Gaming the risk model
When I worked in finance, there was a pretty well-known (and well-used) method of working around the pesky requirements of having a risk model and paying attention to risk limits in your group.
Namely, you’d let a risk guy in the group for a while, long enough to write a half-decent risk model, and then you’d say thanks, and we don’t need you anymore we’ll run with this, and then you’d kick him out of the group. You’d then spend the next few years learning how to game the risk model.
In particular you’d know exactly what kind of trades you could put in that the risk model can’t “see”: things like interest rate risk or counterparty risk, that the poor risk guy didn’t think of at the time, or even better the market you trade in would have developed and changed in the last few years so you were applying the risk model to instruments it wasn’t even meant to measure.
That way you could always stay within your risk limits, as a group, even while you took larger and larger bets on things that were invisible to the risk model. As long as the world didn’t blow up, this method returned higher-than-expected profits, so your “Sharpe ratio” looked great. You got rewarded for this, and in the meantime the company you worked for took on the risk (and they typically didn’t see it as coming from your trading group but rather as some amorphous systemic risk). It’s not clear how many people how high up were in on this method, but it seemed pretty clear that they also enjoyed the ride as long as it lasted.
The CDO market
CDOs were the reason we had a global economic crisis and not just a housing bubble. The CDO market is complicated, and you can learn a lot about it by reading the book. Suffice it to say I’m not going to be able to explain the whole thing, but let me simplify the story thus.
At the beginning (late 1980s through mid-late 1990s) there were not that many securitizations outside of the federal arena (Freddie Mac, Fannie Mae, and FHA), and they were pretty useful because they made piles of riskier but still viable-looking mortgages more predictable than individual mortgages. The top of the pile (they were separated in to groups called “tranches” depending on possible defaulting actions) were rated AAA by the big three ratings agencies (Moody’s, Fitch, and S&P) and probably deserved it, because they had a big cushion of loss protection beneath them. The lower tranches were lower rated and harder to sell, which limited the size of the overall market.
Starting around 2003 the lower-rated, harder-to-sell tranches from the BBB to the junior AAA tranche started getting resecuritized into instruments called CDOs. In fact there were riskier CDOs, called mezzanine CDOs, which consisted mainly of the BBB tranches, and “high grade” CDOs consisting mostly of old A and AA tranches. These CDOs were again tranched, with around 75% of the par value getting an AAA rating.
Yes, you heard that right: if you took a bunch of easy-to-imagine-they’d-fail low rated mortgage bond tranches (especially if you knew anything about the terms of those mortgages and how much they were counting on the housing market to continue its climb), and bundle them together, then the resulting package would, at its highest tranche, be deemed AAA. It made no sense then and it makes no sense now.
The CDS and synthetic CDO markets
Enter the credit default swaps market. The ability to buy CDS protection (insurance on the underlying bonds) on a higher tranche of the mortgage bonds (the first generation securitization) while purchasing a lower tranche made it possible for lots of people to bet that “if things go bad, they will go really bad”, while limiting their overall exposure. Moreover, the income on the lower rated tranche would fund an even bigger short position on the higher rated tranche, so this was a self-financing bet.
The demand for more cheap credit default swaps led some clever traders to realize they could create CDOs largely or entirely from credit default swaps rather than actual bonds. No need to be constrained by finding real borrowers! And you could bet against the same crap BBB bonds again and again, and have them packaged up and have most of the value of the “synthetic” or “hybrid” CDO rated AAA (again with the collusive help of the ratings agencies).
At first, the big protection sellers in the CDS market was AIG and the monoline insurers. But they only wrote CDSs on the least risky AAA CDO tranches. Later, after AIG stopped being involved, that side of the CDS market was entered into by all sorts of really dumb people, with the help from the complicit ratings agencies who kept awarding AAA ratings.
Even so, there was still a bottleneck for this re-rebundled synthetic/heavily synthetic CDO market. Namely, it was hard to find people to buy the so-called “equity tranche”, which was the tranche that would disappear first, as the first crop of the underlying loans defaulted.
That’s when Magnetar Capital came in. They set up deals to fail. They did this through explicitly designing the synthetic CDOs (banks gave this privelege to whomever was willing to buy the equity tranche) and by, in addition to buying the equity tranche, they bought up all of the CDS’s in the synthetic CDO.
The overall bet Magnetar Capital was taking was similar to the one above: when the market goes bad, it will go really bad. The difference is that Magnetar’s exposure was altogether very short: they set up the equity tranche to pay lots of cash in the short term (a couple of years), which would finance the cost of all of the CDSs in the hybrid CDO, which meant they didn’t just cover the exposure but magnified it multiple times. And it was again a self-financing bet, as long as they were right about the market exploding rather than slowly degrading.
How big was this? Magnetar Capital made the majority of the market in 2006, which was one of the biggest years in this market. And everything they did was legal. They also drove demand in the subprime mortgage market, during its most toxic phase, by dint of a combination of leverage and the clever manipulation of investors, specifically convincing them to post cash bonds.
