How can lawmakers reduce the skyrocketing rate of gun deaths in the United States? How can they stymie the spread of viral fake news stories designed to undermine our elections? Certain constitutionally protected activities—like owning a gun or speaking online—can generate social harms. Yet when lawmakers enact regulations to reduce those harms, they are regularly struck down as unconstitutional. Indeed, the very laws designed to most aggressively reduce social harms—like total criminal bans—are the least likely to be upheld. As a result, regulators appear stuck with an unpleasant choice—regulate constitutionally or effectively, but not both.

This Article proposes a novel solution: Pigouvian taxation. A Pigouvian tax is an economic tool whereby people are required to bear the social costs of their own activity, rather than forcing others to do so. Pigouvian taxes can thread the needle that traditional regulations have not, reducing serious social costs while respecting constitutional protections of individual rights. This is because many constitutional tests—for example, strict scrutiny’s “narrow tailoring” requirement—implicitly reflect the very kind of economic thinking on which Pigouvian taxes rely. In short, constitutional doctrines protecting individual activity do not require society to implicitly subsidize such activity by absorbing any and all costs it generates. Legitimate social costs may be regulated. But regulations must maintain a careful proportionality between the constitutional burdens they impose and the social harms they seek to eliminate. Pigouvian taxes, unlike traditional command-and-control rules, are inherently well-suited to such tailoring. Thus, in areas where traditional rules have been difficult or impossible to adequately tailor—like guns and speech—Pigouvian taxation presents an important new regulatory tool.



Gun deaths are on the rise in the United States, recently reaching levels not seen since the 1970s.1 Fake news is spreading like wildfire across social media, damaging reputations and confusing voters.2 Constitutionally protected activity—like owning a gun or speaking online—can generate important benefits, but it can also cause serious social harm. And frequently, legislative attempts to avert these harms—or perceived harms3 —are struck down as unconstitutional. The First Amendment prevents states from imposing tort damages on newspapers that print grossly negligent falsities about public officials.4 The Second Amendment forbids municipalities from banning the deadliest variety of firearm—handguns.5 The list goes on.

As a result, policymakers and scholars often assume that only second-rate solutions are available for avoiding social harms from protected activities. The policies that seem most likely to survive constitutional scrutiny are precisely those least likely to have the desired effect. Consider again guns. After District of Columbia v. Heller,6 regulators retreated from advocating broad firearms bans—a regulation applicable to every potential gun owner—to advocating much more limited measures.7 These proposals—background checks, magazine size limits, or assault weapons bans—can, by design, produce only small effects. Even if perfectly enforced, they forbid activity comprising only a few small slices of the social-cost pie.8 One can tell the same story for other protected activities.9

This Article proposes a solution. Bringing economic analysis to bear on questions of substantive constitutional law, it argues that the apparent tension between effective regulation and constitutionality is not fundamental. Rather, it can often be resolved using careful policy design. This is because, properly understood, constitutional protections embody the kind of welfarist thinking familiar to economists. Although the Constitution forbids certain kinds of restrictions on protected activities, it does not grant citizens an unlimited license to harm others while engaging in those activities. Put another way, the Constitution does not require that society implicitly subsidize protected activities by absorbing any and all social costs that those activities generate. On the contrary, constitutional law regularly allows, under the right conditions, regulations that reduce those costs. Consider tort liability for negligent shootings or for fraudulent speech.10

Based on this limiting principle of constitutional protection, this Article suggests a promising new tool for regulating protected activity: Pigouvian taxation. A Pigouvian tax is a fee imposed on an actor engaging in some activity, equal in amount to the expected social costs of that activity.11 A carbon tax is a Pigouvian tax. It forces carbon emitters to account for the harm they cause so that they engage in carbon-emitting activity only when the benefits exceed the costs.12

Pigouvian taxation of constitutionally protected activity would work the same way. Gun ownership creates a risk of social costs—from accidental or intentional shootings. A Pigouvian tax on gun ownership, then, is one equal in amount to those expected harms. The tax might take the form, for example, of an annual fee in the range of, say, $1,000 per gun-owning household.13 Alternatively, it might be imposed, in whole or in part, on gun manufacturers, on purchases, or on a consumable complement of guns, like ammunition. Similarly, “fake news” stories generate measurable costs in the form of damaged reputation—for those lied about—and decision-making against interest—for those lied to. A Pigouvian tax here might be imposed on social media giants, extracting some small fee for each interaction with fake news—a “share,” a “like,” etc.—that their algorithms generate. In both examples, many choices exist for precisely how the tax would be designed, sized, and collected. But in each case, the key is to pick a design that—to the best of our regulatory abilities—tracks actual social costs, including variations between entities and over time.14

In important cases, Pigouvian taxes can thread a needle that traditional laws have not, achieving both effectiveness and constitutionality. That is because, in at least three important doctrinal areas—guns, speech, and abortion—the Constitution requires proportionality between a law’s constitutional burdens and its regulatory goals.15 This is often called a tailoring requirement. If a law imposes a substantial burden on some benefit the Constitution protects, but does little to reduce social harms, it fails the tailoring test. To be sure, poor tailoring is not the only reason a regulation may be struck down. In these doctrinal areas, the Constitution also imposes what one might call a legitimacy requirement. It categorically restricts which effects of protected activity may legitimately be treated as social costs to be eliminated.16 Historically, however, the constitutional difficulty has quite often been tailoring, not legitimacy.17

Tailoring is where Pigouvian taxes shine. Compared with traditional regulatory alternatives, the constitutional burdens imposed by Pigouvian taxes are easier to balance with reductions in social costs. Indeed, a right-sized Pigouvian tax is identical to the regulated activity’s social cost.18 Moreover, unlike traditional command-and-control rules, Pigouvian taxes never demand that anyone cease engaging in a constitutionally protected activity. Instead, people themselves weigh the costs and benefits and act accordingly. Thus, a Pigouvian tax preserves people’s freedom to access benefits that the Constitution protects. It deprives them only of unprotected social subsidies for those activities in the form of externalized costs.

