– July 17, 2020
open gate

Paul Romer’s plan for reopening America (his title, not mine) is a call for vastly increased testing, so we can find and isolate people infected with the novel coronavirus, in order that we can expand economic reopening and make the economy as functional as possible until such time as a vaccine is approved.

I’m for increased testing, I’m for figuring out how to manage this pandemic, and I’m also for opening the economy as much as we can with an eye to the marginal benefits of economic growth and the marginal costs of death and debility from COVID-19. But from a policy analyst’s perspective, his plan is fatally flawed. In fact, it is such a superficial top-down approach that I do not think that it can be properly called a plan at all.

To summarize, the plan calls for testing every American every two weeks and isolating those who test positive. With a population of 328 million, that means 164 million tests a week, or 23.4 million per day. Over a year that’s 8.5 billion tests. That’s a lot of production, but perhaps less than it seems at first. It’s dwarfed by the number of plastic bottles made each year, for example. So the only real production worry is the ramping up. (OK, there’s also the displacement of other goods, but for the argument here, let’s assume that we know making tests is worth the opportunity cost.)

Romer proposes to get to that level of testing by a) spending more money to incentivize testing labs; b) removing regulatory barriers to test production; c) establishing a network of national and university testing labs; d) expanding testing in stages, beginning with essential workers who are most likely to be infected, then as production ramps up successively extending testing to other essential workers, to everyone who needs to return to work, and finally to the rest of the public; and e) isolating those individuals who test positive until they are recovered.

Let’s consider the good parts of the plan. Romer expects it to cost $100 billion a year, based on an estimate of $10 per test and 20 million tests per day (7.2 billion per year), so a cost of $74 billion per year, plus “other costs for infrastructure, training, and production of supplies.” He notes that economists have estimated the costs of lost economic productivity to be $500 billion per month, which means the cost would be recouped in a matter of days once the economy was largely reopened. His math is marginally understated, but not enough to challenge that claim. If testing everyone really means testing everyone, then we’re talking about 8.5 billion tests per year, adding up to $85 billion for the tests, plus the other costs. That is still quickly recovered by economic gains.

We can extend this farther by noting that the size of this proposal puts it into the realm of megaprojects, which are notoriously over budget, frequently tripling in cost from the first estimates. San Francisco politician Willie Brown famously said of the disparity between a megaproject’s proposed costs and the real cost, that of course they lied to the public, because the true price would have scared people away. He even implied that the public was at fault for believing what they were told.

I don’t wish to accuse Romer of intentionally misleading people, but for marketing purposes, perhaps he has unconsciously focused on the lowest cost estimate. So following the megaproject rule of thumb, let’s say the real cost would be $300 billion. Even then we would be talking about cost recovery within weeks, so on the assumption that universal testing would allow for economic reopening, the worst case cost estimate still favors this plan. 

The plan also shifts the focus of testing away from symptomatic people toward regular testing of asymptomatic people, beginning with healthcare workers and first responders, so as to identify and quarantine infected people more quickly. Symptomatic people are to self-quarantine automatically, and people needing medical treatment should be assumed to be infected and treated accordingly without testing, reserving the tests for the asymptomatic. Apart from the data loss of not identifying some COVID-19 cases, this is a good strategy for a serious pandemic, as it appropriately prioritizes testing needs.

Romer also proposes reducing regulatory barriers to test production. This is not a new idea, but is an obvious and necessary inclusion in any serious pandemic testing plan. It is so important that it ought to be a default position for any proposal.

However, despite these good elements, the plan is fatally lacking in substantive detail and attention to real-world human behavior. In critiquing it, though, I want to be clear that I am not critiquing the general goal of increasing testing. Whether that is a good idea or misguided is outside the scope of this essay and of my expertise. It is also possible that in the time it takes to get Congressional approval of spending for such a plan and then get the testing ramped up a vaccine may be developed, making the plan largely irrelevant. But that is an unknown, so for the purpose of my argument here I share Romer’s implicit assumption that an effective vaccine is not on the near horizon.

Romer proposes to establish a network of national and university laboratories to do the testing, noting that there are 36 such laboratories equipped to do this testing, and recent research indicates that each could potentially “scale up to millions of tests per day.” That sounds promising, but if the goal is to scale up as rapidly as possible, why begin with such an artificial restriction? Private labs are already doing testing, so why wait for the national and university labs to scale up instead of including this other existing and expandable capacity?

Writing at Harvard Business Review, Nikhil Bhojwani and Atul Gawande argue that the testing network should include “national chains with many retail locations,” local health providers, and even ambulance companies. We should bear in mind also another rule of thumb of megaprojects; they tend to take twice as long as predicted. Despite Romer’s confidence in these labs, a cautious policy analyst would suggest we plan for the scaling up to take longer than predicted. The private sector side of the testing network would alleviate any bumps in the scaling up of national and university labs. This detail sets Romer’s plan at cross-purposes, undermining its declared purpose.