Let’s go back to the groups gaming their risk models from the beginning of this post. Same thing happened here, except the group was this entire market, and the risk guy was the combination of the ratings agencies and AIG, as well as the greedy fools who wrote CDS on mortgages in 2006. And instead of the hedge fund being on the hook for their trading group’s games, in this case it was the United States and various European governments who were on the hook.
How predictable was this whole scheme? My guess is that Goldman Sachs knew exactly what was happening and what was going to happen. They made a very intelligent bet that if and when the housing market went under, AIG would be backed by the government. In essence this entire market was an enormous bet on government bailout. Not everyone knew, of course, especially the guys who were long the market when it collapsed, but lots of people knew. The same people who right now know where the dead bodies are on the books and who aren’t coming forward with a plan to resuscitate the financial system, in fact.
At the very least I think this story argues for the treatment of CDS as insurance, with the requisite regulation. In different terms, Magnetar chose buildings where they saw arsonists enter with gallons of gasoline and matches, and bet everything on a fire in that building. The question then is, how many fire insurance claims should one entity be allowed to buy for one building?
Check out this outrageous article about what Freddie Mac has been doing. Seriously makes my blood boil!!
Update: Yves Smith on Naked Capitalism posted this morning about how this is maybe not such a big deal.
In this recent article in the Wall Street Journal, we are presented with two sides of a debate on whether there should be a unique medical identifier given to each patient in the U.S. healthcare system.
Both sides agree that this would help record keeping problems so much (compared to the shambles that exist today) that it would vastly improve scientists’ ability to understand and predict disease. But the personal privacy issues are sufficiently worrying for some people to conclude that the benefits do not outweigh the risks.
Once it’s really easy to track people and their medical data through the system, the data can and will be exploited for commercial purposes or worse (imagine your potential employer looking up your entire medical record in addition to your prison record and credit score).
I agree with both sides, if that’s possible, although they both have flaws: the pro-identifier trivializes the problems of computer security, and the anti-identifier trivializes the field of data anonymization. It’s just incredibly frustrating that we haven’t been able to come to some reasonable solution to this that protects individual identities while letting the record keeping become digitized and reasonable.
Done well, a functional system would have the potential to save people’s lives in the millions while not exposing vulnerable people to more discrimination and suffering. Done poorly and without serious thought, we could easily have the worst of all worlds, where corporations have all the data they can pay for and where only rich people have the ability or influence to opt out of the system.
Let’s get it together, people! We need scientists and lawyers and privacy experts and ethicists and data nerds to get together and find some intelligently thought-out middle ground.
The blog interfluidity, written by Steve Randy Waldman, posted a while back on opacity and complexity in the financial system, arguing that it is opacity and the resulting lack of understanding of risk that makes the financial system work.
Although I like a lot of what this guy writes, I don’t agree with his logic. First, he uses the idea of equilibrium from economics, which I simply don’t trust, and second, his basic assumption is that people need to not have complete information to be optimistic. But that’s simply not true: people are known to be optimistic about things that have complete clarity, like the lottery. In other words, it’s not opacity that makes finance work, it’s human nature, and we don’t need any fancy math to explain that.
Partly in response to this idea, I wrote this post on how people in the financial system make money from information they know but you don’t.
But then Steve wrote a follow-up post which I really enjoy and has a lot of interesting ideas, and I want to address some of them today. Again he assumes that we don’t want a transparent financial system because it would prevent people from buying in to it. I’d just like to argue a bit more against this before going on.
In a p.s. to the follow-up post Steve defines transparency in terms of risks. But as anyone knows who has worked in finance, transparency is broadly understood to mean that the data is available. This could be data about who bought what for how much money, or it could refer to the data of which mortgages are bundled in which CDO’s, and whose houses those refer to and what is the credit score of the mortgagees, or all of the above. Let’s just say all of the above, say we have all the data we could legally ask for about everything on the market.
That’s still not a risk model. In fact, making good risk models from so much data is really hard, and is partly why the ratings agencies existed, so that people could outsource this work. Of course it turns out those guys sucked at it too.
My point is this: a transparent system is at best a system that gives you the raw ingredients to allow you to cook up some risk soup, but it’s left up to you to do so. Every person does this differently, and most people are optimistic about both the measurement of risk and the chances of something bad happening to them (see AIG for a great example of this).
I conclude from this that transparency is a goal we should not be afraid of, because first of all it won’t be all that useful unless people have excellent modeling skills, second of all because no two firms will agree on the risks, and third of all we are so far from transparent right now that it’s laughable to be afraid of such an unlikely scenario.