Moreover, a well-designed Pigouvian tax imposes the minimum possible burden among cost-controlling regimes.19 It permits people to employ whatever constitutes, by their own lights, the least burdensome means for reducing social harm while engaging in protected activity. Indeed, it rewards them for doing so. If a gun manufacturer, for example, invents and adopts a cheap, effective new safety feature, it thereby earns a reduction in its tax burden. All of this suggests that if a right-sized Pigouvian tax is not sufficiently tailored to pass constitutional muster, nothing is.

Traditional regulations do not work this way. Consider, for example, a total ban on handguns. For some people—say, those who will use a gun only at a range, rarely, and for fun—the private benefits of ownership may be dwarfed by social costs. But for individuals with a significant need for concealable protection, the total benefits of handgun ownership may outweigh the harms. A ban prevents ownership in both cases. This avoids social costs from shootings, but it also completely eliminates constitutionally protected benefits for those who would choose to enjoy them, despite the costs. Putting the problem another way, the ban places the highest possible constitutional burden on every instance of gun ownership, regardless of the quantum of social costs avoided. Extreme disproportionality of this kind is a constitutional nonstarter.

Traditionally minded regulators can try to be more precise. But command-and-control rules—bans, safety standards, licensing regimes, etc.—are inherently difficult to tailor. To achieve the constitutionally mandated balance, lawmakers must evaluate both the social costs and the constitutional benefits of the activity to be forbidden.20 Pigouvian regulators, by contrast, need measure only costs of the taxed activity.21 Moreover, well-tailored command-and-control rules must forbid only those categories of activity reliably generating more social costs than constitutional benefits.22 Such categories may be difficult or impossible to identify. Accidental shootings, for example, are diffuse and hard to individually predict. Thus, although every individual shooting is devastating, average expected costs may be—on a risk-adjusted basis—modest. It is thus unlikely that regulators could identify any significant category of people whose gun ownership could be restricted for the sake of preventing accidents. Pigouvian regulators, by contrast, can tax all gun owners in the amount of expected harm. Then, people themselves perform the requisite balancing, determining whether they reap enough constitutional benefit from ownership to outweigh the expected costs.

Despite all of this, Pigouvian taxation of constitutionally protected activity is not a panacea. Pigouvian taxes can resolve certain, but not all, tensions between regulation and constitutionality. This Article defines three model taxes—on guns, fake news, and abortions—to illustrate the solution’s promise and its limitations. Pigouvian taxation is an ideal tool when—as with guns and fake news—the primary constitutional impediment to regulation is tailoring.23 But in other cases—as with abortion—the core tension truly does arise from fundamental disagreement about what counts as a legitimately regulable social cost.24 In such cases, shifting from traditional regulatory tools to Pigouvian taxes is unlikely to enable more regulation of protected activity.

Constitutional regulation via Pigouvian tax raises other puzzles, both practical and normative. The foremost practical question is how regulators could measure the social costs of protected activities and thus set accurate tax rates. This Article contends that, in many cases, frameworks already exist—in both law and social science—for estimating such costs.25 When targeted social costs are familiar—as with the model taxes on guns and fake news—those frameworks, though imperfect, are robust and already constitutionally approved. For more exotic costs, experience from administrative agencies’ cost-benefit analyses suggests that adequate measurement is feasible. Of course, no method for estimating social costs—whether from gun ownership or carbon dioxide—is flawless. Any framework will produce some error and brook disagreement. Nevertheless, as this Article argues, modern, good-faith cost analyses are good enough to produce regulation that constitutes good public policy and—crucially—is constitutionally sound.26

As for potential normative problems, regulation by Pigouvian tax is sure to have distributional effects. Pigouvian taxes would raise the price of constitutionally protected activity, making it less available to the poor, but easily accessible to the rich. This is a serious concern. But it is one illuminated, rather than caused, by thinking about constitutional regulation in explicit cost-benefit terms.27 All regulations affecting protected activity—health regulations for abortion providers, gun-permitting schemes, liability for fraud—increase the cost of activity, either implicitly or explicitly. Yet current constitutional doctrine considers this problem only obliquely and inconsistently. This Article suggests how existing constitutional rules could be updated to increase fairness, particularly in this age of skyrocketing wealth inequality.28 One possibility would be to expand the application of existing doctrines that guarantee everyone, regardless of wealth, some minimum quantum of constitutional benefits.29 Another solution would be to redouble our commitment to redistributing wealth, eliminating poverty, and ensuring that no one is denied any basic goods, constitutional or otherwise.

