Apples and Oranges: A Critique of Utilitarianism, Part 4
This is part 4 of a 4-part series on utilitarianism. This material is related to my forthcoming book, WILD PROBLEMS: A GUIDE TO THE DECISIONS THAT DEFINE US, coming on August 9. In Wild Problems, I critique what I call narrow utilitarianism — the day-to-day calculus of pleasure and pain — as a basis for making big life decisions. In the process of writing the book, I realized that utilitarianism is embedded in various ways in how economists think about decision-making at both the societal and individual level. I also realized that the utilitarian approach tends to encourage economists to focus on the measurable at the expense of what is not measurable and that this can lead us badly astray, a theme that is at the heart of my book. So this essay is in some sense the intellectual backstory of the issues that I write about in WILD PROBLEMS.
In part 1 of this essay, I argued that measurement is crucial for science and progress. But we have a craving for certainty. This encourages us to reduce complex things into simpler things so we can compare them especially when making decisions. Here I explore various ways that this desire to quantify complexity seduces us into ignoring things that are less easily measured. We particularly are drawn to scalars — a single number — as a way of summarizing a more complex situation. This phenomenon is an example of what Jerry Muller calls the tyranny of metrics.
In Part 2 of this essay, I gave a number of examples of how our love for quantifying things can lead us to oversimplify an issue and cause us to pursue goals that are not the ones we truly value. I ended that section by introducing Jeremy Bentham, the man who hoped to make morality as scientific as physics. Here I explain why he failed and why his philosophy of utilitarianism while very popular even today is deeply flawed.
In part 3 of this essay, I described Jeremy Bentham’s view of how we should think about decision-making both at the individual and group level. Betham argues that life is a series of pleasure and pain and that we should try to maximize the former and minimize the latter. Economists adopted Bentham’s perspective on human behavior arguing that this is how human beings behave and how they rationally should behave. Unfortunately it is hard to remember that not everything is easily measured or quantified — the calculus of well-being can easily mislead us. I expand on this theme in this final section and give some examples of economists’ conclusions that I think are not warranted.
Outside of a few creative practitioners like Gary Becker, economists have had little or nothing to say about the decisions people make as parents or how they behave in religious communities, or how they decide whether to vote — a decision economists constantly decry as irrational — or why gift-giving is a central feature of human societies from time immemorial and persists even after economists point out its “irrationality.” For the most part, economists focus not surprisingly on our behavior as humans in the marketplace — our behavior as consumers, workers, employers, or investors.
And because economists take preferences as given, there is nothing in the economist’s view of humanity that who we are today need not be who we are tomorrow. In worldview of the mainstream economist, there is little scope for what the philosopher Agnes Callard calls aspiration.
All of this would be fine except that economists easily make judgments about broader life outcomes well beyond the marketplace. Should the minimum wage be increased? Is free trade good for America? Should we allow driverless vehicles that might lead to millions of taxi drivers and truck drivers being out of work? What kind of schools reforms make students better off? These are all questions that go well beyond the material, the commercial, and the easily measurable. Do economists have anything definitive, anything precise, accurate, and indispensable to say about these questions?
Jeremy Bentham certainly thought that someone should have something to say about these questions. He argued that when policy-makers try to evaluate a policy, all they need to take account of is the pleasure and the pain. The benefits and costs.
Alas, there’s a catch to Bentham’s vision. Bentham offers no way to quantify the value of pleasure and pain. While the duration of pleasure or pain could imaginably be quantified, Bentham’s very first category, intensity, which is the heart of the matter, can’t be measured even by the person who experiences it. There are no units to conceive of the measurement.
In the early days of economics, some economists conceived a measure of utility — utils. But where is the util to be found? Where is the utilometer to sense it, measure it, and show it in the printout? Pleasure and pain can’t be quantified. Bentham’s edifice is a castle made of sand, an exercise in description and classification rather than a working tool for an individual seeking happiness or a legislator seeking a guide for designing good public policy.
And if utility cannot be measured in practice, if the value of pleasure and pain cannot be computed, what can Bentham’s famous adage — the greatest good for the greatest number — possibly mean?
