Blow-Ups and How They Happen

On Monday 19 October 1987 the Dow Jones Industrial Average fell about 22.6% in a single session, still the largest one-day percentage fall in its history. No war had started and no bank had failed that morning. Three months later the Presidential Task Force on Market Mechanisms, chaired by Nicholas Brady, published its report, and the Brady Report put a trading rule at the centre of the day: portfolio insurance. A small number of institutions running it sold large volumes of S&P 500 futures in Chicago as prices fell. The selling pushed futures to a discount to the stocks they tracked, index arbitrageurs carried it to New York by buying the cheap futures and selling the stocks, and on the worst afternoon the arbitrage itself broke down (the companion's summary of the report; detail to verify).

Portfolio insurance was not a reckless idea. It promised an institution the payoff of a protective put without buying one: sell futures as the market falls, buy them back as it rises, and the portfolio's value is floored. In option terms it was the delta hedge of a long put that the insurer did not own. On paper it worked, because the arithmetic assumed what Black–Scholes assumes, that prices move continuously and that the market can absorb the hedger's trades. For one user that was close enough to true. For many users running the same rule on the same day it was false. Every insurer had to sell more as prices fell, so their selling became the fall; prices gapped rather than moved; and the replicated put failed at precisely the moment it was bought for.

Three things changed afterwards, and all three are still visible. Index options have since been priced with a skew: out-of-the-money puts trade at much higher implied volatility than calls, because the market began to charge for the crash it had just seen. Exchanges introduced circuit breakers that halt stock and futures trading together (Chapter 19). And the day left a lesson about mechanical selling that recurs in nearly every case in this chapter. A rule that sells into falling prices, whether a stop-loss, an insurance programme, a volatility target or a leveraged product's daily rebalance, is safe for one user and dangerous when widely used.