Known Unknowns


Welcome to Known Unknowns, a newsletter that is always Bayesian—and thinks that you should be, too.

Be a Bayesian

Last week, I went on The Indicator to talk about how to make a risk assessment about traveling for the holidays. I think we’ve all learned a lot about our feelings around risk and what we can tolerate over this past year. It has become popular to judge people for their behaviors, and we’ve seen how one person’s risk choice can impact their whole community. But people are going to go places during the holiday season—we don’t lock them up here. And shame and judgement won’t stop them from doing so. In some cases, it may even embolden them.

The fact that the virus poses a fairly small amount of risk for most people, but catastrophic risk for others, brings up pretty much every bias that behavioral economists write about. It’s no wonder that some people make what appear to be irrational or even dangerous choices. But I maintain that one of the biggest public health failures we faced was absolutely terrible risk communication.

For instance, many people lined up to take a rapid test before traveling for Thanksgiving last week. We were told repeatedly on the news that this wasn’t enough to keep you “safe,” the implication being, “Don’t bother testing, don’t travel, and spend Thanksgiving with your immediate family—and if you don’t have one, spend it alone.” That is certainly the lowest risk option, but the problem with that message is that it suggests that testing isn’t useful simply because it doesn’t eliminate risk entirely. Now, I am a proud Bayesian, so I think that a more helpful message is that testing is great and you should do it all of the time, because it reduces risk significantly—though not entirely. Maybe, if you travel to see your grandmother, there’s a 10% chance that you’ll infect her with the virus. If you get a rapid test beforehand, maybe there’s a 2% chance. Still not zero, but much better. So, if you must travel, get a test. That seems like a point worth emphasizing. Imagine if everyone got tested before they travelled, and then didn’t go anywhere if they were positive for the virus.

Instead, however, our public health officials seem to think that people can’t weigh different risk options and take the necessary steps to reduce risk. I’m actually heartened that the lines were so long, because it suggests that people are capable of being a Bayesian more than they get credit for.

This past year should have forced us all to rethink our worldview, and it has hopefully changed some minds regarding how the world works and what we are capable of. I’ve always believed that people can make better risk choices, but risk must be communicated in a sensible, natural way. In some ways, I still believe this, but there certainly have been examples of bias and inconsistent risk-taking. Lots of people are careful, but only until they don’t want to be anymore.

In the last several years, the idea that people can't understand risk has driven many policies. Some countries even have behavioral economists tasked with nudging people into making better risk choices. But in the last several months, I think the idea that people can’t understand risk has not been helpful. It is why pandemic management tends to be either draconian or way too permissive. A Bayesian would say there’s a big middle ground, where we arm people with data and information, along with reasonable, data-driven restrictions, and they’ll make better choices. Or, as Alex Tabarrok argues, we can’t let the most risk averse among us dictate policy.

This spring, we will all have to be Bayesian in our tactics. As the vaccine is rolled out, risk—at least for those at catastrophic virus risk—will be reduced, so we’ll have to constantly update our risk assessment with new information, and then behave accordingly. It may still be a long time before we’re completely safe, but we will be progressively safer very soon. And I don’t know what that will mean, but I expect dire warnings until the very end, and at some point, many people will start ignoring them completely.

The Labor Market

As expected, the jobs report on Friday was not great—not absolutely terrible, but certainly not great. With more shutdowns and increased virus fear, there are now fewer new jobs. And while the unemployment rate has fallen a bit, the share of prime-age people with a job has remained pretty flat. Considering what the economy is currently experiencing, that’s not too bad in the grand scheme of things. According to the New York Times, at this current rate, it will take 29 months to get back to where we were in February.

But I’m skeptical that we’ll get back to less than 4% any time soon. We might, of course—we may have a huge amount of growth next year. After all, this was not a normal recession. It was truly an exogenous shock, with what appears to hopefully be a definitive-ish end date.

But I’m a student of Ned Phelps, so when it comes to unemployment, I think it’s important to think about what’s cyclical and what’s structural. I’m less concerned about scarring workers, or with people getting fed up, because we’re all thinking that there will be a big change next year when things go back to normal. Nor am I worried that people will never fly or travel again—there may be lots of pent up demand there, which means that there’s money waiting to be spent.

But I do think that we’ve seen a big shock, with the fast adoption of existing technology that will quickly make some labor obsolete. It was already happening, of course, but gradually—and now it has happened very fast because of the sudden change of labor circumstances globally. It’s sort of like how when we opened trade to China, it was a much bigger shock that caused more disruption than we had seen before from traditional trading relationships. So I expect that many low-skilled jobs may not come back. Perhaps there is some scope for retraining, but there may also be lots of displacement that we don’t fully appreciate for a few years.

And as much as economics Twitter likes to deny it, monetary policy can't fix structural unemployment. I’m still hopeful that we’ll pass a stimulus deal before the end of the year. If we spend that money now, it will have a much bigger impact than it will months from now. And people and small businesses need some relief from their current hardship. But in the longer term, we need to start thinking about the structural issues, too.

Some links

My City Journal feature about the red state/blue state risk divide is no longer behind the paywall.

Greg Mankiw on interest rates

Merton on trust in finance

Until next time, Pension Geeks!