure) in the event a borrower defaults.
  • In the case of “non-revolving” residential mortgages, these off-balance sheet exposures typically arise because borrowers have got ahead of (“pre-paid”) their contractual loan repayments.

APRA noted that the credit conversion factor (CCF) currently applied to off-balance sheet exposures was much higher for IRB banks than for standardised, thereby partially offsetting the lower risk weights applied under the IRB approach. It had been expected that APRA would address this inconsistency by applying a 100% CCF under both approaches.

Contrary to this expectation, APRA has proposed to revise the CCFs applying to (non-revolving) off-balance sheet residential mortgage exposures as follows:

Current

Standardised 0-50%

IRB 100%

Proposed

40%

100% (unchanged)

The interesting nuance here is that APRA is not saying that standardised banks are likely to experience a lower percentage drawdown of credit limits in the event a borrower defaults. In the “Response to Submissions” that accompanied the Discussion Paper, APRA noted that “Borrowers do not typically enter default until they have fully drawn down on their available limit, including any prepayments ahead of their scheduled balance.

However, APRA also noted that loans with material levels of prepayment are also likely to be lower risk based on the demonstrated greater capacity to service and repay the loan.

Under the IRB approach, the greater capacity to repay the loan is generally recognised through a lower PD estimate which the IRB formula translates into lower risk weights reflecting the lower risk. In the absence of some equivalent risk recognition mechanism in the standardised approach, APRA is proposing to use a concessional CCF treatment to reflect the lower risk of loans with material levels of prepayment. It notes that the concessional CCF treatment will also contribute to ensuring the difference in residential mortgage capital requirements between the standardised and IRB approaches remains appropriate.

Summing up:

  • Looked at in isolation, 100% is arguably the “right” value for the CCF to apply to off-balance sheet exposures for a non-revolving residential mortgage irrespective of whether it is being measured under the standardised or IRB approach
  • But a “concessional” CCF is a mechanism (fudge?) that allows the standardised approach to reflect the lower risk associated with loans with material levels of prepayment

Tony – From the Outside

Low Risk Residential Mortgage Risk Weights

I have posted a number of times on the question of residential mortgage risk weights, either on the general topic of the comparison of the risk weights applied to the standardised and IRB ADIs (see here) or the reasons why risk weights for IRB ADIs can be so low (see here).

On the question of relative risk weights, I have argued that the real difference between the standardised and IRB risk weights is overstated when framed in terms of simplistic comparisons of nominal risk weights that you typically read in the news media discussion of this question. I stand by that general assessment but have conceded that I have paid insufficient attention to the disparity in risk weights at the higher quality end of the mortgage risk spectrum.

A Discussion Paper released by APRA offers a useful discussion of this low risk weight question as part of a broader set of proposals intended to improve the transparency, flexibility and resilience of the Australian capital adequacy framework.

In section 4.2.1 of the paper, APRA notes the concern raised by standardised ADIs…

A specific concern raised by standardised ADIs in prior rounds of consultation has been the difference in capital requirements for lending at low LVRs. Stakeholders have noted that the lowest risk weight under the standardised approach would be 20 per cent under the proposed framework, but this appears to be significantly lower for the IRB approach. In response to this feedback, APRA has undertaken further analysis at a more detailed level, noting the difference in capital requirements that need to be taken into account when comparing capital outcomes under the standardised and IRB approaches (see Box 2 above).

But APRA’s assessment is that the difference is not material when you look beyond the simplistic comparison of risk weights and consider the overall difference in capital requirements

APRA does not consider that there is a material capital difference between the standardised and IRB approaches at the lower LVR level. For loans with an LVR less than 60 per cent, APRA has estimated that the pricing differential that could be reasonably attributed to differences in the capital requirements between the two approaches would be lower than the differential at the average portfolio outcome.

In explaining the reasons for this conclusion, APRA addresses some misconceptions about the IRB approach to low LVR lending compared to the standardised approach

In understanding the reasons for this outcome, it is important to understand the differences in how the standardised and IRB approaches operate. In particular, there are misconceptions around the capital requirement that would apply to low LVR lending under the IRB approach. For example, it would not be appropriate to solely equate the lowest risk weight reported by IRB ADIs in market disclosures with low LVR loans. The IRB approach considers a more complex range of variable interactions compared to the standardised approach. Under the standardised approach, a low risk weight is assigned to a loan with a low LVR at origination.

One of the key points APRA makes is that IRB ADIs do not get to originate loans at the ultra low risk weights that have been the focus of much of the concern raised by standardised ADIs.

