Excess Spread — where the bonds went, E-MAC and cheese, mechanical trouble
- Owen Sanderson
Where did all the bonds go?
As the LDI crisis hit last year, UK real money funds sold a lot of bonds! It was a pretty exciting few days (to sideline gawkers such as myself) and pretty horrifying for treasurers with deals in market, arrangers looking for exits, or investors marking down their positions.
But with the dust well and truly settled and markets back to functioning, what does the landscape for securitised markets look like?
Dealers, we understand, did step up during the tough times and take bonds down. Catching a falling knife is more attractive if you’re buying bulletproof bonds with low capital charges, getting puked out for technical reasons not credit reasons. One suspects that a few “flow trading” desks were quite happy to do a little bit of stealthy prop trading against an excellent backdrop.
But these bonds didn’t really hang about. In CLOs, the buyers of last resort appear to have the big alternative asset managers. Apollo claimed it did north of €1bn in CLOs during the period. If BWIC volume was, let’s say, €4bn, that’s a pretty good backstop. Ares also said it was a big buyer in this period; it’s not a massive reach to think that GSO, KKR, Carlyle were also active. If you spent 2021 buying single B loans at 350 bps, the argument for triple A CLO at >300 in 2022 is easy to make.
In consumer securitisation, and particularly UK RMBS, the buyers of last resort were not the PE shops but banks, particularly the new institutions with more wiggle room to buy non-conforming or BTL. The clearing bank treasuries like cheap prime RMBS in big size, but we hear Starling, Metro and Chetwood were hoovering up triple-A RMBS.
Listed Metro Bank has the best disclosure, so we can throw around some numbers. As of end 2021, Metro Bank had £38m in RMBS; as of end 2022, it had £1.13bn in RMBS and £168m in ABS. It doesn’t specify when this was bought, but if there was, say, £4bn or so on BWIC after the LDI, £1bn+ into Metro represents a very sizeable chunk.
The vibe from Metro’s management is still very anti-fun though. From the latest investor transcript:
“We will remain disciplined about trading amongst the asset side, not to mention we now have a moment, given the base rate rises we've seen, to actually generate really good returns on very low risk weights by buying gilts and/ or structured products. We don't do any balance sheet optimisation today. We don't do any securitisations. We don't have any forward flow agreements. Again, there are lots of ways for us to use this balance sheet in anger that we're not currently doing.”
That’s not true of the other challenger banks we hear took down big slugs of RMBS. Starling Bank’s latest disclosures run to March 2022, and it had around £11m of RMBS at that point, hardly a sniff of the market.
But Starling has been running around signing forward flows, buying portfolios (it discloses a £514.2m closed book purchase from April last year, probably Masthaven), and, of course, it bought Fleet Mortgages in 2021, in the process getting everyone excited about specialist lender M&A multiples. We will, no doubt, find out more later. Following Starling’s Series D raise, the bank’s CET1 ratio is a stonking 43%, so there’s a lot of room “to use this balance sheet in anger” if the returns are right.
Chetwood Financial is perhaps the most interesting institution. We discussed it a little last week, but it’s very much the bank of the moment.
“If you’re selling [a portfolio] right now, you’ve got to be speaking to Chetwood,” as one banker put it. It has a variety of lending products available, including the “Wave” credit card, and the BetterBorrow unsecured loans, which are sold on to a third party.
But it’s focused on the NIM, and is fairly agnostic about where it comes from — own originations, third parties, forward flows, RMBS bonds, portfolio purchases — if the economics are right, Chetwood will take a look
Funding comes from the fixed term deposit market (substantially cheaper than swaps for much of last year, and it remains so today), which allows it to scale very rapidly — you can bring in hundreds of millions a month if you’ve got a banking licence and FCSC support. Capital comes from Elliott Advisors, which will feed in equity as and when needed, and seems to have been doing so in size.
At the last reporting date, 31 March 2022, Chetwood had a £337m balance sheet, but it’s quite willing to look at deals larger than this figure, and has probably grown quite a lot since then. Companies House discloses £244m in capital base, with the latest shares allocated on January 26… a hell of an equity cushion for a £337m balance Match, so it’s a good bet that things have moved fast since then.
