Covid-19: What’s Next?

By Matt Gordon

This is an internal memo written between April 21 and April 28. Some Behaviornauts asked to share it with friends and family, so it’s being published with minor changes for clarity: links to Instrumental graphs about the spread of Covid-19 now link to our public graphs instead of our internal project page.

Black Swan

The problem with a black swan is that there’s no precedent to guide decision making and increase confidence you’re doing the right thing. Instead of littering this document with uncertainty phrases, please assume that everything I say is prefaced by one or more of “Based on what I could find”, “With limited data”, and other common phrases that mean there’s a lot of uncertainty about what’s happening.

I welcome new sources of information and alternative interpretations.

What’s this article about?

This article is about figuring out what’s next. That’s a big question to answer, so consider this a seed for discussion with a goal of generating a common knowledge base. Primarily, it’s intended to help us figure out what’s next for Expected Behavior, but it’s also intended to be helpful for figuring out what’s next for us as people.

To have a conversation about what’s next, we need to make sure we have a shared understanding of the current state of the pandemic. First, I’ll talk about the medical and economic facts. After that, I’ll talk about what they could mean for Expected Behavior and for you.

Medical

There are only a few ways to mitigate the health effects of the pandemic.

• Mutation — The virus could mutate into something less lethal.
• Vaccination — We could develop a vaccine that renders the virus harmless.
• Treatment — We could develop a method for treating the disease if it becomes serious.
• Herd Immunity — Enough people could get the virus and develop antibodies that it can no longer spread effectively.
• Avoidance Behavior — We could use things like social distancing and PPE to reduce the spread.

Let’s look at each of those mitigations and some related information.

Behavioral controls will be diminished or eliminated soon

As you’re likely aware, President Trump has started advocating for reopening the United States. There are many different plans, but they have a lot in common. I’ll stick to the Federal guidelines.

The guidelines emphasize certain standards to “mitigate risk of resurgence” and “protect the most vulnerable”:

Develop and implement appropriate policies, in accordance with Federal, State, and local regulations and guidance, and informed by industry best practices, regarding:

- Social distancing and protective equipment

- Temperature checks

- Testing, isolating, and contact tracing

- Sanitation

- Use and disinfection of common and high traffic areas

- Business travel

There are problems with each of these, which I’ll discuss below.

Social distancing will soon stop.

By Phase 2 of the federal guidelines for reopening, social distancing will no longer have serious government support. Groups of 50 people will be permitted to gather. Non-essential travel will be allowed. Bars and gyms will be open with “diminished standing-room occupancy” and “strict physical distancing”.

Gym rats aren’t famous for their courtesy, conscientiousness, or hygiene. Neither are drunk people. It’s much easier to determine if a group is larger than 10 than if a group is larger than 50, so I expect enforcement will be down. Non-essential travel means out-of-state visitors can begin spreading the virus in our area again. In practice, I believe Phase 2 will be the effective end of any social distancing sufficient to mitigate the virus.

There is not enough medical-grade Protective Equipment (PPE) and there won’t be for a while.

I think it’s pretty well known that there’s a shortage of PPE in the United States right now. I’ll expand this section if there are questions.

It’s not clear that cloth masks would be enough without other protection.

Earlier this month, the CDC started encouraging the general public to wear cloth masks, but it’s unclear if masks alone would provide enough protection to contain the spread.

One review of the literature found wide variance in the efficacy of masks, including a filtration rate between 3% and 60% for masks made from household materials. The fit of the mask also impacts its effectiveness. While homemade masks can have good fit, it’s up to the individual creator.

Still, the studies on this topic generally conclude that wearing a mask is better than not. Effective enough to remove the need for social distancing? It’s not clear.

An even bigger question is practical adherence: Would enough people wear one without a government directive? Would they continue to wear masks at bars or restaurants when they ate or drank? Would they wear them at dinner parties?