In particular, IRB estimates are more dynamic through the life of the loan, for example, they are more responsive to a change in borrower circumstances or movements in the credit cycle. Standardised risk weights generally do not change over the life of a loan. For an IRB ADI, the lowest risk weight is generally applied to loans that have significantly prepaid ahead of schedule. A low LVR loan on the standardised approach is not necessarily assigned the lowest risk weight under the IRB approach at origination.

APRA states that it is not appropriate to introduce “dynamic”factors into the standardised risk weight framework.

APRA is not proposing to include dynamic factors in determining risk weights under the standardised approach for the following reasons:

– the standardised approach is intended to be simple and aligned with Basel III. For the standardised approach, APRA considers it more appropriate to focus on origination rather than behavioural variables as this has more influence on the quality of the portfolio and leads to less procyclical capital requirements; and

– the average difference between standardised and IRB capital outcomes is much narrower at the point of origination, which is the key point for competition. While the difference between standardised and IRB capital outcomes could widen over the life of the loan, APRA has ensured that the difference in average portfolio outcomes remains appropriate

But that it does intend to introduce a 5 per cent risk weight floor into the IRB approach to act as a backstop.

That said, APRA is proposing to implement a 5 per cent risk-weight floor for residential mortgage exposures under the IRB approach, to act as a simple backstop in ensuring capital outcomes do not widen at the lower risk segment of the portfolio. This is consistent with the approach taken by other jurisdictions and will limit the difference in capital outcomes between the standardised and IRB approaches for lower risk exposures. This risk-weight floor is in addition to other factors that will reduce the difference in capital outcomes between standardised and IRB ADIs, such as the higher CCB for IRB ADIs and lower CCF estimates for standardised ADIs.

As always, it remains possible that I am missing something. The explanation offered by APRA however gives me confidence that my broad argument about the overstatement of the difference has been broadly correct. Equally importantly, the changes to residential mortgage risk weights proposed in the Discussion Paper will further reduce the gap that does exist.

Tony – From the Outside

Capital adequacy – looking past the headline ratios

Comparing capital adequacy ratios is full of traps for the unwary. I recently flagged a speech by Wayne Byres (Chairman of APRA) that indicated APRA will be releasing a package of capital adequacy changes that will be more aligned with the international minimum standards and result in higher reported capital ratios.

While waiting for this package to be released, I thought it might be useful to revisit the mechanics of the S&P Risk Adjusted Capital (RAC) ratio which is another lens under which Australian bank capital strength is viewed. In particular I want to highlight the way in which the S&P RAC ratio is influenced by S&P’s assessment of the economic risks facing banks in the countries in which the banks operate.

The simplest way to see how this works is to look at an example from 2019 when S&P announced an upgrade of the hybrid (Tier 1 and Tier 2) issues of the Australian major banks. The senior debt ratings were left unchanged but the hybrid issues were all upgraded by 1 notch. Basel III compliant Tier 2 ratings were raised to “BBB+” (from “BBB”) and Tier 1 were raised to “BBB-” (from “BB+”).

S&P explained that the upgrade was driven by a revision in S&P’s assessment of the economic risks facing the Australian banks; in particular the “orderly decline in house prices following a period of rapid growth”. As a consequence of the revised assessment of the economic risk environment,

  • S&P now apply lower risk weights in their capital analysis,
  • This in turn resulted in stronger risk-adjusted capital ratios which now exceed the 10% threshold where S&P deem capital to be “strong” as opposed to “adequate”,
  • Which resulted in the Stand-Alone Credit Profile (SACP) of the Australian majors improving by one-notch and hence the upgrades of the hybrids.

Looking past the happy news for the holders of major bank hybrid issues, what was interesting was how much impact the revised assessment of the economic outlook has on the S&P risk weights. The S&P assessment of the economic outlook is codified in the BICRA score (short for the Banking Industry Country Risk Assessment) which assigns a numeric value from 1 (lowest risk) to 10 (highest risk). This BICRA score in turn determines the risk weights used in the S&P Risk Adjusted Capital ratio.

As a result of S&P revising its BICRA score for Australia from 4 to 3, the risk weights are materially lower with a commensurate benefit to the S&P assessment of capital adequacy (see some selected risk weights in the table below):

BICRA
Residential MortgagesBankCorporate, IPCRE, Business LendingCredit Cards
3302375105
4373387118
% Change
18.9%30.3%13.8%11.0%

The changes were fairly material across the board but the impact on residential mortgages (close to 20% reduction in the risk weight) is particulate noteworthy given the fact that this class of lending dominates the balance sheets of the Australian majors. It is also important to remember that, what the S&P process gives, it can also takeaway. This substantial improvement in the RAC ratio could very quickly reverse if S&P revised its economic outlook score or industry rating.

