The New Normal for the Bank Levy

A significant package of changes to the bank levy rules comes into force on 1 January 2021, which will affect almost every bank subject to the levy. The principal changes remove liabilities of non-UK subsidiaries from the bank levy base, with the knock-on consequence that intra-group liabilities owed to those subsidiaries fall into the bank levy charge for the first time. The changes also allow UK banks to elect to remove their non-UK permanent establishments from the bank levy base, update the relief for netting arrangements, and cater for some developments in regulatory capital rules in the last ten years.

On 1 January 2021, a major reform of the UK bank levy rules will finally enter into force more than five years after its original announcement, and almost three years since the changes were enacted in FA 2018. This is by far the most significant package of changes to bank levy in the ten years of the tax’s existence, and it will change the way all banks calculate and manage their bank levy liabilities.

This article summarises the existing bank levy rules, explains what will be changing, and looks at how banks should be reacting to these changes.

How it started

The bank levy was originally introduced by the coalition government in FA 2011 Sch 19. (All legislative references in this article are to Sch 19, unless otherwise stated.) The bank levy rules gave effect to the view that banks needed to make a greater tax contribution, to take account of the fiscal support they had been given through the 2008 financial crisis, and that, implicitly, they would expect to receive in any future crises.

The bank levy was charged on the amount of a banking group’s ‘chargeable equity and liabilities’, as appearing on the group’s balance sheet. It has always been clear that one purpose of bank levy is to encourage banks to reduce the size of their balance sheets, and to shift towards lower risk funding models. To support this policy, the bank levy rules:

• Do not tax tier one equity and liabilities (principally, ordinary shares, and capital raised as additional tier 1 debt).
• Do not tax funding provided through customer deposits protected under the UK financial services compensation scheme, and comparable overseas schemes.
• Do not tax liabilities owed by a bank to a counterparty, where those liabilities can be net settled against liabilities owed in the opposite direction. Importantly, bank levy netting applies to cases in which a bank, in substance, borrows money on a secured basis by selling shares under a repo, or by lending stock in return for cash collateral; in that situation, the bank can normally net its liability to return the cash against the value of the securities it has provided as (effective) collateral.
• Allow banking groups to deduct their “high quality liquid assets” from their chargeable equity and liabilities. These essentially comprise the highest-quality assets which are eligible for inclusion in the banks’ liquidity buffer (e.g., deposits with a central bank, bonds issued by EU member state central governments, and some very high-quality covered bonds).
• Apply a 50% per cent reduction in the levy rate when dealing with long-term liabilities (broadly, liabilities which do not need to be repaid within the next 12 months).

Additionally, although the bank levy code includes a targeted anti-avoidance rule, which disregards arrangements entered into with a main purpose of reducing a bank levy charge (para 47), safe harbours in para 47 mean that it cannot apply to any transaction which, on an ongoing basis, uses any of the methods above to move the bank towards a lower-risk funding model. As HMRC acknowledged in its Bank Levy Manual (at BKLM641050), the bank levy code ‘seeks to encourage banks to adopt lower risk funding strategies’ – and it would clearly be counterproductive if banks which responded to that incentive were then denied the bank levy saving because of the TAAR. (As banks are also subject to HMRC’s code of practice on taxation for banks, which requires them to comply with the spirit, as well as the letter, of tax law, it is helpful that HMRC has thus confirmed that, if a bank takes steps to reduce its bank levy liabilities by (for example) converting short-term funding into long-term funding, the resulting bank levy saving is consistent with the intentions of Parliament.)

But this policy aim, of encouraging banks to reduce the risk of their funding models, often conflicted with the second policy aim of bank levy, which was to raise a fixed revenue target each year. From 2011 onwards, banks responded to the bank levy incentive (and pressures from prudential regulators), by reducing the size of their balance sheets, and moving to lower risk funding models. To keep hitting the revenue targets, therefore, the government needed to increase bank levy rates repeatedly, with the rate quadrupling from 0.05% in 2011 to 0.21% in 2015. As Richard Milnes argued in these pages in 2014 (‘Lessons from the bank levy’, Tax Journal, 13 March 2014), using bank levy as a means of incentivising good behaviour, whilst clawing back the incentives through regular rate increases, was unlikely to be sustainable in the long-term.

