The Wall Street Journal has a story today on a new type of credit market transaction described as “nonlinear finance”. That label alone should send off alarms, since one assumes it is truth in advertising, a rare commodity in Big Finance. “Nonlinear” says that under certain scenarios, the price of the instrument goes “nonlinear,” as in behaves in a radically different, ungraceful manner or can be expected to have its price gap out if particular conditions are met. That would also suggest the instrument would be hard to hedge.
One of the reasons I can’t be as specific as I’d like, as Wall Street Journal readers pointed out, is the actual article is thin on details. However, that isn’t as surprising as it should seem. These trades sound a lot like the old CDOs, the ones that blew up so spectacularly in the crisis. Technically, those were asset-backed securities, or ABS CDOs.1 If you were reading the financial press before the crisis, the only reporter who recognized the importance and riskiness of CDOs was the Financial Times’ Gillian Tett, who doggedly kept after them and managed to ferret out critical bits of information. CDOs also became large enough as a product that there was some aggregate data, but it wasn’t terribly reliable (one huge problem was the potential for double-counting).
And it also makes sense that financiers would find a new bottle for the old CDO wine, since any investor would probably have a lot of ‘splaining to do if he were to invest in something that was sold as a CDO, even if that was a straight up description.
First to the critical bits of the Journal story, then more discussion as to how worried to be about this development. From the Journal:
Putting together deals in what some dub “nonlinear finance” is a growth business for investment banks’ big bond-trading arms and is helping clear unwanted assets off some balance sheets. However, such private deals, which aren’t publicly traded and don’t have public credit ratings, are a challenge for regulators keeping track of the growth of shadow banking and understanding whether such activity is driven by regulation or its avoidance.
The business isn’t new, but it is heating up as banks hire specialists and commit balance-sheet capacity to feed investor demand. Goldman Sachs Group GS -1.51% said in September that it could double its financing of “bespoke collateral” by giving its fixed-income trading arm an extra $5 billion of balance-sheet capacity. This would bring in at least an extra $100 million of revenue, which likely only counts the net interest income Goldman would earn on debt it keeps and not all the other deal fees involved.
Deutsche Bank is a market leader in this business, earning roughly €400 million ($464 million) each quarter from all the financing linked to fixed-income trading, including nonlinear trade, while others such as Credit Suisse Group and BNP Paribas are more focused on certain products or regions.
So what is this business? It starts with lending to private-equity or hedge-fund clients who want to buy assets that are hard to value, hard to sell, or low quality. Such assets can’t be financed in markets by fixed-income trading arms in the traditional way that liquid, high-quality assets are.
The hot assets right now include pools of European bad loans; other private loans; large property deals, especially in the U.S.; and things like infrastructure assets in emerging markets. Some come from weaker banks’ balance sheets, but many are being found by investment bankers, or the hedge funds and private-equity firms that anchor the deals. Investor demand for such assets is outstripping investment banks’ ability to find the assets, according to one banker in the field.
Banks slice the financing into tranches. The riskiest equity slice is owned by the anchor hedge fund or private-equity firm and gets the biggest payoff if the assets perform well. The safest slice pays steady coupons and gets paid first.
Yves here. This is clearly a structured finance product. A bunch of what sound like pretty drecky loans are throw into a legal entity, and then the cash distributions are tranched. The investors do not have ownership rights to the assets. They have claims on the cash flows. The most senior tranche gets the first payment (say X% return per annum, after meeting any fees or costs) and only when the investors in that tranche have gotten everything they are supposed to does the next person in line get anything. This “waterfall” process continues all the way to the bottom, most junior layer, usually called the equity tranche. Note that there are usually separate waterfalls for interest and principal payments.
Even though the rating agencies comported themselves badly, the lack of rating agency involvement is a mixed bag. Rating agencies imposed some discipline on the process. More important, they kept RMBS and CDOs from being an opaque market by issuing ratings on them, providing market commentary and downgrading them, admittedly after the market had already started repricing them, but that was still an important communications mechanism to people who were really behind the eight ball, like the Fed. The fact that they were rated also allowed for the compilation of aggregate data.
However, the lack of ratings will presumably constrain the size of the market. A lot of institutional investors have tight limits on how much they can invest in non-rated paper.
Here are the things to watch:
Will this “non-linear finance” get to be anything other than a niche product? Recall that there was both a subprime RMBS and a related CDO market in the 1990s. They both blew up. They didn’t do much damage because they weren’t that big.
Will the structures have too many correlated risks? We don’t know. These deals sound like they are stuffed with so much exotica that particular deals might wind up with unexpected correlations and go spectacularly bust, but the premise seems to be that enough of the constituent assets are so exotic that they won’t trade in a highly parallel manner, save in a Big Crisis where everything risky falls off a cliff all together.
