(__Attention conservation notice__: 1600 words reviewing two books from the perspective of securities traders. Run right out and read [book: Diary of a Very Bad Year], and skip Das.)
These books need to be reviewed together, because they overlap in a lot of ways. For one, the author of [book: Traders, Guns, & Money] is unbelievably self-aggrandizing, while the subject of [book: Diary of a Very Bad Year] is just literally unbelievable.
[book: Traders] came out in 2006, before the world had fully melted down, so it gets some credit for being out in front about how incomprehensible certain derivatives, particularly the famed Collateralized Debt Obligations (CDOs) and Credit Default Swaps (CDSes) are. To review: a Collateralized Debt Obligation is essentially a piece of a mortgage (or some other asset backed by collateral, as opposed to something like credit-card debt). Typically these are assembled into “tranches,” which are groups of mortgages containing similar risk of default. If many mortgages default, one tranche — the “equity” tranche — gets wiped out first; its risk is therefore higher than that of the other tranches, so its return is correspondingly higher. As the defaults mount, the other tranches get wiped out in sequence. This is how you can end up with a collection of poor mortgages bundled into a security that gets labeled “AAA” (investment-grade): the later tranches, which are less likely to get wiped out, are AAA, while those which are first in line when the revolution comes are higher-risk. It looks like magic, but it’s actually sensible.
Mathematically, there are a few troubles with this. One is that you need to know some things about the correlation of the assets in the mortgage pool. That is, does knowing that one mortgage is in default tell you anything about whether another is in default? Suppose all the mortgages in your pool came from the same neighborhood; it’s likely that their defaults would be highly correlated. If the one mortgage defaulting means that all mortgages will default, then we say that their correlation is 1; if there’s absolutely no relation between whether one defaults and whether another does, then we say they have correlation zero. Obviously a lot depends on the correlation: if the correlation is 1 between the defaults of all the mortgages in your pool, then dividing into tranches doesn’t matter in the least: all mortgages will default at once, so all tranches will be wiped out at the same time, so it doesn’t make sense to call one tranch AAA and another junk. And it’s hard to estimate correlations when few people typically default on their mortgages. Recent experience suggests that correlation is near 0 most of the time, but near 1 when the economy is in a certain kind of recession; this is not helpful information. But in any case, you need to know the correlation if you hope to get any sense of how risky each tranche is. Since higher risk should yield higher return, you need to know the correlation to figure out what the yield on each tranche is.
A second, related problem with this sort of tranching is that it’s very sensitive to slight mis-estimation of the correlations. This is especially the case if you build new securities from a collection of CDOs, which are called “CDO-squared.” An excellent paper called “The Economics of Structured Finance” gives the clearest examples I’ve seen of this phenomenon. Bottom line: getting the correlations, or the individual default probabilities, just slightly wrong can drastically change the value of the security.
The people who assembled these complicated things are called “quants,” though [book: Traders, Guns & Money] and many other books make clear that complicated securities existed before quants did. So you’d think that [book: Traders, Guns & Money] would go easy on the quants. But no. Like a lot of books from the crisis, Satyajit Das likes to take cheap shots at the nerds hovering over their computers and their formulae. And like all the rest (I’m thinking, [foreign: inter alia], of [book: When Genius Failed], Roger Lowenstein’s depiction of the Long-Term Capital Management crisis), it misses the crucial question: maybe quantitative modeling is bad, but what’s the alternative? Does “going by gut feel” really have a better track record than using numbers?
Das’s own argument strongly suggests that the answer is no. [book: Traders, Guns & Money] is essentially a long litany of catastrophic explosions in the finance industry. Underlying all of them is the basic idea that you never destroy risk; you just shift it around. Or take the most recent mortgage meltdown. One problem seems to have been that the process went like this:
1. A bank issues a mortgage.
2. The bank immediately sells that mortgage to another company.
3. The company packages up many mortgages into tranched CDOs, as discussed.
4. The company constructs something called a Credit Default Swap (CDS) that’s sort of like, but importantly different from, an insurance policy. The CDS pays off if the mortgagee defaults. The company is now “hedged”: if they did the math right, they carry no risk at all — the insurance policy will cancel out any losses on the mortgages.
