Now I am pleased to see the publication of his new book--Apollo's Arrow: The Profound and Enduring Impact of Coronavirus on the Way We Live. It's a remarkable achievement, because he has written a comprehensive study of the COVID-19 pandemic, in all of its biological, sociological, psychological, and economic dimensions; and he completed it only eight months after the pandemic began in January.
This book helps me to think about the many questions I have raised about the pandemic in previous posts (here, here, and here). In this post, I will consider two questions. When and how did the experts decide that the COVID-19 pandemic justified a lockdown imposed by government? And is it true that the benefits of COVID-19 lockdowns exceed the costs?
LOCKDOWNS WITHOUT PRECEDENT
We must begin by recognizing one startling fact about the response to this pandemic--that there is no historical precedent for the general lockdowns of society ordered by governments in 2020. Christakis does not give enough weight to this fact. He writes about the history of public health policies in responding to infectious diseases in a way that suggests to the reader that the COVID-19 lockdowns were nothing new. In only one sentence does he implicitly recognize the uniqueness of the lockdowns. Speaking about the shutdown of almost all of China at the end of January, he writes: "It was the largest imposition of public health measures in human history" (9).
As measured by the number of people who died in two years, the flu pandemic of 1918 was the deadliest infectious disease outbreak in American history or in world history. So if we were looking for historical precedent for pandemic lockdowns, we should find it here, but we don't.
So, for example, as Christakis indicates, the Health Commissioner for New York City in 1918--Royal Copeland--"generally favored keeping things open," although the sick were quarantined in mostly voluntary isolation, the city ordered staggered business hours to avoid rush-hour crowding, and there was a campaign against public spitting as unsanitary. The schools remained open (73-74). Christakis relies on an article by Francesco Aimone (2010). And Aimone reports that Copeland "reasoned that more restrictive means, such as ordering all businesses and municipal offices closed, was unwarranted because of the low incidence and concentrated prevalence of the disease" (72). Theaters were also open, although they were subject to public health regulations.
Christakis praises Copeland. "Copeland endured enormous criticism, but the strategy was vindicated by the results. Public health decisions always involve difficult, utilitarian trade-offs between benefits and costs to different people." The results were clear in that "New York City had approximately half the excess deaths that Philadelphia had" (74). So here the "utilitarian trade-offs between benefits and costs" justified keeping schools, businesses, and theaters open, and relying on mostly voluntary quarantining of the sick.
In contrast to his praise of Copeland for keeping schools open in New York City, Christakis says that two studies of 43 major U.S. cities in the 1918 flu pandemic show that the earlier that schools were closed, the lower the number of excess deaths (Bootsma and Ferguson 2007; Markel et al. 2007). He presents a graph comparing St. Louis and Pittsburgh on the timing of school closures and public gathering bans--showing that the earlier imposition of nonpharmaceutical interventions (NPIs) was correlated with lower mortality rates (124). Graphs like this comparing cities in 1918 have been the single most commonly used evidence apparently justifying the COVID-19 lockdowns in 2020. The flaw in this reasoning, however, is the fact that no city in 1918 imposed anything close to a general lockdown.
Christakis' graph comes from Merkel et al. (2007). But Christakis does not tell his reader that this article points to "two outlier cities"--Grand Rapids and St. Paul--that had low excess death ratios even though they imposed few restrictions for very short periods. In fact, Grand Rapids--the least restrictive of the cities--had the lowest excess death ratio of all the 43 cities!
Nor does Christakis tell his reader that these cities refused to lockdown businesses. In Los Angeles, the city council debated this and decided not to do it. In St. Louis, something like a lockdown of businesses was done for only about two days. So people in 1918 judged that the likely costs of a prolonged lockdown would be greater than the likely benefits. Has anyone in 2020 demonstrated that now the benefits of doing this exceed the costs?
THE UTILITARIAN CALCULUS OF COST-BENEFIT ANALYSIS
As already indicated, Christakis stresses that "public health decisions always involve difficult, utilitarian trade-offs between benefits and costs." Remarkably, however, Christakis never shows that any of the people who initiated the COVID-19 lockdowns had actually worked through any precise cost-benefit analysis to justify this. Christakis himself presents his own cost-benefit analysis in one paragraph of his book, but it's hard to understand his calculations.