The CDC guidance on masks makes it clear that you should not touch your eyes, nose, or mouth after touching your mask. Can most people avoid touching their mask for a full day in the office? I have my doubts.

Overall, there are more questions on this topic than there are answers. It could work. In my opinion, though, any solution that depends on a very large number of people doing a tedious thing for a long time is bound to fail.

We do not know the true asymptomatic rate.

At the start of the pandemic, it was thought to be close to zero. Recent studies show much higher, but widely-varying, results. A study from Iceland that tested 6% of the population found that 43% of those with positive tests were asymptomatic at the time of testing. A more recent study from New York indicates it could be as high as 88%, though apparently 71.4% developed symptoms later.

All of the studies I’ve read about the asymptomatic rate have only tested each person once. If the p95 of symptom production is 5 days, the difference in results could be a matter of test timing.

Temperature checks are only effective on symptomatic individuals.

As far as I know, temperature checks are a practical measure for symptomatic cases. Given what we know about asymptomatic cases, temperature screening will probably not be an effective method for general containment.

Even if we grant temperature checks are effective on the symptomatic, 5 days is still plenty of time to pass a temperature screening and spread the virus.

There are still significant test shortages.

It was originally estimated that we would need to conduct between 750,000 and 1,000,000 tests per week to control the virus when it is not in outbreak. In the 7 days from April 22–28, the United States conducted about 1.6 million tests. Dr. Fauci has this to say about the current state of testing:

Dr. Anthony Fauci, director of the National Institute of Allergy and Infectious Diseases and member of the White House Coronavirus Task Force, says “we are not in a situation where we can say we are exactly where we want to be with regard to testing” capacity for COVID-19 in the U.S. — via time

He elaborates that some specific materials are scarce and/or not evenly distributed, making it impractical to test everyone that needs to be tested.

Some private organizations are calling for a drastic expansion of testing. Nobel prize-winner Paul Romer suggests we need 22 million tests per week to effectively contain the virus. The Rockefeller Foundation suggests we ramp up to 30 million, a project that would cost 100 billion dollars.

UPDATE ON 4/29: A recent paper from Google and Tel Aviv University says

To create containment, we need to test 30% of the population every day. If we only test 10% of the population every day, we get 34% of the population infected — no containment

There are not enough contact tracers.

The most popular estimate is that we need between 100,000 and 300,000 contact tracers. I can’t find my source for this, but the last estimate I saw was that we had about 10,000.

Even if we could test everyone that might have been exposed, we lack the infrastructure to find and quarantine people in a timely manner.

Sanitation and disinfection only mitigate a secondary transmission mechanism.

While it’s technically possible to get coronavirus from touching an infected surface and delivering the virus, it’s currently considered to be a secondary transmission mechanism. Coronavirus primarily spreads through droplets expelled during human-to-human contact.

There is currently no evidence that sanitation practices played a major role in the spread of the virus, so even perfect sanitation is unlikely to significantly affect the spread.

We still don’t know the actual death rate.

The death rate remains unknown. To determine the death rate, we’d need to have some confidence in the number of infections and the number of deaths from infection. We have neither.

Lack of test kits means that some people are refused testing if they don’t present as significantly ill, reducing the accuracy of our infection rate count.

In the United States, the stated death rate ranges from 0.7% to 7%. Given the uncertainty around asymptomatic cases and other compounding factors, those numbers are considered very unreliable.

There are no reliable predictive models.

There are several popular models of what will happen in the near future, though the lack of solid data has made all of them unreliable. They currently estimate deaths in the 60,000 to 100,000 range through May 23. However, all of them assume we’ll continue to social distance for the entirety of their projection. As it seems likely that some states will choose to open soon, I expect the models to be even less reliable over the near term.

UPDATE ON 5/1: The New York Times reports about half of the states are in the process of reopening.

It will be a long time before enough people are infected to confer herd immunity.

The idea of herd immunity is that enough people have antibodies that the virus can no longer spread effectively. The exact percent of the population isn’t agreed on, but I’ve seen numbers between 40% and 80%.