This Article proceeds in five parts. Part I briefly recounts how the Constitution has clashed with regulation in three areas: guns, speech, and abortion. These three examples ground the rest of the Article’s analysis. Part II turns to Pigouvian taxation. It first describes Pigouvian taxes generally, including their basic features, their practical superiority to command-and-control regulation, and their historical successes. It then defines three model Pigouvian taxes—on gun ownership, fake news, and abortion. Part III takes up constitutional law. It argues that, in at least these three doctrinal areas, a single, transsubstantive framework drives the constitutional analysis. The Constitution blesses regulations targeting legitimate social costs arising from protected activities. But such regulations must balance reductions in social cost with burdens on constitutionally protected benefits. Part IV then shows why Pigouvian taxes are generally more likely than their traditional analogues to achieve such a balance. It also deals in specifics. The Part argues that the model taxes on guns and fake news would avoid the particular constitutional pitfalls of their predecessors. And it describes the methods by which social costs could be measured in order to set constitutionally sound tax rates. Finally, this Part illustrates Pigouvian taxes’ limitations. Here, the model tax on abortion serves as an example of where taxation would be unlikely to achieve additional regulation. Part V takes up practical and normative challenges, arguing that, while important, they can largely be overcome.

I. The Regulatory Dilemma: Effectiveness or Constitutionality?

This Part briefly recounts the tension between constitutional protections and lawmakers’ regulatory goals in the realms of guns, speech, and abortion. In each area, it has seemed, the regulations likely to have the biggest effects are those least likely to pass constitutional muster.

Why should these three case studies anchor the Article? First, the set includes two cases—guns and speech—where Pigouvian taxes can facilitate additional regulation of protected activity and one case—abortion—where they likely cannot.30 Second, these examples are salient, representing instances where some lawmakers very much wish to regulate, but the Constitution seems to stand in their way.31

What can these case studies tell us about other constitutionally protected activities? Should regulators also tax them? That depends on whether they raise the regulatory dilemma at issue here. Speaking online, owning a gun, and obtaining an abortion are all individual activities. They are things people can simply go out and do on their own. And regulators worry that when people do these things, they generate harm. This harm invites regulation, despite the activities’ constitutional protections, raising the tension that Pigouvian taxes might help resolve.

But most of the Constitution’s provisions do not generate tension of this kind. They do not protect activities that people can individually engage in and thereby generate costs. Some provisions instead forbid the government from engaging in certain activity—for example, the Equal Protection Clause’s bar on discrimination.32 There is no constitutional dilemma inherent in regulating discrimination by the government. Legislators can ban it entirely without raising any constitutional concerns. The same goes for, say, unreasonable searches.33 The Constitution’s structural requirements—for example, the separation of powers34 —are even further afield from the problem Pigouvian taxes can help solve.

But the Constitution does protect other individual activities—like religious exercise35 or free movement36 —that—like guns, speech, or abortion—might generate social costs. It also protects some activities that citizens and governments perform in tandem—like voting37 or using the courts.38 These could pose similar regulatory challenges. Pigouvian taxes might be useful in any of these areas. But their viability depends on whether substantive constitutional structures protecting these activities mirror the rules protecting gun ownership, speech, and abortion.39 Likely, some will not (or will not entirely).40 This Article’s analysis—legal, practical, and normative—thus provides a foundation for further research about Pigouvian taxation of additional protected activities.

Begin with guns. In the second half of the twentieth century, lawmakers in cities like Chicago and Washington, D.C., were concerned about the social costs of gun ownership. They had—and still have—good reason. Gun injuries and deaths from homicide, suicide, and accidents are a serious problem in the United States. Overall homicide rates are roughly seven times that of other economically developed countries, driven by gun homicide rates twenty-five times higher.41 When you add accidents and suicides, which outnumber homicides,42 gun death rates are ten times higher than other rich countries.43 Moreover, for the first time in decades, U.S. gun deaths are on the rise.44 Empirical research suggests that gun ownership rates have a causal, not merely correlative, relationship to these tragic figures.45 Thus, by the late 1980s, both Chicago and D.C. enacted what amounted to total bans on handguns46 —the type of gun involved in the vast majority of killings.47

However, such broad and aggressive regulations of guns are unconstitutional. In Heller, the Supreme Court overturned D.C.’s ban. It did so precisely because the law was so far-reaching.48 The Court reasoned that, unlike other traditional—and more limited—regulations, D.C.’s law imposed constitutional burdens that could not be justified by its reductions in social costs.49

In the aftermath of Heller, many mainstream, high-profile politicians have dialed back their regulatory ambitions. They have retreated to proposals that, by design, regulate only a small sliver of guns or owners, thereby passing constitutional muster.50 Among the most popular of these are proposals to outlaw particular subclasses of guns or accessories—assault weapons, large-capacity magazines, or add-ons like bump stocks.51 But assault weapons, for example, are used in only about 2%—or at most 8%—of gun crimes.52 Thus, even assuming perfect implementation and enforcement, along with zero substitution of other weapons, the effects of such a ban are sure to be minor. Research on the now-defunct federal assault weapons ban concludes as much.53 Background checks54 likewise operate as bans on relatively small slivers of people—like convicted felons—buying guns. Like assault weapon bans, they leave most of the cost-producing population unregulated. It is not surprising, then, that a recent RAND meta-study found only limited evidence that background checks produce any decrease in violent crime or homicide rates.55

