Robert H. Frank at NYT:
EVERY year at tax time, libertarians indignantly denounce government income transfers from rich to poor. Society’s income distribution, they argue, should reflect as closely as possible what people would earn in unregulated private markets.
When critics on the left counter that income transfers are required in the name of social justice, libertarians yawn — and the debate goes nowhere.
There is a more fruitful way to look at the issue. Suppose, for the sake of discussion, that we grant the libertarian premise that private pay systems provide the best ethical template for society’s income distribution. As closer scrutiny of that premise will make clear, the libertarian denunciation of income transfers fails on its own terms.
The main problem is that private pay patterns embody an implicit tax that is actually far more progressive than the federal income tax. To understand why, first consider some background about the way these patterns work.
Economic theory holds that in competitive labor markets, workers are paid the market value of what they produce. In actual markets, pay does rise with productivity, but not by much. The most productive carpenter in a framing crew, for example, might produce twice as much as his least productive colleague, but is rarely paid even 30 percent more.
To see the pattern at first hand, consider groups of co-workers who perform similar tasks in your own company. In one case, suppose that your two most productive co-workers leave the job; in the other, suppose that the three least productive leave. Which group’s departure causes a greater loss of value? Most people would answer that losing the top two hurts more.
If so, economic theory holds that their combined salaries should be higher than the combined salaries of the bottom three. Yet the typical pattern is the reverse: any three workers in a group performing similar tasks earn substantially more than any other two.
In short, the startling fact is that private businesses typically transfer large amounts of income from the most productive to the least productive workers. Because labor contracts are voluntary under United States law, it would be bizarre to object that these transfers violate anyone’s rights.
David Friedman on Frank:
It is an ingenious argument, but there are at least three problems with it:
1. Frank’s analysis of the effect on pay of concern with status implies that what people care about is not their status relative to the rest of the world but their status relative to those near them—in his case, their fellow employees. The more distant someone is from me, the weaker the effect—a fact some of whose implications I discussed long ago on this blog. Most of the people who benefit from income redistribution are very far from me. That does not eliminate his argument, but it weakens it. When he writes
“For starters, high-ranking members of society, who also tend to have the highest incomes, know they will be able to send their children to the best schools and have access to the best health care.”
he fails to distinguish between absolute and relative values. Getting good health care is very important to me, but its value does not depend, so far as I can judge by introspection, on other people having worse health care.
2. Back when Lyndon Johnson was pushing the War on Poverty, his claim was that it would eliminate poverty, result in poor people no longer being poor. If Frank is correct, that claim should have been a death sentence for the program. On his argument, after all, I benefit by other people being poor, since it raises my relative status, and so would be worse off if they stopped being poor. The great majority of voters, then and now, are not poor, so if he is right the great majority should have viewed what Johnson claimed to be doing as hurting them, and voted against it. That isn’t what happened. For Barack Obama to center his proposals for health care on making it available to the small minority of uninsured individuals should have been political suicide if Frank is right. But the bill passed.
3. The most serious problem with Frank’s argument, suggested by the previous point, is that if it is correct he ought not to be making it, since its implications are ones of which he clearly disapproves. He puts it in terms of what “enlightened libertarians” think social institutions should be. But as an economist, he surely believes that people’s behavior mostly reflects their perception of their own interest. The implication of his op-ed is that it is in the interest of everyone to make everyone else poorer, thus raising his relative status. The rich ought to be in favor of grinding down the poor, the poor ought to be in favor of pulling down the rich, and the people in the middle ought to be in favor of both. I do not think that describes the policies that Robert Frank, who is a nice man as well as an able and original economist, wants.
Robert Frank e-mails a reply to Friedman:
My column that provoked David’s comment also provoked a lot of angry mail from other readers who identified themselves as libertarians. None of them, however, made David’s good-faith effort to respond to my argument on its own terms. So I’m pleased that such a respected and able member of the libertarian community has responded to it with care. I will attempt to address his criticisms in a similar manner.
Yet there remain several ways in which we seem to be talking past one another. Most important, we disagree about the very nature of concerns about relative position. In my piece, for example, I claimed that high social rank has substantial instrumental value, observing that the highest-ranking members of society “know they will be able to send their children to the best schools and have access to the best health care. Low-ranking members enjoy no such confidence.” David disagrees, saying that this passage demonstrates my failure “to distinguish between absolute and relative values”: “Getting good health care,” he writes, “is very important to me, but its value does not depend, so far as I can judge by introspection, on other people having worse health care.”