As of April 28, the United States has conducted 5,795,728 tests. Of those tests, only 1,005,592 were positive.

The estimated population of the US in 2019 is 328,239,523. That means that we only know for certain that 0.3% of the population has the antibodies needed for herd immunity. The most optimistic estimate I’ve seen is 5% of the US has been infected, though that’s a completely uncited assertion.

Assuming we take the lowest value for herd immunity and the highest value of infections, we’re currently 1/8th of the way to herd immunity.

Those that are infected may not be immune for very long.

An initial study by Fudan University in Shanghai showed a third of people infected have low levels of antibodies, and some have no detectable antibodies.

For those that do develop significant antibodies, they may not last long enough to matter. There hasn’t been enough time to test Covid-19 specifically, but previous tests on coronaviruses indicate that those that survive infection may only retain their antibodies for a year.

If that’s the case, or anywhere close to it, natural herd immunity can never be achieved.

A vaccine will likely not be available for years.

It took four years to produce the Mumps vaccine, the fastest on record.

On the one hand, we benefit from having significantly greater medical knowledge and technology, as well as nearly unprecedented worldwide focus on this disease. On April 10th, there were 62 candidate vaccines in production.

On the other hand, we’ve never produced a vaccine for a coronavirus before, and taking all of the steps to produce a vaccine (even a promising one) takes time. Most commonly, experts are estimating it will take at least 1–2 years before a vaccine can be produced.

Covid-19 seems relatively mutation stable.

Covid-19 is mutating, but relatively slowly compared to other RNA viruses. The slow mutation rate of the coronavirus means that these changes will emerge over the course of years. Slow mutation means any vaccine we develop will likely be effective for longer. It also means it’s less likely to mutate into a harmless variant of itself. Ultimately, this is out of our hands.

The potential for developing a treatment is unclear.

Most of the treatments I’ve seen depend on extracting antibodies from the previously infected. That has some obvious problems considering the foregoing information.

The exception is remdesivir, an existing antiviral drug. From a recent federal trial against Covid-19:

The trial enrolled 1,063 patients who were given remdesivir or a placebo, according to N.I.A.I.D. The time to recovery averaged 11 days among those who received the drug, compared with 15 days for those who received the placebo. (via NYT)

Dr. Fauci’s opinion on the study:

“It is a very important proof of concept, because what it has proved is that a drug can block this virus,” Dr. Fauci said. “This is very optimistic.”

The FDA indicated it would grant emergency usage of remdesivir. However, the actual data for the study is not available, even for peer review. Other scientists were skeptical of the political nature of the reveal.

A Chinese study of remdesivir found:

Remdesivir use was not associated with a difference in time to clinical improvement (hazard ratio 1·23 [95% CI 0·87–1·75]). Although not statistically significant, patients receiving remdesivir had a numerically faster time to clinical improvement than those receiving placebo among patients with symptom duration of 10 days or less (hazard ratio 1·52 [0·95–2·43]). Adverse events were reported in 102 (66%) of 155 remdesivir recipients versus 50 (64%) of 78 placebo recipients. Remdesivir was stopped early because of adverse events in 18 (12%) patients versus four (5%) patients who stopped placebo early.

Or, to further summarize:

“Unfortunately, our trial found that while safe and adequately tolerated, remdesivir did not provide significant benefits over placebo,” said the lead investigator of the new study, Dr. Bin Cao of the China-Japan Friendship Hospital and Capital Medical University in Beijing.

It’s not clear how likely it is for a treatment to be developed before a vaccine. Current expert opinions seem to be mixed.

The number of cases and deaths in the Indianapolis area has not yet significantly decreased.

I loaded the case and death data for every county in the United States into Instrumental. I then graphed the cases and deaths by day for Marion county and for the Greater Indianapolis Area (Marion county and every adjacent county).

