NICK EICHER, HOST: It’s Wednesday the 20th of May, 2020. Glad to have you along for today’s edition of The World and Everything in It. Good morning, I’m Nick Eicher.
BRIAN BASHAM, HOST: And I’m Brian Basham. First up: recommendations for reopening.
EICHER: In mid April, the White House Coronavirus Task Force recommended states have a 14-day decline in COVID-19 cases before easing stay-at-home restrictions. Few have actually met that benchmark. But by Memorial Day—less than a week from now—every state will have started lifting limits on businesses and public gatherings.
BASHAM: With tens of millions predictably unemployed, the pressure to get people back to work is driving the push to reopen. Some governors say they know they’re taking a risk but one they say is both calculated and necessary.
But there’s also been some confusion about how to define a 14-day decline. Does an isolated outbreak of cases, say at a nursing home, mean states need to reset their count? And how should officials factor in increasing test capacity, which leads of course to more positive cases?
EICHER: Scott Ganz is a research fellow at the American Enterprise Institute and an assistant professor of public policy at the Georgia Institute of Technology. He ran the numbers to make his own recommendation for how states should interpret the data. And he joins us now to talk about it. Good morning, professor!
SCOTT GANZ, GUEST: Hey, thanks so much for having me.
GANZ: So, I think that when the White House Coronavirus Taskforce came out with this idea that we should be looking for 14 days of reductions in cases, people thought that seems pretty intuitive. We look at a line and if that line goes down for 14 days, then we’re all good. But then you realize when you’re actually looking at a line in real time and asking is that line really actually going down for two straight weeks, it’s not as easy as that. So, for example, if there’s just some normal randomness in the data, right, we might see a little blip upwards in the middle of a week. And then you need to ask yourself, well, do we now not have 14 days of declining cases? Or, if you want to think about it sort of in a different way, imagine all we’re looking at is is the data today less than the number was 14 days ago? We face the same problems where if it just happened to be the case that 14 days ago the data was a little bit higher or a little bit lower, you’re going to get very different answers. Luckily, thinking about this from the perspective of statistics, we’re pretty comfortable taking into account that kind of random variability in data and that’s what I was trying to do.
EICHER: How do you propose—I mean, look, I looked at your paper and it reminded me of some economics textbooks of many, many moons ago, but we’re explaining this as simply as possible. Without going into the full equation, how do you propose it? What’s your thumbnail rule?
GANZ: So, the way that I think about it‚—and I think this is consistent with the way that others who are looking at the data are talking about what we mean when we’re talking about 14 days of consistent decline is the data that we’re looking at should be well-modeled, sort of roughly approximated by a downward sloping—what I’m going to call a stair-step function, which I call in the paper a monotonically decreasing function. What that means is we should be able to draw a line through the data where that line, as time goes along, it moves down. It doesn’t need to move down at a constant rate—it can be going down at a slow rate, a fast rate, it could be going down slow at the beginning and then a little bit faster and a little bit slower. But sort of over the course of the 14 days, we want that line trending downward. And for me, I think that’s what I look for in the data and that’s actually what I think that most people look for when they think about “What are we talking about when we mean 14 days of consistent decline?”
EICHER: Now, the model that you created, you applied it to some of the data that was publicly available. Can you just talk briefly about what you found and how states are applying the data we have?
GANZ: Yeah, for sure. So, I applied this criterion—what I’m going to call a monotonically decreasing criterion—to eight different states. And I chose those eight states just purely because I thought the data was good for those states. You could apply it to more states if you wanted to. And I asked the question: Do positive test rates—so think about that as the ratio of the total number of positive cases divided by the total number of positive tests—does that have this consistent decline? Is it well characterized by a downward sloping stair-step function? And the answer I found was that in New York it appeared to be the case and in the other seven states that I analyzed, the data didn’t seem to say that. The data was sort of better characterized by a different shape.
EICHER: Well, now, as I mentioned, more or less, every state is going to be opening or at least beginning to reopen by Memorial Day. So, are you saying that there are some states that are acting in rash ways? Are you concerned about all of that? And, maybe the better question is, how does this guide policymakers?
GANZ: So, I would say to start, I’m more concerned actually with the state of the data. So, if we think about all of the different variables or the different sort of indicators that we might want to look at when it comes to deciding when should I reopen my state? We could think about how many cases are there. We could think about what is the overall rate of infection in the population. We could think about conditional on getting infected, what is the probability of having a serious negative reaction or even dying? And between different states, they’re also sort of on the nature of the whole, I just don’t think we have a really good grasp of what those numbers are saying.
And so as a result, we’re sort of asking policymakers to make a decision with their gut when we would hope that instead they would be able to make decisions based on objective analysis.
EICHER: And speaking of which, isn’t it a little bit arbitrary even to say, well, we’re going to draw these lines and we call these state lines? Wouldn’t it be better, really, county by county, or city by city, area by area because, I mean, outstate Missouri is not the same as St. Louis, Missouri, for example.
GANZ: So, that’s absolutely right. Right, so the way that the policy response has been organized in the U.S. is on a state-by-state level and so as a result you’re seeing some states treating some counties different than others, so in Virginia for example they’re opening up the areas of Virginia outside of the D.C. suburbs earlier than the areas in the D.C. suburbs precisely because those two regions are very, very different in terms of both the intensity to which they’ve been affected by COVID-19, but also in terms of sort of simple things like the density of the population. But, nevertheless, the question of whether the state is the right unit to be managing the response is a really good one and one that my understanding is the U.S. is quite unique in terms of managing the crisis sort of in this individual way.
EICHER: That kind of vindicates the idea of federalism, I mean, just sort of saying we’re not going to have a top-down solution. Local policymakers need to do the best they can because there is a lot of gut going on here, right?
GANZ: So, that’s definitely the approach that we’re taking, right? And it has some benefits and it has some costs, right? One of the benefits is that you do at least have the opportunity to have solutions which are tailored to the local environment, but there are huge costs, right? If we think about the fact that people, for example, can easily travel across state lines, that means that we have a public health response that’s being managed on a state level, but the pandemic is global but definitely national.
EICHER: Scott Ganz is a research fellow at the American Enterprise Institute. He’s also an assistant professor of public policy at the Georgia Institute of Technology. Thanks for joining us today!
GANZ: Hey, this was great. Thanks so much for having me.