Do the Negatives of the Economic Shutdown Outweigh the Benefits?

A CONVERSATION WITH PROF. ELIZABETH BOGAN

By Arman Badrei; contributors: Ben Gelman, Tiger Gao and Arjun Mani

Professor Elizabeth Bogan

On Mar. 18, in an email to the students of her ECO 101 class, Princeton University Professor of Economics Elizabeth Bogan offered analysis on the recent economic shutdown in the U.S. as a result of COVID-19. She argued, in what she characterized as “somewhat contrarian” position, that the shutdown was premature, excessive, and would ultimately result in an exceedingly high amount of unnecessary economic loss and death.

From a simple cost-benefit analysis perspective, Prof. Bogan believed the shutdown’s negatives far outweighed the benefits at the time. She also expressed frustration for such a high degree of response despite the unavailability of information and data on the severity of COVID-19: “What is disturbing is that we are risking lives and suffering from shutting things down about a disease that we never gathered the data to know how deadly it really is.” Her full thoughts and the entirety of the emails can be read below after the Q&A section of this post.

In a recent interview with Policy Punchline on the ethical dilemmas surrounding the Covid-19 crisis, Prof. Peter Singer of Princeton framed the similar debate from a moral-ethical dimension: especially in the context of developing and underdeveloped countries, it is not at all unlikely that the government-induced economic shutdowns would actually end up causing more loss of lives due to poverty, starvation, and migration than otherwise by Covid-19. From an utilitarianist perspective, therefore, are the lockdowns the more optimal choice?

Dedicated to further explore those issues, on April 1, Policy Punchline reached out to Prof. Bogan to pose some follow-up questions on her policy ideas. In our conversation below, which transpired via email, we discuss and assess the U.S. economic response to the global pandemic as well as possible future actions.

Q&A WITH PROF. BOGAN:

PP: There was an initial debate a few weeks ago about mitigation versus suppression strategies. The U.K. initially adopted the mitigation one but quickly found itself overwhelmed by patients and public panic long before a possible “herd immunity” could emerge. They adjusted their strategy based on some projections by Imperial College that questioned the short-term efficacy of mitigation strategies. Does this case demonstrate how mitigation strategies would not work? Or is the U.K. example not indicative? 

EB: I am well aware of Neil Ferguson’s work and prediction of 2.2 million deaths in the U.S.  But his death rate estimates behind the prediction were way off. Deaths/cases have little meaning if you are getting case data only based on testing the very sickest patients. What if those without symptoms or with only mild flu-like symptoms who had COVID-19 were not tested, as we know was the situation?  Then the deaths predicted from the very sick only sample would be grossly too large compared to the infected population. In fact, Neil has walked back his own numbers from getting more case numbers (which are still too low). Now his model predicts 80,000- 120,000 deaths in the US. I believe this number is still too large and am awaiting a study that Eran Bedavid at Stanford is trying to do. He wants to get 2,500 people to do drive through blood samples of healthy people in Santa Clara County, CA to ascertain how many people carry the antibodies that show they have had COVID-19 and didn’t know it. That is to statistically estimate the actual cases and get a new estimate of the death rate to put into the models.

President Trump has taken to quoting the 2.2 million, so that when our deaths are much lower, he will take credit. If we had tested sooner, we could have isolated the carriers of the disease who didn’t stay home because they didn’t know they had COVID-19.

PP: Are you concerned that unfreezing the economy would potentially overwhelm hospital systems as we’re seeing in Italy? 

EB: Not if we had isolated the carriers and the vulnerable. Unlike earlier pandemics this one in Italy and New York was 99% on people with serious chronic health issues like COPD, diabetes, etc. and old. As an aside, we know an Italian doctor who says fewer would have died if they hadn’t tried to hospitalize so many. The doctors didn’t know at first that they were dealing with COVID-19, so they weren’t careful enough to separate them from the really sick people with other ailments in the hospital. Many of those already in the hospital got COVID-19 and died.

PP: Wouldn’t that be a worse outcome, both economically and from a public health perspective? Wouldn't the sudden influx of COVID-19 patients deprive poor Americans of their routine healthcare?

