No but actually, what is it in practice?

Structurally, it’s a securitisation, but no assets change hands. It’s a credit default swap, but it’s completely custom. It’s an insurance policy, but not like any insurance policy you’d recognise. It might just be a financial guarantee. It might have an SPV, it might be a customised bank bond, it might be collateralised, or not.

C’mon, answer the question. 

Let’s start with how one is made

Step 1: Find a portfolio of loans on bank balance sheet. Can be anything; popular asset classes are corporate lending in funded or RCF format, but subscription lines, mortgage warehouses, CRE lending, agricultural loans, project finance, trade finance, leveraged loans, mid-market lending, mortgages, derivatives counterparty exposures, wealth management loans have all been done.

Step 2: Draw a financial box around them, using whichever contractual tool suits you best.

Step 3: Write a contract that says “pay out any time one of these loans defaults/enters restructuring, up to a maximum value of X%”. X varies a lot depending on the asset class; 5-6% of the total notional in the box might be usual in investment grade, leveraged finance would be more like 20%.

Step 4: Find someone to take the other side of that contract and face the bank (sell protection). 

Step 5: Figure out a price that works for both sides

Why bother?

A generic fondness for hedging aside, the main point of it is to free up the bank’s regulatory capital. 

Structure the deal right and the bank’s regulators will conclude, rightly, that most of the risk has been transferred to the hedge fund/private credit/supra/insurer and what’s left is a much less risky senior position.

This is the Significant Risk Transfer in the title — it’s a regulatory term of art saying essentially “yes, you have transferred enough risk to slash the capital on this portfolio”.

‘Slash the capital’ sounds like a thing banks like to do — what does that mean?

What actually happens, through a bank capital-regulatory lens, is that the risk weight of a portfolio of whole loans is replaced by the risk weight of a senior securitisation tranche (the one the bank is left holding). The loans don’t move, but hedging changes the risk weight.

Risk weight?

Risk weights are the main route through which banks allocate their regulatory capital. Think of them like a lens that sits over the accounting balance sheet of a bank. A risk weight of 100% on a corporate loan means that it is as large, in risk terms, as its accounting value. A risk weight of 20%, on, say, an owner-occupied mortgage, means it’s only 1/5 the size of its accounting value.

For a given quantity of capital, therefore, a bank could write 5x as many mortgages as corporate loans.

You can also do a back-of-the-envelope calculation to arrive at some solid cash figures. If a bank with 15% CET1 ratio (all equity) holds a corporate loan with 100% risk weight, then each dollar of the loan is supported by 15 cents of equity. If a bank could convince its supervisor that these loans in particular were much less risky that normal, and only needed to be risk-weighted at 66%, the bank could hold 10 cents of equity against each dollar of the loan. And thus is economic value created.

(The above is wildly simplified but it will do as a approximation)

Simple is good, continue

So to continue wildly simplifying — let’s do a worked example. Assume a €1bn portfolio of corporate lending, risk weighted at 100%, owned by a bank with 15% CET1 as the initial position. That gets to a capital requirement of €150m (€1bn x 1 x 0.15)

Now assume you strike a contract with a hedge fund to cover the first 10% of defaults. The bank is left with 90% of the exposure, but much less than 90% of the risk, so the remaining piece can be given a 10% risk weight instead (this is the risk weight floor for senior securitisation positions under the Basel III SEC-SA approach; more of this later).

So the bank now has a €900m securitisation piece, attracting a 10% risk weight, for a €13.5m capital requirement — €900m x 10% x 15%.

Sounds great, back up the truck

Well, this doesn’t come for free.

Let’s start with the headline price. In the model example above, the bank has bought €100m of protection. Risk transfer funds aren’t charitable institutions, and usually target double-digit spreads, so let’s assume this deal cost Euribor+1,200bps, and round that up to 16%.

You can think of it at a portfolio level though. These deals can reference any kind of assets; all you do is draw a box around them on a bank’s balance sheet.

If the €1bn corporate book was paying 4%, you’d be getting paid €40m a year on the initial €150m of regulatory capital. Take off €16m to pay for the protection and you’re getting €24m on €13.5m of regulatory capital; brilliant!

That said, €40m is still a bigger number than €24m…banks do want to optimise their regulatory capital, but many of them are still over-capitalised compared to where management would be comfortable running the business. They may not have attractive alternative opportunities to deploy regulatory capital into, they may be constrained by other factors; liquidity, leverage ratio or similar. And what they most want are earnings to pay out, not incremental capital.

That 16% figure, too, may not be such a good deal. It’s below a lot of banks’ implied cost of equity, but it’s well above what many banks are actually paying on their already-existing equity. If it was easy to do a rights issue in small increments, would a bank actually want to go and raise at 16%? Crudely, these deals give a benefit to the capital base but a hit to the P&L. 