As I’ll argue in a moment, the failure to grasp the relevant distinction is his, not mine. He is correct, though, that demands for health and safety are among the least positional of all demands. Suppose, for example, that you ask someone to choose between two hypothetical worlds: World A, in which he has a 5 in 10,000 probability of dying on the job each year while the corresponding probability for other workers is 10 in 10,000; or World B, in which his annual probability of dying is 2 in 10,000 and the corresponding probability for others is 1 in 10,000. Almost everyone picks World B, the absolutely safe but relatively unsafe one.
But that fact has absolutely no bearing on my claim that persons of high social rank have greater access to the best health care. Resources are finite. The most renowned practitioners cannot provide the most costly medical treatments to every seriously ill patient. Indeed, as things now stand in the United States, not everyone has access to even minimal basic care. (We are first among industrial nations in “preventable deaths” per capita—deaths that would not have occurred if the afflicted person had received competent and prompt medical attention for the ailment that later caused his death.) When there’s not enough for everyone, can there be any doubt that society’s lowest-ranked members are least likely to get the best care?
The importance of social rank for education is even clearer. We care about the absolute quality of education, to be sure. Even so, a “good” school in every society is an inescapably relative concept. It’s one that compares favorably with other schools. Such schools are almost always located in neighborhoods with relatively expensive houses. This is hardly surprising, since most parents want to send their children to good schools, and it’s relative income that predicts which parents will be able to outbid others for houses in the better neighborhoods that surround those schools. In contrast, if all we knew was a family’s absolute income, we’d have no idea whether it would be able to send its children to good schools.
The most renowned practitioners cannot provide the most costly medical treatments to every seriously ill patient.
There is not a fixed number of skilled practitioners or good hospitals; the richer the society is, the more of both it can support. If everyone else’s real income doubles and mine increases only fifty percent, that means that the total supply of goods and services has gone up–roughly doubled. There is no particular reason why that shouldn’t include the supply of medical services, permitting everyone else to consume twice the medical services he consumed before and me to consume fifty percent more than I consumed before.
Indeed, as things now stand in the United States, not everyone has access to even minimal basic care. (We are first among industrial nations in “preventable deaths” per capita—deaths that would not have occurred if the afflicted person had received competent and prompt medical attention for the ailment that later caused his death.) When there’s not enough for everyone, can there be any doubt that society’s lowest-ranked members are least likely to get the best care?
I am shocked to see an economist talking about “enough for everyone.” What do those words mean? Additional medical services would produce some benefit well past the point at which the entire GNP is spent on them.
As you have just conceded, what matters is not whether I get the best care but how good the care is that I get. When my absolute real income increases but my relative income decreases, I can afford better care than before, even if other people can afford care better still. So, in the case of medical care, it is absolute not relative that matters.
The importance of social rank for education is even clearer. We care about the absolute quality of education, to be sure. Even so, a “good” school in every society is an inescapably relative concept.
Schooling is a more complicated case because schools produce both education and status. So far as the education is concerned, a good school is not an “inescapably relative” concept, it is an absolute concept. If all schools get better, that means that students are learning more—becoming better doctors, lawyers, businessmen, parents, farmers, consumers of art and literature, or whatever they are going to use the education for. The result is a more productive economy with more stuff for people to consume and consumers better able to take advantage of what is available to them.
Do you really want to argue that schooling was as good in England in the tenth century, or in Cambodia or Ethiopia today, as it is in a modern developed society? That’s what your “inescapably relative” would imply.
It’s one that compares favorably with other schools. Such schools are almost always located in neighborhoods with relatively expensive houses. This is hardly surprising, since most parents want to send their children to good schools, and it’s relative income that predicts which parents will be able to outbid others for houses in the better neighborhoods that surround those schools. In contrast, if all we knew was a family’s absolute income, we’d have no idea whether it would be able to send its children to good schools.
If everybody gets richer, the quality of schools can increase everywhere. You are again confusing absolute and relative. The richer people will still, on average, have better schools–but the “worse” schools can be producing education as good as, or better than, the best schools used to produce.
Your claim is correct only with regard to the status output, not the schooling output. If you go to Harvard and I to Cornell, that may result in your winning out over me in our courtship of the woman both of us wish to marry. It may result in people who happen to know our backgrounds treating you with more deference than they treat me.