EB: Not just poor, but everyone faces lower hospital access. Not good, but perhaps not preventable. We can’t put COVID-19 back in the bottle. We can only try to optimize policy for the most people given the reality of COVID-19.

PP: This WSJ piece arguing against economic shutdown says that we wouldn’t have a functioning health system without a functioning economy. How would the economy function if people were worried about the virus and believed that they wouldn’t have access to care when they get sick? 

EB: The public has been panicked by nonstop news coverage and predictions of millions of deaths. Imagine if starting in January the news showed every automobile accident and the number of deaths from auto accidents everywhere with crying family and friends interviewed. And highlighted cumulative death numbers since Jan. 1 on every news show. And predicted the death rate from driving on the highway from the death rate only from auto accidents. Closing the roads would reduce deaths but destroy civilization as we know it.

PP: A recent economics paper titled “Pandemics Depress the Economy, Public Health Interventions Do Not” finds that “cities that intervened earlier and more aggressively do not perform worse and, if anything, grow faster after the pandemic is over.” Do you think this lends evidence to how nonpharmaceutical interventions are necessary now and will not hinder economic growth in the long run?

EB: There is an earlier paper on the 1918-19 flu pandemic that shows that the hardest hit cities were Philadelphia, Scranton, and Pittsburgh. It doesn’t say, but I will speculate that they were the worse because coal was mined there, so lung disease would have dominated.

Much of the argument for shutting down today is based on a 2007 study of the Great Pandemic flu of 1918-19 published on May 1, 2007 in Proceedings of the National Academy of Sciences (PNAS). The following quotation is from this article: “Early implementation of certain interventions in 1918, including closure of schools, churches, and theaters, was associated with lower peak death rates…. These findings support the hypothesis that rapid implementation of multiple non-pharmaceutical interventions can significantly reduce influenza transmission, but that viral spread will be renewed upon relaxation of such measures…. cities that had low peaks during the first wave were at greater risk of a large second wave.”   

The cities with the interventions flattened the curve but didn’t reduce the cases or death rates significantly overall. Of the 17 cities studied that used interventions, they did so for about 6 weeks. Maybe if they had done it for months the result would have been different. When would it have been safe to stop? No one knows.

PP: What do you say to critics who contend that the choice between economic ruin and public health sacrifices is a false one, since the government could simply step in and pay people’s wages until the economy is unfrozen, which would mitigate the economic damage while also enabling lockdowns to continue?

EB: This is the strongest argument for mitigation. It confirms the tremendous economic costs of a shutdown, but definitely reduces the alternative deaths I was most concerned about. Given the huge real suffering from the shutdown, I am very thankful that monetary and fiscal policy have been so aggressive this past week. It is scheduled to cost about $4-5 trillion or roughly a fifth to a quarter of GDP. But that is necessary to keep even more people from dying of stress-related diseases, suicide, or not qualifying for food stamps and shortening their lives and their children’s lives from hunger. However, even unemployment insurance and small business aid in the stimulus package cannot save all the dreams of people who had just started successful personal businesses. It is not clear that they can be made whole. The longer we have lock down, the longer it will take to recover, and the more dreams that will be permanently ruined; and the more lives that will be lost prematurely over the next few decades from stress induced illnesses that shorten their lives.

For example, we know there were significant deaths among the millions who lost their houses to foreclosure during the Great Recession.

PP: What are your specific recommendations for how we should be handling this? 

EB: The government should have been ready for mass testing and done it and quarantined the ill. Leaders should have pushed washing hands, keeping our hands off our faces, staying home if at all sick, extending sick leave benefits, isolating vulnerable people, personally avoiding crowds, sanitizing metal and plastic surfaces, wearing face cover, and cancelling events that draw large crowds into close contact.

PP: Should some industries or events, such as large sports/entertainment gatherings remain shut down while restaurants for example continue to operate?

EB: Yes.

PP: What about public transportation?

EB: No. It would not have been as expensive as what we are doing now to have hired people to walk through subways with sanitary wipes for all hand grips and requiring riders to wear masks or scarves over their nose and mouth.

PP: Given your opinions on the scale of this crisis, what are your thoughts on the $2 trillion relief package passed by Congress?