In a related issue, Romer makes the common policy advocate’s mistake of ignoring the principal-agent problem. Even given capacity and payment for the tests, what incentivizes these organizations to operate in a way that meets the needs of the tested individuals? I have a friend who is moving out of the country and has to be tested within 72 hours of his flight’s departure from the U.S. He has had to reschedule his flight because the rush analysis that he paid for did not happen in a timely manner. Mistakes are made, of course, but his attempts to get an explanation from various participants in the process have been fruitless, with each giving him a different excuse. The most straightforward explanation is that his problem is not their problem.

Romer’s implicit assumption of a rapidly expanded but smoothly operating process waves away such problems, even though they are familiar to economists. Private sector labs are not immune to this problem, either, but the more labs that are available, the easier it is for individuals and firms to shift away from the unsatisfactory ones to those that better meet their needs. This is especially true as Romer proposes bi-weekly testing for the indefinite future, giving test consumers a strong incentive to shop around.

But what really makes this plan problematic is the absence of any mechanism for getting people to get themselves tested other than the simultaneously vague and threatening “implementation and enforcement.” That is to say, this “plan” is not a plan; it’s merely a goal, and setting goals is much easier than making meaningful plans to achieve them. Romer calls this a “simple roadmap,” but it is more than simple; it is simplistic. It makes the typical mistake, far too common among policymakers, of ignoring the actual individuals whose compliance is necessary: the 328 million people of the United States.

One of the rules of public policy that I tell my students is to make policies for the type of troublesome humans you really have, not that ideal type you wish you had. This “plan” is typical of expert planners, all top-down and with the assumption that the bottom level simply complies, or if they do not, can be simply compelled. If this were true, we’d all obey the speed limits and Prohibition would have been a smashing success. The bottom-level people envisioned by such plans, despite the pretense of concern for their well-being, are not real homo sapiens but automatons, mere clockwork images of people.

We can distinguish between Romer’s reasonably well-defined groups based on their incentive to comply with a testing regime. Curiously, Romer’s scheme is built on levels of risk, yet he fails to account for people with lower risk having lower incentives to comply. Front-line health care workers are likely to be eager to comply because their risk is greatest. Other first responders, such as police, paramedics, and firefighters are somewhat less likely to do so, although perhaps still willing. They are also employed in hierarchical organizations that can demand compliance with relative ease (assuming their unions don’t object). Other essential workers, such as grocery store workers, delivery drivers, and so on are at less risk, and so have less incentive.

Again we can work through employers for many of them, but their employers are also aware of the diminished risk. Then there are those who can work from home, the self-employed, those employed under the table, and the non-employed. Here we lose ability to work through employers, and the incentive to make the effort to get tested declines even further. Romer’s only suggestion here is “enforcement,” which is implicitly treated as costless. It is not, of course, and he makes no effort to address the cost of enforcement on this margin to the marginal gains.

We should also bear in mind that the cost of enforcement, as with mask and social distancing enforcement, is likely to fall disproportionately on the poor and minorities. It is past time for all serious policy advocates to consider the class and race consequences of their proposals as a matter of standard practice.

Another factor related to compliance is ideology, and this also is ignored by Romer, who treats the public as an undifferentiated mass. A substantial proportion of the public believes coronavirus is not serious, or even that it is a hoax. How do you persuade them to comply if they have means of evasion? It is likely that up to a third of Americans would resist the scheme out of the belief that it’s a foolish overreaction or a government plot to strip away their liberties (and we don’t really know what forms enforcement might take, so that fear is not entirely without basis).

Finally, if this regime has to be conducted for an extended period of time, then we will get dropout, which will increase over time until it reaches the limit where only those who subjectively feel at great risk or those who can be easily compelled are being tested. This is certain. Longitudinal research studies consistently suffer dropout of participants. Ballot and survey fatigue, when there are a large number of ballot options or too many survey questions, is a known and regular phenomenon.

Consider, for example, a small business owner who requires tests for his employees, and after two months has zero positive results while she sees the local case rate decline. Eventually the bother of requiring her employees to test will outweigh her expected gains. As an economist, Romer knows about marginal costs and benefits, so it is a mystery why his plan fails to account for them. I suspect his purpose is to get the ball rolling, and to count on smart people to solve all the problems as they arise. But as with all planning, the more we game out the problems ahead of time, the better prepared we are to deal with any that arise.

The weakness of the plan is also revealed in Romer’s inept Twitter responses to critique. In one he said, “When you strip away all the noise and nonsense, note that once we cover essential workers, it’s easy to test everyone in the US once every two weeks. Just do it.” But “just do it” is a marketing slogan, not a plan.

