We should not put a value on human life

This is a draft script of a talk I gave at the Cambridge Union as part of a debate on the motion: "this house believe we can put a value on human life". I had to abridge it quite a bit for the actual event, but I can't be bothered editing it here and some people may be interested in a fuller articulation. 

In Stephen Spielberg’s cinematic masterpiece Jurassic Park, Scientists clone dinosaurs from prehistoric DNA. It’s all wonderful until the dinos get free and start eating everyone. Jeff Goldblum has a lot of iconic lines in Jurassic Park, but the one that made him meme famous was: “your scientists were so preoccupied with whether they could, they didn’t stop and think whether they should”. I put it to you that the allegory of Jurassic Park and the logic of Goldblum’s dialogue applies when it comes to putting a numerical value on human life. We can, but we shouldn’t.

There are a few ways analysts approaching determining the value of a life in economic analysis.

First, there are willingness to pay methods. These involve eliciting, either through direct questions or some sort of ingenious survey, how much money someone will pay to achieve appreciable reduction in their risk of dying.

There are two related methods. In the labour market approach, we look at the differences in pay between relatively more or less risky jobs. How much ‘danger money’ do you get for working in a warzone, that sort of thing.

And in the consumer preference approach, we look at how much people pay in real market transactions for safety features like airbags. We can then back out how much safety is worth to them.

Second, there is the human capital method, where we try to estimate the earning potential of an individual. We think about how long a person is likely to live, how much time they will spend working in that life, and their typical hourly wage. Then we multiply it all up to get their value as a human being.

So we can and do put a value on human life. Or so analysts will tell you. But it’s not clear to me that any of these techniques produce meaningful values. These numbers are actually illusory.  

One set of reasons for why these estimates are meaningless pertain to technical issues. Let me list some of these off quickly. First, it is well established that humans are terrible intuitive statisticians. We badly misjudge the base rate with which some events occur, we’re highly susceptible to the gambler’s fallacy, and our attitude to various risks depends substantially on how those risks are framed, rather than on the underlying math. Pointing out all this errors in our common statistical reasoning was one way that behavioural economics got started.

Why then would we trust people’s stated or revealed preferences for risk?

This is especially true with respect to events that rarely occur. Wilfredo Pareto, one of the main intellectual progenitors of the rational choice theory that underpins this sort of economic analysis, stressed that people’s behaviour could only be taken to reveal preference in cases where they have repeated the choice many times. Otherwise they simply didn’t have enough information.

Now with respect to risky thing, like car crashes and airbags, very few of us have any opportunity to learn what it’s like for those things to happen. That’s great. But it does mean that willingness to pay judgements are not rational, informed, or wise in such contexts.  

A second technical issue relates to long term predictions in complex systems, which is basically impossible. The butterfly effect is a catchy name for the idea that a small change in a deterministic nonlinear system can result in very large changes over time in that system.

This is a big problem for the labour market approach. I’m 37. Let’s assume, generously, that I’ll live to the classic academic retirement age of 65. So I’ve got 28 more years to work. What are the odds that I will be promoted to Professor in that time? Or that my next popular book will be a smashing success? What are the odds that my trade union will successful lobby for higher wages? What about the odds that the government will improve funding for universities? Or that university management will finally address the colossal administrative waste that drives much of our cost base? What are the odds that there will be an apocalypse in that time from AI, climate change, some rogue dictators, or whatever. All these probabilities swirling but we just take my wage and multiply it by my life expectancy.

These problems of projection and prediction bedevil nearly all cost benefit analysis. Take a very simple case – whether to expand a bus network in medium size city, or build a tram system. The tram system completely changes the long term structural development of that city by encouraging densification along a narrow corridor. The bus system instead encourages a proliferation of hubs. Neither of these long term developmental pathways can be predicted today, but they matter immensely for the wellbeing of the living in that city.

So we get some numbers of out these methods that seem ‘reasonable’, but it’s not clear to me that they are meaningful. They rely on assumptions stacked upon assumptions, many of which are little more than sticking a finger to the wind. And these assumptions are often immensely decisive. Setting an interest rate high or low can easily turn a long-lived infrastructure project from a bargain into a white elephant, and we’ve seen in the past year how unexpectedly and unpredictably interest rates can fluctuate.

