Jerry Della Femina entitled his 1971 satiric look at the advertising industry “From Those Wonderful Folks Who Gave You Pearl Harbor”. The title came from an adman’s tongue-in-cheek idea for a new campaign during a brainstorming session on behalf of the client, Panasonic. Della Femina obviously knew that the idea of such an advertising campaign was funny because it was an outrageous parody of something characteristic of the industry. Certainly, the attack on Pearl Harbor in 1941 is a kind of tribute to the skill of Japanese organization and precise execution, but it could hardly be used to sell Japanese-made products to Americans!
Would that bankers and regulators got the joke. From the financial press we have learned recently: (a) that computerized trading has provided about 70% of the volume recently observed on major markets, (b) that exchanges such as the NYSE are building new facilities to accommodate it and help it respond even faster, and (c) that the quants who are promoting it (but who of course are otherwise disinterested observers) are threatening that if computerized trading is regulated or otherwise reined in, that capital markets will lose liquidity, and suffer unimaginable consequences. Where have we heard this before?
Remember portfolio insurance? Automated programs were supposed to sell futures and options in a market decline, generating cash and protecting the investor against loss. When a big market decline came along on Monday, October 19, 1987, the New York Stock Exchange registered its largest one-day loss ever—22.6%. As the market fell, computers, obeying the programs that had been fed into them, vomited up more and more sell orders, accelerating the debacle.
The problem was that the assumption lying behind portfolio insurance was that you would always have continuous, two-way markets. When panic selling started in Hong Kong, and spread around the world, what you had by the time Wall Street opened was neither continuous nor two-way: stocks gapped down, and even at these lower prices, there were no buyers. The computers, doing what they had been told to do, kept trying to chase prices lower, and in ever-rising volumes to compensate for the mounting losses in underlying stocks. This selling pressure forced prices still lower. The decline was accelerating into the closing bell. If trading could have gone on for another hour or two, we might have gotten the Dow to 00.00.
OK. We learned from that one, didn’t we? ‘Portfolio insurance’ became a dirty word (well, two dirty words) among conventional equity managers. The quants went back to their algorithms, worked out the bugs, and came up with bond arbitrage. Especially suited for mortgage bonds, the mathematicians’ ultimate triumph was the firm that couldn’t miss: Long-Term Capital Management. The smartest guys in the room broke free of the institutional constraints that had bound them at Salomon Brothers/Traveler’s/Citicorp. Wow could those guys trade! Armed with all those sophisticated quantitative techniques, they were able to translate minuscule pricing discrepancies in the credit markets into huge profits through the magic of leverage. You didn’t have to do homework, you didn’t have to wait weeks or years for the story to ripen, you just had to have the right algorithm. They kept breaking the bank and breaking the bank and breaking the bank . . . until one year it broke them. And virtually destroyed their prime broker, (the old) UBS. At the time, the financial stability of the whole Western World seemed threatened. The Fed got involved, lifelines were tossed to various organizations, guarantees were given, and somehow we muddled through (except for the shareholders in Long-Term Capital, and in UBS, that is.)
Fast forward nine years. This time, the mathematicians persuaded almost everyone that if you sliced and diced a pile of low-quality mortgages often enough, you could come up with a thousand-layered confection rated AAA. And, for those who still weren’t completely convinced, they had an instrument for you that would allow you to insure your credit instruments up to 100% of the amount lent. Hell, you could insure to 200% or 500%, or any amount you wanted. Why get your hands dirty with equities, and accept their piddling 8% or 15% or 30% returns, when you could double your money in the debt markets? The music was playing, man, and you gotta dance, dance, dance.
And everyone got rich, and we all lived happily ever after. No, wait. The housing market eventually rolled over, as it does every 15 or 25 years, the AAA confections behaved like the CCC paper they were made of, and banks began to emit cries of distress. Then the insurance everybody had, $62 trillion of it on $12 trillion of actual credit, began to bear down on all those credit-worthy financial organizations, and the lights started to go out, popping in a shower of sparks like toy cars under the feet of Godzilla. One of the largest writers of the insurance required $180 billion of taxpayer money, and it still wasn’t solvent. Something wrong with the algorithm?