Moreover, bank levy was charged on all of the liabilities of a consolidated group (or sub-group) headed by a UK entity, even where those liabilities were owed by a non-UK entity. This deterred non-UK banks from locating a regional holding company in the UK; and meant that UK-headed banking groups had to pay UK bank levy on their worldwide liabilities. This was particularly challenging for UK-headed banks with significant overseas operations and a small UK domestic business; and led to concerns that banks in this position might migrate their HQ from the UK, so that they paid levy only on their (relatively small) UK operations. If that had happened whilst there was a fixed revenue target for bank levy, the rate charged on other banks would then have needed to increase substantially to hit the target, which would have risked driving more bank HQs towards the exit.

How it’s going

In 2015, these concerns over the long-term sustainability of the bank levy model caused the Chancellor to announce that bank levy rates would be gradually reduced, from 0.21% in 2015 to 0.1% in 2021; and that, from 2021, bank levy would apply only to UK liabilities, and would cease to be charged on liabilities of overseas subsidiaries of a UK banking entity. (This wasn’t all good news for banks: these gradual changes were ‘paid for’ by the immediate introduction of a banking surcharge on banks’ profits, and the expected net effect was an increase of £1.5bn in tax revenues from banks between 2016 and 2021.)

Following consultation on the detail of the rules, the bank levy changes were then enacted in FA 2018, even though they would not become effective for a further three years. (The Hansard report of the Public Bill Committee debate on the bank levy reform is, incidentally, a quite extraordinary read: the provisions were discussed for over two hours, in a debate which encompassed, among other things, Marxism, Scottish nationalism and the economic policies of Francois Hollande, but which spent no time discussing the words of the legislation itself - and which, sadly, therefore casts no real light on Parliament’s intentions.)

What, then, has actually changed in the new rules?

1. Limiting the levy to UK-based equity and liabilities

Obviously, the most substantial package of changes are those which remove overseas equity and liabilities from the bank levy base.

For standalone UK resident entities (with no UK resident parent, and no subsidiaries), and for UK sub-groups with no overseas subsidiaries, the new rules work in the same way as the existing rules - you can simply use the equity and liabilities appearing in the relevant balance sheet (FA 2011 Sch 19 paras 15H and 15J). Similarly to the old rules, you can ignore any liabilities owed to other UK resident members of the broader group of which the entity or subgroup forms part, to avoid double-counting (para 15R).

Where a UK group/sub-group has non-UK resident subsidiaries, however, the default position will now require it to prepare notional consolidated accounts, covering only the UK resident members of the sub-group. Bank levy will then be charged based on the equity and liabilities in that notional balance sheet (para 15K). The group can, alternatively, make an entity-by-entity election under para 15L, under which it would compute the bank levy position of each entity separately, rather than producing notional consolidated accounts. However, although a banking group can make and revoke these elections at any time, they can only be made where this merely simplifies the calculation process, as they are deemed to be ineffective if they are made with a main purpose of reducing the bank levy charge (paragraph 15L(5)). Entity-by-entity elections are, therefore, likely to be unusual in practice.

Importantly, one effect of these changes is that some intra-group liabilities, which were previously disregarded for bank levy purposes, can now give rise to a bank levy charge. For example, if a non-UK resident subsidiary has previously lent money to its UK resident parent, that liability would have been ignored under the old rules, so the bank would not have needed to check (for example) if the liability qualified for bank levy netting. Under the new rules, the liability which the UK parent owes to its non-UK subsidiary would, in the first instance, form part of the bank levy base. Banks will, therefore, want to identify any intra-group liabilities which could form part of their chargeable equity and liabilities for the first time in 2021. And, for those liabilities, they will want to investigate (a) if the liability already qualifies for bank levy relief (for example, under the netting provisions), and (b) if not, whether steps can be taken to bring the liability within the relieving provisions. (For example, can it be brought within a valid netting agreement? Can it be converted into a long-term liability?)

