#### Need to know

• At least three large European banks are exploring synthetic securitisations as a way of reducing spikes in loan-loss provisioning under IFRS 9.
• The trades work by transferring the risk of a rise in provisioning for a book of loans to investors.
• Unlike traditional credit risk transfer securitisations, the trigger point for IFRS 9 securitisations is a rise in provisioning after a loan has become impaired, rather than a credit event.
• The amount of capital relief banks can obtain hinges on the composition of the loans going into special-purpose vehicles and the timing of payments by the SPV to the bank.
• Auditors warn that regulators might view transactions sceptically, adding banks will need to argue that they are using freed-up capital to diversify their portfolios.
• Some believe that securitisations for IFRS 9 purposes should be exempt from regulatory requirements that accompany significant risk transfer trades.

Dealers are putting a new twist on old-school risk transfer trades in a bid to reduce the sharp capital swings they expect to encounter under new accounting standards.

Several large European banks are understood to be in the advanced stages of structuring bespoke synthetic securitisations with the intention of smoothing expected spikes in loan-loss provisioning following the industry’s transition to IFRS 9, the new accounting regime set to enter into force in most jurisdictions on January 1, 2018.

In doing so, banks are breaking fresh ground: “When people have done securitisations up until now for regulatory capital purposes, they’ve got the capital rule book out and followed the guidelines,” says Adrian Docherty, global head of financial institutions advisory in the fixed-income division at BNP Paribas. “The tranching and terms and conditions currently are built around those requirements. When you design a transaction for a different purpose, you’ve got to throw that rule book away and get out the accounting rule book.”

The trades are among a number of innovations being explored by dealers prepared to go to great lengths to avoid the capital-sapping effects of the new framework – from simple loan book sales to more complex agreements to sell off assets deemed impaired, before subsequently repurchasing them (see box: The asset swap alternative).

Yet, like the market for any new product, it might take a while for banks to stoke interest in IFRS 9 synthetic securitisations, observers warn. Investors may struggle to get comfortable with the risk/reward dynamics of an untested instrument; banks may end up having to offer up too much yield to make the risk transfer economically viable. Similarly, regulators and auditors may baulk at the idea of banks using complex risk transfer mechanisms to offload loan-loss provisions that they are obliged to reserve for.

Navigating these hurdles may be worth it, though, considering what’s at stake. The International Accounting Standards Board estimates loan-loss provisions will increase by 44% on average when the new rules take effect, which would translate into a drop in banks’ Common Equity Tier 1 ratio of 10–20%, according to some estimates (see box: Capital punishment).

IFRS 9 requires that firms recognise expected losses for a loan over its entire lifetime once a “significant deterioration” in credit quality is determined. These determinations have to be recalibrated each quarter based on economic forecasts using the banks’ own credit models.

Stage 1 assets, those that have not undergone credit deterioration since being placed on the books, need only reserve for 12 months of expected losses. Stage 2 assets, where a “significant deterioration” has occurred, must have provisions assigned to cover lifetime expected losses. Stage 3 assets, those that are actually impaired, need to be assigned lifetime expected losses and have their expected interest payments docked, too.

“Most people typically focus on the impact on January 1, 2018 of switching over,” says Amnon Levy, head of portfolio and balance sheet research at Moody’s Analytics. “There is a different dynamic coming from the forward-looking nature of the allowance, and the impact it has on the volatility of provisions.”

### Volatility unavoidable

The realisation that volatility of loan-loss provisions, and consequently capital ratios, is a feature, not a bug, of the accounting regime, is driving some banks to explore structured credit solutions to ameliorate the pain.

“The new accounting rule may make portfolios, even though they’re very solid loans, very expensive to keep on the books without transferring some of the credit risk, because firms will have to start providing for future credit losses rather than incurred credit losses,” says Assia Damianova, special counsel in the capital markets group at law firm Cadwalader.

The capital impacts will be especially acute for banks with concentrated credit portfolios, where the majority of the exposure is to a handful of names. Moody’s Analytics simulated the IFRS 9 loss allowance for a well-diversified portfolio and a portfolio with highly concentrated exposure to the oil industry. The loss allowance for the well-diversified portfolio at the one-year horizon never exceeded 10%, while the loss allowance for the concentrated portfolio at the one-year horizon exceeded 10% in 3.7% of trials.

