[Revised November 3, 2011]
There are still a lot of details to be worked out, but the €250 – €275 billion EFSF first loss insurance facility is starting to take shape.
The amount of exposure that the EFSF can take in any form and retain the AAA rating is capped at €452 billion Euro – the amount of guarantees provided by the AAA entities. 6 of the 17 EFSF countries are AAA (Germany 211, France 158.5, The Netherlands 44.4, Austria 21.6, Finland 14 and Luxembourg 1.9). The agencies have made it clear, that if any of those countries go on watch, or get downgraded, that would impact the rating of EFSF. If something happened to the ratings of Luxembourg, or Finland, or even Austria, they could adjust the size down marginally. The key will be for Germany, France, and the Netherlands to maintain their ratings – those 3 countries alone represent €414 billion of the AAA guarantees.
It looks more and more like the EFSF guarantees will be used in 3 different ways. A portion will be used to raise money to meet commitments already made to Greece, Ireland, and Portugal. Another portion will be allocated to provide additional capital to banks. Finally, a portion will be used to back first-loss insurance. There is some discussion about using it for secondary bond purchases, but at this stage, I am not sure how the EFSF would change the capital allocations or structures to do that, and in the end it may remain in the hands of the ECB anyway.
The EFSF First-Loss Insurance Program is like Nothing We Have Ever Seen Before
There are lots of examples of “guarantees” or “wraps” or “insurance” out there. But, providers of such “insurance” normally guarantee the full notional of the underlying asset(s). In this case, as a purchaser of such guaranteed paper, you can (if you so choose) largely ignore the credit quality of the underlying asset(s) and look directly to the credit of the guarantor for repayment. You would do better looking at the underlying assets, but it isn’t critical as would be the case if you were buying them outright. Fannie Mae is an example of such first-loss guarantor. All of the “monolines” (MBIA for instance) also have this business model. There is, however, a big difference between a full “wrap” and this partial wrap. Since it is only partial, you need to do a greater assessment of the underlying assets than you would need to do if it was a full wrap. The “first-loss” wrap allows the EFSF to leverage their guarantees, which is a benefit, but it doesn’t eliminate the need to focus on the underlying assets.
There are also lots of examples of credit enhancements where “first-loss” risk of a pool of assets is assumed by one group, making the “second-loss” more acceptable/safer. Credit Card deals, CMO’s, and CDO’s all have this characteristic. In all of these, the first-loss tranche is created for a pool of assets. In effect, this tranche lowers idiosyncratic risk and lets the “second-loss” investor focus on whether they are being compensated appropriately for the systematic risk. Credit Card deals are a great example (largely because senior investors haven’t been burned). If an underwriter has a good track record, you can assume that the credit cards they have issued will have a certain loss rate. You don’t know in advance which customers will or won’t pay, but you don’t care so long as a big enough percentage of the losses are taken by the first-loss provider. A CLO is similar. You may not know enough about individual leveraged loans, but if someone takes a 25% first-loss on a pool of 100 loans, it is easy to get comfortable that you won’t have losses on the other 75%, and you accept a return based on that. The EFSF “first-loss insurance” is different from these structures because the “credit enhancement” is on a single specified asset. Once again, it potentially reduces the possible loss and provides leverage, but investors cannot readily discount the underlying asset.
CDO’s and traditional “wraps” both ensure that the default of a single underlying asset doesn’t impact the second loss provider. That is just not the case with the EFSF program.
If EFSF isn’t like a “Monoline” and isn’t like a “CDO” what is it?
Without knowing the exact format, the EFSF first-loss insurance looks a lot like a Binary or Digital CDS. Those trades have a fixed payout if a Credit Event occurs. They never became that popular in the regular CDS market because you were more exposed than ever to some form of technical default, and recoveries in general are very volatile – ranging from 0 to par. We don’t know the exact format (insurance, wrap, guarantee, structured note, etc.) that the EFSF will take, but it looks like they will be providing something that pays a fixed amount if there is a default. Recovery plays a key role here. If there was a default and recovery was higher than 75% for a sovereign – although this doesn’t exactly seem likely – the payment under the “first-loss” insurance would be less than the maximum of 25% and the “second loss” holder wouldn’t have a loss. If a default occurs with respect to a European Sovereign with recovery below 75% (which seems a pretty safe assumption at this stage), the first-loss insurance would be the full 25%. Thus, a “digital” or “binary” outcome. For now also, lets assume that they find a way to make everyone believe that the guarantee will act as the market expects it should (as opposed to all the questions raised by the EU manipulating the situation to deliberately avoid a Credit Event in respect to Greek CDS specifically, and Sovereign CDS in general).
