Today, six U.S. states will allow business to resume, joining 21 others that have already begun lifting restrictions put in place to contain the coronavirus pandemic.

Internationally, Italy begins phase two of its reopening today and next week France will lift its strict social distancing rules and allow schools to resume. Politicians from these places have generally argued the economic cost of an ongoing lockdown outweighed the risk to lives that a resurgent epidemic might pose. Few louder than President Donald Trump who's cautioned that “we can’t have the cure be worse” than the disease.

Elsewhere, though, politicians have made the exact opposite argument: When the British government extended its lockdown in late April, it said it was doing so not just to save lives, but to save livelihoods too: that extending the lockdown would reduce the chances of a second peak that would prove even more economically devastating. U.K. Prime Minister Boris Johnson has said he will only unveil a plan for a phased reopening the country’s economy later this week.

What the economists say

Which set of policymakers is right? So far, most economists who have modeled the costs of various policy responses to the Covid-19 pandemic side with the British: that an extended lockdown saves lives and is better for the economy in the long-run.

But, there’s a catch.

As always with economics, much depends on the assumptions built into the models. And some of these assumptions—about government’s ability to monitor infection rates accurately and about the behavior of workers and consumers—may be incorrect.

There’s also a vocal minority of economists who argue any lockdown is a mistake: it devastates the economy in exchange for very little benefit in terms of lives saved in the long-run. Moreover, they argue a more limited set of social distancing policies, such as those being pursued in Sweden, is closer to the ideal policy.

The debate stems in part from the fact that, given how unprecedented the situation is, there isn’t much established research on the economics of lockdown policies.

What the epidemiological models say

Emanuel Ornelas, a professor at the Sao Paolo School of Economics in Brazil, says most economists have only in the past six weeks started to learn about epidemiological models and how to incorporate them into their own economic forecasts.

The only similar event that has received any scrutiny from economists is the Spanish Flu pandemic, which began in 1918. One of the lessons of those studies is that timing is crucial: cities that imposed social distancing measures too late suffered economic pain, but saw little reduction in deaths.

Another lesson is that cities that lift social distancing measures early, almost always faced a second wave of infections, says Martin Eichenbaum, an economics professor at the Northwestern University in Evanston, Illinois. That’s because while lockdowns suppress the epidemic, they don’t usually reduce the infection rate to levels where it dies out completely. As soon as restrictions are lifted, infections climb rapidly again.

This second peak harms economic activity, even if containment measures are not reimposed, Eichenbaum says, both because more people become sick and die, but also because people become afraid to work and shop.

In a working paper Eichenbaum co-authored with fellow Northwestern economist Sergio Rebelo, and Mathias Trabandt, a professor at the Free University of Berlin, the researchers modeled various lockdown scenarios. In their research, an early end to containment policies initially brings about a big economic boost, driven by a 17% surge in consumption. But this boom also results in a jump in infections that soon plunges the economy into “a second, persistent recession.”

“Prematurely abandoning containment brings about a temporary rise in consumption but no long-lasting economic benefits,” the researchers write. In addition, “tragically, abandonment leads to a substantial rise in the total number of deaths caused by the epidemic.”

In Eichebaum’s most sophisticated model, for a country the size of the United States, lifting social distancing restrictions 12 weeks into an epidemic would cost 500,000 additional lives. Compare that to Eichenbaum's more optimal policy decision: gradually ramping up containment measures as infections rise. At the peak, restrictions imposed would inflict some pain—most notably, by reducing consumer and business consumption by 76% at week 32—but then the impact would gradually ease as infections wane.

Eichenbaum says there is also another big problem with relaxing social distancing measures too soon. If infections come surging back, requiring that social distancing be reimposed, “trust in policymakers is eroded, and then you get an added recessionary effect because the public from then on refuses to trust the government that it is actually safe to go back to work.”

Another paper by researchers from Goethe University in Frankfurt, Vivantes Hospital in Berlin, and the University of California at Berkeley, comes to a similar conclusion. “A non-trivial outcome of our study is that strong suppression strategies lead to lower total costs…when containment efforts are not relaxed with falling infection rates,” the authors write. “A short-term control approach of softening containment with falling numbers of new cases is likely to lead to a prolonged endemic period.” In other words, the illness, and its attendant economic pain, will continue.