Lawmakers appear similarly stuck when it comes to regulating fake news. During the 2016 election between Hillary Clinton and Donald Trump, just 156 election-related fake news stories were shared nearly 40 million times on social media, resulting in as many as 760 million click-throughs to read a fake story.56 About 74% of all stories and 80% of shares were pro-Trump.57 These stories claimed, for example, that Clinton ran a child sex ring based in a pizza shop and that Democrats planned to impose Islamic law in Florida.58 One fake pro-Clinton story claimed that thousands of people at a Trump rally chanted, “We hate Muslims, we hate blacks, we want our great country back.”59

Moreover, some of 2016’s fake stories were promulgated by Russian intelligence agencies to disrupt the U.S. electoral process.60 The effects may have gone well beyond disruption. At least one study suggests that fake news alone might have convinced enough Obama voters to defect to Trump to swing the election.61 To reiterate, an online fake news campaign, run in part by a hostile foreign government, may have changed the outcome of the 2016 U.S. presidential election.62

Fake news can be harmful even when not political. Cutting-edge technologies, like “deep fakes” (wholly concocted videos that look real), are enabling new and disturbing methods of online abuse among private individuals. Deep fakes allow anyone to, for example, convincingly insert anyone else’s face into a pornographic video.63 Victims of these fakes range from celebrities to former romantic partners and are almost always women.64

As with guns, even aggressive critics of fake news have despaired of lawmakers’ ability to enact laws meaningfully curbing its spread. Professor Cass Sunstein has recently argued that First Amendment decisions like United States v. Alvarez65 and New York Times Co. v. Sullivan66 bar laws that could effectively regulate online falsehoods.67 This leads him to advocate constitutional reform, updating precedents that are “dinosaur[s] in light of what is happening online.”68 Other scholars have agreed that constitutional change is needed.69 In the absence of such change, still others have proposed minor tweaks to the system—media literacy education, additional fact checkers, or codes of ethics for political campaigns—that are unlikely to make much of a difference.70 Here, as with guns, the choice seems to be between ineffective regulation and longshot constitutional reform.

Finally, consider abortion. Nearly fifty years ago, Roe v. Wade71 held that criminal bans on abortions before fetal viability were unconstitutional. Pro-life lawmakers see the decision as fundamentally incompatible with their regulatory goals. They therefore advocate a rollback of a woman’s constitutional right to terminate her pregnancy. For decades, a concerted conservative movement in the United States has focused on appointing Supreme Court justices who will overturn Roe.72 And with Justice Anthony Kennedy’s replacement by Justice Brett Kavanaugh—and Justice Ruth Bader Ginsburg’s replacement by Justice Amy Coney Barrett—some believe the project has succeeded.73 Time will tell. In the meantime, antiabortion lawmakers have enacted numerous abortion restrictions that fall short of criminal bans. Many of these less restrictive laws are putatively aimed only at ensuring abortion procedures are safe,74 though there is ample reason to suspect pretext.75 Either way, the most aggressive of them have been struck down as unconstitutional.76

Everywhere, then, lawmakers appear trapped between the Constitution’s vital guarantees of individual liberty and the putative harms that follow when that liberty is exercised. The following parts argue that a regulatory design from the economist’s toolkit—the Pigouvian tax—is uniquely suited to resolve this tension, at least in some important cases.

II. Pigouvian Taxes: An Overview

Pigouvian taxes have long been favored by economists for pure policy reasons, aside from their constitutional advantages explored in this Article. This Part briefly explains why. It then goes on to define the three model Pigouvian taxes—targeting specific social costs from gun ownership, fake news, and abortions—that frame the rest of the Article’s analysis.

A. The Structure and Policy Advantages of Pigouvian Taxes

To understand Pigouvian77 taxation as a regulatory tool, consider that human activities often produce both benefits and costs. Imagine, for example, a factory producing both useful products and harmful pollution. The rational, self-interested factory owner will produce a marginal unit of her product only if the costs she bears (like materials and labor) are less than the benefit she will receive (the price consumers are willing to pay). Crucially, absent regulation, the owner does not bear most of the cost of pollution. Instead, third parties who live near the factory and suffer ill health effects bear it. Thus, the factory owner’s costs remain artificially low, and she produces some units for which total social costs exceed total benefits.

A Pigouvian tax is one that forces the factory owner to account for costs that she would otherwise ignore.78 The tax for each marginal unit of product equals the harms from the pollution emitted in the production of that unit. A properly sized Pigouvian tax raises the product’s price to reflect true social costs, and the factory owner runs the factory only when doing so, on net, benefits the world.79

Economists have long argued that Pigouvian taxes are better policy tools than traditional alternatives.80 For one thing, they present fewer opportunities for mistakes. Consider the most common traditional regulatory scheme: command-and-control. Quotas are command-and-control scheme, placing a ceiling on the maximum number of units a factory may produce. But to set the right quota, regulators must estimate both the total costs and benefits of production, determining at what point the former exceed the latter.81 This problem becomes especially acute when benefits vary from factory to factory—for example, if some produce better products than others. Then, an efficient quota requires separate, factory-by-factory benefit estimates. By contrast, Pigouvian regulators need to know only the harm from the marginal unit of pollution, not its private benefits.82 They can then set the tax equal to the harm, and factory owners, who know their own private benefits, will individually balance benefits and costs.