EB: As mentioned above, it is absolutely necessary. Since the government induced this recession, it owes the public trying to counter the worst outcomes. Also note that the Fed will spend about another $2 trillion or more to keep the financial system afloat, which is essential to avoid a Great Depression. 

PP: How do you think it compares to the responses of other nations such as the U.K. or South Korea?

EB: South Korea did a better job than the U.S. It was prepared to roll out tests quickly and quarantine sick people.

Our biggest failure was we were not prepared to roll out massive numbers of tests and thought 200 tests that had to be sent to the CDC was okay. Had we tested and done what I suggested in the second part of Question 6, I do not believe we would be seeing surging numbers today.

I am also intrigued by the suggestion of President Trump to get a bill to spend $2 trillion on infrastructure. I don’t know how long US Treasury bonds will only need to pay less than 1% to get people to buy them. But this is the time to run up some debt if the funds are well spent to make the economy stronger so that the borrowing is self-financed in the long run from higher tax revenues on a stronger economy. The time to do infrastructure projects like roads and railroads is in a recession when public spending doesn’t crowd out private spending.

PROF. BOGAN ORIGINAL EMAILS TO HER STUDENTS:

I. “COMMENTS ON POLICY TOWARD CORONAVIRUS” - SENT ON MARCH 18

First, I want to emphasize that I am complying with everything the administration and governmental units have asked of us, as you should also do. Still I am an academic economist and my training is to examine policy with cost-benefit analysis based on available data. Economics has a lot to offer in understanding today’s policies. Our objective should be devising policies that maximize social welfare given the constraints.

My review of the costs and benefits of current policies is somewhat contrarian. I present my analysis here because it is my job to make you think about policy evaluation. (I am totally comfortable that some of you will not agree with my analysis. I don’t have all the data I’d like and may change my own mind in real time as new data become available.) In summary, the data look to me like nearly closing the economy will cause more suffering and deaths than would the COVID-19. And far more young and middle age people will suffer. 

People completely miss that the world is full of trade-offs. They say every life matters and I say they don’t really understand there are trade-offs. Cars kill about 40,000 people a year. The seasonal flu in 2017-18 killed more than that and we didn’t decide to shut down the economy in 2017-18. Hospital mistakes are estimated to kill 40,000-80,000 patients a year. The point is these awful costs can’t be eliminated. But each is out-weighed by superior benefits. The question today is: Should we continue shutting down the economy? 

I believe we should have told people over 65 that they were endangered and should self-isolate. Old peoples' homes should have stopped taking visitors. Certainly, it was correct to get people washing hands, etc. Then we should have left the schools open and let the corona virus run through the schools. It appears that most kids who get this virus don’t even know they are sick. But that would build “herd immunity” to the virus. What we have most likely done with closing the schools is guaranteed that the virus will spread again whenever we open the schools. That is what happened in 1919 to the 1918-19 flu epidemic.

Much of the argument for shutting down today is based on a 2007 study of the Great Pandemic flu of 1918-19. https://www.pnas.org/content/104/18/7582. It is a study published in PNAS (Proceedings of the National Academy of Sciences) on May 1, 2007. The following quotation is from this article:

“Early implementation of certain interventions in 1918, including closure of schools, churches, and theaters, was associated with lower peak death rates…. These findings support the hypothesis that rapid implementation of multiple non-pharmaceutical interventions can significantly reduce influenza transmission, but that viral spread will be renewed upon relaxation of such measures…. cities that had low peaks during the first wave were at greater risk of a large second wave.”    

The cities with the interventions flattened the curve, but didn’t reduce the cases or death rates significantly overall. Of the 17 cities studied that used interventions, they did so for about 6 weeks. Maybe if they had done it for months the result would have been different. When would it have been safe to stop? No one knows. 

Let’s look at the macroeconomic policies today, first monetary policy and then fiscal policy:

1) The Federal Reserve is taking stimulative actions as it did in 2008-9. It plans to inject $1.5 trillion dollars into the economy.

The Federal Reserve said it would establish a new lending facility to reduce the strains from the accelerating economic damage from the coronavirus. The Fed dropped the Fed Funds Target Rate a full percentage point to 0-.25%. Zero to .25% is where it was from December 2008 to December 2015. The low rate encouraged businesses to invest and in time the economy recovered. But this time we have a government induced slowdown of many businesses. The Fed’s actions can’t do much about that.