Will the packagers introduce additional leverage? One of the big reasons CDOs became such a destructive product was that there was so much demand for them that there weren’t enough real economy subprime loans to begin to satisfy the appetite. So the cleverest sponsors and packagers (you could “sponsor” a deal if you provided the equity tranche funding, then you got a big say over what went into the deal) began creating CDOs that consisted mainly of derivatives instead of mortgages. We described long-form how this worked in ECONNED. Effectively, the originators were creating synthetic borrowers to stand in the place of live ones. The result was that they created economic exposure to subprime risk that was 4-6x the amount that existed in the real economy. The financial crisis is widely misunderstood as a housing crisis. A housing bust would have been very harmful to the real economy, but it would not have nearly destroyed the global financial system. The 2008 crisis was a derivatives crisis.
Will these deals be a Ponzi scheme? Again as we explained in ECONNED, both the 1990s and 2000s CDO markets were Ponzi schemes. Despite all the seeming demand for the product, what investors wanted were the most senior tranches. For reasons too long to go into here, there was also demand for the equity tranche. Due to the high structuring costs, the interest payments weren’t high enough to adequately compensate the risk of investing in the junior tranches. CDO salesmen were very adept at finding stuffees (really dumb investors) as well as making liberal use of hookers and drugs to make sales at somewhat less clueless investors. But even then, they had parts they couldn’t sell.
Banks didn’t want this dreck on their balance sheets. It was permissible to put a portion of the unsold garbage in a supposed first generation CDO. This makes me wonder how much of the assets in these “nonlinear” deals are effectively CDOs squareds, which also means risky and blow-up prone. The junior tranche of a CDO can easily fail 100%, so if CDO contains 20-30% of junior CDO tranches as its assets, its supposed most senior tranche has a pretty high risk of having that 20-30% of its payout not materialize.
The other way those tranches got “placed” were in “CDO squareds” where the constituents were mainly junior tranches of other CDOs, with some better quality assets mixed in to make it look a bit better.
The large point is that this development is a sign of how desperate search for yield has become. In and of itself this “new” product appears not to be significant in aggregate terms, so in isolation it is not too alarming. But it is the sort of thing that can burn the unwary badly. If it becomes a meaningful-sized activty, the large-scale hazards go up too. Stay tuned.
1 The ones made of corporate loans, called CLOs, have similar structures (as in are also technically CDOs). Even though people have also been worrying about an increase in CLOs, they only performed garden-variety badly during the crisis. The models for measuring the diversification of risk in the CLOs worked pretty much as expected. The big problem was that the banks were still stuck with a lot of them on their balance sheets, and the tranches they held did fall meaningfully in value (there was no reliable reporting on this since the banks didn’t want this exposed, and they were trading microscopic amounts with friendly hedgie counterparties so they could mark them at much higher values than what they would have sold at). The sense I had was the CLOs went to 80-85 cents on the dollar when the banks weren’t prepared for that (due to the severity of the crisis, not due to the CLOs performing out of line with what investors understood about the structures). And they were in no position then to show any additional losses. So even though CLOs are highly levered structures, they were nowhere near as highly leveraged as CDOs, which were resecuritizations and hence were mind-bogglingly geared. So while they are in the “watch with some concern” category, since “innovative” financiers could find ways to amp up the risk of CLOs, they have to date performed in line with what
Bernie Sanders: The business of Wall Street is fraud and greed.
No truer words were ever spoken.
I just may be wrong here but it seems to me that given the potential toxicity of the assets being pooled, this “new” innovation/instrument could spawn an even bigger secondary CDS market. Thanks Yves, will definitely be staying tuned.
Agreed completely and it would have helped if I had made that point explicitly. The unsatisfied demand for the old ABS (read “heavily subprime”) CDOs sand the recognition that investors would buy heavily synthetic CDOs led to a big increase in CDS referencing really drecky subprime RMBS. As I explained in ECONNED, that had the effect of driving demand for really bad RMBS, and in turn really bad loans, and the masquerading of the risk (all that ever sold were the AAA tranches, the rest were stuffed to non-economic investors or rolled into other CDOs) not only prevented the normal risk pricing mechanisms from working, they actually undermined them.
So if people start creating CDS to help satisfy this demand synthetically through a similar process, you could see all real economy distortions due to large scale credit mispricing.
CDS market, single names, has shrunk (notional outstanding) into a third of what it was less than a decade ago, and the trend is down (http://www.swapsinfo.org/charts/swaps/market-risk-activity TIW doesn’t capture all, but is a pretty good indicator I believe).