Another advantage of Pigouvian taxes is that they are technology forcing. A factory owner whose cap on production is fixed has no incentive to reduce per unit emissions. Nor does a factory owner whose regulatory scheme requires, for example, a specific minimum suite of emission-reducing technologies, but no more.83 By contrast, a Pigouvian tax keyed to the factory’s actual emissions creates an incentive for the company to figure out ways to reduce its per unit pollution. The promise of lower taxes for less emissions acts as a payment to the factory for investing in green technology.

Real-world experience bears out economists’ enthusiasm for Pigouvian taxes. Where they have been tried, Pigouvian taxes and their close cousins84 seem to produce significant behavioral changes.85 The highest-profile example from the United States may be the Ozone Depleting Chemicals Tax. That policy was implemented by Congress in the late 1980s to comply with the Montreal Protocol and avert the destruction of Earth’s ozone layer.86 Once early enforcement issues were resolved,87 the tax was a substantial success. Under it, the United States has exceeded its targets for reducing emissions of ozone-depleting chemicals.88

Other taxes aimed at socially costly activities have likewise generated strong effects. A handful of U.S. cities now have taxes on sugary drinks,89 ranging from $0.34–$0.68 per liter. A recent empirical study of such taxes suggested that the low-end tax of $0.34 per liter leads to a 22% reduction in consumption.90 If such a tax were implemented nationwide, its estimated effects on obesity and related diseases would produce a net reduction of $1.4 billion in social costs each year.91 Similarly, some local governments have imposed taxes on disposable grocery bags.92 In its first year, Chicago’s $0.07 fee per bag led to a 27.7 percentage-point reduction, from a baseline of 82%, in customers using disposable bags.93 Other localities have seen even bigger effects, like Montgomery County, Maryland, where a $0.05 fee produced a 42 percentage-point reduction.94 While economists have not yet comprehensively evaluated the net welfare effects of such taxes, preliminary results suggest they are positive.95

B. Three Hypothetical Pigouvian Taxes

This Article anchors its analysis to three model taxes on constitutionally protected activities generating hard-to-regulate social costs: owning guns, disseminating fake news, and obtaining abortions. These examples are selected to show the contours of Pigouvian taxation as a tool for regulating constitutionally protected activity. Two promising examples—guns and fake news—will show when and how Pigouvian taxes can resolve the tension between effective regulation and constitutionality. The other—abortion—will illustrate the kind of constitutional tension Pigouvian taxes cannot overcome.

The three example taxes are defined as follows: The Pigouvian tax on gun ownership would target bodily and financial harms from intentional and accidental shootings.96 The Pigouvian tax on fake news would target reputational harms like lost revenue or professional opportunity. It would also target fraud-related harms like the lost opportunity to make a truthfully informed choice, including a vote. The Pigouvian tax on abortions would target loss of potential life and medical risks to the woman. Note these policies’ specificity. They are not designed to reduce protected activity per se, but rather specific harms stemming from those activities.

These definitions enshrine each tax’s core regulatory objective. Implementing them would, of course, necessitate many detailed decisions about policy design. Who should pay the tax? When? How should rates be set and varied? Part V discusses the constitutional and normative principles that should guide such decisions.

III. Constitutional Structures Favoring Pigouvian Taxation

If traditional regulations have been unable to redress important social harms while remaining constitutional, when and why should Pigouvian taxes fare better? This Part lays the legal groundwork for answering these questions. The Part aims to demonstrate three things true in at least the law of guns, speech, and abortion.97 First, policy design matters immensely to constitutionality. And the Constitution’s strictures for policy design reflect, to a surprising degree, the kind of cost-benefit thinking familiar to economists. In short, these constitutional rights protecting individual activity are not absolute. Protected activities can create social costs, and the Constitution does not require society to implicitly subsidize such activities by absorbing any and all of those costs. Rather, it blesses properly designed laws aimed at reducing them.

Second, this Part argues that, across these constitutional doctrines, regulations of protected activity live or die by two factors: legitimacy and tailoring. First, legitimacy: For a law to survive scrutiny, it must be aimed at controlling effects of protected activities that can be legitimately treated as social costs. Under the First Amendment, for example, reputational harm to individuals from defamation is a legitimately regulable social cost.98 But damage to the government’s reputation from criticism is not. Rather, the freedom to criticize the sovereign must be treated as a core benefit the First Amendment protects.99 Moreover, if a regulation is particularly burdensome on or odious to protected benefits, it can only be justified if it targets particularly important social costs. Second, tailoring: Even laws passing these legitimacy thresholds must be properly tailored. Principally, this means that they must reduce social costs without too disproportionately undermining the benefits that the Constitution protects.

This legitimacy-tailoring framework should sound familiar. After all, the term “tailoring” appears explicitly in one of the best-known constitutional tests—strict scrutiny. In fact, the strict scrutiny test contains both elements. Under it, a law is constitutional only if it furthers a “compelling Government interest”100 —a particularly important variety of legitimately regulable cost. And the law must further be “narrowly tailored” to that interest101 —carefully avoiding constitutional burdens in excess of the targeted social cost. Many other constitutional tests exist. The discussion below shows how the legitimacy and tailoring requirements are embedded everywhere.102

Not everyone will agree that the legitimacy-tailoring framework pervades constitutional analysis—even in the three doctrinal areas explored here. Proportionality, in particular, is the subject of ongoing debate.103 Recently, conservative members of the Roberts Court—including the late Justice Antonin Scalia—have attempted to shift constitutional law away from balancing and toward more categorical rules.104 And on the other side of the political aisle, Professor Ronald Dworkin has argued that rights should act as “trumps,” not as interests to be balanced against other societal goals.105