These numbers all seem suspiciously easy and attractive…

That’s absolutely right. We’ve walked through a toy deal struck on fairly attractive terms, and there are plenty of other variables to consider.

For another, slightly more complicated worked example, have a look at the European Systemic Risk Board’s October 2023 paper on the topic — the below shows a saving of 63m in capital on a 2bn portfolio.

One complicated aspect, touched on by the above, is how to size the protection, which itself depends on the riskiness of the assets in question. The smaller the slice of protection, the smaller the cost. Banks don’t want to overhedge their books, but they may well be happy to hold positions that, in a normal securitisation, would be single-A rated, rather than just the nearly-risk-free triple-A equivalent. 

They may find it efficient to buy a small slice of protection on the most risky portion from a hedge fund (say, the first 5% of losses) and then buy insurance on losses from 5-9%. The asset class matters, because this determines how fat the slice of protection needs to be.

Sometimes, the bank will hold the 0-1% slice itself — this is the “expected loss” in the portfolio — and buy protection on the unexpected losses above that.

These structures are determined by the fearsomely complex securitisation capital requirements. Big banks with the requisite authorisations can use the Internal Ratings Based approach to calculate the riskiness of their own assets.

Applied to securitisation, this becomes SEC-IRBA, for Securitisation-Internal Ratings Based Approach. If this doesn’t work, they can use SEC-ERBA (Securitisation External Ratings Based Approach) and if that doesn’t work, SEC-SA (Standardised Approach).

These formulae spit out different figures for the risk weights on different securitisation tranches, which are also sensitive to the maturity or expected average life of these tranches. Basically, there’s a lot of structuring work in figuring out how much protection to buy, and optimising cost and structure.

This stuff all looks at the capital treatment post-transaction, but the status quo before transaction is more complicated than the dummy model above. 

Most regulatory regimes for big banks allow them to risk weight their own assets, and they’ve usually tried to squeeze these down as much as possible before even thinking about securitisation. So figuring the right portfolios for this kind of deal means looking carefully at their existing IRB risk weights, considering the suitability for securitisation, which in turn means looking at the diversity and correlation of the assets, their lumpiness or granularity, their historical performance, their existing PD and LGD……

If that wasn’t enough, these regimes get another couple of overlays — a special European subsidy called STS, which cuts down securitisation capital requirements, a special factor called the P factor which raises them up again, and coming down the tracks, something called the Output Floor, which limits the benefits of banks using their own internal models to come up with capital requirements at all.

I think I’ve had enough of regulators for now, please stop

No can do.

In most jurisdictions, SRT transactions need some kind of regulatory approval — the Significant Risk Transfer we talked about needs supervisory blessing. If banks are going to cut their capital requirements, their supervisors want to look over their shoulder.

This sets up a tension. Supervisors want to see conservatism — lots of protection bought, which will work well and bear risk effectively. They want to see lots of risk transferred.

The actual boilerplate says banks must transfer 50% of risk weighted assets in a mezzanine tranche of a three tranche deal, or at least 80% of the first loss in a two tranche deal. This risk must stay transferred for the lifetime of the deal, and must also meet a “commensurate risk transfer” test; the capital relief is smaller than the losses transferred.

Investors in these deals, though, want to get paid for assuming as little risk as possible, and there is a longstanding suspicion that the structuring banks are on their side. They’re not anticipating their banks falling over; they just want the capital benefits as cheap as possible.

So structuring banks can tweak trigger events, positively select portfolios, adjust amortisation, callability and even include “synthetic excess spread”. 

In a normal loan portfolio, the interest coming off the performing assets will cover some of the losses from defaulted assets; this just recreates the same thing artificially. 

The deals are heavily negotiated on both sides; structuring and syndication (or bilateral placement) is a long and difficult process. Structurers also need familiarity with how regulators think, and how they will view transactions. Here’s the ECB’s document on requirements for banks; they’re supposed to engage at least three months before executing a deal.

This sounds like a lot of work

Yes, and it’s fairly specialist too. There’s a high barrier to entry for both issuers and investors. 

The first deal, for a bank, is always the hardest — big players in the market tend to be global systematically important banks with their own securitisation teams in-house, with large and sophisticated treasury and capital management teams staffed by former structuring people. Think of a big European bank with a markets business and you’ll probably find it has a big SRT programme.

….and how exactly would I find that?

By knowing where to look. Banks subject to the Basel III regime publish a “Pillar 3” report with all kinds of details in. Go there and search for SRT securitisations.

