Regulation Asia contributor, Jamie Lloyd Evans, explores the possibility of a policy change on AT1 capital in light of recent banking system distress.
In a previous article, I suggested that policymakers might at some point choose to eliminate AT1 as a regulatory category within the capital stack in order to simplify the Basel framework. Since that time one prominent regulatory voice has suggested a longer-term move in that direction, but the probability of some type of policy change has increased following Credit Suisse’s AT1 write-down in March of this year.
The main problem with AT1 capital is its complexity. AT1 instruments are hybrids that combine the characteristics of equity and debt, on top of which regulatory bells and whistles have been added over time. Complexity in turn creates behavioural uncertainty which affects investors and bank supervisors alike.
A distant antecedent of AT1 was Innovative Tier 1 which was formally recognised as a subset of Tier 1 capital by the Basel Committee in 1998 and limited to 15% of total Tier 1 capital. This was a political compromise but entailed a certain intellectual commitment to the notion that this is “going concern” capital, ie. capable of absorbing losses in the normal course of a bank’s operations, as opposed to a liquidation or resolution scenario[1].
Today loss absorbency is hard-wired into AT1 in the form of a Basel 3 pre-specified trigger, set at a minimum CET1 level of 5.125%, which results in a write-down or conversion to equity when breached; this applies to all instruments accounted for as liabilities [2].
Additionally, all instruments are subject to a discretionary loss absorbency provision, known as the point of non-viability (PONV), which occurs either when government support is provided to the bank or the bank is deemed to be at risk of insolvency by the authorities.
Critics of AT1 argue that there is little or no evidence that these provisions have ever functioned effectively on a going concern basis, and this article attempts to explore this point of view.
Swiss finish
The Credit Suisse situation is a very current example and is likely to become perhaps the most important AT1 case study for many years to come.
In summary, Credit Suisse merged with UBS in March of this year following a USD 17 billion permanent write-down of Credit Suisse’s AT1 capital.
The Swiss authorities subsequently explained that in their eyes a viability event had occurred according to the PONV provisions when Credit Suisse received liquidity support from the Swiss government. PONV was triggered outside of a bank resolution scenario, and therefore the merger with UBS proceeded ostensibly on a commercial basis.
The positive interpretation of this case is that AT1 instruments provided a lifeline to the bank while it was still a going concern and enabled it to avoid a more rapid deterioration or slide towards insolvency: QED.
Critics however might point to the fact that the non-viability event had already occurred at the time of write-down and for that reason the bank was already in a distressed condition from which it would have been unlikely to recover.
A non-viable bank is not necessarily equivalent to a bank that is a gone concern or insolvent in legal or accounting terms, but it seems difficult to disagree with the view that a non-viable bank is – conceptually – much closer to a gone rather than going concern.
Perhaps the more important question is why none of the pre-specified triggers were activated prior to PONV, but we will return to this question.
American exceptionalism
Given the idiosyncrasies of the Credit Suisse case, a richer historical precedent might be found in the experience of the Troubled Asset Relief Program (TARP) in the USA. TARP was introduced in 2008 at the height of the Financial Crisis as a multi-faceted attempt to stabilising the financial system.
One component of TARP[3] imposed a requirement on the largest US banks to issue preferred stock – AT1 in today’s language – to the US government. Banks such as JP Morgan protested that this form of mandatory recapitalisation was unnecessary and sought to repay the TARP funds as soon as possible. However, other banks, such as Citigroup, actually needed more capital with the result that additional government investments were made and subsequently converted to common equity.
TARP represents something of a Rorschach test for policymakers and two schools of thought emerge from this episode. On the one hand, TARP validates AT1 as a form of going concern capital because it enabled the largest US banks to continue to operate throughout the crisis. On the other hand, cynics would argue that the need to convert the preferred stock investment to equity in the case of Citigroup is a proof-text example of AT1’s inability to absorb losses on an on-going basis.
TARP had many moving parts and it may be difficult to draw general conclusions. One important observation, however, is that the US authorities – rightly or wrongly – identified AT1 as the best entry point in the capital stack to support the banks while seeking to limit costs to taxpayers and direct equity ownership (or creeping nationalisation) of the banks.
European heights
In the aftermath of the Financial Crisis a new wave of capital instruments was led by the European banks, often referred to as contingent convertibles. Most of these instruments were intended to qualify as AT1 but also introduced much higher pre-specified loss absorbency triggers, eg. at 7% or above.
These instruments in one sense are more consistent with the notion of going concern loss absorbency, because the higher the trigger the greater the distance from the point of non-viability or risk of insolvency. Consequently, the exercise of a high-trigger could provide more flexibility to a bank to remediate its balance sheet problems before it becomes distressed.
High triggers therefore seem to be very reasonable but certain assumptions underlying this approach are open to question.
Firstly, this approach assumes that the capital ratio accurately reflects the insolvency risk of the bank and is readily available to stakeholders on a timely basis.
Secondly, it suggests that the pre-specified trigger would be exercised prior to PONV, ie. PONV ought to be determined at a lower capital intervention point.
Thirdly, it potentially ignores the negative signaling effect that a breach of the high trigger could have on all parts of the capital stack above and below AT1 (if they were not already reflecting the distress of the issuer).
