Should regulators care that we buy so many things we wish did not exist?

More than two centuries have passed since the publication of Adam Smith's justly celebrated account of how the invisible hand of market forces often promotes the common good (1776). Smith was more circumspect than many of his modern disciples, who continue to insist that individual and collective interests almost always coincide. Even in the relatively isolated cases in which they acknowledge market shortcomings, their skepticism about government inclines them to oppose regulatory intervention. As Milton Friedman was once said to have remarked, “If you put the federal government in charge of the Sahara Desert, in five years there'd be a shortage of sand.”

In “Goods That People Buy But Wish Did Not Exist,” Professor Cass Sunstein challenges his former University of Chicago colleagues by identifying a broad class of cases in which individually rational choices lead to outcomes that no one favors (Sunstein, 2024). In one, he describes why the reluctance to compromise important social relationships might motivate someone to attend an upcoming social gathering, despite having a clear preference for it to be canceled. If other guests felt the same way, all would attend the party even though all would be happier if it had never been scheduled in the first place.

Such examples are akin to the familiar stadium metaphor, in which all stand to get a better view, only to discover that no one sees any better than if all had remained comfortably seated. Market failures of this sort are, in fact, extremely common. They are also fundamentally different from those caused by cognitive errors, shortsightedness and loss aversion, which were Professor Sunstein's focus in his 2008 bestseller, Nudge, coauthored with Richard Thaler (Thaler and Sunstein, 2008).

Nudge problems can be attacked unilaterally, as when someone who learns that taking sunk costs into account is a mistake can then try to ignore them. Solving collective-action problems is more difficult. Even if candidates for an investment banking job realized that it would be better if all spent less on interview suits, it would be difficult to enforce an agreement to reduce outlays, since each candidate could gain by spending more than rivals. Professor Sunstein acknowledges this difficulty, noting that “Collective action, private or public, is necessary to eliminate goods that people consume but wish did not exist.” (2024, p. 1)

The losses from standing to see better, or from attending boring parties, or from some of the other potentially regrettable choices Professor Sunstein identifies – such as the purchase of Barbie dolls, high heels and men's ties – are relatively small, which is likely why few societies have attempted to regulate them. Such examples, however, greatly understate the scale of welfare losses caused by other members of the broader class of problems he has identified.

Workplace safety is a case in point. Because safety measures are costly, we face an unavoidable tradeoff between wage income and safety on the job. Orthodox economic theory assumes that workers navigate this tradeoff rationally. They accept additional risk if and only if they deem what they could buy with extra pay sufficient to compensate for the corresponding reduction in safety. This view led Milton Friedman and other free-marketeers to object that regulators force workers to buy too much safety.

But Friedman's view ignores the social dimension of safety choices. Consider the decisions facing workers who aspire to send their children to the best possible schools. In a significant measure, school quality is, like the utility of an interview suit, context-dependent. A good school is one that's better than most other schools in the area, and in virtually every jurisdiction, the better ones are in more expensive neighborhoods. From any single worker's perspective, then, a clear motive for accepting a riskier job at higher pay would be to bid more effectively for a house in a better school district. But when all workers respond to that incentive, the result isn't what they'd hoped.

The problem is that their additional bidding power serves only to push up the prices of houses in better school districts. Half of all children still attend bottom-half schools, as before. This simple logic thus answers a question posed by Elisabeth Warren and Amelia Warren Tyagi in their 2004 book, The Two-Income Trap (Warren and Tyagi, 2004): Why could parents in the 1950s get by on a single paycheck while two-earner couples later struggled to make ends meet? The reason, they suggested, was that the second paycheck financed a largely fruitless bidding war for houses in better school districts.

The plausibility of positional concerns as a rationale for safety regulation is buttressed by the transparent implausibility of the various alternative rationales offered by others. Critics on the left have argued that it's needed because corporate power enables employers to withhold even safety devices that would pass any reasonable cost-benefit test. But that argument overlooks that a firm's incentives to provide such devices are exactly the same as those facing self-employed workers. For example, if a self-employed metal worker would be willing to sacrifice $50 a week for the additional safety provided by a guard rail for his lathe that costs only $30 a week to maintain, he would install it. But we'd expect exactly the same outcome if he'd instead been employed by a firm. If the employer installed the guard rail and cut weekly wages by $40, both it and its employee would be $10 a week better off than they would have been without the device. Failure to provide it would thus be to leave cash on the table.

That's not the kind of behavior that critics on the left normally ascribe to greedy corporate actors. Nor do those critics recommend the provision of safety devices whose value to workers is less than their cost. Even the dominant employer in a company town would find it advantageous to install any safety device that passed the cost-benefit test.¹ Taken together, these observations suggest that viewing safety regulation as an antidote for corporate exploitation is logically inconsistent.