The doctrinal arguments below are thus both descriptive and normative. This Article contends that the existing law of guns, speech, and abortion is best understood as revolving around legitimacy and tailoring. It acknowledges that certain members of the Court may be skeptical of at least the latter half of that claim. But the Article argues that even rules that facially eschew proportionality almost inevitably end up reincorporating its logic.106 Moreover, this result should be normatively acceptable across interpretive philosophies. As shown below, even originalism—a preferred conservative theory—can incorporate legitimacy and tailoring as the yardstick for constitutionality.107

This Article therefore joins a growing chorus of constitutional scholars arguing that, despite some protestations to the contrary, proportionality analysis is—and should be—crucial to constitutional law. Professor Jamal Greene has recently written that “[t]he U.S. Supreme Court balances pervasively, and what categor[ical] [rules] it maintains are riddled with exceptions.”108 Professor Vicki Jackson has likewise identified proportionality as a core principle in the Fifth, Eighth, and Fourteenth Amendments, as well as wherever strict scrutiny is applied.109 Both favor increasingly explicit proportionality analysis, both for its policy consequences and because it comports with a Founding-era understanding of the Constitution.110 Professor Michael Coenen has catalogued how First Amendment doctrine looks favorably on regulations that impose light penalties while rejecting similar laws that impose heavy ones.111 Professor Joseph Blocher has argued that Second Amendment analysis must inevitably include balancing, since a categorical approach “neither reflects nor enables a coherent account of the [ ] Amendment’s core values.”112 Professor Josh Blackman has likewise argued that Second Amendment jurisprudence should be firmly rooted in proportionality, in part because other individual-rights doctrines are.113 And Professors Jud Mathews and Alec Stone Sweet have explored the historical roots of proportionality analysis in U.S. constitutional law.114

This Article builds on that scholarship. But it advances a framework that includes both a proportionality requirement and a categorical, rule-like legitimacy requirement. And, for the first time, it frames both requirements explicitly using the economic concepts of social cost and benefit.

The resulting model most closely resembles Professor Richard Fallon’s. In both a law review article and a recent book, Fallon carefully reconstructs the core features of strict judicial scrutiny and argues that less-exacting constitutional tests often share those features.115 According to Fallon, any determination of whether someone has an “ultimate” constitutional right to be free from a constitutionally salient law proceeds in phases.116 First, courts must identify what interests both the constitutional right in question and the regulation under review serve.117 The stringency of judicial review depends on how odiously the regulation burdens constitutionally protected interests.118 When the constitutional burden is particularly heavy, only the most qualitatively important regulatory interests can justify the regulation.119 Those steps map roughly onto this Article’s legitimacy framework, as further explicated below.120 Then, in Fallon’s view, courts must ask whether the regulation is sufficiently tailored—an inquiry fundamentally requiring a proportionality between regulatory and constitutional interests.121 That analysis corresponds to this Article’s formulation of the tailoring test, with proportionality at its center.122

The examples below also demonstrate this Part’s third key point: where traditional regulations have stumbled, the problem has often been tailoring. Even when lawmakers target social costs that the Constitution treats as legitimately regulable, they repeatedly struggle to balance those costs with constitutional burdens.

A. The Constitutional Law of Guns

In Heller, the Supreme Court held for the first time that the Second Amendment enshrines an individual right to own guns.123 The Court invalidated D.C.’s law banning possession of handguns and its mandatory trigger lock provision.124 Second Amendment law remains nascent, with only one major Supreme Court case since Heller.125 Even so, constitutional law protecting gun ownership already contains the key two-part analysis relevant here. First, the Second Amendment affirmatively permits lawmakers to target certain legitimate social costs by regulating gun ownership. Second, it imposes a tailoring requirement, disapproving laws that work a disproportionate reduction in constitutionally protected benefits as compared with their reductions in regulable costs. The difficulty, at least in cases that have reached the Supreme Court, has been tailoring.

On a first reading, Heller appears to reject proportionality as a relevant standard. The majority excoriated Justice Stephen Breyer, in dissent, for advocating a “freestanding” and thus “judge-empowering ‘interest-balancing inquiry.’”126 But the Court cannot have meant that proportionality inquiries have no place in Second Amendment jurisprudence. For one thing, the Court wrote that it “kn[e]w of no other enumerated constitutional right whose core protection has been subjected to” a test like Justice Breyer’s.127 Yet the Court was aware of tests, like strict and intermediate scrutiny, that explicitly ask about the balance between constitutional burdens and a statute’s prosocial effects. The majority’s objection must then have been that proportionality cannot stand on its own, as it might in some other constitutional democracies.128 Instead, additional rule-like constraints—like the legitimacy requirement—must apply.

