Trying to Get a Handle on the Cost and Benefits of Covid-19 Containment Measures


The focus of public discussion has been on lowering mortality (and to a lesser extent morbidity) rates, and a lot less attention has been focused on the countervailing costs of drastic containment measures. A friend of mine sent me what I consider at least a decent framework for thinking about such questions, though of course the results one gets determines on assumptions for which we lack sufficient data to have any certainty about. I've posted, with permission, below what my friend sent me. I would add one more thing: when I've raised such issues with people, they raise the question of unlikely but at least remotely possible worst-case health scenarios. I counter that we also have to consider unlikely but at least remotely possible worst-case economic scenarios, such as the shutdowns plus government intervention causing a massive long-term contraction, or, alternatively, hyperinlation:

Someone needs to do a cost-benefit analysis for the containment measures being considered and in some states (like my own) implemented. The benefit, obviously, is saving lives that would otherwise be lost due to CV19. The cost is drastic reductions in economic activity, which will result in loss of life as well as reduced quality of life. For harsh enough and lengthy enough measures, the latter will eventually outweigh the former. I haven't seen anyone do this math yet, so I'm going to take a stab.

To do the math, I'll have to make a lot of assumptions, and you may disagree with some of them. That's fine. I'm just trying to wrap my head around the problem with some actual numbers. But if you want to challenge my assumptions, please suggest better assumptions so that you (or I) can redo the math and see what changes.

For this kind of analysis, government regulatory agencies typically put a value on human life. This will offend some people, but refusing to do so results in absurd conclusions, such as shutting down activities with levels of risk that normal people willingly accept on a daily basis in their everyday lives. So what's the value of a human life? Agencies are not consistent, but the range is usually somewhere between $5 million and $9 million.

Now, how many lives will be saved by the measures in question? I don't know. The answer depends on infection rates, fatality rates, and the efficacy of social distancing and other measures. It also depends on the capacity of the healthcare system—the whole "flattening the curve" argument. But it's definitely wrong to consider all deaths as the potential benefit of containment measures, because many deaths will happen regardless. We have to consider the reduction in deaths. According to one estimate I found, the overall number of deaths could be around 480,000 in the US. Let's assume that the containment measures will reduce this by half, meaning 240,000 lives. That's the potential benefit.

Now consider the costs. Let's focus on GDP (which means we're ignoring any human cost in terms of lost personal contact and so on). Let's suppose the containment measures will reduce GDP by 50% for one quarter of the year (or equivalently, by 25% for half of the year, or by 12.5% for a whole year). I don't really know if this is a reasonable assumption. Maybe a lot more of the economy can just operate on autopilot. But when everyone not in an "essential profession" is told to stay home, reductions of this magnitude don't seem out of the ballpark. The GDP of the US is about $21.5 trillion, so based on the above reduction, the cost of the containment measures would be about $2.7 trillion.

Dividing that $2.7 trillion cost by 240,000 lives saved, we get a cost per life of about $11.2 million per life—which is beyond the high-end estimate of the value of a human life used by regulatory agencies.

What happens if we change the assumptions? Depends on which ones we change, obviously. If we halve the GDP loss, then containment measures will pass the cost-benefit test after all. On the other hand, if these measures only save half as many people, or 125,000, then they will be even more unjustified. Feel free to sub in your own numbers to see what you get.

NEXT: Trial by Videoconference

Editor's Note: We invite comments and request that they be civil and on-topic. We do not moderate or assume any responsibility for comments, which are owned by the readers who post them. Comments do not represent the views of or Reason Foundation. We reserve the right to delete any comment for any reason at any time. Report abuses.

  1. Actually, it’s a tad more morbid than this — in this case, you also have to take into account life expectancy of those whose deaths you prevent.

    The “value of a life” calculation is based on random persons, while the people who are dying from the Wuhan Virus have serious underlying health issues, which would otherwise have killed them in the near future. Hence the value of the life is that time the person has left before imminent death from the other condition.

    Yes, this time is of great value — but it isn’t the same as the person who has 30-40-50-60-70 years left to live. And that needs to be recognized — your valuation is way too high.