Suppose you have two tickets to the musical Hamilton and you invite me to come with you, your treat. Alas, I’ve already seen it and unknown to you, it’s my 25th wedding anniversary, so I turn you down and take my wife to dinner at an elegant restaurant. You find a different friend and have a great time. Who had more happiness from the evening, me or you?
The right answer is I have no idea. No one does. Not me. Not you. And certainly not a philosopher king economist looking on from the outside. No amount of data on the frequency of our smiles during our two very different experiences can answer the question. It’s a meaningless question that falls outside the purview of science or social science.
Or try this variation. You can’t find anyone to go with you so you watch Hamilton alone. You discover you don’t like hip-hop musicals. Given the hype and the expectations, you leave the theater badly disappointed. I, meanwhile, enjoy my anniversary dinner tremendously. The food’s spectacular. The wine, divine. My wife and I spend the evening in rapturous conversation savoring our marriage and each other.
Did my pleasure outweigh your pain? Again, a totally meaningless question. It’s apples and oranges all the way down. I was delighted to discover that Bentham himself used fruit to illustrate the difficulty:
’Tis vain to talk of adding quantities which after the addition will continue to be as distinct as they were before; one man’s happiness will never be another man’s happiness: a gain to one man is no gain to another: you might as well pretend to add 20 apples to 20 pears.
Bentham himself was deeply frustrated by this practical difficulty. My kingdom for a scalar. But how?
Modern utilitarians argue that in some cases, the benefits are so much greater than the costs that measurement is irrelevant. Imagine a policy that doubles the income of every poor person in the United States but as a result, Jeff Bezos has to pay an extra dollar for a cup of coffee. Surely the gains to all those poor people outweigh the trivial cost to Bezos who already has so much.
Or consider Peter Singer’s example — a child is drowning as you walk to work. Surely the gains from saving the child outweigh the cost of ruining your fancy shoes and being a few minutes late. Surely, the ethical choice is save the child.
I think most people would agree that you should save the child and that Jeff Bezos should willingly pay more for coffee if he could help millions of people. And happy people would be happy to legislate these solutions — taxing Bezos a small amount if the proceeds could be spent in a way to lift millions out of poverty and taxing people who are well off with jobs and nice shoes if it can save the lives of children. But these two examples are reductio ad absurdums. What they actually imply for harder questions in the real world is much murkier.
Let’s take the Bezos example. The utilitarian argument is that the world is a better place if a nearly trivial cost paid by Bezos can lift millions out of poverty. The argument is that the pain endured by Bezos is smaller than the pleasure received by millions of people improving their material well-being. You probably agree and I do, too. But what are you willing to do to Bezos to help millions? Take half of his income and his wealth? Take more than that and turn him into a pauper? Destroy his ability to create Amazon? Torture him on pay-per-view TV along with his extended family?
Somewhere in there, you would probably draw a line and say that such a policy makes the world a better place but more than that makes the world a worse place. Some would draw the line at the very beginning — no one has a right to harm someone in order to help others. Others might never draw the line arguing that hurting one person no matter how grotesquely is worth it if millions can be lifted out of poverty. And still others might be content to say say they don’t know what it means to make the world a better place. It’s too complicated.
But the promise of Bentham, the promise of utilitarianism is that the question of where to draw the line can be drawn objectively, scientifically, precisely. That promise cannot be fulfilled. Of course if you have a choice between saving ten lives instead of one life, everything else held constant, then save ten lives. But that’s an apples and apples problem. Most of the world where the tough questions are inevitably ubiquitous is apples and oranges. No calculus is available for those challenges.
I’m not saying data is irrelevant. When you are deciding how to spend your charitable contributions, try to find out what the impact is on the people you are trying to help. Find out how many will be helped. The amount of good and how many are helped are both relevant. This is what is admirable about the Effective Altruism movement — charity should help people not merely signal virtue or bring comfort to the giver.
If you are a legislator trying to decide how to vote on the minimum wage, you will surely want some measure of how many people will be helped and how many might lose their jobs even if you know that these scalars cannot easily be combined. But don’t pretend that all the costs and benefits can be reduced to a scalar or that the things we cannot measure are unimportant.
