Join our community of smart investors

Coordinating distancing

A bar in New Mexico shows how we might observe social distancing efficiently
May 12, 2020

We all face a problem: as the lockdown eases, how can we comfortably maintain social distancing in shops and on public transport? The question raises profound issues in economics, philosophy and politics.

Let’s say it’s Saturday morning. You think: “I’ll not go to B&Q this morning because the queues will be massive as only a few are allowed into the shop. I’ll go on Monday afternoon instead.” That’s perfectly reasonable for you. But if it’s reasonable for you, it’ll be reasonable for countless others. And if they all think this way B&Q will be empty on Saturday morning but have a massive queue on Monday afternoon. Your reasonable thought is then self-defeating.

And this is true generally. Any forecast for when a shop will be empty will be self-defeating if enough people share it. The same is true for public transport. The government wants us to stagger our journeys. But how? If everybody thinks “I’ll not get the train until 10 o’clock” all that happens is that the rush hour crush is pushed back a couple of hours.   

We can only observe social distancing efficiently if we all think differently. But this defies standard economics. It says that we all try to maximise utility and in doing so reach much the same solution. In trying to maintain social distancing, however, this is self-defeating.

There’s a name for this puzzle. Economists call it the El Farol problem, named by Brian Arthur at the Santa Fe Institute after a local bar that sometimes got uncomfortably crowded.

But here’s the good news. He showed, via computer simulations, that the numbers going to El Farol would quite quickly converge upon a comfortable level – enough to give patrons a good time without being overcrowded. This is consistent with a finding by Ido Erev at the Institute of Technology and Harvard’s Alvin Roth. They have shown that simple learning allows us to solve quite complex coordination problems such as how to avoid crowds. We don’t need the fancy optimisation used to torture economics students – trial and error works nearly as well.

Or does it? Ernst and Young’s Duncan Whitehead has pointed out a problem here. Say people turn up only to find a place too crowded. Some stay away next time while others decide to give it another go. Those who give it a go find it pleasantly empty and so keep going back. But what about those who stayed away? Some will continue to stay away, having learned their lesson. And others might turn up after a while but find it still busy, as other people have had the same idea as them. The upshot, says Mr Whitehead, is that some people will keep going and others will stay away. Which means that some will remain in self-imposed lockdown even after the legal lockdown has been lifted.

This isn’t necessarily catastrophic, though, as it’ll be different people locked down with respect to different stores: the people staying away from B&Q will be different from those staying away from Gates’ Garden Centre, and so on.

You might think all this is trivial. In fact, the exact opposite. The issues here are deep ones. How widely applicable is the standard economists’ conception of rationality, given that it is no use in this context? When that rationality is no help, is the result chaos and unpredictability or is it instead convergence to some kind of order? And how socially desirable is that order? Do we need state intervention to achieve decent levels of coordination, or does it emerge from decentralised decision-making? These questions are central to economics and politics. The El Farol problem is actually the heart of economics – vastly more so than the forecasting that passes for economics in so much of the media.