Going on to the second post of Steve now, he has some good points about how we should handle the very dysfunctional and very opaque current financial system. First, he talks about the relationship between bankers and regulators and argues for strong regulation. The incentives for bankers to make things opaque are large, and the payoffs huge. This creates an incentive for bankers to essentially bribe regulators and to share in the proceeds, which in turn creates an incentive for the regulators to actually encourage opacity, since it makes it easier for them to claim they were trying to do their job but things got too complicated. This sounds like a pretty good explanation for the current problems to me, by the way. He then goes on:
… I think that high quality financial regulation is very, very difficult to provide and maintain. But for as long as we are stuck with opaque finance, we have to work at it. There are some pretty obvious things we should be doing. It is much easier for regulators to supervise and hold to account smaller, simpler banks than huge, interconnected behemoths. Banks should not be permitted to arrange themselves in ways that are opaque to regulators, and where the boundary between legitimate and illegitimate behavior is fuzzy, regulators should err on the side of conservatism. “Shadow banking” must either be made regulable, or else prohibited. Outright fraud should be aggressively sought, and when found aggressively pursued. Opaque finance is by its nature “criminogenic”, to use Bill Black’s appropriate term. We need some disinfectant to stand-in for the missing sunlight. But it’s hard to get right. If regulation will be very intensive, we need regulators who are themselves good capital allocators, who are capable of designing incentives that discriminate between high-quality investment and cost-shifting gambles. If all we get is “tough” regulation that makes it frightening for intermediaries to accept even productive risks, the whole purpose of opaque finance will be thwarted. Capital mobilized in bulk from the general public will be stalled one level up, and we won’t get the continuous investment-at-scale that opaque finance is supposed to engender. “Good” opaque finance is fragile and difficult to maintain, but we haven’t invented an alternative.
I agree with everything he said here. We need strong and smart regulators, and we need to see regulation in every part of finance. Why is this so hard? Because of the vested interests of the people in control of the system now – they’ve even invented a kind of moral philosophy around why they should be allowed to legally rape and plunder the economy. As he explains:
I think we need to pay a great deal more attention to culture and ideology. Part of what has made opaque finance particularly destructive is a culture, in banking and other elite professions, that conflates self-interest and virtue. “What the market will bear” is not a sufficient statistic for ones social contribution. Sometimes virtue and pay are inversely correlated. Really! People have always been greedy, but bankers have sometimes understood that they are entrusted with other people’s wealth, and that this fact imposes obligations as well as opportunities. That this wealth is coaxed deceptively into their care ought increase the standard to which they hold themselves. If stolen resources are placed into your hands, you have a duty to steward those resources carefully until they can be returned to their owners, even if there are other uses you would find more remunerative. Bankers’ adversarial view of regulation, their clear delight in treating legal constraint as an obstacle to overcome rather than a standard to aspire to, is perverse. Yes, bankers are in the business of mobilizing capital, but they are also in the business of regulating the allocation of capital. That’s right: bankers themselves are regulators, it is a core part of their job that should be central to their culture. Obviously, one cannot create culture by fiat. The big meanie in me can’t help but point out that what you can do by fiat is dismember organizations with clearly deficient cultures.
Hear, hear! But how?
Please consider purchasing a 55 gallon tub of lube from Amazon.com (pictured below). And before deciding, I suggest you read the reviews (hat tip Richard Smith via Yves Smith).
Also, please be sure to take this quiz to differentiate (if you can) between Newt Gingrich and a comic book supervillian.
I love music. I work in an open office, one big room with 45 people, which makes it pretty loud sometimes, so it’s convenient to be able to put headphones on and listen to music when I need to focus. But the truth it I’d probably be doing it anyway.
I’m serious about music too, I subscribe to Pandora as well as Spotify, because I’ll get a new band recommendation from Pandora and then I want to check their entire oeuvre on Spotify. My latest obsession: Muse, especially this song. Muse is like the new Queen. Pandora knew I’d like Muse because my favorite band is Bright Eyes, which makes me pathetically emo, but I also like the Beatles and Elliott Smith, or whatever. I don’t know exactly how the model works, but the point is they’ve pegged me and good.
In fact it’s amazing how much great music and other stuff I’ve been learning about through the recommendation models coming out of things like Pandora and Netflix; those models really work. My life has definitely changed since they came into existence. I’m much more comfortable and entertained.
But here’s the thing, I’ve lost something too.
My oldest friend sent me some mixed CDs for Christmas. I listened to them at work one recent morning, and although I like a few songs, many of the them were downright jarring. I mean, so syncopated! So raw and violent! What the hell is this?! It was hip-hop, I think, although that was a word from some far-away time and place. Does hip-hop still exist?
I’ve become my own little island of smug musical taste. When is the last time I listened to the radio and learned about a new kind of music? It just doesn’t happen. Why would I listen to the radio when there’s wifi and I can stream my own?
It made me think about the history of shared music. Once upon a time, we had no electricity and we had to make our own music. There were traveling bands of musicians (my great-grandmother was a traveling piano player and my great-grandfather was the banjo player in that troupe) that brought the hit tunes to the little towns eager for the newest sounds. Then when we got around to inventing the radio and record players, boundaries were obliterated and the world was opened up. This sharing got accelerated as the technology grew, to the point now that anyone with access to a browser can hear any kind of music they’d like.
But now this other effect has taken hold, and our universes, our personal universes, are again contracting. We are creating boundaries again, each around ourselves and with the help of the models, and we’ve even figured out how to drown out the background music in Starbucks when we pick up our lattes (we just listen to our ipods while in line).
Keep in mind, as well, that there’s really one and only one goal of all of this, namely money. We are being shown things to make us comfortable so we will buy things. We aren’t being shown what we should see, at any level or by any definition, but rather what will flatter us sufficiently to consume. Our modeled world is the new opium.