When I ask you which movie you’d rather watch tonight, This is Spinal Tap or When Harry Met Sally, you can probably come to a decision. In the economist’s world, we pretend that you weigh the prospective pain and pleasure from watching each movie, or perhaps the expected pain and pleasure from watching each movie, and come to a decision. Is that what people literally do when choosing a movie? It may not be a bad approximation. But that does not imply that we can determine which movie is better for the family if three members vote for Spinal Tap and only two vote for Harry Met Sally.
Majority rule is one way to deal with the apples and oranges problem. But there is nothing utilitarian about using that method. It leads to a decision as long as the number of family members is an odd number and everyone votes. But that does not imply that somehow the family is better off watching one movie rather than another.
You can’t even argue that it would be a better world if the family split up and three of the family members watched Spinal Tap while the other two went into another room to watch the movie they prefer. Maybe something that can’t be quantified is lost when the family splits up in this way. It’s not a question that can be answered by philosophy or social science in a decisive way. Each family deals with these challenges in its own way. Those solutions inevitably take account in some way of what members prefer and how intensely. But there’s no Benthamite arithmetic in the literal sense of adding up pain and pleasure to measure the gains or losses to the family from choosing one movie over another.
The apples and oranges problem has bedeviled economists ever since they decided to weigh in on public policy. Let’s return to the question of increasing the minimum wage. Millions of poor workers get a raise. But what if that means that the poorest, least-skilled workers possibly can’t find work, can’t provide for themselves through their own efforts, can’t provide for their family, can’t have dignity, and face a lifetime of dependency.
Can you even make the case for raising the minimum wage even if all it does is ruin the life of one worker, let alone the thousands or hundreds of thousands who might actually lose their jobs or who struggle to find one in the first place? Sure, you can count or at least estimate the dollar gains in higher salary to the workers who keep their jobs. Can you weigh that with a clean conscience against the loss of dignity of that one worker?
You might say yes. Someone else will say no. Dostoevsky would say no, I think, based on a similar example in the Brothers Karamazov. Ursula Le Guin would say no based on her story, “The Ones Who Walk Away from Omelas.” You might say yes. And if you have to make that decision whether to raise the minimum wage, certainly the number of workers who are helped and hurt would matter. But it’s not a scientific assessment. It’s a judgement that is inevitably loose, vague, and indeterminate. What I have been trying to say is that if you are not careful, you may come to ignore the intangible gains and losses in dignity and pride that cannot be quantified.
But even if you remember those intangible factors, people will differ as to whether raising the minimum wage is good for the poor or good for the country. That is not a scientific question amenable to measurement. But if all you economists can say is that some people win and some people lose, what’s the fun of being an economist?
Thomas Sowell likes to say, quoting, I think, George Stigler — that the essence of economics is understanding there are no solutions only tradeoffs. Any intervention comes with costs and benefits that cannot be easily compared in order to come to a judgment about what is best.
Harry Truman longed for a one-armed economist because then he wouldn’t be able to say “but on the other hand” because there’d be no other hand. In other words, Truman wanted to turn an economist from someone prone to loose, vague, and indeterminate observations into someone who is precise, accurate, and most importantly, indispensable. But that’s like saying I’m having trouble deciding between moving to Austin and moving to Boston. Which is going to be better for me? There’s no straightforward answer.
An economist who has two arms and can only focus on tradeoffs becomes a mere functionary in the palace of the king. The one-armed economist can be the power behind the throne.
Here is how it’s done. Like Waldfogel’s analysis of gift-giving, you want to be able to convert the intangible factors like how much please a sweater gives you, into some dollar measure. That will let you combine apples and oranges.
Consider one small piece of the North American Free Trade Agreement that allowed brooms from Mexico to come into America without tariffs. Allowing that competition will lower the price of brooms in the United States and destroy some or all American broom-making companies.
By importing brooms from Mexico, 100 million broom buyers gain a dollar, the savings per broom from lower prices now that brooms from Mexico are allowed without tariffs. As a result let’s say a thousand American broom makers will lose their jobs. Profitable American businesses will disappear and the towns where those businesses are located will struggle. How do you weight those changes? In theory, you can convert the losses into a dollar amount. Would you want to just add them all up to decide whether to rid of tariffs on brooms?