Indeed, the Heller majority’s reasoning did turn on the legitimacy of regulated costs and the law’s tailoring. The Court began by agreeing that, “[l]ike most rights, the right secured by the Second Amendment is not unlimited. . . . [T]he right [is] not a right to keep and carry any weapon whatsoever in any manner whatsoever and for whatever purpose.”129 On the question of which costs are legitimately regulable, the Court agreed that the social costs from “handgun violence in this country” could be regulated using “a variety of tools . . . including some measures regulating handguns.”130 The Court also approved of Founding-era laws aimed at controlling the “great Damages . . . frequently done . . . by persons . . . with Guns and other Fire Arms.”131

As for tailoring, the Court favored Founding-era laws that “d[id] not remotely burden the right of self-defense as much as an absolute ban on handguns.”132 Those laws, the Court thought, were more balanced, regulating moderate costs by imposing moderate burdens—for example, “punish[ing] the discharge (or loading) of guns with a small fine and forfeiture of the weapon (or in a few cases a very brief stay in the local jail), not with significant criminal penalties.”133 But the Court also looked favorably on modern laws that impose heavy burdens—criminal bans—only on particularly costly instances of gun possession. It wrote that “nothing in our opinion should be taken to cast doubt on longstanding” laws restricting “the possession of firearms by felons and the mentally ill,” “forbidding the carrying of firearms in sensitive places,” or outlawing “dangerous and unusual weapons.”134 The Court considered these “only as examples” of permissible, well-balanced restrictions.135 It did not “purport to be exhaustive.”136 Other longstanding, properly tailored regulations—like tort liability for negligent shootings—are also almost certainly allowed.137

Certain readers may balk at this characterization of Heller, worrying that it willfully ignores the majority’s declared originalism.138 But Heller is best understood as endorsing the legitimacy-tailoring paradigm because it comports with the Second Amendment’s original public meaning. The analysis above acknowledges Heller’s heavy emphasis on Founding-era and longstanding firearms restrictions. The key question for originalists is what those laws tell us about the Second Amendment’s original meaning. What about them comported, in the Founding generation’s eyes, with the Constitution? And what principles can we therefore carry forward and apply when analyzing new kinds of firearms regulations? Heller’s answer, as demonstrated above, is that Founding-era laws squared with the Second Amendment for two reasons. They regulated legitimate costs, and they were well-tailored, imposing only modest burdens on modestly costly activity and heavier burdens on costlier activity. There is nothing unoriginalist about this line of reasoning.

The problem with D.C.’s law was tailoring. The Court recognized the legitimacy of D.C.’s targeted social cost—gun violence.139 But it held that the law was too burdensome on the constitutionally protected benefit of self-defense.140 Though D.C.’s law targeted harmful gun use, it did so by “prohibiti[ng] . . . an entire class of arms,” including every instance where handguns are purchased “for th[e] lawful purpose” of self-defense.141 Moreover, the law imposed its burden on self-defense in “the home, where the need . . . is most acute.”142 The Court therefore held that the law would fail “[u]nder any of the standards of scrutiny that we have applied to enumerated constitutional rights.”143 Such overbreadth could not be tolerated under any version of the tailoring test.

Since Heller, the lower courts have largely converged on legitimacy and tailoring in the Second Amendment. They generally apply strict or intermediate scrutiny—which contain both elements144 —to gun-related constitutional claims.145

B. The Constitutional Law of Speech

Free speech case law is rich and varied, yet legitimacy and tailoring pervade it. The examples here show how different regulatory designs—including fees on speech—live or die by these twin requirements. They also show how the strictness of the requirements varies with the kind of law under review. The examples even demonstrate that legitimacy and tailoring underpin rules that do not invoke them explicitly. As with guns, tailoring is also where laws regulating speech often stumble.

1. Fees imposed on speech.

This Article’s proposal for imposing fees on speakers to make them bear their own social costs is not unprecedented. In Forsyth County v. Nationalist Movement,146 a county levied fees on public demonstrations so that “those participating [would] be held accountable” for the increased “cost of law enforcement” during their march.147 The Supreme Court evaluated the law for legitimacy and tailoring,148 finding deficiencies in both. The fee scheme failed the legitimacy test because it treated the controversial character of certain ideas as a cost to be regulated. The county administrator would “examine the content of the message that is conveyed, . . . estimate the response of others to that content, and judge the number of police necessary to meet that response.”149 Thus, unpopular ideas incurred higher fees. But the freedom to express disagreeable ideas is a benefit the First Amendment protects, so disagreeableness cannot legitimately be treated as a social cost to be reduced by regulation.150 The county responded that it did not illegitimately treat controversial ideas qua ideas as costly, but rather focused on the secondary effect of people’s response to them.151 But since at least its ruling in Brandenburg v. Ohio,152 the Court has held that listeners’ disorderly response to unpopular speech is usually not a legitimate reason to regulate speech, either.153

Forsyth County’s fee scheme’s tailoring was also woefully deficient. “The decision how much to charge for police protection or administrative time—or even whether to charge at all—[was] left to the whim of the administrator.”154 The record showed that the administrator’s fees varied wildly, adhering to neither “articulated standards” nor “objective factors.”155 The resulting incongruence between social costs regulated and constitutional burdens imposed supplied a further justification for striking down the law.156

This need not be—and has not been—the fate of all fees imposed on speakers. In Cox v. New Hampshire,157 the Court upheld a law “contain[ing] much of the same language as” the later ordinance in Forsyth.158 The key difference was that, although that law imposed “a permit fee for the ‘maintenance of public order,’” it did not “charg[e] a premium in the case of a controversial political message delivered before a hostile audience.”159 The intuition here is that any demonstration—controversial or not—can generate costs. Protestors may make a mess, spill into streets, divert traffic, provide cover for opportunist crime, or become unruly. These costs are legitimately regulable; “[t]here is nothing contrary to the Constitution in the charge of a fee limited to [such] purpose[s].”160

To avoid Forsyth’s tailoring problem, regulators can size fees to reflect reasonable estimates of these legitimate costs. They may vary rates based on crowd size or the length of the event.161 They may also account for location, route, date, or time of day.162 Adhering to these principles, many cities and counties currently do impose cost-internalizing fees on public demonstrations. Every circuit to consider the issue has determined that such well-tailored fees are consistent with the First Amendment.163

2. Strict and relaxed legitimacy and tailoring.

Regulations of speech extend far beyond licensing fees for demonstrations. A wide-ranging set of First Amendment cases shows that legitimacy and tailoring drive constitutional analysis for regulations of many types. These cases also show how the stringency of legitimacy and tailoring requirements can ramp up or down, depending on the character of the law under review.