The other aspect of this process that is worth noting is the way in which the risk weights are anchored to S&P defined downturn scenarios and an 8% capital ratio. In the wake of the Royal Commission into Australian Banking, there has been a lot of focus on the idea of large banks being “Unquestionably Strong”. APRA subsequently determined that a 10.5% CET1 benchmark for capital strength was sufficient for a bank to be deemed to meet this test.

In that context, the S&P assessment that an 8-10% RAC ratio is “adequate” sounds a bit underwhelming. However, my understanding is the S&P risk weights are calibrated to an “A” or “substantial” stress scenario which is defined by the following Key Economic Indicators (KEI)

  • GDP decline of up to 6%
  • Unemployment of up to 15%
  • Stock market decline of ups to 60%

The loss rates expected in response to this level of stress are translated into equivalent risk weights using a 8% RAC ratio. The capital required by an 8% RAC ratio may only be “adequate” in S&P terms, but starts to look a lot more robust when you understand the severity of the scenario driving the risk weights that drive that requirement.

Summing up

I am not suggesting that there is anything fundamentally wrong with the S&P process, my purpose is simply to offer some observations regarding how the ratios in the capital adequacy assessment should be interpreted:

  1. Firstly to recognise that the process is by design anchored to an 8% capital ratio and risk weights that are calibrated to a very severe (“Substantial” is the term S&P uses) stress scenario, and
  2. Secondly, that the process is very sensitive to the BICRA score

This is not an area in which I will claim deep expertise so it is entirely possible that I am missing something. There are people who understand the S&P rating process far better than I do and I am very happy to stand corrected if I have mis-understood or mis-represented anything above.

Tony – From the Outside

APRA reflects on “… a subtle but important shift in regulatory thinking”

Wayne Byres speech to the Risk Management Association covered a range of developments but, for me, the important part was the discussion of the distinction between strength and resilience referenced in the title of this post.

This extract from the speech sets out how Mr Byres frames the distinction …

… in the post-GFC period, the emphasis of the international reforms was on strengthening the global financial system. Now, the narrative is how to improve its resilience. A perusal of APRA speeches and announcements over time shows a much greater emphasis on resilience in more recent times as well.

What is behind this shift? Put simply, it is possible to be strong, but not resilient. Your car windscreen is a great example – without doubt it is a very strong piece of glass, but one small crack and it is irreparably damaged and ultimately needs to be replaced. That is obviously not the way we want the financial system to be. We want a system that is able to absorb shocks, even from so-called “black swan” events, and have the means to restore itself to full health.

In saying that, financially strong balance sheets undoubtedly help provide resilience, and safeguarding financial strength will certainly remain the cornerstone of prudential regulation and supervision. But it is not the full story. So with that in mind, let me offer some quick reflections on the past year, and what it has revealed about opportunities for the resilience of the financial system to be further improved.

APRA Chair Wayne Byres – Speech to the 2020 Forum of the Risk Management Association – 3 December 2020

To my mind, the introduction of an increased emphasis on resilience is absolutely the right way to go. We saw some indications of the direction APRA intend to pursue in the speech that Mr Byres gave to the AFR Banking and Wealth Summit last month and will get more detail next week (hopefully) when APRA releases a consultation paper setting out a package of bank capital reforms that is likely to include a redesign of the capital buffer framework.

This package of reforms is one to watch. To the extent that it delivers on the promise of increasing the resilience of the Australian banking system, it is potentially as significant as the introduction of the “unquestionably strong” benchmark in response to the Australian Financial System Inquiry.

Tony – From the Outside

Separating deposit-taking from investment banking: new evidence on an old question

The BOE has released a paper exploring the question of how ring fencing deposit taking from investment banking impacts the banking market. I have included the abstract of the paper below and you can find a summary of the paper here on the “Bank Underground” blog. I don’t see this as the final word on these questions but it does offer a perspective worth noting.

Abstract

The idea of separating retail and investment banking remains controversial. Exploiting the introduction of UK ring-fencing requirements in 2019, we document novel implications of such separation for credit and liquidity supply, competition, and risk-taking via a funding structure channel.

By preventing conglomerates from using retail deposits to fund investment banking activities, this separation leads conglomerates to rebalance their activities towards domestic mortgage lending and away from supplying credit lines and underwriting services to large corporates.

By redirecting the benefits of deposit funding towards the retail market, this rebalancing reduces the cost of credit for households, without eroding lending standards. However the rebalancing also increases mortgage market concentration and risk-taking by smaller banks via indirect competition effects.