One key insight of this study was that the best overall results came when policymakers tried to reduce the overall number of cases across the entire span of the epidemic, rather than simply trying to drive down the current rate of infection below a certain threshold.

Eichenbaum’s research indicated that a gradual ramping up of restrictions, culminating in a very strict lockdown at the height of the infection, followed by a very gradual easing of restrictions, worked relatively well. But there was an even better policy, which he calls “smart containment.” This involves setting policies for different people based on infection status—those currently infected would need to be strictly quarantined, while those who have been infected, but recovered from the virus, are allowed to resume normal activities. This is similar to the “immunity passport” idea that the U.K. and Germany have reportedly toyed with.

Ornelas says the easing of restrictions should not apply to all industries equally. He says it is better to start with those sectors that have the biggest economic impact, for both consumption and employment, for a given geography to achieve the smallest possible health impact. In some places, this may be easier than others: for instance, a location where tourism is the most important industry—critical for both consumption and wages — will have a harder time reviving the economy than one dependent on the operation of a few large factories that might be able to resume operation with social distancing measures in place.

Eichenbaum says the idea of different rules for different industries makes economic sense but raises “ethical issues,” since it means citizens lucky enough to work in some sectors will be spared further financial pain, while others may be forced into bankruptcy.

The importance of testing

In all these models, the biggest assumptions are that government actually has a decent view of infection rates that it can use to calibrate the severity and duration of social distancing restrictions. But that requires extensive and accurate testing. This is especially important when so many infected people may be asymptomatic, as they seem to be with the coronavirus.

Unfortunately, that’s not the reality we live in. The U.S. and the U.K, as well as France and several other European countries, have struggled to put enough testing in place. Most countries don't have the capability to conduct the sort of randomized testing that Eichenbaum says would provide policymakers with enough insights to set the least economic damaging policies.

Even countries with very strong testing regimes are struggling to achieve the kind of careful policy calibration that the economists assume is possible. Germany has one of the best testing regimes in the world, with the capacity to swab as many 800,000 people each week. Yet, after it began easing lockdown measures two weeks ago, allowing many small businesses to reopen, it found its reproduction number—a metric for the average number of other people each infected person passes the virus on to—shot back up, going from 0.7 to about 1, the point at which the epidemic begins to grow again. That has prompted the country to postpone plans to reopen schools.

The danger of lockdown policies

Given that lockdowns only suppress an epidemic and that the virus is likely to come roaring back once they are lifted, a few economists question whether strict restrictions are actually worth the economic damage they cause. Alexis Akira Toda, an economist at the University of California San Diego, is one of them. “People might think I’m crazy, but I am thinking through models and playing with these models and I always come to the same conclusion: that the current lockdown policies in most countries are probably wrong,” he says.

Toda says that, without a vaccine, the only way to end the epidemic is for enough people to become infected, recover and develop immunity that the transmission rate, the R, naturally drops below 1. The trick, Toda says, is to allow that to happen while shielding the people who are most likely to become severely ill—the elderly and those with pre-existing health conditions—and preventing the health system from being overwhelmed. This is similar to the herd immunity strategy that Sweden is pursuing and which the U.K. considered until new data showed its health system was unlikely to be able to cope with the case load.

To save the health system, Toda says, a strict lockdown might be necessary at some point during the epidemic, but he says to avoid needless damage to the economy, the most stringent lockdown should only be put in place when the epidemic is nearing its peak. He says in many parts of the world, including many U.S. states, lockdowns were put in place far too early.

Even more surprisingly, Toda says that many of the beneficial aspects of social distancing policies can be achieved with less damage to the economy by simply publicizing the level of infection in a given area.

In places with high infection rates, people will voluntarily stay home, avoid public transport and shun crowded bars and restaurants out of fear of catching the virus. Elsewhere, they will go to work as normal. That means economic activity won’t be as severely depressed as it is during mandatory lockdowns that cover entire regions, states or countries.

But again, this policy works best if there’s an effective testing regime, so government can publicize accurate infection rates for different areas.

All of the economists agree on one thing: except for very small countries, setting national policies probably doesn't make sense. “It should be decided at more local level, with each city or region setting a policy based on the level of infection in that area," Toda says.