What is the Value of a Binary CDS
Let’s look at a trade where if a default on a specific entity (or asset) occurs, and the owner of the insurance receives 25% of the bond notional he owns. There are two concepts here: the “theoretical price”, and the “practical price”. The theoretical price is based on the likelihood of a payment being triggered (in this case default). So if the insurance were to be written on an extremely safe entity, the value of the first-loss insurance would be close to 0%, since default is very unlikely and thus no contingent payment under insurance is likely to occur. On the other hand, if the insurance were to be written on an extremely risky entity, the value should approach the 25%. In this case, the risk of default of the underlying entity would drive the theoretical value.
For the practical price, what someone is willing to pay for this option is a function of the quality of the entity that provided the guarantee. If the guarantor is unlikely to pay because it too is near default, then even if the “theoretical” value was 25%, the price such first-loss insurance would trade at or “practical” value would be close to 0%.
So a first loss guarantee on Portugal written by the EFSF would have a value close to 25% (high probability of default, coupled with high quality insurance provider). A first loss guarantee on Italy will have some mid-level value, and could over time trend to zero if Italian credit improves, and could drift towards 25% if Italy worsens while the AAA guarantors remain strong. A first loss guarantee on France would have little value, and would never have a value much greater than about 16.2% (the portion of 25% that the AAAs without France are responsible for paying).
Finally it is important to note that if recovery were to look with certainty more than 75%, this first-loss insurance would cease to be a binary instrument because the payout now could range from 0% to 25% depending on actual recovery rate, but the owner of the insurance would be fully covered on his losses.
These details are important, because whoever is managing the EFSF can use them to maximize the benefit of the EFSF guarantees, or to waste the guarantees.
With €250 Billion of First Loss Guarantees to Issue, how should the EFSF do it?
There are a few big questions that the EFSF must answer as they embark on this scheme. Each decision is important and no choice is obviously superior to others, each has its own drawbacks and benefits. Some will help more if what the market really needs is a one-time shot in the arm (which seems rather doubtful at this point), and some are better as a way to deal with another big round of weakness. We see 3 key decisions that need to be made – form of guaranty payouts, asset specific or country specific, and tied to new issues only or not.
Form of Guaranty Payouts – Wrap or CDS
The EFSF can choose to set up its guarantee either (1) to make payments that cover “originally scheduled” interest and principal payments on a specific bond (in effect, a guarantee of the specified cash flows) or (2) a CDS-style one time lump sum settlement.
The first way is how a typical “monoline wrap” works. The idea is that the guarantor puts the owner of the bond into a position where he would be if the default had not happened (but not in a better position which would be the case if the principal is immediately repaid). If the underlying bond fails to pay a coupon, then the EFSF would pay the coupon on 25% of the bond notional. This works much better if the guaranty is linked to a specific bond and isn’t freely tradable. Each time a new bond is issued with a wrap, it is that bond’s cash flows that are guaranteed. This can work with the guaranty being able to be stripped out, but works best when it cannot be. The big benefit of this format for guarantors is that they don’t have to come up with as much cash up front. The big 25% payment is only paid when the “wrapped” bond matures. This type of guarantee reduces immediate rating pressure if one of the borrower countries defaults. On the other hand, the market will not like this form of wrap. Delayed “wrap” payments don’t usually cause much concern for purchasers of “wrapped” bonds, but given the correlation between the guarantor and the defaulting bond, delay of guaranty payments increases the risk that future guaranty payments will not all be made. For instance, if Italy were to default, as a bond owner you want the EFSF to pay now, because the likelihood that the EFSF itself is in worse shape in the future, has increased.
Other benefit of this “wrap” approach is that the EFSF could just decide to make the payments even if there wasn’t a default. This would directly shift 25% of the debt burden from the underlying country to the EFSF. I’m not sure the EFSF would ever do that, but I could see circumstances that it would make sense – if France and Germany actually return to growth, while Italy or Spain remain stagnant.
However, since this is not a full wrap, any failure to pay coupons would still be a failure to pay with respect to 75% of the bond notional. The wrap also avoids having to have too much focus on defining default. So long as investors are receiving coupon there is no default, and if the country doesn’t pay the coupon, the EFSF pays the coupon on 25% of the issue. Simple and really just Failure to Pay trigger.