    And then the stress of confinement will be causing deaths. Deaths from high blood pressure & stroke, deaths from domestic violence, and deaths from all the mental health issues it creates.

    1. And as this only is going to delay — not prevent — the spread of the virus, the only lives that are actually saved are those that would be lost due to the shortage of medical resources.

      1. The most important part.

    2. Think of the gigantic amount of wealth created if there were mandatory euthanasia for anyone with a terminal disease or in a long term care facility.

    3. Ed, lots of folks with substantial underlying health issues—issues which put them at substantial risk of dying from a coronavirus infection—have not previously been shown to have any notable reduced life expectancy. There are some, including people with serious auto-immune disorders, who seem to benefit by use of pharmaceuticals which suppress systemic inflammation—with a result that they actually gain life expectancy compared to normal expectations.

      So I don’t think someone who is 65-years old, with a 20-year life expectancy despite (or because of) a suppressed immune system, is someone accurately described as having, “health issues which otherwise would have killed them in the near future.” But of course coronavirus can put a quick stop to that kind of thinking.

      Other than that, your ice-flow approach to triage for the old is the same morally as it always has been.

    4. The other framework is to look at the reduced life expectancy during/after a depression, in communist countries, etc. If we plunge the country into a depression, which reduces the overall life expectancy by 5 years, that’s a lot worse than reducing the life expectancy by a few weeks.

      Thing is, old people need us to be able to support them for years to come, and for that, we need a strong economy. This isn’t an “economics versus people’s lives” thing. Social Security and Medicare are in precarious shape; if we have a massive recession, we will not have the funds to keep elderly people alive.

  2. Solid prediction: there will be fewer deaths over any time period you choose from this form of the virus than from abortion.
    So why the panic?

    1. I know you thought this was very clever, but it demonstrates (efficiently) why the moral arguments against abortion are unpersuasive. The deaths of blastulas are not concerning to anyone besides nosy busy bodies.

      1. Abortions are not confined to to the blastula stage which ends before the woman will have missed her period.

      2. I’m glad we’ve determined that the value of human life is contingent on whether its termination is concerning to anyone (setting aside that you excluded 50% of the population from “anyone”). I look forward to the euthanasia of millions who have no friends or relations.

        1. The moral arguments for not killing humans who have no friends or relations are different from the moral arguments for not killing blastulas. If you can’t see that, it is no wonder that you are confused by moral indifference to blastulas.

          1. I’m not confused by it. There is moral indifference to humans with no friends or relations, as well as some with, regardless of the size or developmental stage of the human. I understand that. Most abortions aren’t blastulas, but that’s beside the point.

            1. At what point does a blastula become a separate and distinct human?

              That is admittedly bait, but let’s see if anyone bites.

              1. Trick question: it already is a separate and distinct human being.

                Unlike coronavirus, medical or surgical abortion is not a natural death.

              2. Sure, I’ll bite. A fetus acquires personhood when it’s viable. IMO Roe is bad jurisprudence, but good policy. It makes the best available compromise between the conflicting interests of a pregnant woman and the blastula→fetus/potential person→actual person developing inside her.

                1. Leo,
                  MK asked about human beings and you responded as if he asked a completely different question about “persons.” I assume that was a cute, intentional change of subject. “Person” just means someone who has legal rights, such as the right to life under the Constitution.

                  It’s always funny when somebody misuses the term to make circular statements like “abortion should be legal because it’s not a person” which means the same thing as “the life of an unborn human should not be legally protected because the life of an unborn human is not legally protected.”

                  1. It’s always funny when someone purports to know that what I really meant is what conveniently proves that people who disagree with them are semantically ignorant, circular arguing, subject changers.

                  2. Most people use the term “human being” to connote a person. (see for example, theobromophile below who said that a non-person human life is a contradiction in terms). To keep us from taking past each other, we need a new term that means “human” without connoting “person.”

                2. Many states have laws on the books that allow a person to be charged with double homicide when the victim is a pregnant woman, i.e. her baby also dies. There generally are not requirements that the foetus be viable.

                  Do you think that a man who, for example, kills his pregnant girlfriend ought to be charged with one or two counts of murder?