Here’s Santander, with €36bn of synthetic SRT transactions

BNP Paribas, with €49.9bn

and Barclays with £40bn

So north of €120bn across three banks only?

Well, these are some of the biggest. But yes, it’s a big market. Bear in mind that these figures are the portfolio notional, not the size of protection bought — but this could still represent €10bn in placed notes.

Who is buying all of it?

Well, it’s quite specialist. Some funds only do SRT. Christofferson Robb practically invented the market, and still has a huge market share, though it’s diversified into NPLs and a few other securitisation-adjacent businesses. Chorus Capital, ArrowMark and Newmarket Capital have also made a specialism of this market. 

A few others are well known securitisation shops. Think Chenavari, Polus (fka Cairn), Orchard Global, Axa IM, SPF, CQS, 400 Capital for investors that might do CLO tranches, equity, ABS, specialist lending, and related activities, as well as SRT. Magnetar Capital offers another example. Pemberton and M&G have started funds relatively recently, while Elliott has dabbled. DE Shaw has a fund, CarVal, Angelo Gordon and so on.

The largest, and perhaps most unusual, is Dutch pension specialist PGGM, which manages the healthcare sector pensions among others. For years it was the largest account in the space — here’s a few tombstones (and a rather more sophisticated explanation of the market).

Other players include a few specialist insurers (RenaissanceRe is the largest), some public sector bodies such as the European Investment Fund, British Business Bank, and International Finance Corporation (the World Bank group’s private sector body). These all tend to sell protection without putting up collateral, relying instead on their credit strength. 

The protection slice is then risk-weighted as a triple-A supranational, instead of a risky slice of a pool of loans.

More recently, the big beasts of alternative asset management have become more active in the market — private credit heads at KKR and Ares have named SRTs as a top trade, while Blackstone has also been mentioned. Apollohas been active in past, and will surely return.

Ok, I believe the hype — what am I actually buying again?

As we said, it can come in a variety of wrappers and structures. Part of it depends on who you are!

Bank regulators (and banks) want the protection to be bombproof, and always-available. When default rates are spiking and banks really need capital, hedge funds also tend to start failing. So these deals tend to be fully collateralised, usually with cash, very occasionally with bespoke matched maturity notes from a supranational.

So you might have a special purpose vehicle which issues multiple tranches of credit-linked notes, referencing different tranches of a loan portfolio on a bank balance sheet. The bank subscribes for the senior notes, and the fund subscribes for the junior or first loss, purchasing a note with the proceeds held on escrow in the SPV.

Here’s the ESRB again:

So you have two layers of credit protection contracts, one between SPV and bank, and one between bank and fund. This is fairly burdensome to structure, thanks to the double layer, the extensive lawyering involved, and the costs of SPV administration.

Considerably cheaper and more common is a credit-linked note issued by a bank directly. 

Most big investment banks with structuring desks have a CLN shelf, which they can use for repackaging all sorts of investments. Want your French government bond exposure in yen format? Get a repack. Want a fixed coupon bond with exposure to the upside if Brent fixes above 90 at year-end AND the S&P is up more 10% in Q4? It can be done!

So these can also be used for issuing an SRT transaction. The credit protection comes from the note having a variable principal amount linked to losses on the underlying portfolio. So you buy a customised bank bond for $100m, the bank holds onto the money, and pays back $100m minus any losses incurred under the protection contract.

This wrapper tends to come with a more attractive coupon, because the structuring bank passes the lower transaction costs through to the investor.

If you’re an insurer, it works a little differently. Insurers can in theory invest from either side of their balance sheet; they could buy SRT as an asset for their investment book, or use their liability side to insure a portfolio of loans. In practice, insurance capital rules, a nightmare of complexity we will not discuss today, mean that they’re pretty much exclusively doing these deals as liabilities.

That means writing a bespoke insurance contract to pay out any credit losses between X and Y% of a portfolio. Insurance rules tend to favour taking less credit risk than the funds, so insurers will often prefer to cover the “mezzanine” position. A hedge fund might cover the losses from 0.5-5%, with an insurer covering losses of 5-8%.

The capital treatment also works a little differently — the credit rating of the insurer substitutes for the credit risk of the securitisation position, which is more likely to be in the single-A range. So the bank doesn’t get rid of the risk entirely, as will a fully cash-collateralised position. It just cuts the risk position.

Not sure I like bank bonds, this year of all years

That would be a very reasonable concern. SRT tranches are riskier than most bank debt, but part of their attraction is that they should be relatively pure exposures to a specific portfolio, not to generic bank credit or the capricious whims of banking supervisors.

CLNs generally have some kind of ratings-based trigger in them, but this isn’t much use; banks, when they fail, fail much too fast for these to be much use. You can hedge out the risk with an equity short or CDS position (both imperfect hedges), but this adds to the cost. Investors that bought a CLN because the economics were better would be giving up that benefit and more so that they could put on a hedge.