I do not know if it has occurred to you, but one implication of your argument is that spending on schooling, insofar as it produces status, imposes a negative externality on others, so private individuals will tend to buy a more than optimal quantity of schooling for their children. It follows, on straightforward economic lines, that instead of subsidizing schooling, as we do on an enormous scale, we ought to tax it. If schooling is “inescapably relative,” we could cut every school’s expenditure in half and still produce the same amount of education, relatively speaking, while saving many billions of dollars. Perhaps that should be the subject of your next op-ed.
Many thanks to David for his spirited reply. In responding to it, I’ll begin with another attempt to clear up some issues I thought had been settled in our first exchange.
For example, David again spends considerable energy arguing that health care is not a positional good—that what matters primarily is its absolute quality, not its relative quality—and emphasizing that we can raise its absolute quality by devoting more resources to it. Well and good. But so what? As my original response to David’s comment made clear, I am well aware that the demand for health and safety is among the least positional of all demands. And I agree completely that the absolute quality of it is what matters. But those points do not challenge the fundamental claim on which my argument rests—namely, that high social rank has substantial instrumental value.
When a serious health problem strikes, a person wants the highest absolute quality of care possible, but because the quality and quantity of care are limited at any given moment, not everyone seriously ill patient can have the best. My claim, which is surely completely uncontroversial, is that someone with high social rank is more likely than others to get the best care.
Access to the best care is of course not the only reason that high social rank has instrumental value. It also commands access to the best education. In David’s response, he again devotes considerable energy trying to establish that the absolute quality of education matters. But here, too, I ask myself, why would anyone think I disagree? If everyone were better educated, our economy would be more productive. Our incomes would be higher, and that would be a good thing! In the 19th century, a family with five children typically saw several of them die before reaching their 10th birthday. That this no longer happens is primarily a consequence of the fact that our absolute incomes are so much higher now, which is in part a consequence of better education. So of course the absolute quality of education matters.
But the relative quality of education also matters. In modern labor markets, the absolute salary gap between the best-paying jobs and other jobs is larger than at any point in history, and there are almost always many more applicants for the top jobs than employers could possibly interview. Surely it is uncontroversial to note that educational credentials are one of the most important screens that employers use to whittle their applicant lists. For a candidate even to land an interview, his absolute educational quality must be high, yes. But that’s not enough. It must also compare favorably with that of other applicants. So unlike the health care domain, in which absolute quality is the main concern, the educational domain is one in which both absolute and relative quality matter. But here, too, the important point for my argument is that persons of high social rank are more likely than others to be able to send their children to the best schools. That point is completely uncontroversial.
The health and education domains are hardly the only ones in which high social rank has instrumental value. That value is the basis for my claim that in the libertarian’s ideal world of zero transaction costs, people would not be able to claim positions of high social rank for free. As in the analogous case of high-ranked positions in private work groups, they would command positive implicit prices.
In short, notwithstanding the fact that the most important interpersonal comparisons are local, the poor have experienced substantial costs because of the additional spending of the rich. Far more than difficult-to-document claims of psychological damage used by inequality, it is these concrete costs that constitute grounds for saying that in a world without transaction costs, high-ranked positions in the social hierarchy would not be available free of charge. The rich are not paying for the right to compare themselves directly to the poor. They are paying to maintain a social structure from which they benefit greatly.
Everyone gains, for example, from greater opportunities for specialization and exchange. But as international experience amply demonstrates, social stability cannot be taken for granted when income and wealth inequality grow beyond a certain point. Diverse societies are efficient, but will not remain stable unless the terms of the social contract are perceived as fair. And as every country on the planet has decided—implicitly or explicitly—part of such a contract entails income transfers from rich to poor.
To the libertarian’s objection that such transfers are morally unjust, I have argued that they are consistent with the libertarian dictum that the best social arrangements are those that mimic as closely as possible the arrangements people would have negotiated in a world of zero transaction costs. I continue to invite attempts to rebut that argument in its own terms.
Reading Robert Frank’s most recent post, I have the impression that he is puzzled and frustrated by my failure to be convinced by his obviously correct arguments for his position. If so, our feelings are similar, since I remain puzzled as to why, after I have pointed out the gaping errors in his argument, he continues to repeat them. Instead of making another attempt at a point by point response, I am going to try to step back, summarize his argument as I understand it, and try to show what is wrong with it. That should give him an opportunity either to understand my argument, if the problem is that he doesn’t, or to point out how I am misunderstanding his.
The basic claim made in his Op-ed and defended in his posts here is that poorer people are worse off due to the existence of richer people, richer people better off due to the existence of poorer people, hence that it is only just to compensate the poorer at the expense of the richer. That claim has two parts. One is his old argument about the effects of the fact that humans care about (among other things) relative status. The other is the claim that, concern with relative status aside, the existence of rich people makes the less rich materially worse off and that the existence of poor people makes the less poor materially better off.