                  Also, in all seriousness, why do we need to balance the “conflicting interests” of the mother and the rights of the foetus? I’m a woman and am honestly unclear as to what right of mine would be vindicated via abortion.

                  1. The liberty right to determine whether to bear a child.

                    1. Absent coercion (e.g., rape or forcible insemination), I already have that liberty. What I don’t understand is extending that liberty to the point after where the decision to be pregnant has already been made.

                    2. I understand the argument that a fetus is a person, and thus a woman’s liberty to determine whether to bear a child takes a backseat to the fetus’ right to life. But, it appears you think the woman outright forfeits her liberty right after pregnancy (except for rape and forced insemination *) even if the fetus is not a person. That strikes me as very wrong. She still should have the liberty right of changing her mind. Whether the fetus’s right to life trumps her liberty right is a separate question.

                      (*) In the case of rape, since you agree the woman maintains her liberty right after pregnancy, and you believe a fetus is a person, do you believe the woman’s liberty right trumps the fetus’ right to life?

                  2. I don’t doubt you honestly feel that way. I might too if I believed abortion was the moral or actual equivalent of murder.

                    To understand what right an abortion could vindicate, try putting yourself in the mind of somebody who doesn’t share your beliefs. A person who sincerely believes an early stage fetus is at best no more a human being than a caterpillar is a butterfly, and at worst a kind of parasite. Is it hard to believe that person could sincerely find it a violation of their liberty to prevent them from removing the fetus from their body?

      3. It’s a utilitarian argument, so I appreciate it. From a similar point of view, I’m waiting for someone on the left to make the case that since more people die by gun homicides (about 15k/yr) than have been killed worldwide by the Chinese Coronavirus, that we should go to the same extreme lengths to combat that as well.

        1. mad_kalak, crank up the gun homicides to the plausible upper values for a coronavirus pandemic in the U.S., and maybe even you would go to extreme lengths. I hope you would.

          1. Probably not, and neither would many Americans, because we had such an event already, the Civil War, and gun rights actually expanded somewhat afterwards. And the mechanism for say, a million gun homicides, absent a complete societal collapse, just doesn’t exist. But interesting hypothetical nevertheless.

            Year over year, gun homicides at 15k/yr in one human lifespan will exceed most worse case projections of this pandemic anyway, if we are talking about the U.S.

            All this said, treating guns like a virus to be eradicated is just the reason why the CDC had it’s funding for gun control research cut and redirected to traumatic brain injuries.

        2. Weird no one on the left has yet made such an argument, huh.

          Almost as though such bare utilitarian calculus is much more common on one side than the other.

    2. Perhaps because many of us believe abortions do not end the life of a person.

      1. “life of a person”

        What life is ended?

        Kangaroo? Fish? Worm?

        1. A non-person human’s life (recognizing that most people think “human life” has the connotation of person).

          1. “A non-person human’s life”

            Care to explain more? That seems to be a contradiction in terms.

            1. Because human life and personhood are different at law, and under the morality of most people.
              Including most pro-life people, if polls about whether the woman getting an abortion is a murderer have any truth to them.

              1. How are those different under law, except as applied to abortion?

                1. The law permits the destruction of embryos in an IVF clinic (unused embryos) and therapeutic cloning (stem-cell research).

                  1. So why are the very youngest human beings not given the right to life, and is the law just in doing so?

                    1. Again, you are using “human beings” to connote “person,” which clouds the argument. The pro-choice argument is the right to life of the non-person human can take a backseat to the liberty right of a woman to choose whether to bear a child.

                    2. No, you’re clouding the argument by making up a new category of humans: non-persons. If you’re going to exclude one category of human beings from human rights, then you need to explain why, and saying “They aren’t persons” is not enough. Perhaps explain exactly why it’s okay to exclude some members of our species from “personhood,” and explain what “personhood” means in a way that is not simply code for “abortion is morally acceptable.”

                      Sophistry aside, you and I both know that we both were blastocysts, embryos, foetuses, newborns, infants, toddlers, adolescents, and are now adults. You know that a “foetus” is not some weird different entity that is not a person but will be a person if a woman wants it to be a person. You’re finding what you think is a clever argument to ignore the fact that you’re advocating for murder of an entire group of human beings, based on the fact that they are inconvenient.