But the bigger issue around meaning is really the normative assumptions – the ethical judgements –  that go into putting a value on a human life. The most obvious one being that everything is denominated in money. My willingness to pay. The market value of my labour.

I’m reminded of one of my favourite Saturday Morning Breakfast Cereal comics, called the ethical fourier transformation. This a joke method where when you’re confronted with an ethical conundrum, you convert it to the realm of economics. Having solved it there, you declare it a solved ethical problem.

In the comic, an economist advocating for the ethical fourier transform is challenged to explain how it can solve even basic ethical dilemmas, like whether it is ethical to steal bread to feed your family. The economist checks grain prices and declares that it is ethical until the projected price dip in Mid-November.

The interlocutor replies that this is a stupid method – it should always be ethical to steal to eat.

But the economist is crafty and asks “is it ethical to steal truffle mushrooms and champagne to feed your family?”

I guess not. The interlocutor reflects on this awkwardly for a while before saying that the conversation is weirding them out.

The economist shakes their fist: “come to the dark side!”

I put it to you that this is in fact the dark side. That it drives us towards dystopia. And that there’s plenty of evidence for that already.  

A central reason why we use money in cost-benefit analysis and the other methods that utilise the numerical value of a human life is that money is unidimensional and cardinal. This makes mathematical comparisons across options relatively straightforward.

But there are obviously very many things that we care dearly about that are not traded in markets, nor are they straightforwardly describable numerically in terms of magnitude or intensity. Things like love, freedom, agency, social harmony, neighbourliness, democracy, prestige, etc. Human wellbeing, especially at the social level, is made up of incommensurate, multidimensional factors that are often indivisible.

The consequences of brushing these complexities out of economic analysis are voluminously documented, especially in the policy space. For example, in Hilary Cottam’s book Radical Help, which concerns how to fix the British welfare state, she explains how aged care services are overwhelmingly commissioned on the basis of efficiency. Standard economic logic. We need value for the taxpayer dollar. We need to get aged care done at rock bottom prices so that we have more money for homelessness, or tax breaks, depending on your politics. Certainly we need ‘rigorous’ evaluation, and that requires measurement.

What’s measurable? The number of old people washed, wages, time.

What can’t we measure? Human care. Sensitivity. Being personable. Relationships.

So old people are washed brusquely by underpaid workers whose every incentive encourages them to rush. The old and vulnerable are left feeling violated and discarded by society, and the aged care workers much the same.

What is the value of these feelings?

Have we commissioned good ‘aged care services’ in this case? I think not. But economic analysis is frankly unable to commission good aged-care services because precisely what makes these services good, makes them humane, is something qualitative.

More broadly, the fundamental weakness of economic modelling is that it must assume sociology, psychology, and political science away for tractability. This can often result in powerful policy insights that improve human lives. An example is quota-trading systems for fisheries management, which have resulted in highly profitable and biologically sustainable fisheries in Australia and New Zealand.

But there are also many negative results, including some of today’s most pressing social problems, and most of these have some relationship to the value of human life, and certainly to economics’ impoverished understanding of human wellbeing as mere desire fulfilment.

In the US rust belt and to a lesser but still significant extent in the UK’s deindustrialised North, deaths of despair from opioids are accelerating at a frightening rate. Much of the academic literature on this topic is converging on the disintegration of social networks as a key driver of this hopelessness. Community pubs, scout halls, sports clubs and other sources of social cohesion were systematically defunded from the 1980s onwards because they contributed insufficiently to productivity.

Meanwhile, the prevailing economic orthodoxy was that if economic opportunity was leaving a place then we shouldn’t fight those market forces. It would be inefficient. Dying regions didn’t need help; people in them needed luggage so that they could move to opportunity. This again overlooks the centrality to human wellbeing of roots, connections, familiarity, identity, networks, history, tradition, culture, on and on. All that is considered is where wages are higher – that’s all that people can possibly want. So the brightest are incentivised to leave while rest stay behind in increasing misery.