The Credit Suisse case arguably calls into question some of these assumptions. For example, the last publicly disclosed CET1 ratio of the bank prior to the merger was 14.1%; this was published with the bank’s full year results on 9 February 2023 but by 19 March 2023 the merger with UBS had been announced.
Perhaps the gentlest criticism then of the high-trigger approach is that it inhabits the first part of the famous Ernest Hemingway quote on bankruptcy: “gradually”, as opposed to “then suddenly”.
Cost control
Higher-trigger AT1s are more costly for banks than low-trigger AT1 and Tier 2, and this brings us to the cost of capital argument in favour of AT1.
Banks would generally claim that AT1 provides the industry with greater flexibility to manage their aggregate cost of capital and that its elimination would raise costs for the industry. This is rooted in the sense that AT1 is located on a continuum of capital – or total loss absorbing capacity (TLAC) – and in the absence of AT1 the more senior creditors (eg. T2 capital providers) would require the ‘gap’ to be filled by common equity. An extension of this argument sometimes includes investor diversification and financial flexibility.
Policymakers are unlikely to disagree with these arguments or wish to see banks’ cost of capital increase arbitrarily. However, regulators are principally concerned with ensuring the safety and soundness of banks rather than optimising returns.
There is also a possibility that the cost of capital argument – when examined in more detail – could lead regulators to draw the opposite conclusion from the banks. Specifically, that if AT1 pricing has historically been closer to Tier 2 capital rather than equity, then the market already intuits that AT1 is more akin to gone concern rather than going concern capital.
At this point it may be helpful to clarify what the removal of AT1 could mean in practice. On the one hand it could mean the elimination of AT1 as a sub-category of Tier 1 going concern capital, or on the other hand the exclusion of AT1 instruments from the regulatory capital stack altogether. However, the latter position is difficult to maintain because it assumes that AT1 instruments have no loss-absorbing capability whatsoever.
Re-designation of AT1 instruments from going concern to gone concern capital (ie. re-classification from Tier 1 to Tier 2 capital) therefore seems to be the more reasonable option that policymakers might consider.
Re-designation not design
What would be the implications of a re-designation of AT1 as T2 capital?
The best outcome from a regulatory perspective would be a reduction of behavioural uncertainty. In practice this would entail a ‘relocation’ of AT1’s loss absorbency to the context of bank resolution and liquidation.
Some consequential modifications would however be required.
Firstly, ‘CET1’ as a specific category within Tier 1 capital would essentially be redundant and the capital framework might redefine Tier 1 capital as synonymous with common equity. Given the experience of TARP, policymakers might consider holding open a Tier 1 sub-category for government-led recapitalisation efforts, but there is unlikely to be a ready consensus on this point.
Secondly, assuming the minimum Tier 1 and Total capital ratio requirements are maintained at 4.5% and 8%, other requirements (eg. the leverage ratio, large exposures base) that currently reference Tier 1 capital might be recalibrated downwards to be aligned with CET1 levels.
Thirdly, if AT1 is reclassified as Tier 2 capital then distribution constraints related to these instruments might be removed. These constraints arise if a bank chooses to draw down its capital buffers and this is regarded as an obstacle to buffer usability.
One disadvantage of this approach however is that the expanded Tier 2 category would initially be more complex given differences in instrument seniority, tenor, cost, etc. This could lead to the adoption of sub-categories (eg. upper or lower T2) but applying sub-limits would certainly create additional complexity – and may not be warranted given current disclosure standards.
What’s in a name?
If regulators choose to re-designate AT1 it’s difficult to anticipate how the market for such instruments would react.
Superficially, investment bankers and private bankers would need to relabel their marketing materials from AT1 to Tier 2, but assuming no significant change in yield/pricing, greater clarity on loss-absorbency and distribution constraints might even enhance the attraction of these instruments.
Investors in traditional Tier 2 capital might also have an interest in ensuring that banks continue to issue more junior AT1-type instruments within the capital / TLAC stack regardless of regulatory classification.
This implies that redesignation could be more of a cosmetic challenge rather than a change in economic realities. But policymakers would need to be mindful of unintended consequences, if the aggregate supply of AT1-type instruments should shrink or if recapitalisation options are narrowed.
However, one reason why redesignation might work in practice is that the Tier 1 ratio is accorded much lower informational value than CET1-based metrics by investors and supervisors, and possibly lower than Total Capital measures as well.
This discussion is probably guilty of over-simplifying the issues related to AT1, but momentum seems to be with those who are questioning its complexity and the resultant behavioural uncertainty.
A change in policy is by no means inevitable, but arguably the status quo provides us with a version of AT1 that is like a friend that joins us for dinner at an expensive restaurant; until the bill arrives we’re never sure whether they have brought their wallet.
—
[1] For the purposes of this article, loss-absorbency will refer to a loss of ‘principal’ value of the instrument as opposed to capital conservation features such as the suspension of distributions.
[2] Differences in the mechanics of loss-absorbency for liability AT1 capital versus equity AT1 capital are not discussed here, but assumed to have similar levels of efficacy in a distressed scenario
[3] The Capital Purchase Program (CPP)