That safety regulation has little impact on firms with the greatest economic power provides yet another reason to question the exploitation argument. Engineers at Apple and Microsoft labor under far safer conditions than required by OSHA legislation. Safety regulation binds most heavily in low-wage labor markets, such as those for fast-food employees and unskilled factory workers. Those markets come closest to the perfectly competitive ideal described in textbooks – the very markets in which workers least need protection from corporate elites.

Other rationales for safety regulation also wither under scrutiny. Many have argued that safety regulation is needed not as a remedy for exploitation but because workers often don't understand the risks they face. Although that's sometimes a valid concern, it clearly misses the mark in many of the industries most heavily affected by safety regulation. If regulation was required to protect coal miners from black lung disease, it was surely not because miners were ignorant of that risk, since family members and others in their communities had been dying of the disease for generations.

In 2001, Professor Sunstein and I collaborated on a paper in which we attempted to estimate the extent to which positional concerns might lead workers to sell their safety too cheaply (Frank and Sunstein, 2001). In our paper's abstract, which he wrote, we concluded,

If relative living standards matter, an individual will value an across-the-board increase in safety more highly than an increase in safety that he alone purchases. Where the government currently pegs the value of a statistical life at about $4 million, it ought to employ a value between $4.7 million and $7 million. A conservative reading of the evidence is that when government agencies are unsure how to value regulatory benefits along a reasonable range, they should make choices toward or at the upper end.

Our estimates suggest that positional concerns of the kind that motivate many of the examples in “Goods That People Buy But Wish Did Not Exist” offer the most parsimonious explanation we have for why even the poorest countries have enacted at least limited forms of workplace safety regulation.

Positional concerns are also central to Professor Sunstein's example of people who agree to attend parties they would prefer to have canceled. That's because many of the social relationships they're hoping to protect are their most important sources of information about valuable hiring and promotion opportunities (Granovetter, 1973).

Because context shapes people's evaluation of some goods more heavily than of others, social forces can also generate a variety of other individually rational purchase decisions whose consequences we don't like. Following Fred Hirsch (1976), goods whose evaluations are relatively context-sensitive are called positional goods, while those whose evaluations depend less heavily on context are nonpositional. As Professor Sunstein writes,

There has been a great deal of discussion of the difference between positional and nonpositional goods (Frank, 1985, 2005). Health is plausibly a nonpositional good; people want to be healthy, whether or not most other people are healthy. Motor vehicles are plausibly positional goods; you might want what now counts as a very good car, or a fancy car, only because of the current vehicle mix.

Military arms races occur because nations' relative spending on armaments matters more for their security than their relative spending on roads, bridges and toasters. Similarly, what I've called positional arms races occur because our relative spending on positional goods matters more (by definition) than our relative spending on nonpositional goods.

Positional arms races entail welfare losses that dwarf those associated with every other form of market failure. Fortunately, those losses can be greatly attenuated by relatively unintrusive policy measures. As I recently summarized the argument for this claim (Frank, 2024),

Behavioral scientists have been studying the determinants of human flourishing for centuries. The resulting literature is both large and contentious. But one of its most uncontested and consistent findings is that, beyond a point long since passed in the industrial nations, across-the-board increases in many forms of private consumption yield no measurable gains in either health or life satisfaction. When all mansions double in size, those living in them become neither happier nor healthier than before. Nor are marrying couples any happier today than in 1980, even though constant-dollar outlays for wedding receptions are now three times what they were then.

Most income gains since 1980 went to people in the top fifth of the income distribution, and within even that group, the lion's share went to the highest earners. Spending levels for these people were already well past the point at which further increases serve merely to shift the frames of reference that shape what's deemed adequate. A large body of careful scientific research thus provides no reason to believe that that Americans were meaningfully better off in, say, 2019 (the last full year before the pandemic) than in 2012, even though the inflation-adjusted total value of the nation's goods and services was more than $3 trillion higher in 2019.

That there is waste on such a grand scale would be of little interest if there were nothing practical that could be done about it. Yet just a few simple, unintrusive policy changes could improve matters greatly. For instance, we could scrap the progressive income tax in favor of a far more steeply progressive tax on each family's annual consumption expenditure. People would report their incomes to the tax authorities as they do now and then document how their stock of savings had changed during the year, as many already do for tax-sheltered retirement savings accounts. Taxable consumption would then be calculated as income minus savings minus a generous standard deduction – say, $10,000 per person. Tax rates would start low, then escalate as taxable consumption rose.

Taxing only spending would require that rates on the highest levels of taxable consumption be higher than the highest current tax rates on income. They could indeed be much higher since rates under the current income tax are constrained by the concern not to inhibit savings and investment. (Under a progressive consumption tax, higher top rates actually encourage savings and investment.)

