As we move into the next phase of the covid-19 crisis, trying to restart economic activity before the shutdowns wreak overwhelming damage, it is essential that we learn from the experience of the last few months to avoid repeating some key mistakes.
Below I highlight what I see as the ten key lessons and their implications for our next steps.
1. Let’s be clear on what the goal is
The stated goal of the covid-19 strategy was to “flatten the curve”. That meant spreading out contagion over a longer period, avoiding the “peaks” that would otherwise overwhelm the health care system. Then we seem to have fallen in love with the numbers: policymakers took to competing with one another on the numbers of cases and praising their citizens for driving the numbers down, and the goal somehow morphed into driving infections as close to zero as possible—but that cannot and should not be the objective.
The goal cannot be to let the human, health and economic costs of the shutdown rise exponentially while we wait for a vaccine that might take 1-2 years to manufacture at scale if we are lucky. Having empty hospitals forced to furlough doctors and nurses while people die untreated of non-covid conditions should not be saluted as success—it’s an abject failure.
The goal should be to contain contagion within the capacity of the health care system while protecting the most vulnerable and getting the rest of the country back to work. With hospitals across the country sitting empty and a revamped system to produce and allocate personal protective equipment and other medical supplies, this is now a fully realistic goal.
2. The costs of the shutdown will rise exponentially
Economies across the world have been plunged into deep recessions. In the US, 30 million people have already lost their jobs—getting close to one-fifth of the labor force, with the pain falling disproportionately on those with lower skills, lower wages and lower savings. I argued a month ago that the economic and human costs of the shutdowns would rise exponentially after a point as more businesses go bankrupt and temporary layoffs morph into long-term unemployment. The point of inflection is getting closer.
3. The economic shutdown kills people too
In the initial response phase we have largely ignored the adverse health impact of the shutdown; Now this is openly recognized even by the shutdown’s strongest supporters like the UK’s Professor Neil Ferguson. The lockdown kills people through the lengthy postponement of essential screenings and treatments for a range of serious illnesses, the people dying of heart attacks because they are reluctant to go to the hospital, and the suicides and drug overdoses that will come with a deep recession.
We should factor these human costs into our decisions.
4. The models have failed us
The performance of the covid-19 projection models has been embarrassingly poor—and their poor performance can be seen in real time, day by day. The IHME model that the White House heavily relied on predicted 160,000 deaths in March, then it revised it down to 90,000; to 80,000; to 60,000; back up to 72,000…and now nearly doubled to 135,000 because some states are relaxing social distancing measures. A Cornell University study found that the model’s next-day predictions fell outside their 95% confidence interval 70% of the time… Let that sink in for a moment. This model gives an enormously wide range of possible outcomes (the 60,000 projection meant “with 95% probability it will be between 30,000 and 130,000”). The promise is that while the interval is enormous, reality should almost always (95% of the times) fall within its limits. Yet it almost never does.
If I offered you an economic model that gave projections like: ‘next year’s GDP growth will be between +3% and -13%’ and still was wrong two-thirds of the time, would you buy it?
The IHME projections of hospital resource use have also been hugely off the mark, fueling panic over potential shortages that were never even approached outside of a very few hotspots.
The IHME model is a purely statistical model—it does not simulate the spread of contagion in a population of Susceptible, Exposed, Infected and Resistant (recovered and immune, or dead) like standard epidemiological models. But let me be clear: those are not impressive either. The Imperial College London model forecast the UK would suffer 250,000 deaths with the initially announced policy of moderate social distancing; the government decreed a full lockdown and the forecast changed to 20,000 which, Prof. Neil Ferguson admitted, was partly because of the lockdown but largely because they had underestimated the health care system capacity (he added that two-thirds of the 20,000 would probably have died anyway even in the absence of covid). UK recorded fatalities are now around 30,000.
The models’ poor performance is understandable: we don’t know most of the key parameters, we don’t know how exactly social distancing works; the underlying data are very poor. But this is no excuse, and the models’ appallingly poor performance has to be acknowledged. We should stop using these models to scare people and to give an appearance of scientific certainty which has little basis in reality. All this will do is hinder the policy response and further undermine the public’s confidence in the models themselves (a lesson we should have learnt already from Brexit, where economic models confidently and wrongly predicted a massive recession).
At the beginning, when we had no data, the temptation to rely on these models was perhaps excusable. Now we have data—and as Dr. Fauci, who helps guide the US strategy, likes to say, the data trump the models every time. So let’s focus on the data and the evidence, and take the models for what they are worth—which is not much.
5. The experts disagree and don’t have all the answers
“The experts” are not a monolithic and infallible entity. They know a lot, but not everything; they make mistakes; they disagree with each other. Take these twin interviews with a top UK epidemiologist, Professor Ferguson, and a top Swedish epidemiologist, Professor Giesecke. They disagree on almost every key aspect of the covid-19 crisis and the needed response—and yet they are both highly regarded subject matter experts.
We must allow for an open debate that acknowledges the experts’ different views. Hiding behind our favorite expert and branding anyone who disagrees as “anti-science” is not helpful.