With the new FRTB regulation coming on (if it does, it’s not clear what is going to happen in the US), majority of non-index CDSes would be subject to very large capital requirements (the regulation does not specifically target CDSes, but capital for illiquid instruments is punishing in many ways), so it’s going to be even more down in a few years.
The plus and minus of these is that the assets stuffed into the SPV are really illiquid and hard to price. You can think of them more really as a holding company that has bond financing (senior tranche) and equity financing (junionr tranche), where the bond holders hopefully get the first part of the cash, and the equity guys get the rest as divi (if any) – except of course it’s more tax efficient to structure the junior tranche as a junior debt instead.
I’d fully expect that the investors who hold them don’t really value the junior tranche (it’d be beyond mark-to-make-believe level). You can probably do some magic to value the senior tranche, but I’d have to see how exactly the waterfalls work (does it accumulate principals, or just treat principals as normal cash? Do the senior tranche amortise? etc. etc.)
From that perspective, I suspect it’s likely not much – unless, as you say, they found investors for the leveraged variety of this (I can think of a number of ways how to do so, principally with total return swaps, and given some hedgies are betting on end of the world for a decade now, I’m sure they would be ok to stand on the other side)
Since we don’t know how these deals are being structured, we have no idea what exactly is happening.
But you would not have “investors who hold” junior tranches unless they bought them. You buy a particular tranche. You would not buy it unless you thought the risk/return was adequate. The history of securitization of risky assets says the structurers have trouble finding enough stuffees + buyers of junior tranches.
However, both the first gen (RMBS) and second gen (CDO) securitizations had the same problem: there was not enough interest income in the entire deal after the structuring costs to pay enough to make every tranche saleable. The equity tranche was perceived to be absolutely critical to the deal getting launched (AAA investors would not commit with no equity tranche investor). So the equity tranche gets sweeteners. I don’t know exactly what they were in the CDOs (and remember, unlike with RMBS, they had veto rights over the assets, which allowed them to dictate structure and set up CDOs where they could make more money from them failing. But we’ll leave that aside).
But with RMBS, the equity tranche got overcollatearlization and excess spread. The equity tranche in a decent deal would pay out in 18 months. So if they didn’t go tits up immediately, they were really good paper. Banks used to keep the equity tranche in RMBS but then packaged them and sold them as NIM (net interest margin) bonds. There was a group of hedge funds that bought them actively.
So the equity tranche was not the problem. It was the junior tranches over that. There’s no reason to think these are two-tranche deals. Pretty much every securitization has the risks more finely grained than that. So I would expect that there are risky tranches that are hard to sell.
I agree that we don’t know how the deals are structured, so it’s hard to know what is happening. My point is that the assets in these are really hard to value even as discounted-cash (and thus pretty much impossible to price), and hence even any tranche will be hard to price.
So I don’t expect any tranches sold to be liquid either, and hence “price” is likely going to matter not much. I would not be surprised at all, if a lot of these were sold to family offices for example, who would care about the cash thrown off, not the price.
On the RMBS especially – the UK used (in most cases, not all) a very different securitization method (master trust, which in US is mostly for cards or similar short-term assets, vs standalone which was the US standard for private-label). The master trust securitization performed very well even for fairly distressed assets like Northern Rock mortgages (Granite had, from memory, four tranches, although lowest was still BBB+ rated, and no investor lost a penny on any of those, despite the credit losses reaching >10% on some months, and the total loss run at more than half a billion pounds few years back), as it tended to be significantly overcollateralised due to how the seller’s share (which was technically not meant to act as credit buffer, but in reality can) behaved post NR default .
So, there are securitization methods that work reasonably ok, but, to go back to the original point, unless we know how it works – and we don’t – we can’t say.
Something else that occurred to me. These deals have to be private placements (which restricts the market somewhat, especially in Europe), as to avoid some regulatory steps (like publishing the docs). For private placement, the buyers usually don’t care about rating, but that means it’s even more likely that the end buyers are family offices, and likely some Asian institutions.
Thanks for that info re UK. securitizations since I’ve often wondered why they held up so much better – Granite specifically but UK RMBS in general. Was the master trust method also used in other European securutizations e.g. the Irish or Spanish housing markets that blew up so spectacularly.?
If the investors didn’t lose despite high credit losses who actually “ate” those credit losses ? For Granite that was, presumably, HMG/BoE/No11 but what about all the other RMBS issued before Northern Rock morphed into Northern Wreck and all the other pretend banks collapsed shortly afterwards.
The tranches of CDOs were not liquid either. Some were called “trades” because they were never meant to be liquid.
The issue is the asset quality. The AAA tranches of subprime RMBS were pretty much OK but everything below that was generally toast, which was why CDOs went bad top to bottom.
