Changing How the World Thinks

An online magazine of big ideas


Were the Covid measures worth it?

The limits of cost - benefits analysis

rohit eCspQqd yDw unsplash 1

Answering the question can seem to be a simple matter of adding up the costs and benefits resulting from the measures. But things are a lot more complicated than that. What counts as a benefit or a cost is far from obvious. Even if we agreed to what they were, finding a common measure to compare them is riddled with objections. Ultimately, no algorithm can tell us whether the policies were worth it, it’s all down to human judgement writes Stephen John.


According to some analysts, a year of lockdown measures cost the UK economy £251billion. Was it worth it? You might think that the 251billion figure is too high or too low. Maybe the true cost was "only" £25billion. Still, would that have been a price worth paying? These aren't just questions for future historians. Rather, as worries around vaccine effectiveness against new variants grow and previously untouched countries go into lockdown, policy-makers around the world need to decide: when is the cure worse than the disease?

There's a way of thinking about the question, central to health economics, thrown around in debates around the pandemic: that we should adopt lockdown policies only when their benefits outweigh their costs. This answer is simple and appealing. Unfortunately, it's also false. Or, at least, it's an answer which hides far more than it illuminates. While it may be tempting to think that there must be some number - lying, say, between 25.1million and 251trillion - at which the costs of lockdown outweigh the benefits, this is an illusion.


The complications of cost – benefit analysis

To see why, it's useful to take a step back to think about the appeal of the notion that we should choose policy according to the net balance of costs and benefits.

Imagine a highly stylised example, associated with the philosopher John Taurek: you are a lifeboat captain; on one island, one stranger is about to die; on a second, nearby island, five strangers are about to die; you can save the one or the five, but unfortunately, not both. What should you do?

Even if we only focus on the health-related consequences of lockdown, most health economists think we should also consider "quality of life".

Faced with this horrible dilemma, most people think you should save the five. There is an obvious way of spelling out this intuition: in general, all else being equal, we should do as much good as possible. Plausibly, a situation where five lives are saved is better than one where one life is saved. Therefore, we should save the five, rather than the one. Of course, this is horrible for the one, but, still, the alternative would be worse.

This kind of stylised example is a long way from decision-making about Covid-19 and lockdowns. Still, it is easy to see how the same basic logic can be used to defend lockdown measures: had we done nothing, many more lives would have been lost. Therefore, lockdown measures were like the captain's choice: tragic, but necessary.

However, once we have made this move, matters quickly get murky. Imagine that the lifeboat captain gains some more information and learns that the one stranger is 80 and can expect to live ten more years if saved; each of the five strangers is 89 and can only expect to live one more year. Now, we might reason as follows: saving the one gains ten "life years", whereas saving the five only gains five "life years". So, saving the one would generate greater benefit.

This is precisely how some people think we should assess lockdowns: they worry that, even if lockdowns saved some lives, they also cost some lives, say, because of delays to cancer screening programmes. Even if they "saved" more lives overall, typically, the lives they saved were of older people. A proper accounting of lockdown, they argue, should have taken account not only of lives saved, but life years.

But once we have taken this step, we open a Pandora's box. Even if we only focus on the health-related consequences of lockdown, most health economists think we should also consider "quality of life". It would be better to cure one person completely of her fatal disease than to keep two people alive in a comatose state. Indeed, organisations such as the UK's NICE, which makes decisions about which drugs the NHS should purchase, compare interventions in terms of the number of "Quality-Adjusted-Life-Years" they generate. So, perhaps our assessment of lockdown should include not just lives or life-years saved, but include other health-related concerns, such as the number of people saved from "long Covid" on the one hand, or the effects of lockdown on mental health on the other.

Why stop there, though? Lockdowns don't just affect health, but children's education and adults’ economic well-being. Of course, education, financial well-being and physical health are linked, but it seems bad for kids to miss school or people to lose their jobs regardless of whether this affects their health. So it seems we should also take account of these other effects. But why stop there? Most of us make money to enjoy doing other things, such as buying a cup of freshly brewed coffee. Shouldn't we count the costs of all the small pleasures we missed when cafes were shut?

Sunetra Gupta, Hugh Montgomery and David Spiegelhalter weigh up risk, evidence and lockdown.


Money as the measure of all things

By this point, you can probably already see one of the core problems with the notion of comparing policies in terms of whether their "benefits" outweigh their "costs". Doing so requires us to find some way of measuring entirely different sorts of outcomes along a single scale. We need to find some common coin by which to assess, say, the badness of death, of contracting long Covid, of losing your job, and of missing your cup of coffee.

There are in fact tools for making this sort of comparison. For example, in some domains, such as health-and-safety, influential economists think we can calculate the “Value of a Statistical Life” by looking at people’s willingness to pay for risk reducing measures. (So, to take a toy example, if you are not willing to pay more than £1,000 to reduce a risk of death by 1%, then you should not be willing to pay more than £100,000 (1,000 x 100) to reduce a 100% risk of death. In our giant Cost Benefit Analysis, then, each life lost would be equivalent to a loss of £10,000.) We can, at least in principle, commensurate everything in monetary terms. However, the theory and practice of commensuration is contestable.

Some of these concerns arise from worries about "reducing" complex human goods to the financial realm. However, the underlying issues here aren't really about money at all. Consider the "Quality Adjusted Life Year". This measure is supposed to capture such intuitive thoughts as that a drug which eliminates debilitating pain does “more good” than one which cures ingrown toenails. Unfortunately, there are no straightforward biological reasons for ranking some diseases as "worse" than others. So, QALY scores for different diseases are typically constructed using the "standard gamble" technique. To simplify a little, this technique involves asking people questions like "if you could spend one year in full health, then die, or X years being blind, then die, what would X need to be for you to prefer blindness to full health?" Their answers to these questions are then used to calculate the "badness" of being blind.