For primary securitised products markets, the question is really how sticky these accounts are.
If you liked triple-A RMBS at >200 bps, do you still like it at 110 bps? Was it a nice trade for a few months, or is securitisation a habit you acquire and struggle to give up? If it’s the latter, the LDI shock may have proved a longer term blessing to securitisation markets.
The UK real money accounts that were forced to sell to meet margin calls are back (the securitisation habit is hard to kick), and new buyers are now there too. I assume Metro is not going to be doing a billion a year every single year, but even at a lower run rate it could still be a very relevant source of demand. We’ve only discussed three accounts here, but a big sterling RMBS distribution is only 25 investors, and a presound is 10. Proportionally, this is big!
In other Elliott news, Elliott-owned West One / Enra has decided to test the quality of the UK RMBS buyer base with Elstree Funding No. 3, announced on Thursday via arranger NatWest Markets. This is the proper front book specialist lender stuff we need, after the wave of euro STS paper a couple of weeks back — £321m of mixed BTL and prime second lien. There is, however, a nicely size reverse enquiry in the senior tranche. The deal update specifies that minimum £140m of the £265m triple-A is available to buy, so new investors or not, it won’t be scraping the barrel to find senior investors.
Haus party
When my colleague Chris Haffenden, who runs our distressed coverage and writes the Friday Workout, starts asking questions, watch out. He’s been heading over to my desk of late to nose around CMBS. There’s a heavy schedule of CRE loan maturities coming up in 2023 and 2024, many of which are securitised. Some of the most complicated situations (Vivion Investments’ Ribbon Finance) have already been refinanced; Frosn 2018 has already been passed to special servicing. So how seriously should we take this?
First off, European CMBS is small. The loan maturities coming up are from the relatively better years 2017-2019, but it only represents a tiny sliver of large cap CRE financing activity,
Second, it most likely skews to the better end of the spectrum. Far from being the pre-2008 machine for originating-to-distribute (with some juicily mispriced swaps on top), CMBS is now a shop window for banks to show off their top drawer financing mandates. These days it mostly finances big portfolios or trophy assets bought by top drawer sponsors. More complex assets tend to be levered using debt from the real estate funds, which may in turn obtain their own loan-on-loan or fund level leverage in private.
The most reassuring part, looking down a list of near term CMBS loan maturities, is the relatively low LTVs in most transactions — here’s a useful summary courtesy of Barclays (from last year, so missing some developments like the Ribbon refi, Frosn default, and Pietra Nera revaluation):
But that’s only reassuring if you actually believe the LTVs! Interest rates have been on quite the journey since the middle of last year, which you’d think would have to read across to cap rates and hence valuations at some point. Let’s take Haus (ELoC 39), a German multi-family transaction, as one illustration, since I happen to be looking at a downgrade notice from Wednesday.
The transaction's loan to value (LTV) ratio has risen to 75.2% from 67.9% at closing based on a reported October 2022 valuation of €423.7 million, just under 10% lower than at closing. This compares to a Moody's LTV ratio of 91% which reflects a Moody's term value for the portfolio of €350.3 million based on a net cash flow of €20.6 million and a cap rate of 5.9%
Rating agencies are supposed to be conservative, so maybe 91% builds in a lot of caution, and there are some important deal specifics here (big refurb plans, high vacancy). But the deal, in common with other CMBS transactions, already has a bunch of triggers at 75% LTV (blocking extension options, for example), and numbers north of 80% could be hard to refi.
The point is, there’s a lot of imprecision in the reported LTV levels, especially if they’re a bit stale, and with some prudent haircutting those “reassuring” LTVs suddenly don’t look so friendly.
Cheesed off with E-MAC
There’s a rather large fight lining up between NatWest Markets and CMIS Investments, a Dutch mortgage service provider which was once part of GMAC but sold to Fortress in 2010.