Tony – From the Outside

Bank capital adequacy – APRA chooses Option 2

APRA released a discussion paper in August 2018 titled “Improving the transparency, comparability and flexibility of the ADI capital framework” which offered two alternative paths.

  • One (“Consistent Disclosure”) under which the status quo would be largely preserved but where APRA would get involved in the comparability process by adding its imprimatur to the “international harmonised ratios” that the large ADIs use to make the case for their strength compared to their international peers, and
  • A second (“Capital Ratio Adjustments”) under which APRA would align its formal capital adequacy measure more closely with the internationally harmonised approach.

I covered those proposals in some detail here and came out in favour of the second option. I don’t imagine APRA pay much attention to my blog but in a speech delivered to the AFR Banking and Wealth Summit Wayne Byres flagged that APRA do in fact intend to pursue the second option.

The speech does not get into too much detail but it listed the following features the proposed new capital regime will exhibit:

– more risk-based – by adjusting risk weights in a range of areas, some up (e.g. for higher risk housing) and some down (e.g. for small business);
 – more flexible – by changing the mix between minimum requirement and buffers, utilising more of the latter;
 – more transparent – by better aligning with international minimum standards, and making the underlying strength of the Australian framework more visible;
 – more comparable – by, in particular, making sure all banks disclose a capital ratio under the common, standardised approach; and
 – more proportionate – by providing a simpler framework suitable for small banks with simple business models.

while also making clear that

… probably the most fundamental change flowing from the proposals is that bank capital adequacy ratios will change. Specifically, they will tend to be higher. That is because the changes we are proposing will, in aggregate, reduce risk-weighted assets for the banking system. Given the amount of capital banks have will be unchanged, lower risk-weighted assets will produce higher capital ratios.

However, that does not mean banks will be able to hold less capital overall. I noted earlier that a key objective is to not increase capital requirements beyond the amount needed to meet the ‘unquestionably strong’ benchmarks. Nor is it our intention to reduce that amount. The balance will be maintained by requiring banks to hold larger buffers over their minimum requirements.

One observation at this stage …

It is hard to say too much at this stage given the level of detail released but I do want to make one observation. Wayne Byres listed four reasons for the changes proposed;

  1. To improve risk sensitivity
  2. To make the framework more flexible, especially in times of stress
  3. To make clearer the fundamental strength of our banking system vis-a-vis international peers
  4. To ensure that the unquestionably strong capital built up prior to the pandemic remains a lasting feature of the Australian banking system.

Pro-cyclicality remains an issue

With respect to increasing flexibility, Wayne Byres went on to state that “Holding a larger proportion of capital requirements in the form of capital buffers main that there is more buffer available to be utilised in times of crisis” (emphasis added).

It is true that the capital buffer will be larger in basis points terms by virtue of the RWA (denominator in the capital ratio) being reduced. However, it is also likely that the capital ratio will be much more sensitive to the impacts of a stress/crisis event.

This is mostly simple math.

  1. I assume that loan losses eating into capital are unchanged.
  2. It is less clear what happens to capital deductions (such as the CET1 deduction for Regulatory Expected Loss) but it is not obvious that they will be reduced.
  3. Risk Weights we are told will be lower and more risk sensitive.
  4. The lower starting value for RWA in any adverse scenario means that the losses (we assume unchanged) will translate into a larger decline in the capital ratio for any given level of stress.
  5. There is also the potential for the decline in capital ratios under stress to be accentuated (or amplified) to the extent the average risk weights increase in percentage terms more than they would under the current regime.

None of this is intended to suggest that APRA has made the wrong choice but I do believe that the statement that “more buffer” will be available is open to question. The glass is however most definitely half full. I am mostly flagging the fact that pro-cyclicality is a feature of any risk sensitive capital adequacy measure and I am unclear on whether the proposed regime will do anything to address this.

The direction that APRA has indicated it intends to take is the right one (I believe) but I think there is an opportunity to also address the problem of pro-cyclicality. I remain hopeful that the consultation paper to be released in a few weeks will shed more light on these issues.

Tony – From the Outside

p.s. the following posts on my blog touch on some of the issues that may need to be covered in the consultation

  1. The case for lower risk weights
  2. A non zero default for the counter cyclical capital buffer
  3. The interplay of proposed revisions to APS 111 and the RBNZ requirement that banks in NZ hold more CET1 capital
  4. Does expected loss loan provisioning reduce pro-cyclicality
  5. My thoughts on a cyclical capital buffer