The first claim is, I think, correct; humans do care about relative status. But, as I pointed out in my initial post and as each of us has agreed since, those effects are local, hence provide no justification for income redistribution on a national scale. In his most recent post I think Robert Frank to some degree concedes that, referring to his own argument as “difficult-to-document claims of psychological damage used [caused?] by inequality,” and recasts the argument in terms of “concrete costs” due to a cascade of competitive expenditures, in effect substituting for his first claim a version of the second. Doing so makes the first half of the argument in his Op-ed, his old analysis of the effect of concern with status on wage differentials in the market, almost entirely irrelevant to the second half, his defense of governmental income transfers. Yet it was the first half which was supposed to persuade libertarians that they should approve of the transfers defended in the second, since they only reflected, on a larger scale, what would come out of voluntary interactions in a world without transaction costs.
It is the second claim that seems to me strikingly mistaken. He has now twice—in his most recent post and the one that preceded it—first agreed that what matters in health care is the absolute level and then proceeded to put his argument in terms of the relative level, not how good the health care is that someone can get but whether or not he can get “the best” health care.
Frank accepts that most status externalities are local, and so are internalized by firms, clubs, homeowners associations, etc., leaving little net market failure to be addressed by larger policy. Instead Frank says the rich should give to the poor to keep things “stable,” which seems to refer to violent revolution. But the chance of a violent revolution in our society seems very low, and not obviously correlated with the level of redistribution. The data Frank accepts suggesting little envy of the rich would seem to support this. I wonder if Frank would endorse a futarchy giving a large negative weight to violent revolution? I suspect not.
In other words, if Robert Frank follows his own logical argument, he is the one with the “dark conception of human nature.”
What do people spend their time doing? To the extent I care about relative position, what do I do? I take actions to raise my preoccupation with areas I am good at, which is my version of spending long hours observing people in poor neighborhoods, yet without having to feel I am deliberately slumming, which would lower my self-esteem.
In this sense I think Frank should accept Friedman’s attempted reductio. That said, I don’t accept Friedman’s claim that Frank’s theory predicts that ”the rich ought to be in favor of grinding down the poor.” Subtle self-deception about one’s own merits is, at least in today’s America, a more effective recipe for generating utility. So I believe in the ubiquity of status-seeking, though I see the process as more positive-sum than Frank does. I can think my area of expertise is really important and you can think the same, without it being zero-sum.
That all said, what exactly is the problem with redistributive taxation? Why can’t we, for Benthamite reasons, have two rates of twenty and thirty percent instead of one rate of twenty-five percent? Status-seeking or not?
Let’s put it as a dilemma.
The poor’s perception of unfairness is caused either by (a) the “concrete costs” imposed upon them by the excessive spending of the rich or (b) by the extent of inequality itself.
If it’s (a), no progressive transfer is logically required to prevent the imposition of those costs. Some policies, like a progressive consumption tax, logically can fix the problem and increase the level of income and wealth inequality.
If it’s (b), all the positional externality stuff is a distraction, and some policies meant to prevent positional externalities, such as a progressive consumption tax, could logically make inequality rise and potentially make the problem worse.
His talk of “income transfers from rich to poor” suggests Frank is choosing (b), which does look like a fold. Frank might as well have argued straightforwardly that inequality past a certain level offends our sense of justice, and that the perception of injustice threatens to destabilize society and thereby produce great harm. I don’t think this is a great argument either, but at least it has a fine pedigree.
My exchange with Robert Frank over his NYT piece seems to have come to at least a temporary halt. Unless he is going to concede that I am right or I am going to concede that he is, neither of which seems likely any time soon, there doesn’t seem to be a lot more to say—although if he wants to respond to my most recent post in the series he is of course welcome to do so.
I thought, however, that it would be worth exploring a different feature of his argument—his view of what libertarians believe, offered as part of his argument for why libertarians ought to support governmental income redistribution. Frank writes:
“Society’s income distribution, [libertarians] argue, should reflect as closely as possible what people would earn in unregulated private markets.”
Off hand, I cannot remember ever hearing a libertarian make that argument. A better version, later in the piece, is:
“Enlightened libertarians believe that the best social institutions mimic the agreements people would have negotiated among themselves, if free exchange had been practical.”
