Lately, the press has had a field day pointing out just how wrong epidemiology models can be. This shouldn’t be news, especially since most modelers never hid the fact that they had made some very bold assumptions. But it’s still important to ask just where the estimates went wrong. Most of the errors, it turns out, have to do with “social distancing.”
Their success is only as good as government’s capacity to provide timely, accurate, and above all clear evidence of where the epidemic is headed.
Now social distancing is mostly science. For anyone persuaded by the germ theory of disease, physical barriers can’t really fail. So in practice almost all of the uncertainty over “social distancing” is concentrated in the “social” component that most modelers call “compliance.” Early predictions that COVID would kill 200,000 Americans set this figure at fifty percent. But in fact the country did much better than that, with actual compliance exceeding ninety percent. This looks like an obvious error, and indeed social scientists in the UK had already warned that the government that it was underestimating compliance. Even so, it’s hard to blame the modelers. After all, no Western nation had tried social distancing in more than a century. This would have made empirical estimates almost impossible.
But there’s still something strange about ninety percent compliance. Unlike China, Western distancing has been almost entirely voluntary. Whatever the formal rules, there is practically no chance of getting arrested, and the disease itself never even began to approach levels that might terrify Americans into compliance. Then again, nothing in the public opinon surveys would predict this result. At the peak, just over forty percent of Americans said that COVID was a “severe health risk” to their communities, and sixty-five percent were “very concerned” that it would spread. On these numbers, most experts would have said fifty percent was a better guess than ninety.
So what happened? Collective behavior is more than the sum of peoples’ opinions. Before we choose, we also interact. And in this case, at least, the interactions greatly amplified compliance.
Anecdotally, this seems obvious. When history tells the COVID story, basketball will play a leading role. In ordinary times, practically all of us treat the mainstream media as so much white noise. How could we do anything else? But occasionally there’s an event that breaks through and gets our attention. For anyone living in the early 21st Century, the idea that our sports-crazed country would forego March Madness or the NBA would have been unthinkable. That made cancelling the season into a huge turning point, the one unique signal that finally got our attention. But just who, exactly, ordered it? Those who were formally responsible, the colleges and leagues were never more than figureheads. Once players began testing positive, each athelete would have made his own decision to play. Yes, the games could have continued with one or two dropouts, but any substantial number would have forced a shutdown. In this sense, at least, something like majority vote emerged automatically.
The more general point is that hardly any of us work, play, or socialize by ourselves. We do those things with others, and when those people express concern we generally listen. Just why we do this varies. Sometimes the activity is less valuable without the dropouts. Sometimes we are afraid that defying the dropouts would offend them and destroy the group. And sometimes, especially in business, it’s physically impossible to do the task without them. Regardless, we need each other – and negotiate groundrules to make that happen.
Economists typically analyze this kind of dynamic under the heading of “network effects,” i.e. situations in which people derive value from using the same products and standards as their friends. In Silicon Valley, the network effects are so strong that the first widely-popular standard can sometimes drive out all the others.
The good news is that scholars understand network effects pretty well these days. This offers at least four broad lessons for COVID:
- Amplification Works Both Ways - The COVID debate has always pitted people who favor a lockdown over those who prioritize the economy. So far, the former have had the upper hand. But there is no reason why those who favor reopening could not also become a majority, in which case the need for common standards might equally suppress compliance to, say, ten percent. At that point, the country could find itself caught in a bang-bang dynamic where social distancing was always too much or too little.
- Private Regulation Is Powerful - Politicians and bureaucrats generally assume that “official” action is the only kind that matters. And it’s true that private regulation can’t send anyone to jail. Usually, though that isn’t necessary – merely “economic” sanctions like going bankrupt are sufficient. Meanwhile, the possibility of private regulation offers important advantages. Government, after all, doesn’t have nearly enough resources to write, let alone enforce social distancing rules for the whole economy. Fortunately, that doesn’t seem to be necessary. Capitalism has always specialized in the kind of backhanded altruism that thinks constantly about pleasing others to help the bottom line. Today, that means simultaneously persuading consumers that it is safe to shop, workers that it is safe to work, and your corporate trading partners that you won’t cause a scandal. As I have argued elsewhere the results often end up approximating public opinion in much the same way that a general election would.
- Competing Standards Are Inevitable – And Good - Even in Silicon Valley, very few standards grow into monopolies. More often, several networks can coexist in the same physical space, each catering to a slightly different community. From this standpoint, reports of surfers and skateboarders defying shelter-in-place orders probably shouldn’t worry us. Those communities were, after all, small and contrarian to begin with. And when the networks don’t overlap, the existence of different standards becomes a virtue. The reason, to paraphrase the late Tip O’Neill, is that all epidemics are local: What works in rural Georgia won’t work in Atlanta and vice versa.
- Standards are “Sticky” - Starting a new standard is hard. By definition, the first consumers to join a new standard don’t get much in the way of network effects. The result is that existing standards tend to be “sticky” in the sense that they often remain dominant even after technologically superior alternatives enter the market. This same dynamic is already visible in social distancing, where consumers and firms seem reluctant to reopen even after government says they can. At the same time, frustration with an outdated standard tends to accumulate. So when consumers do decide to switch, change tends to happen suddenly. Dominant Silicon Valley firms have traditionally tried to protect themselves by continually morphing their standards into something better. For the rapidly changing COVID situation, this kind of continual revision seems all to the good.
Readers might be forgiven for thinking that my account makes government superfluous. But in fact, it’s central. On the one hand, developing standards is costly and most businesses have little or no medical knowledge to draw on. So even though we expect private actors to pick and choose which standards they use, it’s still government that supplies the menu. On the other hand, we have already said that private standards evolve. But that’s only possible if private actors know how well they’re doing at any given instant. In the end, their success is only as good as government’s capacity to provide timely, accurate, and above all clear evidence of where the epidemic is headed.
Stephen M. Maurer
The Berkeley Blog