The rate of deaths in the area is seeing improvement, but there is a recent spike in case rate that does not look good.

You can look at the data for state and/or counties here.

Economics

More than 30 million people are unemployed.

As of April 23, more than 26 million people have filed jobless claims. By comparison, the nine-and-a-half-year stretch between the last recession and the pandemic’s arrival created 22 million jobs..

To put that number in further context, 26 million people is the same as the working population of the 25 least populous states.

The official unemployment percentage for March was 4.4% with 7.14 million people unemployed. That means that 26 million unemployed is about 16%; still short of the Great Depression’s 24.9%, but growing quickly. It is absolutely the worst rate in our lifetime.

UPDATE ON 4/30: An additional 3.8 million filed for unemployment, bringing us to 30 million.

Consumer sentiment has the largest drop on record and shows dangerous, unwarranted optimism.

In March, consumer sentiment had the fourth-largest drop in the 41 years it has been recorded. The three greater drops were December 1980, October 2008, and Hurricane Katrina in 2005. Of those, 2 out of 3 indicated the beginning of serious recessions.

The mid-month report for April was much worse than March, becoming the worst drop on record. I’ll quote Richard Curtin, Chief Economist for the Survey of Consumers:

Consumer sentiment plunged 18.1 Index-points in early April, the largest monthly decline ever recorded. When combined with last month’s decline, the two-month drop of 30.0 Index-points was 50% larger than the prior record. Of the two Index components, the Current Conditions Index plunged by 31.3 Index-points, nearly twice the prior record decline of 16.6 points set in October 2008. In contrast, the Expectations Index fell by 9.7 points, a substantial decline, but not nearly as steep as the record 16.5 point drop in December of 1980. This suggests that the free-fall in confidence would have been worse were it not for the expectation that the infection and death rates from covid-19 would soon peak and allow the economy to restart. As noted in last week’s special report, anticipating a quick and sustained economic expansion is likely to be a failed expectation, resulting in a renewed and deeper slump in confidence. Indeed, the peak decline in the Expectations Index recorded in December 1980 reflected a relapse following the end of the short January to July 1980 recession, signaling the start of a longer and deeper recession that lasted from July 1981 to November 1982. Consumers need to be prepared for a longer and deeper recession rather than the now discredited message that pent-up demand will spark a quick, robust, and sustained economic recovery. Continued declines in the seven-day average Sentiment Index can be expected in the weeks ahead (see the featured chart). Sharp additional declines may occur when consumers adjust their views to a slower expected pace of the economic recovery.

April’s final report was released at 10am on April 24 (a few hours ago when I’m writing this). It says,

April’s final Sentiment Index reading remained largely unchanged from the mid-month figure (+0.8 points), and households with below median incomes expressed the same level of confidence as those with above median incomes (71.9). This merging reflects somewhat larger April declines among households with above median incomes (-19.8 points) compared with those with below median incomes (-14.0). The seven-day moving average of the Index of Consumer Sentiment indicated a second larger improvement that was quickly reversed (see the featured chart); its cause could not be linked to any direct judgements about the coronavirus. The notable divergence between the two main components of the Sentiment Index remained large. The Current Conditions Index fell by 29.4 points in the past month and by 40.5 points in the past two months, whereas the Expectations Index has posted smaller declines of 9.6 points in the past month and 22.0 points from February. While the decline in both indices indicates an ongoing recession, the gap reflects the anticipated cyclical nature of the coronavirus. In the weeks ahead, as several states reopen their economies, more information will reach consumers about how reopening could cause a resurgence in coronavirus infections. Consumers’ reactions to relaxing restrictions will be critical, either putting further pressure on states to reopen their economies, or exerting added pressure to extend the restrictions even if it has negative consequences for economic prospects. The risks associated with these decisions are not equally balanced, with an incorrect decision to reopen having serious repercussions. The necessity to reimpose restrictions could cause a deeper and more lasting pessimism across all consumers, even those in states that did not relax their restrictions.