This simple policy change would also encourage people to choose smaller houses, spend less on automobiles and interview suits, and reduce outlays on wedding receptions, coming-of-age parties, and the like. Because those changes would merely shift the relevant frames of reference that shape what people consider adequate, they would be essentially painless. In contrast, revenue from the tax could fund medical research, infrastructure refurbishment, climate mitigation, and a host of other things that actually matter.

Most people living in the USA today are, in fact, already subject to a progressive consumption tax because they currently save less than existing caps on retirement savings that are tax deductible. Most households in the right tail of the income distribution, however, save far more than those caps permit. The full implementation of a progressive consumption tax would thus require the elimination of those caps.

Failure to take that step would exacerbate what Adam Seth Levine and Oege Dijk and I have called the “expenditure cascade,” the process whereby spending by top earners stimulates spending by those further down the income ladder. Without invoking this cascade, it's difficult to explain why the median new house in the USA is now 50% larger than in 1980, even though the median hourly wage has risen little since then.

Middle-income households survive by exploiting every available option: saving less, borrowing more, working longer hours, and moving farther from work. Census data reveal clear links between these responses and regional variations in the growth of inequality. In the 100 largest U.S. counties, for example, those where income inequality grew most rapidly were also those that experienced the largest increases in three characteristic markers of financial distress: divorce rates, long commutes and bankruptcy filings (Frank et al., 2014). In European countries, higher inequality is associated with longer working hours, both across countries and over time. Orthodox economic models, which ignore the link between context and evaluation, predict none of these relationships.²

Households with the highest incomes are of course better able than others to take advantage of tax-exempt savings opportunities, and high marginal tax rates would provide a strong motive for doing so. We could expect, therefore, that although a progressive consumption tax would reduce consumption inequality over time, it would also increase wealth inequality, which has already been growing rapidly for other reasons. A progressive consumption tax would thus strengthen the case for maintaining a robust inheritance tax.³

When Professor Sunstein visited Ithaca several years ago to give a lecture in the Cornell Law School, we discussed my claim that reining in the expenditure cascade would yield substantially larger gains than even the huge savings his nudge movement had been generating. With characteristic diplomacy, he took no position on my claim. But he did mention that when he'd led the Office of Information and Regulatory Affairs in the Obama administration, it had been difficult to discuss even the possibility of tax remedies for positional arms races.

The difficulty, apparently, stemmed from what I call the “Boudreaux objection,” which takes its name from the economist Donald Boudreaux. He has argued that because positional concerns are rooted in base emotions like envy and jealousy, it would be an ethical misstep to craft public policies that take such emotions into account. As he once put it (Boudreaux, 2013),

I agree that people are concerned about their relative standing in society. But I don't believe that such a concern should necessarily be embodied in government policy. (I also agree with those who point out that people naturally are biased against foreigners – prejudiced against others whose appearance and language and customs are very different from what is familiar – but I don't want to elevate this natural tribal impulse into government policy.)

For two reasons, it's troublesome that this objection might dissuade people from taking seriously the policy concerns raised by the examples that Professor Sunstein has offered. One is that positional concerns are not fundamentally about envy and jealousy. They're instead a reflection of the plain fact that relative position is an important determinant of our ability to achieve many of life's most important goals.

Even more troubling, however, is the Boudreaux objection's sheer stupidity. Suppose positional concerns really were a consequence of base emotions like jealousy and envy. Would that mean we should ignore them in the design of public policy? Professor Boudreaux would surely agree that bank robbers are motivated at least in part by the base emotion of greed. Would he argue that we should therefore eschew policies to discourage people from robbing banks?

Whatever their source, positional concerns affect people's choices in predictable ways. As in the examples Professor Sunstein has offered, many of these effects are clearly inimical to community interests. Enacting policies that limit the resulting losses implies neither an endorsement nor a condemnation of positional concerns.

Returning finally to the question I posed in my title, should regulators care that we buy so many things that we wish did not exist? As Professor Sunstein writes (Sunstein, 2024),

Legal responses here are limited, but they might be contemplated when someone successfully maneuvers people into a situation in which they are incentivized to act against their interests, by consuming a product or engaging in an activity they do not enjoy, in order to avoid offering an unwanted signal.

For some of the small-bore examples he cites, formal legal remedies would likely be viewed as meddlesome. But as I've attempted to explain, in other cases, analytically similar social forces generate spectacular waste that could be curtailed simply by changing what we tax. And since we must tax something, it's difficult to see how any reasonable person could fail to recognize the issues Professor Sunstein has raised as legitimate objects of regulatory concern.