                    3. Let’s accept for the moment your position that “human life” and “person” are synonymous. Then, as you point out, abortion is murder. So too is destroying embryos in IVF or stem-cell research. The logical conclusion of this position is that a woman who has an abortion is a first-degree murderer (she hired the doctor as the hit man). Ditto for the owners and workers at an IVF clinic or a stem-cell research lab.

                      Perhaps you are OK with these logical conclusions. I am not, and therefore have concluded that personhood cannot begin at conception.

      2. Belief doesn’t always line up with empirical reality.

        That said, I support abortion being legal for eugenic reasons, even if I find it personally abhorrent, and legal murder.

        1. Well, that’s horrible.

  3. This is the right way to approach the problem and policy solution options. I think there’s some gaps in reasoning.

    You want to compare apples to apples. For deaths, you don’t just take the total number of dead from COVID. You would want to know how many would die from COVID but for some containment measure. (Because if some of the people would die in any event, containment has no utility at least for that subset of people.)

    But you need to apply the same rigor to the economic evaluation on the costs of containment. You don’t just find out the costs of containment and measure against that. You have to know the costs of containment compared to economic loss in a non-containment world. That’s going to take the form of (1) normal loss of GDP that you’d expect even in the absence of COVID (if any) and (2) loss of GDP associated with non-containment and the rapid spread of COVID. 240K additional dead in the US is going to have a rather dramatic effect on GDP as well, even if the dead are clustered around the elderly.

    It’s an insanely hard calculus to make and requires generalists. Economists are only going to see economic effects. Epidemiologists are going to be more focused on the spread of disease.

    1. It’s pretty subjective, but I’d like to suggest that Bernstein’s numbers understate the number of deaths that will occur if we don’t flatten the curve.
      1. Some (perhaps many) of the people who have serious cases of COVID but survive may suffer permanent lung damage, which in turn will shorten their lives. Those shortened lives must be factored in.
      2. As Italy found out, if you don’t try to flatten the curve, a whole lot of health care workers die or become seriously ill when hospitals become overwhelmed and lack the resources to protect their employees. This in turn will cause a significantly increased death rate down the line for all sorts of ailments, even after the epidemic is over, and many others will be reluctant to enter the profession, thereby putting their lives at risk. The consequent shortage in health care workers, along with the deaths that will occur because the hospitals run out of beds for serious ailments of all kinds, may increase the total death toll from a failure to act well beyond Bernstein’s numbers.

      1. If it is true that epidemics decrease the attractiveness of health care work, then yet, that should be factored in. I would assume the exact opposite result; epidemics lead to an increase in demand for health care work, and probably inspires more people to enter the profession. But I’d defer to the data. What are you looking at to make your prediction?

  4. If you don’t adopt a containment policy there will be many more cases, including not just deaths but recoveries. There will also be a greater demand for testing.

    Assume a 1% mortality rate. Then by your assumptions there will be 24 million additional cases without containment. Many will require treatment. That cost needs to be included.

    You also need to know not just the reduction in Covid-19 deaths from containment but also secondary effects. For example, the less strain the virus puts on medical systems the fewer will be the avoidable deaths from other causes. Overcrowded ER’s and ICU’s mean more people will die of other causes who wouldn’t have died otherwise.

    Then there is the insurance factor. You have to take variance into account. We have a decent handle on mortality from other causes, but we don’t know what the mortality from the virus is going to be. Maybe 240,000, or 480,000 or who knows. The customary way to deal with risky – high variance – situations is insurance. Is it worth over-reacting a bit – spending more than the expected loss – to prevent a catastrophic loss? What if the uncontained mortality is 1 million, and containment brings it down to 500,000. Note that your $11.2 million figure is 25% higher than the $9 million, which I think is the more commonly used number. How much premium should we pay to cover the uncertainty?

    As NTOJ says, you also have to estimate the GDP loss without containment. The $9 million gets at that, I think, but does not include secondary effects.