Banks with concentrated portfolios are therefore most likely to use securitisations to mitigate IFRS 9’s impacts, says Burcu Guner, team leader for stress testing and IFRS 9 in Europe at Moody’s Analytics.

“Institutions that have relatively long-dated assets in concentrated portfolios are actively exploring this. Beyond the short-term gains from structuring these deals, institutions are looking at how they can use these structures to further diversify their portfolios.”

Market observers say that, while the structures vary depending on the bank, most resemble variants of more familiar credit risk transfer trades – a mechanism used by dealers looking to offload a portion of their credit exposures while retaining the assets on their own balance sheets.

###### Our regulator would look dimly on our engaging in transactions simply to avoid some sort of unwanted accounting effect

Structured credit executive at a large European bank

In a traditional structure, a bank strikes a credit default swap or financial guarantee with a special-purpose vehicle (SPV), in which the latter promises the former reimbursement if it incurs a contractually defined credit loss. The SPV funds itself by issuing credit-linked notes to investors. Following a loss, a cash settlement amount is paid by the SPV to the bank, and the investors’ credit-linked notes are written down accordingly.

Strict conditions govern the use of credit risk transfer trades, however. For example, in Europe under the Capital Requirements Regulation and related guidelines governing “significant risk transfer” published by the European Banking Authority, synthetic securitisations must be structured so that the risk being transferred is “significant”, and reflects the economic value of the underlying portfolio, and is therefore commensurate with the capital relief being sought.