As Christakis suggests, the first mandatory pandemic lockdown in history occurred in China last January in response to the outbreak of the coronavirus in Wuhan. A six-member team of Chinese epidemiologists and physicians advised the Chinese government to impose "closed-off management," using the sort of social control that had originally been developed by Mao Zedong (9). Christakis does not indicate that anyone in this team or anyone else in China had actually calculated the "utilitarian trade-offs between benefits and costs" to support this recommendation.
Similarly, Christakis reports that on March 13, 2020, Dr. Anthony Fauci had publicly stated that a "national lockdown" might be necessary in the United States (90). But, once again, there is no evidence that Fauci or anyone else in the CDC had worked through a cost-benefit analysis to justify this.
Christakis indicates that the CDC "released the aptly titled report 'Community Mitigation Guidelines to Prevent Pandemic Influenza--United States, 2017' three years prior to the pandemic. It was full of sound advice that had been offered for decades" (96).
Christakis does not tell his reader, however, that in the 2017 CDC guidelines for a "Very High Severity" pandemic (comparable to the Spanish Flu), the guidelines provide only that "CDC recommends voluntary home isolation of ill persons," and "CDC might recommend voluntary home quarantine of exposed household members in areas where novel influenza circulates." Notice the word voluntary. And notice that nothing like statewide lockdowns is recommended in this document. Nothing is said about closing "non-essential businesses." (I have written about this here and here.)
Here is the one paragraph in which Christakis presents his own cost-benefit analysis to justify lockdowns:
"These sorts of demographic calculations also allow us to benchmark the financial benefits of saving lives against the financial costs of shutting down the American economy during the development of the NPIs. By using a standard benchmark of five hundred thousand dollars as the economic value of a year of life (or ten million dollars per life, regardless of age), we can estimate the one million coronavirus deaths (at the rough age distribution at which they occur) would be worth about six trillion dollars. Even at the highest end of a range of estimates of the consequences to our economy, including the expenditures by our government, we do not reach that sum. Strictly from an economic perspective, our response was commensurate to the threat posed by the pathogen. It's a bad virus" (304-305).
Christakis here seems to be relying totally on an article by Joshua Goldstein and Ronald Lee (2020). And in turn Goldstein and Lee seem to be relying on estimates made by W. Kip Viscusi (2020) in a blog post. Viscusi estimates the value of a statistical life (VSL) as $10 million, based on estimates of the extra wages that workers would demand for increased risk of dying at work. If a worker receives extra pay of $1,000 to face a risk of 1/10,000 that he will die at work, the VSL is $1,000(1/10,000) = $10 million. Viscusi then estimates that if a U.S. COVID-19 lockdown saves 1 million lives, the economic value of these benefits would be $10 million times 1 million lives--$10 trillion. Goldstein and Lee estimate the value of these benefits as ranging from $6 trillion to $10 trillion. Christakis apparently has chosen to go with the lower estimate--$6 trillion.
How exactly do we derive the number of 1 million lives saved by the governmental lockdowns of society? And how exactly do we calculate the costs of the lockdowns, so that we know for sure that the costs are less than the benefits?
Elsewhere in his book, Christakis says that the lockdown in the U.S. in the spring of 2020 "may have prevented sixty million cases and probably more than three hundred thousand deaths during the acute shock of the first wave of the pandemic" (94). But in another passage, he says that without the lockdown, "it is possible that over a million Americans would have died in the first few months of the pandemic" (91). So which is it--300,000 or 1 million?
Christakis also points to the prediction "that one million people will die from COVID-19 by the time the pandemic is over in the United States after several waves (which is not inconceivable)" (303). So is he suggesting that without any lockdowns in the spring, the U.S. COVID-19 death toll would have reached over 2 million? If so, where did he get this number? One possibility is that he is relying on Neil Ferguson's prediction that if there were no mitigation efforts at all, over 2 million Americans would die in the pandemic. Apparently, Ferguson derived this number by assuming that COVID-19 would be as deadly as the flu of 1918; and adjusting for population growth, 675,000 deaths in the U.S. in 1918 would correspond to over 2 million deaths in 2020. But as Christakis indicates, COVID-19 is not as deadly as the 1918 flu (304).