Further, Richard Curtin had this to say about the time to economic recovery:

An economic recession has already begun, and it is likely to reduce consumer spending for a period that is two to three times as long as the virus crisis (via University of Michigan)

Retail sales have the worst sales slump on record.

March of 2020 saw an 8.7 percent drop, the worst ever recorded. The previous largest single month drop was 4% in the fall of 2008.

Major world economies are shrinking.

The Commerce Department announced the US GDP fell at a 4.8% annual rate in the first quarter of the year. The United States didn’t start seriously reacting to the pandemic until March, so it’s going to get much worse.

Economists expect figures from the current quarter, which will capture the shutdown’s impact more fully, to show that G.D.P. contracted at an annual rate of 30 percent or more, a scale not seen since the Great Depression.

China, the second largest economy, shrank 6.8% in the first quarter.

Analysis

It seems unlikely to me that we’ll have a better solution than social distancing in the near future. Unless a treatment is developed, it may take years before we have an alternative mitigation.

As a consequence, the United States (and the rest of the world) will teeter between economic and medical devastation. Under the unbearable weight of economic crisis, America will reopen enough to trigger a second wave. The second wave will bring more deaths and the realization there is no easy way out. In turn, that will trigger a much harder economic slump as it sinks in that there won’t be a return to normal any time soon. Or, more accurately, this is the new normal.

There will be more waves after that. Some reports indicate there will be waves through 2024, depending on the time to vaccine and immunity duration. Ultimately, the waves won’t stop until we have an acceptable medical response to Covid-19.

Our economic problems, driven by uncertainty, will worsen.

[E]conomists at the Federal Reserve Bank of St. Louis have warned that 47 million people could be thrown out of work and that the unemployment rate could hit 32% by this summer or fall. (via Stanford GSB)

The peak of the Great Depression was 24.9% unemployment.

We won’t start seeing real economic recovery until a few years after we have an acceptable medical response to the virus. Given the length of time that will take, I believe we’re about to enter a period of economic depression that will rival or exceed the Great Depression. A Greater Depression, some would call it.

I realize how extreme that sounds. As I said in the beginning, this is a black swan event and things are evolving rapidly. An unexpected treatment could come out of revolutionary medical work, but the currently available evidence says that’s unlikely.

Next Steps

Assuming medical science does not produce a miracle in the next three months, what can we do? In the face of such overwhelming change, it’s hard to describe everything that should happen next. For now, I’ll keep it to the most immediate things.

Continue to work from home.

Expected Behavior will maintain its recommendation to work from home at least until the Greater Indianapolis Area Weekly Case Rate is 0 for 3 consecutive weeks. At that point, I’ll reevaluate.

Apply for your absentee ballot.

If you plan to vote, and I wish you would, it’s now even easier to apply for your absentee ballot in Indiana. Once you’ve logged into the voter portal, use the “Vote By Mail” link under the “Absentee” section in the grey sidebar on the left to get to the online form.

Make a plan for the people in your life that might be forced back into the old normal.

Your spouse may be required to return to their job. Your young children may have to return to daycare or lose their spot. Older children will likely be expected to return to school in the Fall. If not, there will be more online learning.

Make a plan for all the things you’ve been putting off until “after this is over.”

Relatives that want to visit. Non-urgent medical procedures. Professional haircuts. Consider anything you’ve been waiting to do until this blows over. That might mean finding a way to do it safely under these new circumstances. It might mean trying to time it against a new wave of infections. It might mean admitting it’s not going to happen for a while.

Make a financial plan.

I’m not going to make any specific recommendations here. Consider what it might mean to you if we are about to enter one of the most serious economic events in recorded history and make a plan to weather it.

Make a life plan.

The chance of dying from Covid-19 is directly related to age and comorbidities. We’re not in the most endangered group, but we’re not in the safest group, either. Consider making a plan for serious medical events or death, for yourself and those you love.