As long as Romer refuses to think about the details of getting everyone tested, it may seem simple. But if he could bring himself to the task of a deeper dive, he’d be less sanguine. Romer is making the very common mistake of assuming that if we just have the political will we can achieve it. But as economist Christopher Coyne notes, “will” is an empty term, undefined and unidentifiable. In other words, it’s just a post-hoc excuse for policy failures. The (dis)organizers of the Fyre Festival certainly had will, but what they didn’t have was a grasp of the details, resulting in their catastrophic failure. French soldiers in World War I may have had greater esprit de corps, but it didn’t prevent them from getting bogged down in trench warfare.

In another tweet he said, “When someone tells you that ‘we could never test that many people,’ ask ‘ok, so your plan is to stand by and do nothing as asymptomatic spreaders kill their colleagues?’” This response avoids addressing the question and engages in moral grandstanding. To even question the missing details of the plan is, apparently, to be unconcerned with human suffering. This is the stance of the elite expert who expects the hoi polloi to not question their brilliance. But it’s obvious where the limit of Romer’s expertise is, and it is exactly where citizen-dependent public policy begins. 

We should never simply condemn when we can help, so here are some suggestions to build on, rather than destroy, Romer’s plan.

1. Set a realistic goal. Universal testing, or anything close to it, is not a reasonable goal, both because it is technically quite impossible and because at some point the marginal costs outweigh the marginal benefits. What level of compliance do we actually need to be successful? I don’t know, but epidemiologists should be able to identify some more reasonable target, where testing is sufficient to keep the caseloads low. As a first and obviously rebuttable suggestion I would suggest thinking about 70% because that is the number I have seen booted around for achieving herd immunity. That in itself would be a daunting enough task.

2. Incentivize employers to require testing of their employees. Notably, Romer specifically talks about incentivizing the testers, but not the testees. It’s true that employers will benefit from being able to reopen, but what if that is not sufficient to get to our testing target week after week? The employer’s monitoring and enforcement are business costs that ideally they would prefer not to bear, especially as the benefit is not wholly internalized. An employer who has no positive cases among their employees for a month or two may reasonably decide the cost is not worthwhile. The standard response of policymakers is to mandate that employers test, but this requires monitoring and enforcement costs. But has he looked at whether the enforcement costs would be less than the cost of incentives, or has he simply assumed that question away?

3. Incentivize the self-employed, contract workers, and the nonemployed to get tested. This is your hardest group because we can’t work through employers. Setting aside the issue of ideology – which, unfortunately, may be intractable (although at some price point many will comply despite their skepticism) – this is the population that can keep us well short of universal testing, or even any more reasonable goal. Particularly, high-risk groups in this category should be identified and the greatest effort be put toward incentivizing them.

4. As noted in my critique above, expand the plan to include as many private testing organizations as possible. This minimizes the chances of backlogs and allows firms and individuals to establish working relationships that can make it easier to resolve.

5. Incentivize test-producers to find a way to reduce the individual costs of getting tested. The holy grail of testing is the home pregnancy test, which is available over-the-counter, simple to administer, noninvasive, and gives results within minutes with no expertise required. Whether a COVID-19 test of that sort can be achieved, I don’t know. But it should be our goal. As hockey legend Wayne Gretzky said, you miss one hundred percent of the shots you don’t take. And such a test does not have to be perfectly accurate, because the increased compliance will compensate for diminished accuracy. Even moving closer to this goal by accomplishing just some of its elements would be a huge win. Imagine, for example, if the test could be conducted quickly and noninvasively at your place of employment, with a moderately trained technician on site who could analyze the tests by the end of the work day. That would be a vast improvement over the investment in labs that Romer is banking on.

In summary, Romer has in fact not thought this plan through very deeply, and his response to critiques seems to be that they shouldn’t do so either. His is a superficial plan that relies on magical thinking, believing that if we just mobilize the resources and have the will, that great mass of people whom we actually need to take individual action will simply comply, and we can easily and unproblematically compel those who don’t “just do it.”

One gets the impression that from his years leading the World Bank, Romer has come to think of policy as wholly top-down, only as government action, conducted by official actors with perfect competency and full knowledge, having no need for the local and tacit knowledge of those they desire to control. This is typical of elite experts’ myopic tendencies, as noted by political scientist Roger Pielke, Jr., which is their tendency to have incomplete knowledge of the limits of their own knowledge. But his plan is a good general framework, in which more pragmatic detail-oriented people can fill in the details. Rather than attacking people for questioning the lack of detail, Romer should be inviting them to help fill it in.

James E. Hanley

James E. Hanley

James E. Hanley is the Associate Professor of Political Science at Adrian College. After earning his Ph.D. at the University of Oregon he spent a year as a Visiting Scholar at Indiana University’s Workshop in Political Theory and Policy Analysis where he studied under 2009 Economics Nobel Prize Winner Elinor Ostrom. He teaches courses in Political Science, Economics, and International Studies.

Get notified of new articles from James E. Hanley and AIER.