If you are like me, you might find yourself baffled by such questions. Interestingly, studies suggest systematic differences between the answers given by those who have some condition and those who do not. Who counts as an "expert" in this scenario? Does the blind person have special knowledge of "what it is like" to be blind, or is she biased?

We need to find some common coin by which to assess, say, the badness of death, of contracting long Covid, of losing your job, and of missing your cup of coffee.


Impossible sums

So, the notion of comparing lockdown in terms of whether the "costs" outweigh the "benefits" faces the problem of identifying a common metric. However, it also faces an even more significant, but less obvious challenge: that it falsely assumes that we can aggregate harms and benefits across persons.

To get a grip on this problem, think about my list of possible "costs" of lockdown. Let's assume that, each day of lockdown, some number of people felt a little bit sad about missing their coffee. Let's also assume that, each day of lockdown "saved" 1,000 lives. The logic of comparing policies in terms of costs and benefits seems to imply that we should “add up” all of the little feelings of sadness and compare them to the number of lives saved. If enough people felt a little bit sad, then the aggregate “costs” - say, a zillion units of sadness a day - might outweigh the “benefits” - 1,000 lives saved a day.

This seems odd. Even if we had a cast-iron way of translating deaths and feelings of sadness into some common coin, there is no number of people feeling a bit sad which might justify foregoing a policy which saves a 1,000 lives a day. Twinges of sadness about missing fresh coffee, are simply irrelevant.

More formally, the problem with the notion of comparing the costs and benefits of a policy like lockdown is that it assumes that we can and should aggregate costs and benefits across individuals. We can engage in a moral mathematics where we add the benefits to lots of different individuals into a sum of total benefit, which we compare with the costs to other individuals, leading to an overall assessment about the net balance. This aggregative perspective overlooks the ways in which, in everyday life, we routinely make judgments of ethical relevance. We think that the coffee shop owner losing her livelihood might be "relevant" to thinking about the costs of lockdown, even if the lost pleasure of her customers is not.

In turn, such judgments of relevance shape our claims about the costs and benefits of lockdown. No serious person would ever attempt to argue that we should try to tot up the aggregate sadness caused by closing the coffee shops. However, such judgments are typically made implicitly or unconsciously, reflecting contestable ethical biases. Technocrats may not be aware that they are making ethical judgments, but in choosing what to measure, they are.  

Even if we had a cast-iron way of translating deaths and feelings of sadness into some common coin, there is no number of people feeling a bit sad which might justify foregoing a policy which saves a 1,000 lives a day.


What should be done?

So should we simply give up on the idea of evaluating lockdowns in terms of their "costs" and "benefits"? Of course, not. Even if there is no simple algorithm for calculating when lockdowns are “worth it”, our ethical judgments must be sensitive to, and guided by, the costs and benefits. A lockdown which costs more life-years than it saves is harder to justify than one which saves more life-years than it costs; a lockdown which leads to a £521billion economic loss is harder to justify than one which leads to a £52billion loss. A politician who claims to “know” what is best, irrespective of the numbers, based on her judgment or insight might as well be flipping a coin.

What we also need to recognise, however, is that the numbers we use to measure the costs and benefits of lockdown are always partial, as they rest on implicit, contestable decisions about what to measure and how to measure it. We need to ask whether all of the £521billion "loss" is relevant to thinking about saving lives, because an aggregate economic indicator captures both loss of livelihoods and loss of simple pleasures. And we need to ask whether the £521billion figure captures all the relevant losses of lockdown, because there might be harms, say to victims of domestic violence, which never turn up on economic ledger books. We need to ensure that the inputs to our analysis reflect our ethical concerns.

Unfortunately, it is unclear who is best placed to make these judgments. Reasonable people might differ, say, over whether loess of livelihoods is relevant when lives are on the line. It is not obvious who to trust. Furthermore, these judgments can be shaped by background inequalities. Lockdown may have exacerbated domestic violence, but victims lack the social voice which allows us to recognise the harms they have suffered. More cynically, given voting patterns in the UK, politicians may have short-term reasons to treat younger peoples’ concerns as “irrelevant”. Talk of “judgment” is easy, but good judgment is hard.

So, shaping the inputs to decision-making cannot be entrusted to technocrats or crowd-pleasers. Rather, it needs to be social and deliberative. We must be willing to listen to, and take account of, a multitude of different voices. Such a debate might lead us to recognise that certain numbers are helpful, and others misleading. Importantly, by drawing on widely distributed social knowledge, they might help us identify new options, which retain the benefits of proposed policies, but mitigate some of their costs. Once we have that kind of debate, it may well be that we should simply choose lockdown policies when their benefits outweigh their costs. In the absence of such open debate, I don’t know enough to say whether lockdowns have been worth it. What I do know, though, is that we cannot and should not expect that the epidemiologists or economists can tell us that either.

Latest Releases
Join the conversation

Sign in to post comments or join now (only takes a moment). Don't have an account? Sign in with Facebook, Twitter or Google to get started:

mabel goodrich 29 August 2021

The Novel Covid-19 situation has been damaging all of our everyday activities all over the world. Many schools, colleges, universities have been online due to pandemic situation! This helps people with finding ways to get connected with the people! They have a large amount of wonderful collections from around the world, especially for the adult group of people.