The second way is lump-sum CDS-style settlement. Once a default (however it is defined) occurs, the EFSF pays out 100% minus recovery, up to a maximum of 25% of the bond notional. That would be the simplest form (once anyone agrees on the default language). This form of “insurance” would be the easiest to trade independently of a specifically insured bond. The payout is likely to be 25% since recovery is likely to be below 75% (assuming people are comfortable with how a payout triggering event is defined). The fact that this is easily tradable is offset by the fact that this would trigger fairly immediate EFSF payouts right after a European sovereign has defaulted and might create the systemic contagion that the EFSF is trying to prevent. The need for immediate payouts would mean that EFSF members would need to issue bonds to fund their payments to the EFSF and that will put pressure on their funding, while at the same time losses (if recovery is below 75%) will be hitting the bondholders.
I have glossed over what would be a “triggering” event for the guaranty payout, but with all that is going on already in the CDS market, I suspect that this would be a non-trivial issue in the real world.
I think on balance the “wrap” approach is preferable to the “CDS-like” approach. It has reduced flexibility and isn’t quite as appealing to the market from trading perspective, but gives some hope that the contagion wouldn’t immediately drag down the AAA’s and it doesn’t hinge on some definition of “default” that the market may reject as being too easily manipulated.
Freely Tradable of a New Type of Bond
The original “proposal” that was sent around included the idea of making the guaranty “freely” tradable. If people can really trade the guaranty “freely” it is likely that it would quickly wind up in hands of people hoping to profit from being short. So they will probably try and make sure you hold some bonds in order to hold the guaranty. That might be workable, but will certainly be less effective than being truly tradable.
The main benefit/goal of the EFSF first-loss is to help all of the debt of the country needing support. By making the guaranty applicable to any bond, all bonds should benefit. Since banks and corporations in a country will price debt using the sovereign debt as a benchmark, there is additional benefit to pushing the entire yield curve better. On the other hand, the effect of the guaranty on new issues is very much diluted. New Issues will price at the slightly improved sovereign yield (once the value of the attached guaranty is stripped out). So the country won’t see as much of a benefit on new issues as it would if the guaranty was attached to a specific bond and not stripped out. Freely tradable guaranty may also cannibalize sales. There may well be a class of investor that wants riskier, un-hedged bonds. They won’t be able to get those bonds, so the EFSF may not gain maximum benefit from the guaranty as they are effectively giving it to some investors who didn’t need it to invest in those bonds.
If the guaranty is freely tradable, it won’t be long before the prices of guarantees are in the headlines and drive the markets. “The 5-year EFSF Italy Guarantee just traded at 22% implying a high probability of default.” “As the bond vigilantes attack France, the 3-year EFSF Spanish Guarantee dropped from 24% to 19% in spite of escalating problems in Spain – the market is concerned about France’s willingness to honor its EFSF obligations”. Etc. If the governments are worried that a few billion of net CDS exposure drives the bond markets, just wait until they see the monster they unleash with 250 billion of EFSF guarantees trading daily.
The EFSF can try and link the guaranty to specific bond and try to ensure that it isn’t tradable. We all know, that just means a desk will be set up to try and synthetically strip it, or figure out how to trade CDS against it, etc. So while it could be nice to say it isn’t tradable, the reality is that the market will try to find ways to whip it around anyway. A guarantee in “wrap” form makes it less fungible anyways, so once again, leads to the conclusion that wrap and non-tradable go together, and that CDS style and freely tradable work together, but mixing the other combinations are less effective.
By creating a new bond that is 25% guaranteed, EU can create assets for risk averse investors and still have assets for those investors willing to take on more risk. It is possible that they get better execution on the new issues by offering some EFSF-guaranteed bonds and some traditional bonds at each issue. The guaranteed bonds should provide a much lower coupon for the sovereign borrower, but the overall market for existing non-guaranteed bonds will be weaker and that does flow through the entire system of that country since that is used for a benchmark.
I’m very torn on what is better, but lean towards attaching to specific bonds, mostly because I think the “wrap” form is the way to go, and the “wrap” works better in the asset specific format than as freely tradable. I also think that for a group that spends so much time trying to crush the CDS market, they will realize they are far more comfortable with non-tradable guarantees.
When and How Much to Issue?
This is the absolutely crucial decision. EFSF is notionally constrained. Whether it allocates a €20 billion guarantee linked to Portugal or to France, EFSF’s ability to issue future guarantees will be reduced by the same amount, regardless of the fact that the “value” of the guarantee is dramatically different for these countries. Banks and hedge funds tend to think in terms of “risk”. Both would be more comfortable with a larger position in France than Portugal based on their assessment of volatility, risk, etc. We can argue whether these risk metrics were the root of the problems in the first place, but that is how risk managers are trained to behave. The EFSF guarantees are a scarce resource of €250 billion, that needs to be allocated is such as way as to maximize the value of spending those funds.