By the time the carnage was over, clever people armed with sophisticated equations had turned a cyclical downturn into the worst panic since 1929, caused the first synchronous recession in the world since the creation of the IMF, destroyed two of the five largest prime brokers, driven the other three to their knees (and into the largest taxpayer-funded bailout in history), wiped out over $20 trillion in total market capitalization, and vaporized between 10 and 12% of total world household wealth, but as much as 22% in the United States. Some fun, eh buddy?
The incredible thing is that the surviving financial institutions, after all this carnage, are still at it. You would have thought that the quants would at least have had the decency to hide their heads for a few years, as they did after 1987 and 1998. Nope, computer-generated trading is back, bigger and faster than ever. It isn’t enough to have the biggest, fastest machine, running the quickest program in the world (‘My algorithm can whip your algorithm with its hand tied behind its back’), they want to site their computers in some central marketplace, so that nanoseconds (or is it picoseconds?) can be shaved off each trade. Given the huge volume the computers now account for, the exchanges aren’t about to say ‘No.’
Has anybody had a chance to investigate the implications of this? I don’t know whether the new programs have fixed all the problems of the old, but I don’t think anybody else does, either. I do know that (a) markets keep getting more and more volatile, despite all these volumes supposedly aiding in price discovery, and (b) some pretty horrendous trading mistakes have been made by machines in the recent past. Now, it’s all going to happen in nanoseconds, machine to machine, without any human intervention. Are these programs absolutely bug-free? Who can certify that they are?
Despite the tens of thousands of man-hours spent to de-bug them, and the experience of millions and millions of users out there, new bugs still crop up in the most widely-used software. Suppose some slight modification leads to a fatal mistake? Suppose some hacker manages to break in to one of these machines? How much will prices have moved before some human can press a circuit-breaker? And how do we figure out who should be made whole by whom for all those misleading prices? We’ve had many rogue traders. Is it possible, just maybe possible, that there is a rogue programmer waiting out there, toiling away obscurely within the bowels of some bank?
We had a whiff of Armageddon last fall, when for a while it seemed possible that every major financial institution might succumb to the cascading series of failures, followed by soaring premiums for CDS, followed by a closing off of credit and more failures, and on and on and on. It was a vicious (and accelerating) cycle, fed by a quasi-mechanical process, like a gigantic bank-shredding machine. Is it rational that the solution to this might be more automation? Do we want to see if some crisis can send the market to the circuit-breaking point in two hours? In ten seconds? After all, that’s already 10,000,000,000 nanoseconds—time for a lot of trades.
As Santayana famously observed, “Those who cannot remember the past are condemned to repeat it.” Yes, it’s good that the surviving financial institutions have begun to find their feet, and that animal spirits have begun to return to Wall Street. We need more solid markets, better real (as opposed to illusory) trading volumes, more decent issues. But we haven’t even had a chance yet to assess what exactly went wrong last fall, and how it could be contained in future. And yet some bright minds think that it’s already ‘off to the races and the last one in’s a rotten egg.’ I know, they have a lousy bonus from 2008 to make up for, and time’s ‘a wastin’. But do they think that the devastation of 2008 was caused by a falling asteroid?
Our first priority in trying to rebuild stronger capital markets, should not be to benefit from minute variations in prices. It should not be to try to defeat the remorseless tradeoff of risk and reward by coming up with a tiny asymmetry which permits the proud owners of a novel algorithm a few months of excess returns before everyone cottons on to the strategy and the returns disappear. We have to figure out what was broken and try to prevent it from breaking again. Money has to start circulating through the whole economy again. Capital markets have to return to their function as places where business can raise needed capital, and where households can safely save it. I don’t think anyone at any of the banks involved in this latest triumph of technology over experience is thinking along these lines. They just want to get back to ‘business as usual’ as soon as possible. Not only are they failing to learn from their own history, they are living proof of another famous dictum, this one by Marx: “All facts of great importance in history occur twice . . . the first time as tragedy, the second as farce.”