Libertarians differ widely among themselves on both the basis for views and their implications, so the quote is probably true of some libertarians and certainly false of others—unless it is taken to exclude as unenlightened any libertarian who disagrees with it. But it is worth asking to what degree Frank is correctly describing the implications of the more common grounds for libertarian beliefs. Consider three different alternatives, corresponding to three common approaches to political philosophy.
1. Natural Rights: Probably the most popular position among the hard core of self-identified libertarians, some of whom make opposition to the initiation of force the defining characteristic of libertarianism. For most or all of them, both of Frank’s statements are false. Following arguments along the lines of Robert Nozick’s distinction between desert and entitlement, they hold that what matters is not what you end up with but how you got it. Whether or not the rich would have willingly given money to the poor in a zero transaction cost world is irrelevant to the legitimacy of coerced transfers by the state.
2. Social Contract: The contract is a metaphorical one. What people would have agreed to if … is a common basis for deducing it, which makes this a plausible basis from which to defend the position Frank is arguing for. The only problem is that it is not a position popular with self-identified libertarians, few of whom are Rawlsians.
3. Consequentialism: The argument is that libertarian institutions lead to results that most people would prefer to the results of alternative institutions. This is the position I have usually argued from; while I have some sympathy with the moral intuitions underlying the natural rights position I lack arguments to support those intuitions, so prefer to take other people’s objectives as given and argue that my preferred institutions would better achieve them. Utilitarianism is one version of consequentialism, but not the only possible one.
From a consequentialist standpoint, Frank’s position is at least plausible—how plausible depending on the degree to which one believes that free exchange in a zero transaction cost world would lead to the optimal consequences. It is worth noting, however, that there is no reason to expect it to lead to a utilitarian optimum, for reasons having to do with the difference between economic efficiency and maximum utility.
Regarding Frank’s claim about what libertarians think society’s income distribution should be, like Friedman, I’ve never heard a libertarian make it.
The standard libertarian views on distributive justice are (a) that there is no objection to patterns of holdings that arise from exchange of justly owned goods according to just rules, and (b) because we are in society with our trading partners, and because our trading partners span the globe, the pattern of holdings that arise from exchange is an international pattern. But Frank seems to think the typical libertarian is committed to pattern view of distributive justice as well as to both analytical and normative nationalism, such that it is possible that libertarians have a view of what the nation-level pattern of incomes ought to be.
Consider an example. In an unregulated labor market, skilled immigrants would flock to the United States reducing the wage premium for skilled natives. As it stands now, U.S. immigration restrictions act as a subsidy to native skilled workers. I have never heard a libertarian (or anyone else, for that matter) argue that distributive justice requires wiping out the effect of this implicit subsidy through a transfer of income from skilled U.S. workers to skilled foreign workers. For the typical libertarian, it’s not the pattern that matters, but the justice of the rules that underly them. Libertarians argue that the labor market restriction is unjust and so ought to be removed. The implicit subsidy will disappear as a direct effect. Despite Hayek’s and Nozick’s devastating attacks on pattern conceptions of justice — that constantly intervening to “correct” the pattern is inconsistent with the rule of law (Hayek) and would require the constant violation of individual rights (Nozick) — Frank nevertheless seems to assume libertarians have a pattern theory, but simply prefer market patterns. That’s wrong. Most libertarian accounts of justice do not require that the pattern of incomes be made (by government redistribution!) to mirror the pattern that would be produced by unregulated private markets. Libertarian justice requires unregulated (or lightly regulated) private markets, period. The pattern they produce is not important.
How about Frank’s claim that “Enlightened libertarians believe that the best social institutions mimic the agreements people would have negotiated among themselves, if free exchange had been practical.” This is much more plausible — at least if it is assumed that most libertarians are contractarians, which, as Friedman points out, most are not. In any case, two things: (1) Patterns of holdings are not social institutions. (2) Contractarian libertarians (like me) don’t think the best social institutions mimic the agreements people would make under suitably idealized circumstances. The best social institutions just are the ones people would agree to have under those circumstances.
For an example of what this sort of thing looks like, check out Buchanan and Tullock’s The Calculus of Consent or Buchanan’s The Limits of Liberty. In Calculus, B & T give an ingenious contractarian argument for progressive taxation, not at all based on a patterned conception of distributive justice. If I remember correctly, they say that rational utility-maximizing agents choosing a fiscal constitution behind a veil of uncertainty will want to moderate the impact of negative income shocks by paying higher tax rates when income is high and lower tax rates when income is low. Progressive taxation certainly affects the pattern of incomes, but that’s not the point of it in libertarian contractarian theories.