I reject this calculus. As would almost anyone who isn’t trained as an economist. Are you serious? Ruin the lives of a few thousand workers so 100,000,000 Americans can gain a dollar? Why would you ever thing that adding up the dollars lost and gained is the right way to make this decision? Why would you implicitly assume that 100,000,000 people gaining a dollar is equivalent to say a thousand people losing their income and the pride they get from their work?
Despite this point, I still think free trade is good idea. Why is a different story for a different book — see my book,The Choice: A Fable of Free Trade and Protectionism. But I am in favor of free trade because the broom story misses other intangible benefits from open borders that are even harder to quantify but very important — the ability of each generation to shape life according to the dreams and skills of that generation. But I don’t pretend that’s a scientific judgment, even though there are dollar-denominated calculation that show that the gains from trade that can be quantified exceed the losses.
The dollar gains and losses aren’t irrelevant. But such a calculation ignores everything that can’t be put into dollars and more importantly assumes that a dollar is the same to everyone. A million people who get a dollar’s worth of benefit are then the same as a thousand people who lost $1000. But that doesn’t make any sense unless your only goal is to quantify something that’s hard to quantify. Such a calculation assumes that there is something called “society” apart from the individuals who make up that group, something called society that absorbs the gains and losses in total and becomes better or worse off accordingly.
There is a sense in which this makes sense, and economists use it relentlessly. If the gains are larger than the losses in dollar terms, that means it’s possible to share gains with the losers, compensating them for their losses. If the gains are bigger than the losses, then there will still be enough left over for the winners to enjoy. So everyone is at least as well off as they were before and no one is worse off.
That the losers are in fact not compensated in practice becomes a mere footnote. All that matters is the “net gains to society” — the dollar sum of gains and losses. But this calculus is morally bankrupt, even when compensation does take place. It ignores the problem that we all value money differently. It implies that offering a homeless person $10,000 to appear on television to be publicly tortured is worthwhile as long as the promoter of the event can sell ads that make at least $10,000 in profit.
Plus, politicians aren’t Benthamites. They don’t just care about maximizing gain over pain for the society as a whole and then compensating losers from policies they approve. People with little or no political power tend to be ignored in assessing the virtue of legislation while people with political power tend to get what they want.
But what about life and death? Happiness, pain, pleasure may be ambiguously measured. But death is surely a good thing to avoid, right? Yet here we are in 2020 in the middle of the coronavirus pandemic with people saying that to prevent death, we must close the schools, the bars, the restaurants, places of worship, stadiums, arenas, and so on. It’s a good argument. Death is a bad thing. But so is the despair of not being able to pay your rent. So is the loss of skills from a wasted year of school on Zoom for a ten year old child.
That’s the ultimate apples and oranges problem. How do you weigh them? You can easily hold both of these complex ideas in your mind — the deaths of people on the one hand and the human losses from lost employment, lost dignity, shattered dreams from business started and shuttered, social skills and educational development lost due to elementary school being held via Zoom, on the other. But you can’t get to bigger vs. smaller as our brains would like to and thereby objectively decide which is more important.
You can try. One common response to this incomparability is to use the common denominator of death. Skeptics of lockdown or sheltering-in-place in response to the pandemic pointed to an increased suicide from lost work and social connection, or deaths from cancer treatments missed in the face of fear. But this solution, tempting as it is, misses all the losses from a reduced quality of life when there are are few or any restaurants, bars, concerts, schools, religious services, and so on.
Economists actually pretended to add up these costs and benefits arguing that sheltering-in-place was a good idea. It may indeed have been. But putting numbers on the costs and benefits don’t easily justify that assessment. Back in March of 2020, at the very beginning of the pandemic in the United States, Luigi Zingales of the University of Chicago argued that sheltering-in-place could save 7.2 million lives (including people who would be able to get to the hospital for their cancer treatment and the like). [https://promarket.org/2020/03/13/captured-western-governments-are-failing-the-coronavirus-test/].