6. More testing will get us better data, not eradicate the virus
Most countries have ramped up viral testing (to identify people currently infected), and the first rounds of serology tests are beginning to give a sense of how many people had previously been infected. Over time, more testing will give us a better measure of how many people have already been exposed to the virus, and of the true fatality rate. Greater testing capability will make us better able to identify and contain any new clusters of contagion.
But we should not look for a magic number of daily tests that can help us drive contagion to zero while we reopen the economy—otherwise we are back to step 1.
7. Protect the most vulnerable and focus on clusters—one size does not fit all
We should accept the large degree of uncertainty we still face and use common sense. We know that about 80-95% of covid victims are above the age of 65, so let’s figure out what we can do better to safeguard them—and what we have done horribly wrong in the countries and U.S. states where care homes account for half of all deaths or more.
We know that clusters of heavy contagion can overwhelm the local health care system. So let’s implement lockdowns in the most affected areas. But a moderate degree of social distancing seems enough to stop the virus from spreading like wildfire throughout a country. The rest of Italy did not follow Lombardy’s fate. US states that implemented less draconian policies did not suffer more—in fact in some cases they suffered less.
The greater New York metro area, defined as New York State, New Jersey and Connecticut, has a mortality rate of close to 1,000 per million population (as of May 6); the rate for the rest of the US is about ten times lower—closer to Germany, which is widely acknowledged as one of the best success story among Western countries.
With some areas being hit much harder than others, we have no evidence that we need to apply the same draconian restrictions everywhere—especially given their exorbitant human and economic costs. In fact, whenever you point out that places like Sweden or Florida have implemented milder restrictions and yet have not suffered more, people immediately say “ah, but things there are different”. Very well. But that proves my point. So let’s stop demonizing policymakers who deviate from the path of maximum restrictions.
8. Those who can work from home should keep working from home
Another common sense consideration is that if there are low-cost steps that we can take to reduce contagion risk, we should take them. For example, those who can work (almost) just as well from home, should keep working from home. This will help reduce contagion risk while those who cannot operate remotely go back to their workplaces.
9. Rethink the risk profile of different activities.
Most strategies to reopen the economy assume that an activity that implies closer proximity to the customer is more dangerous and should be avoided. Even hyper-restrictive California has agreed to reopen nurseries; however, it still thinks that allowing hair stylists to operate would mean granting them a license to kill. Yet an 80-year old garden enthusiast might well risk more at a nursery than a 30-year old architect entrusting her head to a masked 20-something hair stylist.
If we accept that we should focus on protecting the most vulnerable—in an intelligent way—we need to look at the risk of different economic activities through a different lens, not purely based on physical proximity.
10. On-again, off-again is not an option
Carefully reopening the economy is an urgent priority. But it’s even more important that once we decide to reopen it, we stick with it.
This lockdown has already caused massive uncertainty for the future that will make the economic recovery harder. A second lockdown might have devastating consequences on people’s determination to keep a business alive or to keep hoping and searching for a job.
We should reopen gradually, keep testing, and monitor the situation while maintaining stricter restrictions and protections for the most vulnerable. We should be prepared to see a pick-up in cases—since almost everyone has been in isolation for nearly two months, this is almost inevitable. But we should not panic as long as the health care system is able to cope.
The decision to reopen the economy should come from having weighed the costs of a prolonged lockdown against those of somewhat higher infection levels. It should not be driven by the hope that the virus will fade away combined with fear for the rising frustration of voters. And the media should stop its fear-mongering that policymakers who allow economic activity to reopen are sacrificing lives for profits (see lesson 3.).
And some hard long-term issues to consider
If we really think this crisis will change the way we live, there are some very serious long-term questions we need to get our head around—but we seem very reluctant to discuss them. Here are three.
Public transport
Public transport has been designed to collapse social distancing in the name of efficiency, speed and sustainability. As a consequence, it appears to be one of the most efficient ways to spread contagion. New York is a case in point.
If, as some experts argue (including Professor Ferguson), some degree of social distancing will become a permanent feature of our societies, then we need to rethink public transport completely. A public transport that complies with social distancing will be very different.
Big cities
Metropolitan centers like New York and London are built on crowded subways and buses, packed elevators and crowded sidewalks. It is not at all clear how they will survive in a world of social distancing. And since big cities are accelerators of innovation and economic growth, that would have bigger implications. A corollary: Silicon Valley’s status as premier innovation hub would be reinforced, as its main competitors are urban centers.
Public health
Public health turns out to be a public good in a brand new way. I argued we should focus on protecting the elderly as they account for 80-95% of fatalities. But those who push back against such a targeted strategy point to studies arguing that as much as 40% of the US population is at high risk from covid, because of old age and / or comorbidities. Since people aged 65 and older are 16% of the population, that leaves about one quarter of the US population that is young (in the sense of not old), but at risk because of various heath conditions.
In many cases, these health conditions are related to behaviors such as nutrition physical activity and smoking. If having one quarter of the population in poor health significantly raises the risk of collapse of the health care system in a pandemic and forces us into a full lockdown with disastrous consequences of its own, then these behaviors have much more serious negative externalities than we thought. If so, we need to figure out how to correct them, likely with a set of incentives, disincentives, taxes and subsidies.
If you object that such measures would be intrusive and paternalistic, well, they are less so than the current regime of indiscriminate extreme social distancing, and they would have better long-term outcomes.