A famous joke about economists has a physicist, a chemist and an economist stranded on a desert island with but a single can of beans to eat and no way to open it. The physicist sets about calculating the right angle at which to throw rocks at the can so as to pierce its top. The chemist tries to figure out how to variously sit the can in the sun and then douse it in sea water so as to make it susceptible to pressure. The economist interrupts their calculations with “What’s the problem? First, we assume a can-opener.”
It’s not a fair joke. We economists are actually much more practical than many other “social” scientists. And in one important respect we’re also more practical than natural scientists: In some of their often ill-advised forays into policy recommendations many natural scientists seem completely unfamiliar with the most practical notion of all: “cost.”
Be that as it may, there’s a bit of “assume a can-opener” in a number of scenarios economists are proposing for getting ourselves out of the current social and economic lockdown, whose costs may or may not be rising as time goes by but are becoming more apparent daily.
The scenarios, which usually involve a gradual return to — let’s not say “normalcy” but just “robust economic activity” — usually assign a crucial role to an assumed can-opener, in this case, cheap, quick and easy COVID testing.
There are two reasons testing is key. The first to see how safe a relaxation of restraints would be. If lots of people have been exposed to the virus, we’re closer to “herd immunity” than if they haven’t been. No one knows what immunity requires. One hundred per cent would obviously work. But lower numbers — Angela Merkel’s much-quoted 70 per cent is a number many people now use — would do pretty well. (Merkel earned a doctorate in chemistry in an earlier life but is not an epidemiologist. Her credibility on immunity emerges only because she heads a government that has presided over COVID-19 numbers that are pretty good so far.
If widespread testing finds most people have had the virus and we’re already near a cumulative 70 per cent infection rate, well, it’s much safer to go back in the water. But if we’re more like 10 per cent, relaxing restraints means lots more people will become infected and that may put the health care system under threat again—though with every passing week most countries’ health care systems should be expanding their capacity while their manufacturers replenish and even add to stockpiles of health equipment and supplies.
The second reason widespread testing is important is to enable policing—figurative “policing” though maybe also literal policing by officers of the law—the post-lockdown regime. New York University’s Paul Romer, who won the Nobel Prize in economics in 2018, demonstrates this on his website with some interesting little “dots in a box” simulations of the disease’s course. He uses red circles to represent people who are infectious, blue triangles for people who have never had the disease and are therefore vulnerable, and purple squares for people who have had it and recovered. (He could use little crosses to include people who have died from the disease but they’re not actually germane to what he’s interested in finding out so, on the good economic principle of keeping things as simple as possible, he omits them.
Then he sends off 200 of these figures to wander randomly about their imaginary community—which is a rectangle on your screen. At each instant of the simulation he counts new and cumulative infections, graphing them on separate charts. After he plays the simulation 50 times, getting a slightly different result each time (as is the nature of randomness), he develops a pretty good idea of how this little world behaves. The curves generated look something like those we see in the papers every day. But in fact verisimilitude is not his goal. Intuition is.
Next he runs simulations assuming cheap and easy testing so that people who have had the disease are let back into the economy, along with people who test negative, while people who test positive for it, even if they are asymptomatic, are quarantined. There’s obviously some danger in letting negative-testers back in as they’re still susceptible to infection. But the danger is reduced if the only people they work with are immune from having had the disease (assuming that does make you immune, which people seem to think is the case though contrary evidence in South Korea is being studied intensively). In effect, you’re creating a special, smaller herd to run the economy, one in which the danger of infection is low.
What results do you get from this kind of regime? Much better than with no control, as you’d expect. Overall rates of infection and immunity build up more slowly, which means health care is less stressed and more people make it to an eventual vaccine without having had the virus. And on average only 10 per cent of the population is in quarantine at any one time. It may not be easy to run an economy with a 10 per cent absentee rate. But it’s almost certainly possible.
Finally, Romer runs the simulations using random isolation—what you would do if you had only limited access to testing—which is basically what we’re doing now. Not surprisingly, you slow the rate of infection and reduce cumulative infections compared to the no-control situation. So that’s good. But to get the same lowering and flattening of the curve as with widespread testing, 50 per cent of the population on average has to be in quarantined at any one time. Run an economy with 50 per cent of people absent? Hard to believe that would be possible.
Romer stresses that these are just “toy” simulations. He doesn’t claim any of the assumptions he makes to get them to run are realistic. And he’s not wedded to the exactness of either 10 per cent quarantine vs. 50 per cent quarantine. What is important is that the difference between the two is big even though both accomplish the same degree of curve-bending. Also: that what gets you the better outcome—same health effect, many more people available for economic activity—is testing.
In the version of the model with testing, Romer tests the entire population roughly every seven days. In the real world, there are 325 million Americans. Testing every one of them every two weeks means 23.2 million tests a day. That’s probably more tests than have been administered in the entire world since the virus became known. According to the running compilation by “Our world in data,” it’s 10 times all the tests the US had done up to April 9th. For Canada, it would mean a somewhat less challenging 2.6 million tests a day. Even so, that’s seven times all the tests we’ve done since we started testing (according to the same source). So it’s a lot of testing. And probably more testing than is strictly necessary. As always, in deciding how much testing to do it would be good to figure out the marginal benefit of the next test and compare it to the marginal cost and to stop where marginal benefit equals marginal cost. In the absence of cost and benefit data, a reasonable guess is that those curves won’t cross until we have done a lot of testing.
Economists are pretty much agreed on the importance of testing. The Chicago Booth School of Business’ IGM survey of leading American academic economists recently had 93 per cent of them agreeing or strongly agreeing with the proposition: “Required elements for an economic ‘restart’ after lockdowns include a massive increase in testing capacity (for infections and antibodies) along with a coherent strategy for preventing new outbreaks and reintroducing low-risk/no-risk individuals into public activities.” One of those agreers, James Stock of Harvard, a world-leading econometrician, emphasizes in a short web post of the same name, “random testing is urgently needed.”
Only two per cent of the survey respondents were “uncertain,” which is actually just one person, 2017 Nobelist Richard Thaler of Chicago Booth, who explained: “My gripe is with ‘required.’ Important yes but would I hold up restart if cases are low but tests are still rationed? No.” Unfortunately, that may be the decision we face as we move into May and the economic and social costs of lockdown do begin to grow.
Until then, however, even Nobel Prize-winning economists are essentially united behind the proposition that something that does not yet exist—the COVID equivalent of a desert-island can-opener—is crucial to getting the economy back to something resembling where it was six weeks ago. A saving grace for the profession’s reputation is that a cheap and easy COVID test is not nearly as magical an instrument as the desert-island can-opener. Today’s policymakers have access to a world economy that until they slammed on the brakes a month ago was producing $US80 trillion worth of goods and services a year. With just a tiny percentage of that productive capacity—paid for by a small percentage of all the money suddenly being spent on supporting incomes and businesses—we’re likely to be able to devise an effective COVID test and scale up its output into the tens and even hundreds of millions. Romer argues that, compared to the $US2 trillion of “palliative care” that the US Congress has voted for the economy, spending $US100 billion on testing would make us “far better off.”
In a catchy new phrase that may already have become a meme: “What’s a hundred billion?” Economics is usually about hard choices. But scarcity is not our current problem. Real economic resources are rapidly becoming available on a very large scale. So why not do both? Provide financial palliatives and produce millions and millions of tests.
Of course, ramping up a cheap and easy test will take at least some time. But if we assume time moves infinitely quickly,…