NatWest is suing for €158m of swap payments relating to various old RMBS deals in the E-MAC series (all the Dutch deals 2004-2008; there was also a German E-MAC series that isn’t included).
Most of these were accumulated as “Subordinated swap amounts”, owing in part to the weird structure in the E-MAC transactions. Instead of balance-guaranteed swaps that match the amortisation of the portfolio, the swaps are basically structured as a series of guessestimates on transaction prepayment rates, provided by CMIS as servicer. The basis between real and estimated swaps payments essentially gets dumped into the swap subordinated amount.
Given that it’s been nearly 20 years since the first of these deals were issued, there’s a fair bit of tracking error built up over time, and the performance of the collateral has not been particularly spectacular, at least in cashflow terms. Many of the mortgages backing this series are interest-only (which is why there are still bonds outstanding after all this time), and struck at low interest rates.
These deals were originally supposed to be refinanced regularly, and included a put option to force the issuer to buy back the notes for cash after a certain period (as an active sponsor would be around to put cash in, buy back the portfolio, resecuritise and so forth).
But, surprise! After 2008 there was no active sponsor, and the put options kept piling up, with no money to actually exercise them. The car crash here has been incredibly slow motion: here’s a Houlihan Lokey presentation from 2012 highlighting the need for something to be done. Houlihan basically wanted to get a portfolio sale done, but needed noteholder approvals, which were not forthcoming.
Failure to exercise the puts jacks up the interest due, but there is also no money available to pay this increased interest, so that just rolls up as well. We’re now in quite weird territory on some of these transactions. To take the oldest one, E-MAC NL 2004-I… there have only been €3.2m of principal losses, over almost 20 years, on an €800m portfolio. But there’s now €43m of unpaid interest, and only €54m of bonds outstanding in the whole deal.
TLDR, there is not a whole lot of interest flowing through the transaction and going to pay the subordinated swap amounts. And transaction structures can end up incredibly wonky if you wind them on for decades without any active deal sponsor.
NatWest is suing CMIS, and if successful, the E-MAC issuers warn “could potentially lead to a situation of financial distress for CMIS Nederland… such situation could potentially affect the Issuer's duly and timely performance of its obligations vis-à-vis the Noteholders and the other Secured Parties.”
There seems to have been a previous dispute between CMIS itself and the E-MAC transactions, with Moody’s noting in 2017 that “CMIS Nederland has taken the position that it has a claim against the Issuers for payments it has made to the swap counterparties, when there were insufficient funds in the transactions to pay the swap subordinated amounts. The notice explains that the payments were made pursuant to an indemnity granted by CMIS Nederland to the swap counterparties, and that the corresponding claims have not been acknowledged by the E-MAC NL RMBS Issuers.”
So the question, it seems, is over who eats these losses — NatWest, CMIS, or the long-suffering E-MAC bondholders.
The missing piece
We’ve touched on the NAIC regulatory proposals for US insurers holding CLOs before, mainly to marvel at the extraordinary amount of relative privilege on display. The fight is basically over proposals to crank up the capital requirements for insurers buying CLO tranches, especially towards the bottom of the capital structure.
It’s not a straightforward regulator vs industry fight like many of the European securitisation battles, but divides the insurance community between trad-ish insurers with asset management arms, and PE/sponsor-backed insurers that are woven into the fabric of the market — look behind a fair few US CLO managers and you’ll see an insurer staking the equity capital needed to fund the issuance plans. The big alternative asset managers have pretty much all launched insurance vehicles to provide permanent capital pools.
My US colleague David Bell (he started off doing European securitisation with me at GlobalCapital) has a very nice piece on the issues.
Signatories from the alternative asset/pro-CLO side include Athene (Apollo/Redding Ridge) Clear Spring Life and Annuity (Guggenheim), Delaware Life, Everlake Life (Blackstone), F&G Annuities (Blackstone partnership), Global Atlantic (KKR), Nassau Financial Group (Nassau Corporate Credit) and Security Benefit (CBAM, pre-Carlyle at least).
