(Attention conservation notice: about 1,400 words on a book that you should read — not only because it’s a good story, but because it says a lot about the world we’re in now, eight years after it was written. On the basis of this book, I intend to snap up Lowenstein’s Origins of the Crash, about the dot-com bubble.)

This is one of that rare breed: educational nerd porn. It’s part of a select group of books that tell a compelling story while not shying away from technical details.
Long-Term Capital Management’s specialty was “arbitrage,” which generally means “trying to root out those places where you can get something for nothing.” The canonical example is a stock on two separate exchanges (London and New York, say) with two different prices. If this difference persisted, I could buy the stock in the cheaper city, sell it in the dearer city, and pocket the difference. Piles of money like this just won’t be left laying around for long. Hence the joke about the economist and his friend who find a $100 bill laying on the ground; as his friend reaches to pick it up, the economist exclaims, “Don’t! If that were real, someone would have already taken it.”
$100 bills aren’t laying around very often, but the occasional penny may be. Suppose a share of Microsoft is momentarily trading for $19.36 in New York and $19.35 in London. Buying in London and selling in New York nets me a penny, but what if I have a computer program at the ready to buy a million shares in London and sell in New York? I’ve just pocketed $10,000 for a few seconds’ labor; nice work if you can get it.
A cleverer piece of arbitrage, used to marvelous effect by Thaler, is the 3Com/Palm split. 3Com planned to separate its subsidiary’s stock into a separate issue; at first it would issue 5% of Palm’s shares, then a few months later issue the other 95%. When the 95% came, every 3Com investor would receive 1.5 shares of Palm. A rational market should, then, assign 3Com a value 1.5 times Palm’s price. This did not happen. An arbitrageur’s job is to make sure that it does.
This, and arbitrage of far deeper complexity, was Long-Term Capital Management’s job. It got its start in the world by employing, among others, two legendarily brilliant economists — Myron Scholes and Robert Merton — who granted it instant cachet. Investors ponied up tens of millions of dollars so that two of the men who invented options pricing could invest their money wisely in multiway trades of complex options.
The trick, at LTCM’s start, was to make arbitrage bets in pairs to minimize risk. Lowenstein gives a fascinating example of a security constructed entirely from mortgage interest payments (”IOs”), and another constructed entirely from principal payments (”POs”). When interest rates are falling, people refinance their mortgages, paying off more principal than they normally would; hence the price of IOs would tend to fall and POs would tend to rise. When interest rates are rising, the reverse happens.
To profit off the rise and fall of IOs, while not exposing yourself to too much risk, you want to choose another security that will rise when interest rates rise and fall when they fall. A natural choice is a Treasury. Buy an IO and sell a Treasury, and you have neatly hedged against the interest rate itself. The quantity you’re actually betting on, after deleting that interest-rate component, is the value of the IO. Long-Term believed that IOs were fetching too low a price, because investors were expecting another round of refinancing. Their hedging strategy allowed them to bet only on its price relative to its fundamental value, without getting involved in a bet on interest rates. They cleaned up.
In a market where all participants are acting rationally, Long-Term Capital Management has nothing to do. If everyone bets on the information available to him, and isn’t swayed by panicky crowds, then the price of a stock at any given moment reflects exactly what it’s worth, and the motion of that stock is literally random. We’ve seen examples where the market is clearly not rational — as, for instance, when it assigns a negative value to the part of 3Com stock that’s not tied up in Palm. As these irrationalities disappear, so does Long-Term’s business.
Which is exactly what happened as the years went by. Other hedge funds were doing the same work that Long-Term was doing: seeking and destroying financial irrationality. The profits from individual arbitrage trades were getting smaller for everyone involved. But Long-Term had to make a profit. Their answer: execute lots of trades. In our example above, they’d buy billions of shares of Microsoft stock in London and sell it in New York. And they’d use massive leverage to carry out the trades: buy shares using 50 times the capital they actually possessed.
They were bringing huge returns to their investors, so the money kept flowing in even as they had fewer and fewer places to put it. They started to get stupid: rather than carefully hedging one rising asset against another falling asset, they started taking one-sided trades. They started investing in fields where they had no expertise.
And finally, disastrously, the market behaved exactly as Long-Term’s models forecast that it wouldn’t. Arbitrage opportunities are supposed to disappear, which means that eventually the market will price 3Com and Palm correctly. Investors have every reason to believe that eventually one share of 3Com will be worth at least 1.5 shares of Palm; when that happens, those investors who held on to their 3Com shares will net a pretty penny.
But what if, instead, the market takes too long to “converge”? In the meantime, your lenders’ own weakening balance sheets have forced them to reduce their leverage. They clear debts off their books, and have to start calling in loans that they made to you. But the market hasn’t caught up with you yet. Your creditors are deleveraging exactly when you wish they wouldn’t. You keep hoping you can hold them off another few months until the market corrects itself, but before it does you go broke.
This is exactly what happened to Long-Term. In the space of 5 weeks, it blew through several billion dollars in capital, until eventually it had to be rescued by a consortium of banks brought together at the Federal Reserve Bank of New York. Long-Term is no more.
Lowenstein extracts a morality play from this, about the evils of following mathematical models of markets. Markets are unpredictable, says Lowenstein, and anyone who thinks the future will look like the past is a fool. He takes this as an article of faith, but there’s no good reason to believe him. The trouble at Long-Term wasn’t modeling; if I tell you “from my gut” that Amazon’s stock is certain to rise next year, that’s a “model” — an imprecise, probably unjustified model. If I predict Amazon’s stock-price movements on the basis of statistical patterns derived from data, that’s a better kind of model. If I build into my model the assumption that future stock-price movements will mimic those from the past, that may be a flawed assumption, but at least it’s available for criticism; my gut isn’t. Lowenstein makes a lot of fun of the eggheads, but he never explains what the alternative might be. If the alternative is old, grizzled bond traders with decades of experience, Lowenstein is then obliged to prove that their record is any better than that of the eggheads. This he does not do, and it’s doubtful that he could.
There seem to be two major problems with Long-Term’s models. First, they don’t account for a phenomenon known to economists and mathematicians since the 1960’s: stock-price motions have ‘heavy tails’. That is, there are more large swings in stock prices than the efficient market hypothesis can explain. (For that matter there’s a tension, which Lowenstein never fully resolves: the financiers can’t believe in the efficient market hypothesis with the one hand, and seek out arbitrage opportunities with the other.) Second, the models didn’t take enough of a global view: in the event of a panic, liquidity disappears and irrationalities persist. These call for better modeling; they don’t signal the folly of the whole enterprise, which Lowenstein implies.
It’s a captivating story, told intelligently and energetically. And unfortunately it couldn’t be more relevant today.