2) Today the Trump administration is seeking an $850 Billion Stimulus package, which is what Congress did in 2009.

The problem this time is that financial markets weren’t frozen and businesses were doing well, then the coronavirus caused the government to close much of the economy in the name of saving lives. We are in a government policy induced recession and stimulus can’t solve the workers’ problems, at least not stimulus alone. 

The only thing that will really make a difference is a drug or a vaccine. The best thing the government has done is to speed up drug and vaccine trials. I believe the benefits of shutting down the economy will not exceed the costs.

What are we doing?

Closing schools has disrupted the lives of 30 million kids and their families. Closing bars, restaurants, athletic events, theaters, etc.  is having a negative impact on small businesses and their employees. Yes the government says it is offering small business loans, but meanwhile it will be really difficult especially for their former employees. 

California closed all its restaurants except for take-out. The restaurant business in California employs 1,830,000 people according to restaurant.org. That’s 11% of California’s employment. Probably about a million of them have lost their jobs. How will they pay their rent? How sick will they be from stress? How much will their children suffer? Even if the government plans to compensate them it will be a mess. Many live on tips and there may not be records of what they usually make. How will the government even find these people? Also the incentive to make false claims will be high, while others suffer from not knowing how to get compensated. Most high income workers can at least partially work from home, but service workers can’t. Shutting down the economy will cause terrible harm to hard-working low to middle income people. Many will die from stress related heart attacks, which happened in 2009+ among people who lost their homes.  Someone needs to be weighing the opportunity costs of people suffering and dying from the partial shutdown of the economy against the damage of the possible Coronavirus epidemic. 

What is disturbing is that we are risking lives and suffering from shutting things down about a disease that we never gathered the data to know how deadly it really is.  Since we only had 200 test kits at first and still are only testing people who present with major symptoms, we have little idea what the actual number of cases is. For example, was the light flu running through northern California in February, actually cases of Covid-19?  No one was tested. To my knowledge, the only country where random testing is occurring this week is Iceland.

The first roughly 1000 Icelanders tested by deCode Genetics for Covid-19 are running just under 1% positive (0.899%) and that’s been stable the past two days of widely available testing. These are all people not in their “travel-related exposure plus close contacts” national quarantine, so it is the population-wide “unexpected“ community acquired infection rate. 

Since the introduction of Covid-19 to Iceland was from the Alps in Europe (mostly Italy) and also from Denver, CO. it seems likely that they were seeded weeks after California and Washington.  So, if anything, our general population rate should be higher than theirs. If their number is correct and not subject to sample bias, then in California alone it would be ~400,000 infections right now (and potentially 8x that nationwide). But CA deaths so far are at 6 on 454 reported cases, or just over 1% reported fatalities per reported case. If the number of cases is off by 3 orders of magnitude, which would be true if our true data looked like Iceland; then the new coronavirus death rate would be way less than routine seasonal influenza. I’d like to be certain that was not the case before we continue to shut down the economy. 

Two comments from the rest of the world: In Italy the average age of deaths from Covid-19 is 81. Life expectancy in Italy is 82. How much suffering of their entire population from shutting down the economy is one year more of old age (for a small number of the old people) worth?

The Wall Street Journal just reported, “Marriott International, the world’s largest hotel company with nearly 1.4 million rooms world-wide, said it is starting to furlough what it expects will be tens of thousands of employees as it ramps up hotel closings across the globe.” 

If you are interested in the 1918-19 pandemic, here is a second study of the 1918-19 flu epidemic: https://www.stlouisfed.org/~/media/files/pdfs/community-development/research-reports/pandemic_flu_report.pdf. “Economic Effects of the 1918 Influenza Pandemic” is a study by the St Louis Fed authored by Thomas A Garrett, published in November 2007.