5. Lots of companies follow steps 1-4, so lots of CDOs and lots of CDSes go out.
6. CDOs and CDSes are profitable, so companies rush in to sell them, so banks are strongly encouraged to pump out mortgages as fast as they can. After all, they’re going to sell them right away, so they’ll hold no risk on their books but they’ll collect all the fees that go along with issuing mortgages.
7. Banks are supposed to identify good and bad credit risks; they’re the ones that are issuing the mortgages, after all. But what incentive do they have to identify those credit risks if they’ll be selling the mortgages just as soon as they can? They have no “skin in the game,” as the saying goes. So they start issuing mortgages to people who probably shouldn’t have them. They don’t tell this information to the CDO issuers; again, what incentive do they have to do so?
8. Mortgages start defaulting, and (to skip a bunch of steps) everything collapses.
Now the question for the class: which parts of 1-8 look to be the mathematicians’ fault? The mathematicians’ main nefarious role here, maybe, was to underestimate the default risk of a CDO. But they had nothing to do with the incentive structure that encouraged banks to issue junk mortgages. You can look through that list and find lots of failure points that have nothing to do with the geeks.
When it comes to doling out judgmentally wagging fingers, then, [book: Traders, Guns & Money] is on thin ice. Add in that Das is a remarkably self-serving author: whenever possible, he wants to convince you that he knew all along that finance was a bunch of hocus-pocus. He’s too cool for school, that Mr. Das, while all the rest of the industry are self-deluded assholes. The result is that [book: Traders] is an unsatisfying book that leaves me feeling icky. Its big strength is in describing, at a very detailed level, how various complicated securities work: swaps, swaptions, and the rest of the arsenal that we’ve become all too familiar with in the past couple years.
It was nice timing for me to move right from that to [book: Diary of a Very Bad Year]. The Anonymous Hedge-Fund Manager is everything that Satyajit Das is not. The HFM (as his interviewer at n+1 calls him) is erudite, calm, literary, and panoptic. He’s not stuck down in the muck of individual trades, although those are what he deals with day in and day out; instead he can take a broader view of the economy, and can identify when we should be scared and when we shouldn’t. He explains what commercial paper is, and why we should care when the CP market dries up.
He explains the contagious nature of financial crises, which is really the crucial detail to all of this. In earlier eras, the contagion was the sort of thing we see in [film: It’s a Wonderful Life]: word gets around that a bank is failing, and people line up at the doors to claim their money before it all disappears. So the New Deal created the FDIC, which guarantees that your money will be there if you come calling for it. The certainty that it will be there, as J.K. Galbraith noted in [book: Money: Whence It Came, Where It Went], assures that no one ever needs to run to the bank to check that it’s there. The modern version of banging on the bank’s doors is when there’s a run on an investment bank, which has nothing like the FDIC to insure it.
The HFM explains all of this with almost George Clooney levels of cool. He’s just too cool, too scholarly, too journalistic in his ability to explain complicated concepts to a lay audience. I have a hard time believing he exists; if he does, he needs to drop the anonymity and use his skills for the greater good. By the end of [book: Diary of a Bad Year], we find that the HFM has retired from New York City to Austin with his fiancée, so he’s got time. He’s used that time recently to sketch out his economic plans for Ezra Klein, so maybe he has a future as an educator. (I’m still not convinced that he’s real, even after writing for Klein. It seems entirely plausible to me that the HFM is a clever synthesis of the n+1 writers themselves.)
If you’re interested in the mechanics of constructing derivatives, by all means pick up the Das book. But the fact that the country even bothered to obsess about the details of swaptions and inverse floaters is a sign of great moral rot. Better to talk with the HFM, who unlike Das can see the forest for the trees.