That’s the difference in the MT structures – you pretty much don’t have to sell out the whole capital structure, and I’m not aware of anything below BBB+ to be sold out of those.
Standalones are much more dangerous, to some extent because they are fire-and-forget, while the MT are really expected to live for a long time.
If the sponsor (like Northern Rock) goes down, the cashflows switch so that the seller’s share (which the sponsor, or on bankruptcy the receiver holds) is right at the bottom, which has impact on the reserve funds and also, indirectly, on any potential credit losses (technically the credit losses are shared proportionally, but of course as the seller’s share grows because it doesn’t get any cash which is all used to repay investors or replenish reserves fund, the proportional credit losses are applied more to it than to the investors).
Or, to simplify, you have a semi-permanent equity (well, technically the juniormost tranche) which can’t go away easily, and in fact, in distress, proportionally grows (because it gets any remaining cash as last).
A clarification – it’s actually somewhat (well, a lot) more complicated, as what exactly happens under stress depends on whether asset trigger (underlying assets go bad en-masse) or non-asset trigger (sponsor goes belly up for example) kicks in, as that has different effects (including different accelerations and waterfall rerangement). But overall, MT is much safer for investors than standalone, which can be easily structured in very skewed ways.
….this sort of deconstruction of financial sector “innovation” is exactly what brought us to “Naked Capitalism” since inception, as well as “ECONNED”…so much thanks!
..especially relevant?: “That would also suggest the instrument would be hard to hedge.”
“So what is this business? It starts with lending to private-equity or hedge-fund clients who want to buy assets that are hard to value, hard to sell, or low quality. Such assets can’t be financed in markets by fixed-income trading arms in the traditional way that liquid, high-quality assets are.” -WSJ
And here CalPERS wants to turn over its PE investing portfolio to private PE/hedge investment firm Blackrock. What could go wrong?
‘Nonlinear’ should definitely ring alarm bells.
Some mathematical background: the class of problems known as ‘linear’ are attractive to researchers because they can be completely solved analytically. With the exception of a few simple or special cases (like the Central Limit Theorem in statistics) there is no particular reason to expect systems in the real world to behave in linear fashion. You can however create approximate linear models that match the real ones fairly closely on very small scales, in the same sense that the Earth can be considered flat on small distance scales.
For a long time, a practically universal technique in science was to create local linear approximations, stitch them together to create something that looked like a global solution, and claim they had solved the problem. That all fell apart in the mid to late 20th century, when mathematicians demonstrated that even very small discrepancies could multiply and produce massively different results (the ‘butterfly effect’). So when you ‘solve’ problems using linear approximation you may be fundamentally altering the nature of your model, to the point where it becomes completely useless for modeling and predictive purposes on anything but extremely short time scales. This led to some interest in studying the properties of nonlinear systems, which almost nobody had previously done due to the prevailing technique of mashing everything into a linear reference frame. Although they were nowhere near as tractable as linear problems, people were able to find some special case solutions and general descriptions of behaviour that gave them a bit more idea about what was going on, even if they fell short of actual predictions in most cases. Applications of this work picked up the description ‘nonlinear,’ which for a while became a trendy buzzword for attracting funding.
In economics and investment I would say it’s almost certainly a meaningless term, since nobody has been able to get even remotely useful results (i.e., having predictive value) with linear models that I’m aware. (The exception being Gaussian Markov processes as a model for stock price movement, which works sort of OK when nothing is going on but does a terrible job of modeling tail events or shocks). So it has about as much descriptive value as “not an elephant.” It is quite possible that it has value in the other sense I mentioned though (i.e., attracting funds) and that’s very likely the reason why it has made an appearance.
Regardless of the structures of individual securities in terms of tranche payment priority, “Wall Street innovation” and investors’ chase for yield due to the Fed’s zero real interest-rate policy since 2008 have likely (again) baked losses into the cake IMO. As forever and always: “More money has been lost reaching for yield than at the point of a gun.” —Raymond DeVoe, 1995
The clear desire of monetary authorities for systemic private sector debt growth regardless, a gradual erosion of the institutional quality of underwriting in terms of balancing yield against risk of loss, the transactional nature of much credit, and the huge losses incurred on structured financial products historically, don’t engender a high degree of confidence that there is nothing below the surface of the iceberg regardless of tranche structures. Wonder how much of the debt-levered corporate stock buybacks, loans to highly levered private equity firms for corporate buyouts, and debt on U.S. retail space are in today’s “non-linear” asset mix?
Too, there are the secondary considerations of how highly debt-levered the purchasers of these CDOs are themselves; and whether there is a related potential for creditor/purchaser defaults.
However, it is noteworthy that there also seems to be a view that “It just doesn’t matter.” … i.e., The Fed can and will create the money to cover losses.