He argued that a single life saved was worth $14.5 million. Try not to laugh at the decimal point being used for something that is inherently unmeasurable. A dollar value for a life lost is often used by economists and policy analysts. The number is based on the amount people forego to avoid the risk of death. It is a measure of the value people place on reducing their risk of death. As a result it is highly sensitive to how much money people have. For that reason alone, it is worrisome conceptually but let’s pretend you can actually measure the value of lost life in dollar terms.
Then Zingales tries to correct this number for the fact that saving the life of a 20 year old is not as “valuable” as saving the life of an 80 year old. Because most of the lives saved from stemming the pandemic would be old people, he reduced that per-person number of $14.5 million by 37%, the factor the EPA uses when they try to evaluate saving elderly lives.
Zingales concluded that the potential gain to the United States of avoiding 7.2 million mostly elderly people dying later than via Covid was $65 trillion. And because the GDP of the US was a mere $21.3 trillion, putting in the economy in the deepest of freezes from sheltering in place so that economic output was zero meant to him that three years of zero economic activity would pay for itself in the value of lives saved.
I’m a big fan of Luigi Zingales. I’ve learned a lot from him. But in this case, I think he’s off the mark. Not because 7.2 million lives saved in the United States is off the mark. And not because $14.5 million is a truly imprecise measure of the value of life that doesn’t deserve a decimal point. And not even because 37% is a subjective shot in the dark at making the value of an elderly life saved relative to a younger life an objective measure. I think it’s wrong because the lost dollar amount of GDP doesn’t capture the cost of putting the economy in the deepest of freezes where there is no economic activity. Forget the failure to measure despair and lost meaning from people sheltering in place for three years. If such a freeze led to a dictatorship in the United States, would it be worth it? Foregone GDP just doesn’t begin to capture what would be lost were we to freeze the economy for three years.
To be fair to Luigi, he might have felt this was just a quick back of the envelope calculation to dramatize the costs of letting the disease spread unimpeded. But I think that calculus omits so many things in the name of creating a single number ($65 trillion) to be compared to another single number — foregone GDP that it is an exercise in faux scientific measurement.
Let’s make it even simpler. Should a frail 80 year old grandmother attend the wedding of her grandchild if it increases her chance of getting Covid and dying from it? Forget the potential for masks and social distancing and even nearly-perfect tests for the virus to reduce the risk to the grandmother. Suppose it is indeed very dangerous. What is the right thing for her to do? Would you tell that grandmother, whose life could end any minute from causes other than Covid that she is making a mistake to risk her life in order to celebrate with people she loves and might never see again?
You might argue that that grandmother should be free to risk her own life, but surely she should not do anything that might jeopardize the lives of others. Yet when people took to the streets to protest the death of George Floyd, crowding close together both with and without masks and surely spreading the disease more widely, mainly commentators and politicians refused to censure those efforts. And correctly so. How do you weigh justice vs. an increase in cased of Covid? You might think those protestors were ineffective. Or that if effective, their impact wasn’t sufficient to justify increasing the incidence of the virus. But would you really pretend to be able to measure the value of lives saved against injustice? (And yes, it’s tempting to argue that the goal of the protests is to save even more lives by reducing deaths at the hands of the police. But even if more lives would be lost from the increased cases of Covid, it can still be worthwhile to protest.)
Bentham steered us wrong. And by us, I mean the economics profession and the world. Public policy cannot be decided scientifically. Morality is not a form of applied calculus. Not all costs and benefits can be measured and those that can be measured are measured imperfectly, usually in dollar terms. Such calculations can inform decision-making, but they are rarely decisive in and of themselves. If we are not careful, we forget about the factors we cannot quantify, and we forget that some measures aren’t what we actually care about.
Sometimes quantifying a subset of factors can lead to overconfidence about how well we have mastered the problem we are trying to improve. We forget about what is in the shadows and assume we have an answer when all we really have is information. More data is only helpful when we remember the limits of what we are measuring.
But I think Bentham steered us wrong in a deeper way, and that is his attempt to convince us that a life well-lived is the accumulation of pleasure and the avoidance of pain, even when we broaden this calculus of happiness to include pleasures like friendship and pride in our reputation as Bentham did. A life well-lived, as I argue in my book, Wild Problems, is not a calculus equation to be solved.