My work on these questions arose out of my dissatisfaction with other proposed answers. Here are three of the most popular.

First: The story I just told about how much Amazon is willing to pay its different workers is a crude version of neoclassical economics’ theory of wage determination, according to which workers are paid the value of the difference they make to their firm’s production — their “marginal revenue product.” So some people argue against severe income inequality by objecting to the idea that a worker’s marginal revenue product is a kind of “contribution” that we should care about. But swap in any competing conception of contribution, and you run into the same problem for egalitarianism: some workers seem to contribute dramatically more than others.

Second: Some people concede that point but deny that labor markets track contribution, however it is understood. They look out onto advanced economies and see people making obvious contributions — teachers, construction workers, janitors, farmers, factory workers — getting paid relatively little while people with “bullshit” laptop jobs get paid a ton. On this view of the economy, its winners are just gambling at high-frequency trading firms or fooling around at Google, spending down the money printed by an ad monopoly. Or maybe some of the winners are making a real difference, but a negative one — inventing social media and smartphones and sports betting apps, ruining the world.

“Every rich person’s job is fake or bad” has a kind of populist appeal. But lots of businesses make their money by providing things that we are happy to have. The claim that there is no relationship between productivity and compensation in actual market economies is hard to credit.

Third: Maybe in part for that reason, a different answer is popular within political philosophy (and, in my experience, among economists). Many political philosophers these days are “luck egalitarians”: they think that, to quote Larry Temkin’s summary, “It’s bad — unjust and unfair — for some to be worse off than others through no fault (or choice) of their own.”

In the meritocratic vision, someone born into ordinary circumstances rises in part on the basis of their own natural talents. But what did they do to earn these talents? Why give them credit for winning the “lottery of birth”? This is the central luck egalitarian idea — a radical expansion of the remit of “there but for the grace of God go I,” from the circumstances of a person’s birth into the nature of the person themselves.

On the luck-egalitarian view, what is unjust about the situation of the least-paid workers is that they are compensated so much less than others for reasons beyond their control: because they were not provided the same kind of education, or otherwise through no fault of their own have less productive talents than the higher paid.

Luck egalitarianism has a kind of alien beauty. But it is, I think, alien. As Robert Nozick once observed, if we are to dissociate ourselves from our most basic qualities, like our talents, it is hard to see what is left for us to be.

There is a better answer to the contribution principle’s apparent inegalitarianism. It has never been articulated in a precise way, but it is a version of a familiar thought — the thought that “society” plays some role in the contributions of individual workers, a thought most associated with early twentieth-century British liberals and (“Fabian”) socialists.

Return to the idea of “essential workers.” Lots of people who do essential work are paid little. Why? Why would essential work, if essential, be paid so little? The empirical economics literature addresses a narrow version of this question: How can workers’ wages fail to reflect the differences that they make at the margin, their marginal revenue products?

But there is a deeper question to ask: How can the differences that workers make at the margin fail to reflect the way in which they are essential?

My answer is that the contribution made by many essential workers, and by many low-wage workers more generally, is a collective one, in a strong sense: the workers together make a difference that is much larger than the sum of the differences they each make on their own. This basic fact can explain why essential work is paid so little, and why the contribution principle does not have the inegalitarian implications that it might seem to have.

Collective contributions are ubiquitous. Think about the workers who maintain highways or who staff grocery stores. Their collective abdication would be a disaster, yet you don’t seem to reach this disaster by adding up the differences they each make one by one. More important, think about the division of labor itself. We get the enormous gains of specialization whether we have n or n + 1 workers in each role. That means that the division of labor is a collective achievement, that its significance cannot be captured by adding up the marginal contributions of those whose labor realizes it.

The problem with individualistic versions of the contribution principle, like reward according to marginal revenue product, is that they are blind to collective achievements.

To see this, we can compare two radically simple economies.

The first is an economy of just two workers. One worker makes a contribution equal to all productive output: without their labor, nothing of value would be produced. But this contribution depends on the second worker: without the second worker, the first worker’s labor wouldn’t make a contribution at all. (Imagine, for example, that the first worker makes lamps and the second produces the electricity that powers them.) Reward according to marginal revenue product registers this dependence; making a difference to the difference that the first worker makes to output is itself to make a difference to output. So the workers have the same marginal revenue products. So far so good.

But now imagine that the second worker is instead a group of workers, and that the first worker’s output depends on what they only together do: without the group’s labor the first worker’s labor wouldn’t make any contribution, but no individual member of the group makes much of a difference at all.

Reward according to marginal revenue product says that the other workers get only the scraps that accord with their individual difference-making; almost everything goes to the first worker, even though that first worker’s labor is wholly dependent on the others’. This does not make sense.

It’s not that the contribution principle simply ignores the other workers’ collective achievement. It’s worse: that collective achievement still shows up, but mainly in the “individual” contribution, and therefore the pay, of the first worker. (For economists tempted to dismiss this as an unproblematic and trivial consequence of the fact that wages are set by marginal and not total or average products, the point is that the marginal productivity of the one worker is itself partly determined by the total product — of the other workers.)

