So you could probably try to buy index CDS – sometimes this is called CDX, after a particular index that is traded a lot, think of it as standing for “Credit Default indeX” because in finance, true story, “index” starts with an “x” – on, what, $1.35 trillion of corporate exposure. But you won’t do that for bunches of reasons. One is that there’s not nearly enough of it – all of the index CDS in the world, combined, including investment-grade and high-yield debt, tranched and untranched, American and European and Australian, everything, adds up to about $11.2 trillion of “notional” (just the amount that is “insured”), and for you to get 12% of that seems pretty hard and for you to get the right 12% – the stuff that correlates with your risk – seems impossible. Another is that you would basically be spending all your profits from lending** on the CDS, so you’d have no money left over to do things like buy computers or rent office space or pay bonuses. A third is that, the way CDS works, it is accounted for on a mark-to-market basis, but remember that your loans are not (see two paragraphs ago). So if credit got better, you would have a creepy loss in your income statement because you would “lose money” on your CDS and not “make money” on your loans and securities. If you are a bank, “losing money” for accounting purposes is actually a big deal, not only because people get mad at you and stuff but also because it affects how well capitalized you are – lose enough money, on a mark-to-market basis, and you could get shut down. (This, again, is part of why you don’t take mark-to-market gains and losses on your loans.)
That is your problem – you want to hedge against a disaster, but you can’t just buy insurance against anything bad happening. So what you do is you conceive of a trade that:
- pays you plenty of (real) money on a disaster, but
- doesn’t cost you a lot of (real or fake/mark-to-market) money on a non-disaster, like regular market moves
What is that trade?
Well it starts by buying credit protection on things that make you a lot of money if things go bad. What you’re looking for here is a concept loosely called “leverage,” which means loosely that you don’t pay very much now but get a lot of money if things go bad very quickly. Think of it as: rather than pay for something that moves down a bit when credit improves a bit and up a bit when it worsens a bit, you’re paying for something that moves down a lot when credit improves a bit and down a lot when it worsens a bit. Some types of protection that do that are:
- buying very short-dated credit protection, like a credit index set to expire in December 2012, so that if things get really bad really fast you are ready for it,
- buying protection on something called “tranches,” which pay you relatively more for the next few defaults among the companies in the index, rather than paying you the same amount for all defaults in that index, and/or
- buying protection on high-yield indexes (junk bonds!), which are likely to go bankrupt faster if things get bad
It seems very likely that JPMorgan’s CIO did some or all of the above. It is hard to know! There is a deep mystery – if you like mysteries (and derivatives!) you can read the links above, but the deepest part of the mystery is that when all the hedge funds were complaining to Bloomberg about how JPMorgan was writing lots of protection on CDX.IG.NA.9 (next paragraph), no one was complaining that they were buying the protections mentioned above. I have no solution to the mystery and neither, it appears, does anyone else outside of JPMorgan. [Update: untrue! Meet me in the usual place.***]
Okay anyway though, this hasn’t solved any of your problems except maybe the size one (because you are buying lesser amounts of more intense protection). You are still paying money for protection, and you still lose money if credit improves. So you do the second half of this trade: you “write protection” (sell CDS) on the broad index. This is, again, very approximately like selling an insurance policy: you take in money now, and pay out some money if the companies in that index default. JPMorgan’s CIO very clearly did exactly that, on an index called the CDX.IG.NA.9 10-year, which despite the name matures in 2017. Intuitively – though it doesn’t quite work this way for curve trades so if you know about credit trading you’ll want to skip the rest of this sentence – you need to sell more of this broad protection than you bought in the previous paragraph, because what you bought in the previous paragraph was more intense than what you’re selling, so that’s why the whale sold so very much protection on this index.****
So what have you done? Something in outline like the following:
- You are getting money in on one end by selling protection and paying it out on the other by buying protection, so you are not paying out all of your profits.
- You have a position that is relatively neutral to credit market moves: if credit markets move up a bit, your big CDS-writing trade moves up a bit (you make some money), and your smaller but more intense CDS-buying trade moves down a lot (which, multiplied by the smaller size, means you lose some money, and they sort of offset). This is called being “DV01 neutral” but don’t worry about it. The point is that you don’t have huge mark to market losses when markets go up or down regular amounts.
- You have a position that makes a lot of money on bigger moves. So if a bunch of companies go bankrupt your really intense trade pays off a lot, while your less intense trade doesn’t cost that much. So in the really bad state of the world this is a good hedge for you. This is called being “long convexity” or “gap risk” but don’t worry about it.
There is much complexity swept under the rug here but the concept is a bet something like this:
- Take 100 companies
- For each one that goes bankrupt, you pay me $2
- For the first five that go bankrupt, I’ll pay you $1 each
- For the next five that go bankrupt, I’ll pay you $3
- For the next ten that go bankrupt, I’ll pay you $5
- If more than 20 go bankrupt, you don’t get any more – I’ve paid you the max of $70.
You lose a bit but not a lot if things stay good, you break even if things go mediocre, and you do great if things are really bad. (But there’s some horizon where if things are really really bad, more than 35 defaults, you start to lose again. This may or may not be a feature of JPMorgan’s bets.)
It seems fairly certain that in broad concept this is what JPMorgan’s Chief Investment Office was doing. But the above allows for lots of nuance. You want a hedge that is basically flat when things are basically okay, and pays out a lot when things are terrible. But define “flat,” and “okay,” and “a lot,” and “terrible.” These are hard judgments and are influenced by your expectations about the world. If you think that there are likely to be only 2-3 defaults, then the bet above looks like a real disaster hedge: it loses money in the expected case but pays off a lot if there are a staggering, unlikely 20 defaults. But if you think that there are likely to be 30-40 defaults, then the bet above looks ridiculously optimistic – not a hedge at all, but rather something that loses money if things are at the bad end of expectations.