It’s been quiet around here, so I thought I would weigh in on something that’s in the news and that I actually know something about – financial derivatives, and in particular JP Morgan’s $2.3 billion-plus derivatives trading loss (word is that it could well surpass $5 billion). There are few topics I actually know anything about, so I couldn’t pass this one up. Since I deal with derivatives all the time, I sometimes assume that everybody can follow these things. That’s often a bad assumption, and I often can’t follow discussions on topics I’m not very familiar with, so feel free to ignore this post.
JP Morgan has said that it was a loss on a hedge, which is counterintuitive since hedges are supposed to reduce the risk of loss. It appears that there has been some creative trade definition here, and the hedge wasn’t much of a hedge. It was speculation.
First, what the cockadoodle is a hedge? Well, if a bank lends money to a company (or buys a bond issued by the company), they, at the most straightforward level, intend to profit from the interest they receive. But they have risk – what if the company messes up or the economy tanks, and they can’t pay back the loan? That’s where the CDS market comes in – CDS stands for Credit Default Swap, and it’s these suckers that were behind the big JP Morgan loss.
If I’m going to make $1 million per year interest from a loan, and I can protect myself from default by buying insurance (that’s essentially what a properly structured CDS is) for $250,000 per year, then I have hedged my default risk. If the company defaults, the seller of that CDS (similar to an insurance company) essentially pays off the loan, rather than the borrower paying it off. And the best part is that the insurance cost me less than the interest I receive, so I still make money.
Credit Default Swaps are transactions that aren’t necessarily tied to whether I lent the company money, though. I can buy a CDS on, say, Microsoft debt, and so can you. Anybody can. And there’s no limit to how much anyone can buy, either, which can lead to trouble, especially when banks deal in the CDS market (this isn’t a political statement – this fact is pretty universally accepted in the derivatives market). If Bank A buys a gazillion dollars of Credit Default Swaps and can force a company into default through strenuous enforcement of loan covenants on a million-dollar debt agreement (many covenants aren’t strenuously enforced, but that’s another topic), then we have a classic conflict of interest – the bank can actually make more money by forcing their own customer into default on their loan than they can make by having their customer make regular payments and eventually pay off the loan.
But, let’s return to the concept of a CDS as an insurance policy on existing debt – we’ll get back to the trading of them without any debt that needs to be hedged. It seems simple, but hedging in the CDS markets is rarely simple, for at least two reasons. In Reason 1, the rules for hedging are so poorly defined that almost anything can be called a hedge. If I own a bond, I’m at risk of the price of the bond going down (due to higher interest rates, which can be caused by default risk – German bonds don’t pay as much interest these days as Greek bonds do, for example). So, a CDS which protects me from bond price decreases can be called a hedge. But what if I’m planning to buy a bond? Then I’m worried about the bond price going up, before I pull the trigger. So now I can hedge by a CDS deal which protects me from bond price increases. So, you can hedge in both directions, which means the rules pretty much allow anything. Any trade can be called a hedge against something on the books of a company as huge as JP Morgan, and deciding whether to hedge at all is in the end a speculative decision. I can write pages on that topic, if anyone wants more detail, but I trust no one does. Suffice it to say that governmental and accounting hedge rules are essentially meaningless.
Now let’s get to Reason 2. JP Morgan had so many loans/bonds and was so active in the market for them that managing all of the CDS deals required for a nice clean hedge (of their existing loan book) was close to impossible. So they did what most companies do, and engaged in portfolio hedging. Portfolio hedging is getting a bad name in the press from people who don’t understand it, but if it’s done properly it makes a huge amount of sense (disclosure – I used to do a form of portfolio hedging for a major commodity company, and still do it to a very minor degree for the company I now work with, but I've never traded in the CDS market). In portfolio hedging, you aggregate all of your positions in the market and only hedge the net position.
So, If I own (am “long”) $100 million in bonds, and through other contracts I owe to other people (am “short”) $80 million in bonds, all issued by the same company, then I am really only exposed to default on $20 million of bonds – on the others, my long and short offset and one will make just as much as the other loses. If they’re not all issued by the same company, but the different companies suffer the same fate, then it works out just the same financially, and so long and short positions get aggregated and the net positions are all that’s dealt with. A big question is how you do the aggregating – do you lump everything together, or make a couple of categories of similar companies that you would expect to face similar fates in the marketplace?
Lumping everything together is passingly rare – just about everybody puts together categories of bonds facing similar fates. Why might they face similar fates? The companies might be in the same industry, or located in the same country, for example. If the market for buggy whips going in the tank, all the buggy whip companies feel the same pain. If the Greek economy goes in the tank, all Greek companies have a greater risk of default (I wonder why I’m picking on Greece today….). And since banks are always buying and selling bonds, they’re not just exposed to actual default. Even if the risk of default goes up, without any actual default occurring, the banks will suffer losses because the value of the bonds they have will decline, and when they eventually go to sell them they won’t get as much for them.
Dealing in the CDS market for every company a big bank has some exposure to is a huge and expensive task, and the categorization process isn’t easy, so banks use their statistical models to try to build simpler hedges that work well – they hedge portfolios instead of individual positions, like we just talked about. If bond defaults are driven by general economic conditions as noted above (rather than incompetence that’s limited to a single company), then the default risk of what you have on your book is probably about the same as the default risk of any well put together index of companies – both will suffer defaults at about the same rate. That’s what led to CDS Index contracts – they’re standardized contracts (that makes buying and selling easier than if they were custom-made) based on the debt of a group of companies, and they can do a decent job for you under the right circumstances.