    1. The 1% mortality rate assumption, however, based on some data now out there from more thorough testing and data collection regimes without the propaganda (*cough* China) coming out of Germany and other European countries are showing a higher mortality rate than the flu (.1%) but much less than 1%…more in the neighborhood of .15%. In Italy, virtually all of the deaths are from people with preexisting health issues.

          1. It does not strike me as a good idea to reach a conclusion on the mortality rate based only on the small-sample size of the people on the Diamond Princess.

      1. It’s true we don’t know what the mortality rate is, but if it’s lower than 1%, and it could well be, that makes the costs of failing to contain higher, not lower if we work with Bernstein’s guesses about the number of deaths.

        Say it’s .1%. Then to get to 240,000 incremental deaths you will have not 24 million, but 240 million extra cases. Obviously silly, but at .5% you need 48 million, unlikely but not utterly absurd. So the treatment costs go way up.

        1. I’m pretty sure the 240,000 incremental deaths assumes a mortality rate. That is, it isn’t a constant and we will have more cases with a lower mortality rate.

      2. At the moment, according to Worldometer, Italy has roughly 47,000 total cases and 4,000 deaths. That’s close to 10%. Granted, those numbers may be wildly inaccurate, but in China deaths continued to occur even after the number of new cases began to decline, which hasn’t happened yet in Italy. It’s hard for me to believe it will get down anyway near .15 percent.

        1. Not only that; Italy’s numbers don’t include people who didn’t have COVID at all, but who died because the enormously overcrowded Italian hospitals couldn’t give them the treatment they needed.

  5. Isn’t there a questionable equivalence built into the utilitarian reasoning used in the OP? Seems like once you assign a value to individual lives, you also have to cope with moral implications for the distribution of loss among the population.

    How is it right to compare any particular imputed cost, whatever the sum, which is divided among almost the entire population (the survivors), to the same cost inflicted on only a tiny fraction of the population (the disease victims)? Are those moral equivalents?

    What kind of reasoning would treat as moral equals a $9-million loss inflicted on one person, to an identical $9-million loss divided among 320-million people. Multiplying the victims even a million-fold does close to nothing to diminish the moral problem.

    1. No, because money spent on saving one person is not spent on saving another. We would not spend $20 trillion (our approximate GDP) “saving even one life” because “you can’t put a dollar value on human life.” We would understand that we could spend that $20 trillion on saving many, many more lives than merely one. We also understand that if life is to be worth living, we want to enjoy it, so we would let people spend money going to the zoo, buying toys for their kids , lounging on the beach, or drinking a nice bottle of bourbon – even though that money could be reallocated to life-saving measures.

      The idea of putting a dollar value on a human life is that, beyond that dollar value, the dollars are better allocated to saving many more people’s lives, or spending that is prohibitive to allowing people to enjoy their lives.

      My husband said that the numbers ($5 million to $7 million) are calculated by looking at the money people spend to avoid risks of death. “It does not mean that people are willing to spend $7 million; it means that people are willing to spend $1.50 to avoid a really, really, really tiny risk of dying.” For example, the cost of adding seatbelts to cars: it’s some extra money in car manufacturing, an extra few seconds per day, but it saves a certain number of lives at a certain cost, and that “cost per life saved” is collectively less than $9 million.

      1. But we don’t do that. We spend vastly more on prevention per life lost to terrorism than we do for car accidents.
        This is not a rational calculus, but that’s because we’re not rational beings.

        1. My understanding is that the calculation is done by analysis of the choices people make, not necessarily the laws that require us to be safer. So if you buy a safer car, live in a safer area, pay for a flu or TDaP vaccine as an adult, buy a fire extinguisher for your house, etc., all those things get factored into how much we collectively spend and how many lives are saved through that spending.

          (I can ask my husband to explain in more detail – he’s a social science professor and teaches this for a living. I just know enough to be a PITA.)

      2. theobromophile — your focus on the legitimacy and methods of assigning a value to a human life is not my focus. My focus is on what happens after you assign such a value, and then must consider the implications. I asked some question intended to induce thought about some of those implications. Care to take on the questions asked in my last two paragraphs?

  6. “How is it right to compare any particular imputed cost, whatever the sum, which is divided among almost the entire population (the survivors), to the same cost inflicted on only a tiny fraction of the population (the disease victims)? Are those moral equivalents?”