It’s darkly amusing to watch from this side of the pond, given that the starting point for US insurers is large and active investment in CLOs and asset-backed products generally, and the starting point for European insurers is almost nothing. There are plenty of insurance-aligned asset managers in the market (the likes of Axa IM, M&G, and MEAG), but rather less of direct insurance investment, especially down the capital structure.
The fight with securitisation regulators in Europe is about making securitisation competitive with just buying whole loans; the fight in the US is more of optimisation and economics.
But the presence or otherwise of US insurance capital is vital in shaping leveraged finance in Europe too.
US insurance capital is heavily immersed in the rise and rise of direct lending, and the success of private credit in taking share from syndicated leveraged loans. Many of the private credit funds are levered through private bank lines, which are effectively private CLOs… and these, in turn, are often syndicated out to the US insurance market. A poke around the “fund ratings” offerings of rating agencies both major and minor will give a partial sense of what’s going on
So you could argue that US insurers are currently behind the buoyancy in US CLO markets, and of global private credit. The missing piece of the puzzle, of course, is European CLOs!
Rude Mechanicals
It’s no secret that new CLO arbitrage has not been great. We’ve rehearsed the general reasons a fair bit, but fundamentally it comes down to LBOs and CLO creation being out of sync. LBOs take a lot longer to put together; CLOs need ratings and legal to be done, but they’re basically pure financial structuring and they switch on and off quickly according to market conditions — so the CLO primary bounce this year has run way ahead of leveraged loan supply.
Given the tough arb times, we’ve been wondering just how much value is leaking out of the market because of, essentially, dumb stuff that ought to be fixable.
There’s no shortage of inefficiencies in the loan market. Settlement is a big one, both in primary and secondary markets. Sometimes there’s a solid M&A-related reason for long settlement times, but even in a liquid leveraged loan, standard LMA docs specify a 10 day settlement. Just this year we’ve had Octura, an industry-backed platform for CLO and leveraged loan trading, executing what it claims is the first fully electronic syndicated loan trade.
This is quite possibly true, though I feel like I’ve read similar articles for years now. Last year we had Symphony and Digital Asset announcing a partnership to “generate efficiency and remove manual processes in the broadly syndicated loans (BSL) market”. Back in 2009 we had well-intentioned plans to “kill the fax” in loan trading. Again in 2015. By 2021 banks were saying that they really meant it this time. I’m being a little unfair; these are slightly different elements of the whole leveraged loan business. But they’re all definitely under the heading of “dumb stuff that ought to be fixable”.
KYC issues also play into the settlement issue — the Loan Market Association’s survey for 2021 saw 53% of respondents say that KYC requirements were the issue affecting settlement times the most. The Association said: “KYC requirements invariably top the list when it comes to settlement delays but interesting to see (an albeit distant) second place for resource issues. We hear an increasing amount of noise generally about resourcing at the operational level and this is obviously a concern, especially given elevated trade volumes. Settlement times are beyond any reasonable KYC issue or indeed resourcing constraint so this will be even more in focus next year.”
KYC for new CLOs can be particularly painful, as KYC processes focus on each vehicle, rather than the manager, so even cookie-cutter repeat transactions need to bash through the bureaucracy.
The other inefficiency that most of the CLO industry is less keen to highlight is the outright cost of each transaction.
Because a given CLO may be reset multiple times in its lifetime, you really need to fold arranger and legal fees into the total cost of gathering CLO AUM — not just at the beginning, but as a drag every two to four years. If we assume a cool million or so each time to make the maths easy, that’s 25 bps on a €400m portfolio, over three years, for 8 bps or so per year. More than half the senior management fee! KopenTech has a mechanism called the “Applicable Margin Reset”, which is supposed to trim these costs… but adoption has been far from comprehensive.
Other capital markets have worked out neat solutions like EMTN documents, shelf registrations to cut friction and costs at issuance (these typically have the downside, or arguably benefit, of standardisation). Securitisation laws like those in Luxembourg and France allow Compartment issues from a single master SPV. CLOs are by some way the most standardised product in securitised products; where’s the streamlining?