While most of these securitisations are private, a few have been made public. Nordea, for example, conducted an €8.4 billion ($9.9 billion) corporate and SME loan securitisation in 2016 for the purpose of reducing risk-weighted assets on its balance sheet. Synthetic securitisations undertaken for IFRS 9 relief will differ in several crucial ways, however. The major difference is how payments are activated: in a classic synthetic securitisation, the trigger for payment is a credit event, such as an obligor’s failure to pay through insolvency or a restructuring of debt. As the bank begins to experience losses, the SPV will make payments to the bank, and to the extent that the bank is able to recover any losses, the bank pays back the SPV. The trigger for an IFRS 9 securitisation, however, would not be a credit event, but an increase in lifetime loan-loss provisions. Provisions increase when an asset moves from stage 1 to stage 2, at which point lifetime loan-loss provisions must be made. A securitisation would have to include bespoke triggers that cleave to the sponsor bank’s own stage allocation thresholds. Versus traditional credit risk transfer trades, “it’s a case of banks bringing that trigger point forward in time to the point where you have to start provisioning against a loan under IFRS 9, rather than the point where you experience a loss on it,” says a derivatives lawyer in London “We’re talking about two different things in the sense that you’ve got provisioning and you’ve got regulatory capital, and the two are flip sides of the same coin. These are trying to get at the same thing, but at a different point in time.” ### A question of timing Crafting the triggers is one challenge; ensuring the resulting payments effectively cover the sponsor bank’s loan-loss increases is another. Lawyers say much will depend on the view accountants take of the timing of payments provided by the SPV under the financial guarantee: can benefits be recognised at the point at which payment under the guarantee is triggered – when loss provisions actually increase – or can they only be recognised after the loan has gone into default and fully worked out? The answer will determine whether IFRS 9 securitisations are worth banks’ trouble. “The accounting literature talks about the point when you are virtually certain that you will be able to collect under that guarantee,” says a structured credit executive at a large European bank. “If you’re not able to recognise the benefit of the financial guarantee until you finally work out the loan and determine what the final loss is, then you have a timing mismatch that may not give you that benefit.” The executive adds that the accounting profession is “coalescing around the view that you should be able to recognise the benefit of that guarantee at the point when a loan becomes problematic and moves to stage 2, which would trigger the IFRS 9 loss provisions”. Agreeing with investors on these points will likely be essential to the structures’ viability; observers point out that potential buyers might baulk at the fact that the payment trigger under the SPV may depend on provisioning that is dictated in part by banks’ own models, and also subject to regulatory scrutiny. “The banks will say, ‘we use models that our peers use, they’re stress-tested, they’re very robust, regulators have signed off on them’,” says Azad Ali, partner in the financial services regulatory practice at law firm Fieldfisher. “But regulators are going to be more concerned about things not being characterised as impaired when they should have been than things being characterised as impaired when they need not have been. Those would be the natural concerns of investors, which will be reflected in pricing.” An additional round of negotiations will have to take place over the securitisations’ tranching. With traditional structures, banks will often retain the junior, or first-loss, tranche of risk transfer securitisations and sell the mezzanine, or second-loss, tranche to outside investors. “The risk in a tranched structure is greatest in the second-loss tranche,” says BNPP’s Docherty. “An investor is so certain that they’re going to lose money in the first tranche, they’re not [likely to risk] investing in it. Why would anyone seek protection on something that’s almost certainly going to happen?” Others argue this does not always hold true. Some banks say they will sell the first-loss tranche where the capital relief available offsets any additional premiums they have to pay to incentivise investors to swallow the first loss. ###### Regulators are going to be more concerned about things not being characterised as impaired when they should have been than things being characterised as impaired when they need not have been Azad Ali, Fieldfisher Each tranche in a securitisation is demarcated by the points at which the holder of the tranche begins to absorb losses (attachment) and stops absorbing losses (detachment). The first-loss tranche by definition has a zero attachment. “We do that all the time – deals with zero attachment to 8% detachment,” says the structured credit executive. “Even though embedded in that tranche are the expected losses of that transaction, for which the investor might therefore charge us more for protection, we free up more regulatory capital by selling the first-loss tranche as opposed to keeping it and just selling the mezzanine.” A bank that chooses to retain the first-loss tranche in a synthetic securitisation will not get any relief from the impact of IFRS 9 unless and until losses exceed the level of that first-loss tranche. That is, a bank can’t take a provision benefit on a loan until any retained first-loss tranche is consumed. Accountants are advising banks that in such situations, the deal would need to be funded by a financial guarantee rather than a credit default swap (CDS) in order to qualify for any IFRS 9 benefits. “You have to structure the deal as a financial guarantee, not as a CDS,” says the structured credit executive. “If you do something with a CDS that’s mark to market, that basically defeats the purpose. If you sold the first-loss tranche, then you would be able to get that benefit from the beginning.” ### Balancing act Finding a favourable balance between making savings on loan-loss provisions and making payments to an SPV for this protection will also factor heavily in banks’ decisions as to whether to structure IFRS 9 securitisations. “The price to transfer this [IFRS 9] risk is very high, because of the limited universe of investors willing to buy the first-loss position,” says Salim Nathoo, head of the securitisation practice at law firm Allen & Overy. “Because