2. How are branches treated?

Where a foreign bank trades in the UK through a UK branch, it will continue (as under the old rules) to be subject to bank levy on part of the equity and liabilities of the foreign bank. As before, the equity and liabilities of the foreign bank are allocated to the UK branch based on the relative value of the assets attributed to the UK branch (on arm’s length principles), and the total assets of the foreign bank (para 24).

Where a UK resident entity trades elsewhere through a foreign permanent establishment, the default position is that (as under the old rules) the equity and liabilities attributable to the foreign PE remain subject to bank levy. However, each entity can elect, under para 15D or 15E, to remove the equity attributable to some or all of their foreign PEs from the bank levy base (in the new rules, these entities are pithily referred to as ‘designated FPE entities’). Where a designated FPE entity election has been made, the equity and liabilities attributable to the foreign PEs (again, determined based on the relative size of the assets attributed to the head office, and to the foreign PE) are removed from the bank levy base.

If a banking group includes a designated FPE entity, alongside other UK resident entities, there are knock-on consequences for liabilities owed by the other UK members of the group to the designated FPE entity. Liabilities which ‘correspond’ to the assets attributed to the relevant foreign PE continue to form part of the bank levy base (para 15R(3) - (5)). Again, banks with liabilities falling within this provision will want to assess if the liabilities qualify for, or can be made to qualify for, netting or another kind of bank levy relief.

On the face of it, banking groups are given great latitude over designated FPE entity elections: they can make and revoke them at any time, and they can apply the election to some foreign PEs but not to others. However, although there is no specific purpose test applicable to these elections (unlike entity-by-entity elections), they remain subject to the TAAR in para 47, which could potentially be used to disregard a designated FPE entity election, where it is made in the expectation that it will reduce the overall bank levy charge. Applying para 47 in this way seems difficult to square with the overall purpose behind the 2021 changes; hopefully, HMRC will, in due course, confirm that it would not expect para 47 to apply here.

3. Changes to the netting rules

As part of the 2021 changes, the bank levy netting rules have been rewritten (into paras 15S -15U), and the government has taken the opportunity to amend several aspects of the rules:

• The netting rules have traditionally applied where a liability owed by the banking group to a counterparty can be net settled, against securities provided to that counterparty as collateral by any member of the banking group which is subject to bank levy, so long as the securities appear on the balance sheet (and have not, for example, been borrowed into the group). This rule has deliberately been relaxed, so that the securities can now come from any member of the banking group, whether or not the entity providing the collateral is itself within the bank levy charge.
• There was, historically, some uncertainty over whether the netting rules required netting to apply on the insolvency of either party to the transaction, or if it was good enough that netting applied only on the insolvency of the bank levy entity. Suppose, for example, that a UK bank borrowed money from a lender outside the UK, and provided on-balance sheet securities as collateral for that debt. You would expect there to be a net settlement of the debt and the collateral on the bank’s insolvency; however, lenders would be much less likely to agree to a net settlement on the lender’s insolvency. On those facts, HMRC ultimately concluded that it was good enough if netting applied only on the bank’s insolvency, and this has now been helpfully written into the legislation (in paras 15S(6)(d) and 15T(5)(c)).
• Less helpfully, the amended rules provide that netting relief does not apply unless, on the occurrence of a netting event, a net settlement is either automatic, or can be triggered at the election of either the solvent or the insolvent counterparty (paras 15S(4) and 15T(4)). That is potentially problematic: netting agreements might well provide that, if a party becomes insolvent, the solvent counterparty alone will be able to decide if netting occurs; and that would seem to mean that bank levy netting relief is unavailable. It is difficult to see a policy rationale for this outcome. There are clearly good reasons why netting relief should not be available if, on a bank’s insolvency, the insolvent bank could unilaterally block net settlement; but that should not mean that the relief is removed where the bank has no power to block net settlement, but the counterparty can choose whether or not to net settle. The whole point of the relief is to encourage banks to put netting arrangements in place, because they reduce the risk to counterparties on a bank insolvency -and that is not affected by whether the solvent counterparty is obliged, or merely entitled, to net settle. Why would the solvent counterparty ever take the option which increases its risk?