Does it make sense to offer guarantees when the market is trading reasonably well? If outright bonds can be sold at 5% yields for 5 years, should a guarantee be attached? You don’t get much benefit from the guarantee in that case, but the notional is used up. If the situation gets out of control, like a Portugal, how much does the guarantee really help? If the EFSF issue guarantees that are worth almost the full notional, it will be deemed as very risky counterparty, and it may not help drive much of a new issue. If bonds can be bought in the secondary market at 60% of par, a 25% first loss protected bond, may not get that much interest. The risk/reward would likely favor purchasing the 60% of par priced bond in the secondary market. If you bought the new issue, you could lose 15% more than if you bought the other bond in the secondary market. In theory you would get a higher coupon on the new issue, so at some point, if default is avoided, you can outperform, but it will be a decision based on time to default and just how massive the coupon is (the bond priced at 60% can appreciate in value). That high coupon might not help the sovereign borrower much, but merely delay the time to default a bit. So if the guaranty has a high intrinsic value at time of issuance, the impact of the leverage is much lower.
My gut feel is that this facility shouldn’t be used when bonds can be issued below 6% (or maybe even 7%) since the notional is used up quickly with limited value being received for the guaranty. It also probably doesn’t work if yields get above 10% because the second loss is too risky, so you don’t get the leverage and the benefit to the issuing country is marginal at best.
I would recommend going with a two-tier system: wrapped bonds where the wrap is asset-specific, and traditional bonds. As a country needs to do an auction, the amount of straight debt and guaranteed debt can be varied. If the market is doing well, maybe 100% can be done as a straight bond, thus reserving the notional for future use. If the market is widening, but still in control, maybe a bigger deal can be offered to the market, in order to use the notional while it is still efficient to do so.
As countries deteriorate, they may have to be allowed to default because not only will the EFSF get little or no leverage, it will get attacked. Investors will not wait for the EFSF to run out of notional before going after countries that haven’t fixed their core problems or even going after the EFSF’s good members if they have taken on too much debt. Once the end of guarantees is in sight (if not before then) countries that still have fundamental problems will face renewed pressure. Limiting the use of guarantees when they aren’t being sold for good value would give more time to the Eurozone to correct its problems, but may not be enough to tide us over into a period where PIIGS default is off the table.
Alternatively, they could flood the market with PIIGS guarantees right away. That could spark a big rally in the bond markets. Maybe that is enough to regain confidence and growth. It could equally fail to do anything, and with no dry powder left, the deterioration could accelerate rather than being delayed long enough to fix the underlying problems.
No easy answers, but the EFSF doesn’t even have the right people to ask the right questions
Some decisions will have to be made soon. The form of the guarantee and whether it is freely tradable or not. Those decisions could be changed over time, but they need to be made sooner rather than later to give the market some confidence and a focus.
Managing when to use the guarantees is a far more difficult problem. There is no right answer, but someone is going to have to be responsible, and their decisions on how to use this scarce resource will control how long the program is effective or if it is even effective at all. The EFSF needs people of its own to figure it out – the banks, and even central banks have so many of their own agendas, that someone needs to be strong to make sure this program isn’t completely wasted or used horribly ineffectively. I ultimately don’t think the program does that much, but decision makers at the EFSF needs to be working on it right now to give it any chance. Mr. Regling, while likely a strong choice when the EFSF was going to make highly negotiated bilateral loans to individual countries, doesn’t seem up to the task of managing this now massive, and highly complex structured version of the EFSF. The recent trip to China trying to pitch investing in something, anything, without any idea of what something or anything would be, is just one example.
Why we have wound up at the stage that issuing binary options on sovereign debt is a good solution, I don’t know, but since we are there, it might as well be done as well as possible.
What about SPITALY?
They could try and jointly wrap a big block of Italian and Spanish bonds. Make it somewhat more CDO like. Let’s say they do €350 of Spain and €650 of Italy. If Italy defaults, the second loss holder would likely have a loss, so it still doesn’t protect the 2nd loss holder against even one default, If Spain were to default, then there would be money left over, but the reality is that you would not have much protection on your Italy EFSF hedge, and it would ensure that investors were now looking to hedge Italian exposure and probably again at a time that Italian debt is in deep trouble because of the Spanish default. Something like this would give the biggest bang for the Euro up front, but does nothing but ensure that contagion will happen if it isn’t enough to solve the problems. The two big countries that the EFSF is trying to protect are both too big, default likelihood is too high, and the deep pockets just aren’t deep enough, and the fact that every country involved is highly correlated, makes any wrap or CDO solution marginal at best because it unravels at the first default.