they demand a high premium, banks are looking at the trade-off between the cost of raising equity versus the cost of buying protection from those sellers. At the moment, they see buying protection as being more effective, but that dynamic might change.” While in theory a bank could reduce loan-loss volatility to zero by securitising its entire loan portfolio, the cost would be prohibitive: “The problem with those types of securitisations is they tend to be quite expensive,” says Docherty. “It could be that it’s not efficient to do the transaction because of the cost. There will be a subset of portfolios where it makes sense and a subset where it doesn’t. It’s anyone’s guess what that proportion would be.” For example, it might make little sense for a bank to securitise a mortgage portfolio if the mortgages carry very little risk of deteriorating, and therefore moving between IFRS 9 stages. The cost of freeing up each euro or pound of risk-weighted asset here would be much higher because the bank has a much lower risk asset to start with. On the flipside, a bank with a pool of stage 2 loans – which will comprise up to 15% of a bank’s total loans on average, according to Moody’s Levy – can immunise themselves from the provisioning volatility they would inflict from degrading to stage 3 or recovering to stage 1 by wrapping them all in a synthetic securitisation. Though IFRS 9 securitisations face unique structuring challenges, they may also gain an easier passage through the web of regulations that traditionally complicate the launch of such products. For instance, some have speculated that as these securitisations are created for accounting reasons rather than regulatory capital ones, they should be exempted from significant risk transfer requirements. “What we’re looking at is a structure focused on achieving accounting impact and not on achieving a regulatory RWA impact,” says Docherty. “Demonstrating to the regulator that significant risk has been transferred to the third-party investor is no longer relevant. You can de-risk the expected losses but you don’t necessarily have to transfer much unexpected loss at all, and that can be a good thing.” Ali at Fieldfisher agrees: “In cases where an increase in provisioning is subject to a payout by the protection provider, the bank has a funded source for that increase. So I don’t see how that would jeopardise significant risk transfer.” Whether regulators and auditors will take this viewpoint remains to be seen. “A key caveat to structured deals is you cannot estimate the perception of the regulator and the auditor,” says Moody’s Guner. “The market will have to start thinking about how the regulators and auditors will react to such deals. Typically, these books are concentrated. If they can argue a diversification point to their regulators and auditors, there is definitely merit.” Banks are understandably anxious to avoid even the appearance of skirting the rules: “Our regulator would look dimly on our engaging in transactions simply to avoid some sort of unwanted accounting effect,” says a structured credit executive at a large European bank. “So we don’t have as our primary objective reducing IFRS 9 volatility. Our objective is to engage in a ‘significant risk transfer’ transaction to manage our capital base better, and to allow us to recycle the capital that we free up into new lending to help us address concentrations of risk.” ### The asset swap alternative Alongside synthetic securitisations, banks are considering other means of lowering their IFRS 9 loan-loss hit, such as diversifying highly concentrated portfolios. “Before everyone thinks about securitisation as a way forward, there are a number of things the credit portfolio management functions could start doing by diversifying their credit portfolios away from these sectors, or raising their credit quality buckets that are heavily concentrated or highly sensitive to stage migrations,” says Burcu Guner at Moody’s Analytics. A structured credit executive at a large European bank says his company already employs a variety of credit mitigation techniques, such as buying insurance, selling participation, and repos. “There’s a whole range of credit mitigation tools, all of which may have an IFRS 9 benefit.” One alternative to securitisation is simply to sell off impaired assets and then repurchase them, effectively transforming them from stage 2 assets to stage 1. Amnon Levy at Moody’s Analytics notes that, once a bank purchases a stage 2 asset, on the books it’s recognised as a stage 1 asset, and the amount of allowance will be set accordingly. For example, if it’s trading at 95 cents on the dollar, it will be listed at 95 cents on the books, and the bank will only need to use a one-year allowance even though the previous bank had to use a lifetime allowance. Auditors and regulators, however, may view such transactions as going against the grain of IFRS 9, and a form of accounting arbitrage. “We’ve seen organisations starting to explore this type of accounting arbitrage,” says Levy. “There is allowance relief associated with that. It’s an example of a manipulation that might follow the letter of the rule, but is not an intended consequence of the rule.” ### Capital punishment IFRS 9 obliges banks to reassess their required provisions using a new expected credit loss methodology, meaning these numbers can grow and shrink dramatically between reporting periods. The new framework will have an immediate impact on banks’ capital ratios when it takes effect on January 1. HSBC estimates a day one impact in loan-loss allowances of$2 billion without transitional arrangements, equivalent to a drop of 15 basis points in its CET1 capital ratio. Barclays estimates a day one impact of \$2.4–2.6 billion without transitional arrangements, equivalent to a 40bp drop in CET1 equity.

The actual IFRS 9 impact on day one could change as a result of ongoing work on models and data, as well as changes in balance sheet position, market conditions and forward-looking economic assumptions.

The volatility in quarterly loss estimates is largely a function of the IFRS 9 model requirements. Most expected credit loss models are essentially modified versions of existing regulatory capital models, which banks have been running and fine-tuning for decades. The Basel risk models are based on a ‘through-the-cycle’ approach that measures average losses over rolling 12-month periods, while IFRS 9 relies on ‘point-in-time’ estimates of loan losses based on the prevailing credit conditions at each quarter-end.

#### 7 days in 60 seconds

###### Basel III, bond market liquidity and the Risk awards

The week on Risk.net, December 2–8, 2017