In practice, this net settlement concern is unlikely to arise for transactions effected under market standard ISDA, GMRA or GMSLA agreements. Although these market standard documents give the solvent counterparty a discretion to decide whether to terminate transactions with an insolvent bank, that is a discretion to choose whether or not a netting event occurs (which, under the netting rules, can still be determined at the discretion of the solvent party). Once the netting event has occurred, net settlement is then mandatory. However, the GMRA, GMSLA, and the 2002 version of the ISDA master agreement also allow the non-defaulting party to elect to extend the set-off to any other amounts owed by the non-defaulting party to the defaulter under transactions which are not governed by the relevant master agreement, and that second layer of set-off would seemingly fall outside bank levy netting relief.

For transactions agreed under bespoke contracts, banks should review whether the netting arrangements could be disqualified from bank levy relief under paras 15S(4) or 15T(4); and, if there is a concern here, they will want to consider whether the arrangements can be amended to make netting mandatory, or at the option of either party, once a netting event occurs.

4. Regulatory capital changes

The regulatory capital requirements imposed by the Prudential Regulation Authority (PRA), and equivalent non-UK regulators, have changed significantly since bank levy was first introduced. The levy rules have not always kept pace with these changes. For example, prudential regulation has encouraged banks to replace short-term repo funding with long-term unsecured debt. Assuming the repo funding qualified for netting relief, the effect of this shift would be to increase the bank levy charge, even though the bank has (by PRA standards) reduced the risk in its funding strategy.

The 2021 rules have made very few changes to cater for developments in regulatory capital rules over the last ten years. The only relevant change (in paras 15V—15Y, and the Bank Levy (Loss Absorbing Instruments) Regulations, SI 2020/1188) allows UK entities to exclude liabilities arising from 'loss absorbing instruments' which they have issued, but only if the entity has effectively used the proceeds to fund tier one equity or liabilities, or other loss absorbing instruments, issued by a non-UK subsidiary. For these purposes, 'loss absorbing instruments' mean any instruments issued to comply with banks' UK and EU minimum requirements for own funds and eligible liabilities (better known as 'MREL'), or comparable requirements elsewhere.

It seems to me, however, that there is also a good policy case to introduce some additional bank levy relief for MREL, or other loss absorbing debt issued to meet regulatory capital requirements. It is difficult to justify why this should be taxed in the same way as long-term debt which does not have any loss-absorbing features.

Conclusion

The breadth of the bank levy changes in 2021 means that they will affect almost every bank which is subject to the levy. Banks’ group tax departments will already be preparing for these changes, where the most important tasks will include:

• updating their systems, so that they can produce notional consolidated accounts picking up only the UK resident members of the group;
• determining whether to make entity by entity elections, or designated FPE entity elections;
• identifying any intra-group liabilities which will form part of the bank levy base for the first time, because they are owed to non-UK resident subsidiaries, or are owed to a foreign branch of a designated FPE entity;
• for these liabilities, reviewing whether they qualify for bank levy relief on any basis, or can be amended to qualify for bank levy relief (keeping one eye on the TAAR in para 47, and the bank’s obligations under the code); and
• where a bank is claiming netting relief, confirming that the netting arrangements continue to qualify for relief under the new rules (and, where necessary, seeking to amend the netting agreements to bring them in line with the new netting rules).

Finally, as bank levy is charged based on the position on the balance sheet date, the good news (at least for banks with a 31 December year end) is that they have until 31 December 2021 to put any necessary changes in place.

The author thanks Joss Laverack for research assistance with this article.

This article was first published in the 13 November 2020 edition of the Tax Journal.