The youth mental health epidemic, which any of my academic colleagues can tell you about, is a function of encouraging teens to see their value in terms of their personal ‘achievement’, whether financial or in terms of how hot and cool you are, rather than their self-actualisation or their contribution to a collective. Education is aggressively oriented towards preparing people for work, not life, and then student’s sense of self worth comes to be dominated by their grades, internships, graduate positions, not their interpersonal skills, relationships, wisdom, or happiness.

I’ll give you one story, again from Cottam’s Radical Help. After months of intensive social work, 2 children from a ‘difficult’ home were finally prepared for school. There household had been calmed down, they were being fed and chaperoned, they had uniforms. They were blocked at the school gates by the principal who was worried about the effect on this already marginalised and penalised school’s league table results if these 2 problem children were readmitted. They returned home, their hope and trust shattered. Funding models premised on literacy and numeracy results are the product of analysing human life in terms of the economic factors we can measure instead of the giving due to our full richness.

This is an example of New Public Management, the site of the greatest human carnage caused by economic logic is New Public Management – the application of economic logic to how government goes about its business, especially the commissioning of public services like education, health, social care, and job centres. Again, there were many bright spots, like garbage collection. But there are also so many spots of the darkest blight.

Mental health and social workers spending most of their time filling out forms to meet diagnostic criteria and justify funding. Pissing millions up the wall on spurious business cases in the name of not wasting taxpayer dollars.

Metrics and no-strings philanthropy.

Maybe you think this isn’t about the value of a human life. But it’s all the same central conceit – that the full richness of a complex puzzle is a distraction, an inconvenience that prevents us from rationally analysing a problem. So we have to flatten that complex problem into a few key variables that we can stick into a spreadsheet and run some maths on. Let’s just make some simplifying assumptions. That’s fine until your assume away the point of the exercise.  

I’m starting to drift into public policy, and it is here where I think faith in numerical, material models of human life have their most pernicious influence, which is to abrogate the moral responsibility of politicians and citizens and simultaneously debase the role and value of science in democracy.

Remember COVID? Remember how the politicians kept saying ‘we’re following the science’. The ‘science’ they were referring to were epidemiological models that make ethical assumptions about the value of life expectancy and cost-benefit analyses of the economic burden of lockdown that can’t handle non-market goods. These value judgements aren’t a matter of science. They’re ethical. They’re political. They are precisely the sorts of thing we want politicians to take responsibility for. Instead they abrogate that responsibility to science.

They do this precisely when science is least capable of making the associated judgements – high stakes, unpredictable, urgent situations like pandemics, financial crises, climate change, refugees, wars, etc.Complex systems

Scientists try their best in these situations, no shade on SAGE, but no matter how many caveats they pump into their reports about confidence intervals and ethical complexity, politicians are going to ignore all of that and just ask for a number. Politicians do not want to take responsibility for hard decisions, and scientists are too willing to let them abrogate responsibility, because it gives them power.

Beatrice Cherrier, a historian of economics, has investigated the slow disappearance of ‘welfare economics’ from the field. Welfare economics is essentially the exploration of normative issues in economics. Her conclusion was that economists realised that they obtained more policy influence the more they could pretend that their work was pure science. So the value judgements get buried ever further.  

When economists do cost-benefit analysis that want a dispassionate procedure that can be applied mechanically to spit out a number. Ethics, which is ultimately vibes, contaminates the desired sterility. But politicians and citizens don’t need to be dispassionate. They can make the value judgements. We should let them, not sweep all these value judgements under the rug.

Some economists will say that this is precisely what cost-benefit analysis is – a contribution to public discourse merely meant to stimulate debate. If that were true then economists wouldn’t be so outraged when policymakers disagree with policy analysis, and they certainly wouldn’t accuse policymakers of being irrational.

What this ends up doing is debasing science and corrupting public policy. We undermine democratic processes, reduce faith in technical analysis, and feed populist concerns that scientists sneak values into their technical analysis to arrive at conclusions that suit ‘elites’. So we should instead be much more reluctant to use models that put statistical values on human life in public policy.