In October 2020, David Cutler and Lawrence H. Summers published a brief article in JAMA Viewpoint estimating that COVID-19 would cost the United States $16 trillion dollars, when combining economic damages and monetized health and life loss. This figure has been extensively cited and used in policy discussions. In this article, we update their estimate, using facts about the disease and its costs to society that have become known since their paper was published.
We find that the total harms of COVID-19 to the U.S. are still about $16 trillion (with a range of $10 trillion and $22 trillion) but the components of harm are significantly different than those estimated by Cutler & Summers. The pandemic caused less economic damage than they projected, but more mental health damage.
Our article, like the Cutler & Summers article, is by necessity a very rough estimate. Many important facts about the effects of COVID-19 are still not known. However, we searched for literature on what is known, and combined that with other data sources or estimates as necessary to produce an estimate.
We use the Department of Health and Human Services methodology for valuing health losses. This method focuses on Quality Adjusted Life Years (QALYs) saved (or lost), discounts future QALYs at a 3% annual rate, and values each QALY at about $500,000.
Cutler & Summers estimated, based on a 2020 Congressional Budget Office prediction, that COVID-19 would cause $7.6 trillion in lost output in the 2020s. However, thanks in part to historic fiscal stimulus, current projections are that the US economy will soon return to the pre-pandemic trendline of output. The CBO estimates that the total output gap in the years 2020-2024 will be 5.8% of GDP. When multiplied by the estimated real potential GDP in each year, in 2020 dollars, this is about $1.3 trillion in lost output.
This calculation forms our low estimate of losses, because it assumes a counterfactual of a zero output gap. However, in 2019, output was 1% above potential GDP. If, in the absence of the pandemic, output would have remained 1% above potential through 2024, the total reduction in GDP caused by the pandemic would be 10.8% of GDP, or about $2.4 trillion. This is our high estimate; it seems likely that in the absence of the pandemic, output would have slowly converged to potential GDP over several years.
Cutler & Summers did not include any estimated costs related to the relief packages that were enacted in response to COVID-19, such as the CARES Act and the American Rescue Plan. The total budgetary resources allocated to these measures is $4.6 trillion.
However, most of this spending is not a cost. It is instead a cash transfer from future taxpayers to people today. A cash transfer is not a cost, because it does not reduce the total real output of goods and services the economy produces. Of the $4.2 trillion already spent, $3.8 trillion was cash transfers, and the rest was the purchase of goods and services. Therefore, we estimate that the relief bills will directly consume $0.5 trillion of real resources.
Transfers are not completely free, however. Taxation generates a deadweight loss, because it has side effects that reduce real economic activity. Based on the literature review presented in Table 1 of Boardman et al. 2020, we believe that the deadweight loss, or marginal excess tax burden, of this spending will be between 10% and 30% of its budget cost. This means that the $4.6 trillion in relief spending will generate between $0.5 trillion and $1.4 trillion in real costs.
Psacharopoulos et al. estimate that the net present value of educational disruptions in rich-world countries is between 4% and 8% of current GDP (Table 4, Scenarios 1 and 3). With a U.S. GDP of $21 trillion, this is $0.8 to $1.7 trillion of harm. This estimate assumes the equivalent of 4 to 8 months of learning would be lost.
As a robustness check, we did an alternate estimate of the value of lost education, based on the literature on the net present value of more vs. less effective teachers and our estimates on how teaching has declined due to the hardships of teaching during the pandemic. This estimate yielded similar numbers.
More Americans died of COVID-19 than Cutler & Summers estimated. They believed the pandemic would cause 625,000 deaths, but the CDC estimates there have been 1.08 million excess deaths as of this writing. We estimate there will be between 1.1 million and 1.4 million excess deaths before the pandemic is over. The high estimate is based on expected future excess deaths from other causes over the next several years, for example people dying of heart disease earlier because they were weakened by a COVID infection.
Cutler & Summers’s methodology used a Value of Statistical Life Approach, which is appropriate when we do not know who will be killed. But we now have better estimates of years of life lost due to COVID-19. For our high estimate of harm, we assumed that people killed by COVID had the same expected quantity and quality of life remaining as an average person their age. For the low estimate, we instead assumed that the people killed had a lower life expectancy (10 years instead of 14.4) and worse health than a typical member of the population in their cohort (an average QALY value of 0.7 instead of 0.81).
Therefore, we estimate that the value of QALYs lost due to COVID-19 mortality is between $3.3 and $6.5 trillion.
Cutler & Summers estimated that long-term complications would cause health losses in survivors valued at $2.6 trillion, based on an assumption that many survivors would suffer long-term Chronic Obstructive Pulmonary Disease (COPD). However, ‘Long Covid’, while a serious problem, is not as bad as COPD in most of the people affected. Although there are some people who suffer serious and debilitating symptoms, most of the cases are relatively mild. Based on what is currently known, and the QALY cost of the post-acute consequences of infectious disease, we assume that the average case causes harm valued at about $55,000.
In a systematic review published in February 2022, Kessel et al. estimate that between 10% and 35% of COVID patients have one of a number of long-term symptoms. As of March 2022, there were 124 million U.S. COVID cases, and we assume there will be 150 million before the pandemic is over, so there will be between 15 and 53 million Long Covid cases, for total harms of $0.8 to $2.9 trillion.
Cutler & Summers estimated that mental health costs would be $1.6 trillion, based on a belief that 80 million people would suffer mental health declines for one year as a result of the pandemic, and a value of $20,000 per year per case.
In a meta-analysis published in February 2022, Schafer et al. show 90% confidence intervals for increases in prevalence of anxiety, depression, and eating pathologies as a result of COVID-19. They show that there were between 60 and 160 million additional cases of these conditions, and that the cases lasted for two years. We use the same $20,000 value per case as Cutler & Summers, but we assume that the harms are additive, i.e., that a person who has all three would suffer harm valued at $60,000 per year. With these case numbers and value per case, the monetized value of two years of this mental health harm is $2.3 to $6.3 trillion.
A full and accurate accounting of the social costs of COVID-19 will not be possible for decades, if ever. Many long-term consequences are not known. However, we do know that the social and mental health harms of a pandemic are a significant fraction of the total harms, and are substantially larger than some experts previously predicted.
There are many costs that we did not estimate here, such as burnout of health care workers, disruptions of the socialization and development of children, loss of social trust, higher inflation, and an increase in government debt. Many of these categories of harm may appear small or be hard to measure using standard cost-benefit methodology, but have the effect of making key institutions weaker and more fragile in a way that increases the risk of future losses. For example, the loss of life would have been much larger without extraordinary efforts by health care workers, and the system may not be capable of making similar efforts in a near-future pandemic.
It is likely that a future naturally occurring pandemic will also be a respiratory virus that did not previously circulate among humans, and that its effects on society would be similar to COVID’s. Under the optimistic assumptions that the fatality rate would be lower, and that lessons learned from the COVID response would allow us to respond in a way that cuts the other costs in half, and using our lower-bound estimate, a future pandemic would cause at least $5 trillion of harm to the USA.
This number is a lower bound. A policy or investment that reduces the chance of a future pandemic by just 1% has an expected value of at least $50 billion, and probably hundreds of billions of dollars. But a future pandemic could be much worse — imagine a pathogen that has an infection fatality rate ten times higher than COVID. Now is the time for large public investments in medical countermeasures and metagenomic sequencing so we can prevent — or at least mitigate — the next pandemic.