Another mistake in Ferguson's estimate is that he assumes that people will not voluntarily change their behavior to protect against the virus. As Christakis observes, there is plenty of evidence that people were voluntarily reducing their social and economic activity two to three weeks before the mandatory lockdowns started (19-20, 90, 133-35, 284). He writes: "people themselves knew what to do, notwithstanding the failures or successes of their country's responses. People began to physically distance before being told or ordered to do so" (135). Consequently, most if not all of the reduction in death rates in the spring could have been caused not by the mandatory lockdowns but by the voluntary mitigation behavior. Christakis does not consider this possibility.
Another problem in the paragraph quoted above is that Christakis does not factor in all of the costs of the lockdowns that he mentions elsewhere in the book. Christakis identifies many "medical, social, and economic costs"--including increasing rates of depression, suicide, and homicide, and harm to children who fall behind in their education (94, 100, 120-21). Poverty is deadly, and the global depression provoked by lockdowns could well create famines in which millions of people will die.
Moreover, it's the poorer and weaker members of society who bear most of these costs. "Affluent people are able to protect their health and livelihood more effectively than others. Remember that wealthy people have been fleeing to their country homes to avoid plague for thousands of years" (179). So why should we trust the rich and powerful people to impartially make the calculus of costs and benefits of lockdowns when most of the costs will be suffered by the poor and the weak?
Another problem with Christakis' cost-benefit analysis of lockdowns is that he admits that lockdowns are not sustainable--they must be lifted after two or three months, and then the virus will return (11). He writes:
"In early May 2020, as the United States began to ease the non-pharmaceutical intervention, Thomas Frieden, the former director of the CDC, observed, 'We're reopening based on politics, ideology, and public pressure. And I think it's going to end badly.' It is one thing to determine what the epidemiology of the situation demands but conclude that the economics countermands it or that the public has had enough, but it is quite another thing to ignore the epidemiology and pretend that nothing bad is going to happen" (319).
Well, if "the epidemiology of the situation demands" a prolonged hard lockdown (a year or more?) until a vaccine is available for everyone, then why not do that? Why lift lockdowns if that means that more people will die? Presumably, the answer is that any lockdown of more than two or three months is too costly--the human harms exceed the human benefits.
Now, in the fall, we have seen higher COVID-19 numbers--infection rates and death rates--than we saw in the spring. So why shouldn't we reimpose hard lockdowns? In recent weeks, some U.S. governors and national leaders in Europe have ordered new restrictions, but these new orders are noticeably softer than what was done in the spring. If hard lockdowns now are too costly to be justified, then why doesn't this mean that the hard lockdowns in the spring were also too costly?
Although Christakis did not intend to do so, his book makes clear to any careful reader the logical incoherence of the argument for COVID-19 lockdowns.
Aimone, Francesco. 2010. "The 1918 Influenza Epidemic in New York City: A Review of the Public Health Response." Public Health Reports 125: 71-79.
Bootsma, M. C. J., and Neil Ferguson. 2007. "The Effect of Public Health Measures on the 1918 Influenza Pandemic in U.S. Cities." Proceedings of the National Academy of Sciences 104: 7588-7593.
Christakis, Nicholas. 2020. Apollo's Arrow: The Profound and Enduring Impact of Coronavirus on the Way We Live. New York: Little, Brown Spark.
Ferguson, Neil. 2020. "Impact of Non-pharmaceutical Interventions (NPIs) to Reduce COVID-19 Mortality and Healthcare Demand." London: Imperial College COVID-19 Response Team, March 16.
Goldstein, Joshua, and Ronald Lee. 2020. "Demographic Perspectives on Mortality of COVID-19 and Other Epidemics." Cambridge, MA: National Bureau of Economic Research, April.
Markel, H., et al. 2007. "Non-Pharmaceutical Interventions Implemented by U.S. Cities during the 1918-1919 Influenza Pandemic." JAMA 298: 644-654.
Viscusi, W. Kip. 2010. "Pricing the Lives Saved by Coronavirus Policies." National Economic Education Delegation Blog. April 8.