    They definitely aren’t, but here it’s because you inserted “whatever the sum”.

    There are argument for prioritizing solving devastating consequences for the few in favor of even larger overall consequences diffused to the many. But you have to make the argument rather than just asking the question. And if you get there, what’s the multiplier? 2x? 3x? 100x? How diffuse?

    And your rule or limiting principle has to be consistent. COVID isn’t the only cause of death that clusters among subgroups.

    1. NToJ, can you give me an example of one of those arguments you are talking about, so I can better understand your point?

      1. Sure, go read A Theory of Justice by John Rawls, or the wikipedia page about it. I don’t find his explanation particularly persuasive, but a lot of others do.

        A form of the argument would be: humans are demonstrably risk-averse, to the point where we prefer risk-averse strategies over slightly more optimal risk-taking strategies. We bias towards overspending (arguably) to insure against remote but catastrophic results. People damned by circumstance with an utterly debilitating genetic disorder don’t get to make a social bargain beforehand. But the rest of us, through society, can. And so risk-averse humans may find ordering society in a way that equalizes happiness for the severely disadvantaged, even if it costs more to the advantaged, because most humans would make that trade if given the opportunity before having some debilitating disadvantage inflicted on them. It’s a self-interested trade off that we make on behalf of people who don’t get to make it, knowing that the debilitating disadvantage could just as easily have affected me (or may affect me later, or someone I care about, later), but by blind luck did not. Put differently, we may be willing to spend 2X to improve the lot of some disadvantaged group by only X, as a form of social insurance.

        Even these arguments have limits. You have to know what the trade off is. Because 1000X versus X makes sense only if there’s no shortage of resources. Triage, to the extent it is morally justified, is based on shortage.

  7. This was in response to Stephen. Sorry.

  8. Prof Bernstein, while I agree that we need a cost/benefit approach to this question, I think focusing only on the economic costs is too narrow. The “containment measures” we are implementing will unquestionably stop some coronavirus-related deaths. Those same measures will unquestionably cause some other deaths. You need to identify, quantify and include those deaths in your calculation.

    Sources of those deaths include:
    – homeless and others who are dependent on services and contacts that are no longer available
    – elders who will die isolated and alone because their support providers (or just random neighbors) aren’t visiting and won’t know that they’ve been hurt
    – families pushed into financial distress who skip needed health care
    – increased death rates among those without corona just due to stress (an aggravating condition for many diseases including suicides)
    – police overreactions when enforcing curfews and other restrictions
    – etc

    I believe that once you start to quantify the actual people these measures kill and compare it to the number saved, the economic costs to the survivors will become the smaller component of your cost/benefit calculation.

    1. This is all true, Rossami, but let’s not overlook the deaths, in a non-containment regime, from the strain on medical resources. I’m talking here about deaths from causes other than the virus. If non-containment policies overload ER’s and ICU’s and hospitals in general, then more people are going to die from heart attacks, trauma, stroke, and so on than otherwise.

      In addition, medical workers are likely, I think, to be disproportionately infected, and the loss or temporary incapacitation of lots of doctors, nurses, technicians, and so on will increase preventable deaths.

      In short, it’s a complicated issue, and while I understand that Bernstein was just trying a very broad-brush approach I think the brush was too broad for any meaningful conclusions.

      1. Yes, you will have to account for those second-order lost lives if they occur. So far, there are no examples of such medical displacement. Not even in Italy where the response has been … less than ideal.

  9. It makes sense to look at the costs of the current mitigation measures, but looking only at the number of potential deaths grossly understates the cost of imposing lower or no restrictions.

    First, if there are 240,000 deaths (give or take), there will also have been millions of people who got seriously ill but recovered. The cost of treating those people will be very high.

    Second, if millions of people are becoming seriously ill, it is unrealistic to assume the economy would continue functioning normally. We could easily see a situation where businesses are unable to function because of a lack of healthy or willing workers, which is basically where we are now, but without the massive number of cases. Moreover, food supply chains could become disrupted if there were an insufficiency of producers and/or transport personnel. That would set off a real panic.

Please to post comments