II. “AN EPIDEMIOLOGIST FROM STANFORD AGREES WITH ME” - SENT ON MARCH 19

Please read this by Eran Bendavid, an epidemiologist, infectious disease doctor, and health policy expert at Stanford’s School of Medicine which he wrote last night. You will see that epidemiologist sare questioning the severity of Covid-19 as I have been for weeks. The infection rate seems to be 100 times what we have  been told. That means that Covid-19 is less dangerous than the ordinary flu and we should not be laying off millions of people by shutting the economy down.  (I believe the actual increased death rates will be higher from the shutdown if it lasts very long than from Covid-19. Some Epidemiologist are coming to the same conclusion. Thanks to all of you who forwarded my comments and who told me how much you liked them.)

“We cannot trust the existing models” - by Eran Bendavid

 The response to the Covid-19 epidemic include extraordinary measures, and no part of the world is spared from the reverberations of the social and economic effects. The severity of these measures appears justified based on a “perfect storm” confluence of 2 critical pieces of information:

1. Projected epidemic curves based on R0and early doubling time estimates that suggest over 60% of the world population could get infected in the absence of control measures.

2. Case fatality rates in the region of 2-3% based on confirmed cases and deaths.

These two epidemic estimates are missing an incredibly crucial component: a multiplier that relates the number of cases to the number of infections. If that multiplier is small, then the number of infections is close to the number of cases and we could indeed be heading towards a common and lethal pandemic. However, if that multiplier is large, then the number of infections is much larger than the number of cases, the true fatality rate is much lower, and, moreover, there may be many in the population that are already recovered and immune. The models and projections that are guiding current policy are essentially ignoring the fact that this multiplier may in fact be very large. 

To take an example, if we currently have 100 people that have been infected for every known case in the US, that would imply about 500,000 people have had Covid-19, and the fatality rate is 0.02% instead of 2%. If 50% of the US population gets infected, as it might if we lift all control measures, that could mean 30,000 deaths from Covid-19. This would deserve serious attention, but it should change the calculus that involves turning off the lights on entire communities in anticipation of overwhelmed ICUs.

Is this 100-fold multiplier even remotely plausible? Of course it is. The only people we – and just about every other health system to date - have been testing are those who show up at the doctor or hospital with symptoms severe enough to justify testing. The extremely limited testing capacity everywhere means that the cases give us an extremely unreliable view of the true number of infections. Let’s take a few threads of evidence to help us get a better sense of the multiplier.

First: doubling times. To date, the epidemic has been circulating in the US since early January, possibly even before (certainly a lot of people traveled from Wuhan to the US in December). If one trusts the doubling time of around 4-5 days, an epidemic seed in early January would have had between 16 and 20 generations to double, implying between 32,000 and 1 million infections in the US, a multiplier of 10-155 (2^16 and 2^20 divided by 6,421 cases on March 18).

Second: limited prevalence studies. On or around January 31, several countries sent planes to evacuate their citizens from the greater Wuhan area. When those planes landed, the passengers on those planes were tested for Covid-19 and quarantined. After 14 days, the average prevalence of all those repatriated expatriates was 0.9%. What can we do with that? If we assume that this was the prevalence in the greater Wuhan area on January 31, then, on a population of about 20 million people, greater Wuhan had about 178,000 infections, a ratio of infections to cases of 31. Next, the Italian town of Vò in Padua. On March 6, the entire community of 3,300 was tested, and 90 were identified, a prevalence of 2.7%. On that day, the province of Padua (population 955,000) had 198 cases, a ratio of 130. Finally, the National Basketball Association. Between March 11 and March 17, a number of NBA players were tested, and 7, out of 450 rostered players, were positive. That’s a low bound of 1.6% (since not everyone was tested) prevalence. If we only take that to represent the population of NBA team cities (population 45 million), that’s 720,000 infections. During those few days, the number of cases reported in the US ranged from 1,281 to 6,421, a multiplier of 112 to 562!! This would make fatality from Covid-19 well below that of seasonal flu.

No one is asking whether or not we should act in the face of this epidemic – of course we should act – but the pressing question is what should we do? The answers depend critically on reducing the uncertainty about this multiplier. We can do that: perform surveillance testing on random samples of the US population, and accelerate development of antibody testing so we can tell where we are in this epidemic. The consequences of weeks or months of lockdown will entail a far greater toll of pain and misery than that caused by Covid-19 if the multiplier is 100. At this point reducing the uncertainty about the true fatality rate for Covid-19 could avert a major downturn for humanity.