How Should You Trade a Nuclear War?

Twice in the last two years, I’ve read headlines and thought “Hm. It sure sounds like a nuclear war is more likely than it used to be.” Like you, I wondered: how will this affect my portfolio? Is this bullish? Perhaps you’re wondering other things, but you should

Twice in the last two years, I’ve read headlines and thought “Hm. It sure sounds like a nuclear war is more likely than it used to be.” Like you, I wondered: how will this affect my portfolio?

Is this bullish?

Perhaps you’re wondering other things, but you should think about how to trade an impending disaster, even if you have other priorities, because people use prices as a proxy for how important news is.

Markets are generally good at synthesizing ambiguous information into a specific quantity. Was Brexit good or bad? Lots of people have opinions, but if you weight those opinions by a) how confident people are, and b) how confident they deserve to be, as measured by buying power, you get a defensible answer. Brexit, as it turns out, was bad. Electing Donald Trump was bad the night of the election, but great by the end of the next trading day.

(I didn’t say that this was a perfect system, just that it’s hard to improve on. If you have a better system you should be very rich by now, because you can arbitrage differences between public perception and reality, but if you’re that rich you’ll constitute so much of the market that it will in fact be a good indicator, just like I claimed)

While it’s not obligatory, it’s at least supererogatory. Accurate prices are a public good, so informed trades contribute to society.

Homo Economicus and you

A perfectly rational disembodied being with a RobinHood account and a desire to maximize its net worth would think of the risk of nuclear war like this: if a country gets nuked, its assets are worth zero. So if its assets are worth x today, and the risk of nuclear war rises from negligible to 10%, it makes sense to treat those assets as worth about 0.9x, and to sell at any price above that. Investors perform this kind of handicapping all the time: two companies are in the running for some contract, the contract generates some amount of profit, so each company’s market value is bumped up by (odds of winning the contract) * (net present value of profits if the contract is won). For union negotiations, it might be (cost of strike) * (odds of strike) vs (cost of new pay deal) * (odds of deal)

This disembodied, amoral, view-from-nowhere-but-my-portfolio being would know how to react to a headline claiming that two countries were more likely to get into a nuclear war: it would sell.

But the rest of us have a view from somewhere, and our best course of action depends on where we are when the bombs start falling. If North Korea successfully launched a nuclear missile at New York, a big drop in the S&P would not be my biggest concern. Death would be.1

But this sets up a funny incentive. If the missile is in the air, American stocks will plummet. But is it certain to hit? The chances aren’t 100% — there have been some false alarms, some more worrisome than others. If it doesn’t hit, stocks should obviously return to roughly normal.

So, crucially, my payoff from selling stocks is a lot lower than that of the disembodied homo economicus. If Homo E. is selling, I should buy. Worst-case scenario: the bomb hits, I’m dead and broke. Best case: bomb is a false alarm, and money has been redistributed to me.

In practical terms, overseas investors fulfill the role of this abstractly amoral investor. An American portfolio manager with 2% of their assets in Indian stocks or 3% in South Korean companies might feel some guilt about trimming that exposure in light of the risk that the country in question will be annihilated. But they’ll do it anyway.2 My contention is that it’s prudent for local investors to buy.

It’s the Apocalypse Put.

The General Case

This is a specific example of an interesting subset of the general problem of finance: there are agency conflicts. People manage money, and theoretically their incentive is to do the best for their clients, but in practice, ego, hubris, laziness, cowardice, and other moral flaws get in the way. This specific subset of agency problems might be called existential agency risk: someone managing money won’t reasonably price a bet whose payoff is conditional on some event that makes collecting the payoff unlikely.

We see this in every bubble: if your job is “dot-com analyst” or “Mortgage-backed securities trader” or “professional options seller,” you aren’t going to do a good job pricing anything whose value is contingent on your job being a bad one.

This may be one reason that passive investing-related companies have performed well and publicly-traded active investment companies have largely suffered: the people pricing these securities are betting on whether or not they’ll be out of a job; it’s not a hedge if you’re fired before you collect.3 Conversely, if you make the bet that everything will be fine, you might collect some extra returns — and if you’re wrong, you were going to get fired anyway.

Some people refer to this sort of thing as an “I’ll Be Gone, You’ll Be Gone” trade. (I’m always hoping someone will form IBGYBG LLC, a hedge fund with a mandate to ask every bank to get them into trades that have a 90% chance of producing good-looking results.)

This dynamic doesn’t just apply to depressing situations like nuclear war.

There are positive forms of this, too: markets would be bad at estimating the value of some company if the company’s business leads to unimaginable prosperity. A working general AI, for example. It’s worth a lot, but in a world with AI, will anyone care about money?

Robin Hanson notes that over a long period of history, productivity grew something like 0.01% per year. Then for most of recorded history, it was closer to 0.1%. And since the industrial revolution, it’s been close to 1%. The numbers may be off a little bit, but we’re looking at orders of magnitude here, and the salient point is that productivity tends to 10x over time. And productivity compounds.

A world where productivity is growing 10% a year looks very different from the current world; the value of the average worker doubles every seven years. Given what technology does to income inequality, you might expect this to mean that some people get unimaginably rich, and you’d be right, but if that happens, the rich will invest more of their money on the margin, leading to further capital accumulation and faster GDP growth. In that world, the hyper-rich have a lock on all material resources, so money only matters insofar as they see it as a good way to allocate them.

This is not exactly a “risk,” although it certainly produces some uncertainty. The issue is that an investor would have a hard time putting a value on a company whose business model ended up with this result. The world in which that business succeeds is a world in successful investment matters less, so it will be underpriced.

Overall, you end up with a market efficiency bell curve, with prediction accuracy on the Y axis and whether the prediction is good or bad news on the X axis. The market is very efficient at deciding whether 0.1% more GDP growth or 1% faster housing starts is a good thing, but systematically bad at pricing either a) a 3D printing company that reduces the price of a house to the price of a sandwich and allows the Bay Area’s population to octuple, or b) a bioengineered species of super-termites who devour every single American home.

Why It Matters

Why this matters: turning back to the Cowen tweet, markets are a good way to synthesize information. The price system takes events you don’t know about (labor unrest in one country, a busy port in another, a monsoon somewhere else), and translates them into the information that does matter (maybe coffee is more expensive than last week).

But markets are bad at discounting extreme events, because those events distort the financial and utility payoff of trades.

People are generally bad at recognizing bad news, because you can defer some of the pain if you refuse to admit the problem. Somewhere between happiness and divorce, many marriages go through a phase where both spouses think “Well, as long as we don’t talk about it, we’ll be fine…”

We should remember that markets have the same pathology. The weirder the future is, the less likely it is to meaningfully affect asset prices. And we live in an increasingly weird world, where even heads of state use market indices to measure the impacts of their policies. Proceed cautiously.


  1. Also I’m underweight equities.

  2. Not that I’m judging them. My response to 9/11 was that I bought stocks when the market reopened the next week.

  3. An active investor who bets that fund management companies’ stocks will go down, and that ETF companies’ stocks will go up, is in effect betting that he’ll be one of the last active managers to get fired. There’s an S-curve of adoption for many asset classes, and it’s only in the steep part of the S that you can be confident valuations will be affected.