But what if you don’t have “the right circumstances”? If you hedge that way, you end up with what’s called “basis risk”. What if the stuff you own defaults, but the stuff you hedged with doesn’t? Or even if you don’t get a default, but the risk of default changes – the stuff you own starts looking dicey, but the stuff you hedged with still looks fine? That’s basis risk. That spread – the difference in credit risk between different sets of companies – is a big market on its own. The differences could be any number of things, alone or in combination – different countries, different industries, different levels of initial creditworthiness, or just random chance – any of them could make a hedge turn sour. And if it goes sour, you get stuck holding the bag. Even if your models say it has always worked out just fine in the past (or usually worked – “always” doesn’t exist in these markets), there’s no guarantee that it’ll work when you need it to. But, of course, what can go badly can also go well, and it seems that JP Morgan had a stretch of it going well. Until recently.
From what I’ve seen, JP Morgan took that basis risk, tried to get cute with it, and in the end did everything very poorly. Please note that the full details of what happened have not come out and may never come out, but some information is known and that’s what this analysis is based on. At the time, JP Morgan had a portfolio of loans/bonds concentrated on European companies. You don’t need me to tell you that Europe isn’t a bastion of economic security these days. So they decided to hedge by buying a CDS Index based on high-yield bonds - bonds issued by high-risk companies. It appears that this index was not concentrated on European debt, though. Why didn’t they hedge with Europe-centered indices? Because (lots of informed but not verified assumptions here) their hedging group was expected to hedge in a way such that if the bond portfolio had losses the hedge made more money than the loss, and if the bond portfolio had gains the hedge lost less than those gains. That's a nice hedge, if you can get it to work that way. Thus, they were essentially expected to be a profit center even though hedging is supposed to be done to avoid losses. They made a market call and got it wrong. They engaged in what’s called “spedging” – speculating with their hedges.
It appears that basis risk was their downfall in this portion of the debacle. Europe continued to perform poorly, while the economies of other regions did well – so, their loans were doing poorly but the hedge didn’t protect them very well from that occurrence. Instead, it protected them from an overall economic downturn, and some economies turned up instead. The loans lost money, and the hedge lost money too. The hedge group didn’t like that, and did another trade to try to make that money back, or at least prevent any further losses. Here’s where they really venture into outright speculation, where they were dealing in CDS “insurance policies” without having the insurable risk. Since their CDS purchase was going badly, they sold CDS’s, but this time on an index of high-quality bonds. If you buy a CDS and sell a CDS, it makes sense that the risks offset. So, the second set of transactions look like a hedge of the first, if you ignore the high-quality vs low-quality basis risk (hint: don’t do that). But that still leaves your initial position, all those European bonds, unhedged. So then they went out and bought CDS’s on a bunch of individual companies.
Does this sound like a confused mess? It is – there’s no way around it. Instead of doing a clean hedge, or shutting down the badly done hedge once it was clear that it was indeed badly done, they just kept piling on new deal upon new deal in a desperate attempt to dig themselves out of a hole. The hole got deeper, and it came time to issue a quarterly report – they couldn’t keep it quiet. Disclosure is required, and a certain something hit the fan.
There is one aspect of this loss that I have seen at other companies that have had similar problems (such as Amaranth, Barings Bank, and others). Reports have come out showing that there was a group which made a lot of money for a period of time by taking on very large risky positions and winning on them, and came to command such power within the company that managers were discouraged from questioning the wisdom of what they were doing. This, of course, is the group that messed up. No one wins all of the time, and when complex, illiquid positions go bad they go bad quickly and in large measure (trust me, I know this first-hand). The group’s track record not only made them immune from criticism (for a while) but also made them overconfident in their ability to trade themselves out of a tight spot. This time, the efforts to work their way out instead made things worse. Despite JP Morgan’s reputation for excellent risk management, there is no computerized risk model that can account for hubris.
They made some bets that went wrong, big time. Is there anything illegal about that? No, there isn’t. The only legal problems I can see would come from whether the company and its managers knowingly misled the investing public about the losses in the period before it all was made public.
But there certainly are political and regulatory problems galore. Derivatives trading has risks that no model can measure, and JP Morgan and the regulatory regime seem to be overdependent on their models (JP Morgan actually invented the most widely used risk model, so that’s not a big surprise). Saying that banks can hedge their risks and control them by using these models is an invitation to actually pile risks on instead. Even the best speculators can blow up, and if they do so it should (according to many, and I’m one of them) be done with their own money and the money of fully informed investors, not the deposits of banking customers or the implicit government guarantees that come with “too big to fail” status. Congress will surely weigh in on this. If the control models don’t really allow you to exercise control, then another means must be found, and forbidding anything of this kind in a (too big to fail) bank with government-guaranteed retail customers is certainly one way to do it. I haven’t seen any other proposals that truly reduce risks, but I’m open to suggestions.
I started out on Burgundy but soon hit the harder stuff. Bob Dylan, Just Like Tom Thumb's Blues
How on earth did I get 10 QPs?