Ultimately, the protection for investors in bank debt comes partly from the bank not blowing itself up, and partly from the decisions of regulators when it does. Lawsuits aside, Credit Suisse’s AT1 holders were toasted by the Swiss authorities, not as an automatic effect of the instrument itself. Behaviour matters.

There is precedent. There are now six banks with outstanding SRT transactions which have failed in some way — BES, Banco Popular Espanol, Co-Op Bank, SNS Reaal, Getin Noble and this year, Credit Suisse.

In none of these cases have bank supervisors decided to bring the SRT transactions within the “bail-in” writedown of bank debt. These institutions followed different rescue paths, with good/bad bank splits, creditor takeovers, and forced acquisitions, with different levels of write-down and haircut of different debt-like instruments.

But it would be a bad look for supervisors to bail-in SRTs. It would be instantly counterproductive, for one thing; the bank would need more regulatory capital for its rescue if the SRT was impaired. It would also be a bad precedent. Supervisors sign off SRTs as loan portfolio protection. Threatening the market by deciding to impair them would be giving with one hand and taking with another.

All this bad stuff, and it wouldn’t even raise much capital; the size of the placed tranches is small compared to the notionals protected through the market. 

In fact, structured notes of all kinds, not just SRTs, have generally been OK in bank rescues/bail-ins. The point of a prompt bank rescue is to stop financial contagion in its tracks; making the bank in question an unreliable counterparty for all manners of derivatives held all over the buyside would spread panic far and wide.

“They probably won’t do it” isn’t super-reassuring, and participants in the spicy end of credit markets try not to rely on this. But they really, really, probably won’t.

What’s better than a good investment opportunity? A levered good investment opportunity….

Sure, you can lever these — the advance rate won’t be massive, and not many banks are active, but yep, you can use SRT tranches as collateral for a repo. Least this swirl of derivatives, guarantees, SPVs appear terrifying and leverage imprudent, consider the following — arguably, it’s a risk like a piece of bank equity. First loss in a core loan portfolio. You can YOLO options to your heart’s content on any listed, 20x levered bank stock; what’s wrong with a 50% advance rate on a 10x levered first loss? 

Why’s everyone talking about this right now?

Well, we will cop to some excitable headlines, but we mean it.

The big thing now is that the US market is on the brink of opening up. Unlike most capital markets, which usually start in the US well before EMEA follows, SRT has pretty much been a European thing, periodically enlivened by a Canadian issuer, or a far-flung subsidiary of Santander or Standard Chartered hitting the market. 

Big US banks have done selective deals, but have been more active in referencing their non-US assets. Tax and structuring issues have been super-painful for the US; the economics of the structures don’t work as well under US regulation, and the format is more constrained. Threading the needle between derivatives regulation, securitisation regulation and insurance regulation across different states is a real challenge.

But obviously, the US is enormous. Even a tiny proportion of the US money center bank corporate lending books coming out in SRT would mean a huge increase in market volumes. Recently, the Federal Reserve has issued a new FAQ, making it clear that the (much simpler and easier) CLN structure can be recognised as a synthetic securitization, if approved by the regulator. Two banks so far (Morgan Stanley and US Bank) have been given the formal regulatory nod, but many more are expected.

Big US banks becoming programmatic issuers in the market has also sparked huge interest in the market from LPs and regular asset management firms, rather than the specialists which have dominated the market so far, potentially broadening the pools of capital involved and regularising the market.

If this opening up lasts, it also stands as a symbol of the megatrends which have defined finance since 2008. Banks have become senior financiers not risk-takers, and non-banks of all kinds, funded through private credit and the alternative asset management industry, have become the origination channel of choice. Whether it’s commercial real estate, sponsored corporate lending, aircraft, shipping, credit cards, consumer lending or anything else, it’s usually a fund-backed originator doing the lending, getting senior finance from a bank.

The final frontier is those assets which can’t or won’t come out in other formats, and here SRT is transformational. It separates asset spreads from the cost of protection (so low-returning assets like corporate RCFs can be included) and it allows the privacy essential to other asset classes (warehouse lines, NAV lines).

The US IG corporate lending book couldn’t be more core to the likes of JP Morgan, Citi and Bank of America; if the capital provider of choice (via SRT) is Blackstone, KKR or Ares rather than their shareholders, that’s a huge change.

It’s also a massive thing for the undercapitalised US regional banking system. Doing a rights issue against a backdrop of negative sentiment and scepticism is a terrible idea, but doing an SRT referencing specific assets can inject capital into the system without having to worry about equity sentiment.