The same thing happens in real economies. Think of an Apple employee who develops software for the iPhone. We can assume that their individual labor is very valuable. But notice the crucial fact: the value of their labor wholly depends on collective contributions elsewhere in the economy — on lithium miners, on Foxconn assemblers, and on and on. Even if these other workers lived in a country where wages were as high as in the United States, they would be paid one by one, according to their individual marginal revenue products, which may well be small. But what they accomplish together is obviously important: without that labor, there are no iPhones.

Through the seemingly innocent mechanism of paying workers one by one, the labor market performs an extraordinary trick. It smuggles the importance of this collective labor all the way to the end of the “value chain”; it enriches not those whose labor it is but those who depend on it.

What does this all mean for severe income inequality? Imagine that a highly paid worker objects to paying more in taxes: Why should the government take so much of my pay, when I’m paid the value of my labor? Our reflex may be to deny that their labor really is so valuable. But how do we know? Maybe it is. No one actually believes that every job that requires a laptop is “bullshit.” (I wrote this article on a desktop, for what it’s worth.)

Here’s a better response. The size of this worker’s contribution depends on what other workers only together do. In a labor market that pays workers one by one, this means that the worker’s contribution is swelled by effectively unpaid labor elsewhere in the economy. Money that should go elsewhere instead goes to the one worker.

The point is not that the worker’s labor is less valuable than the market thinks. It is that the value of their labor is less theirs than the market thinks. Many workers play a role in determining its value, but not all of them are rewarded accordingly.

According to a natural objection, the groups of workers I’m talking about, like lithium miners, don’t contribute that much even all together. If they didn’t mine the lithium, then other workers would replace them. And this shows that, even all together, they do not make much of a difference.

This objection rests on a misunderstanding about contribution and causation. Consider a simple thought experiment: I’m throwing rocks at a bottle, and so are you. Eventually, I hit the bottle. It shatters. But if I hadn’t hit the bottle, then you would have. So I’ve made no difference: if I hadn’t thrown my rock, the bottle still would have shattered. Do I, despite making no difference, cause the bottle to shatter? Yes, of course.

Philosophers call this “preemption”: I preempt you, my would-be replacement. The lesson of cases of preemption is that effects need not depend (“counterfactually”) on their causes. (They might still need to depend on their causes when other things are held fixed. If this calls to mind the methods of causal inference used in the social sciences, that is not an accident.)

The notion of “contribution” in the contribution principle must be a causal one. If I bake a cake for your birthday party, then I have contributed a cake; it would be bizarre to claim that I did not actually contribute because someone else would have baked a cake if I hadn’t.

The workers who mine lithium are causally responsible for the mining of lithium; and so their causal contribution to the production of iPhones is maximal. As is the contribution of the Foxconn assemblers, and the product managers in Cupertino, and so on. Any supply chain with multiple necessary processes displays this same equality of causal contribution. (Compare this with the economists’ “value-added” framework, where the value of even a necessary process is just the difference in the market price of what comes out of it and what went into it.)

So the contribution principle, properly understood, does not vindicate severe inequalities of pay, not in very complex economies like ours. In a follow-up paper, I make the affirmative case that compensating workers in line with their productive contributions in fact means making their pay more equal.

The tools for accomplishing this are familiar. We should tax higher-income earners to raise the effective income of the lower-paid by expanding the welfare state; and of course the socialization of the means of production that is socialists’ ultimate goal would free up for redistribution money that is now private profit. But measures that make pretax wages more equal (“predistribution”) are also important, not least because of the persistence of the belief that I have been arguing against — that workers deserve what the market pays them and so should not be taxed (much).

Collective bargaining is one such mechanism, and the one that an emphasis on essentially collective contributions calls to mind. Because the contribution principle points in the direction of equalizing pay, it recommends a form of collective bargaining that looks something like the “solidarity wage” policy that Swedish social democracy achieved at its height. This model involved nationwide agreements between employers’ and workers’ federations that brought high and low wages much closer together. The policy’s proponents argue that it also made companies more internationally competitive. But even if this weren’t true — even if there were economic costs to compressing wages — the question has to be whether avoiding these costs is worth making low-wage workers suffer under injustice (the “equity-efficiency trade-off”).

That is all about a single country’s economy. But of course many of the world’s best paid workers in the Global North causally depend on the collective labor of workers in the Global South who make dramatically less. The market pays the former some of what, in a just world, would go to the latter. The political and technical barriers to addressing global inequality are daunting, recent reductions in extreme poverty notwithstanding. There is a temptation to say, with John Rawls: “Our social world might have been different and there is hope for those at another time and place.” But the direct transfers recommended by effective altruists and advocates of individual duties of aid perhaps start to look more attractive to those of us on the Left if we think of them not as a replacement for structural solutions to global inequality but as something, anything, to do in the meantime.