The constellation of speech regulations that the Court views as “content-neutral”164 are generally subject to intermediate scrutiny’s moderate legitimacy and tailoring requirements.165 Such laws, for example, regulate the time, place, and manner of expression, or they impinge on First Amendment protected benefits only incidentally.166 These laws pass the legitimacy test if they target social costs that the government has a “substantial,” “important,” or “strong” interest in reducing.167 As for tailoring, they are invalid if the “[g]overnment [has] regulate[d] expression in such a manner that a substantial portion of the burden on speech does not serve to advance its goals.”168 Such laws, however, “need not be the least restrictive or least intrusive means of” avoiding those costs.169

“Substantial” or “important” regulatory goals include, for example, reducing costly “secondary effects of adult theaters”—e.g., “prevent[ing] crime, protect[ing] the city’s retail trade, [and] maintain[ing] property values.”170 A zoning law designed to reduce those costs is well-tailored if it “affect[s] only that category of theaters shown to produce the unwanted secondary effects.”171 The cost of “unwelcome noise” from loud outdoor concerts likewise clears the legitimacy hurdle.172 And a law requiring musicians to use a city-employed sound technician is sufficiently tailored to those ends.173

But a municipality may not require a permit before residents are allowed to engage in door-to-door advocacy.174 Potential social costs from such advocacy—fraud, crime, intrusions on privacy—are “important” enough to clear a moderate legitimacy hurdle.175 But when the permit requirement would substantially burden expression,176 while working little reduction in these costs,177 it fails to strike “an appropriate balance between the affected speech and the governmental interests that the ordinance purports to serve.”178 A ban on all speech near abortion clinics similarly fails the tailoring test.179 It places heavy burdens on speech, yet makes few improvements on harassment or public safety that could not be made by other laws unrelated to speech.180

Content-based regulations of speech are subject to stringent legitimacy and tailoring requirements—usually formulated in terms of traditional strict scrutiny. Here, a merely important or substantial social cost cannot satisfy the legitimacy requirement. Instead, the law must target a cost that the government has a “compelling” interest in regulating.181 Tailoring then demands that the proportionality between social costs avoided and protected benefits burdened be narrow.182 The law is also invalid if there are “less restrictive alternatives [that] would be at least as effective in achieving the [statute’s] legitimate purpose.”183

The “negative impact” of a “graphic image . . . on a young child,” for example, constitutes a compelling cost.184 But a law banning adult cable programming during daytime hours is insufficiently tailored when existing scrambling technology already eliminates most such costs.185 Then, the law’s heavy burden on protected benefits, like expression, far outweighs its benefits.186 Moreover, voluntary, customer-requested blocking would constitute a “less restrictive alternative” for achieving similar effects—another failure of tailoring.187

Some commentators have suggested that strict scrutiny does not actually turn on the proportionality between regulations’ important goals and their constitutional burdens. Professor Gerald Gunther famously claimed that such review is “‘strict’ in theory and fatal in fact”—that laws triggering it are, in effect, categorically unconstitutional.188 But this is a misconception. Professor Adam Winkler has shown empirically that, in about 30% of all federal cases—and 22% of free speech cases—laws subjected to strict scrutiny survive.189 And as recently as 2010, the Supreme Court upheld a content-based ban on speech that provides “material support” to terrorist organizations.190 It held that “combating terrorism” was a social benefit “of the highest order.”191 The law was also narrowly tailored—despite prohibiting support for nonviolent activity—because, the Court thought, such support is actually quite costly. It “frees up other resources” and “lend[s] [terrorist organizations] legitimacy,” thus furthering their “violent ends.”192

3. Legitimacy and tailoring analysis in putatively categorical rules.

In some situations, First Amendment rules do appear facially categorical—eschewing the proportionality analysis inherent in tailoring. This Section shows how legitimacy and tailoring can undergird even such doctrines.

The First Amendment rules governing defamation are a good example. After Sullivan, a public official can recover for defamation only if the alleged defamer acted with “actual malice.”193 This rule is facially categorical; recovery is allowed if the defendant knew of or recklessly disregarded their statement’s falsity, but not if the defendant did not.194

Defamation targets social costs—reputational damage from false speech—that pass the legitimacy test.195 But before Sullivan, defamation had a serious tailoring problem. In Sullivan, an Alabama police commissioner sued the New York Times for making misstatements about the police as they violently broke up civil rights protests.196 The misstatements were utterly trivial, doing little if any harm to the commissioner.197 In fact, the commissioner was not even mentioned—either by name or title.198 Despite this, Alabama’s defamation regime allowed—with negligible mens rea requirements—a jury to impose damages on the Times of half a million dollars.199 That would be roughly $4 million today.200

A regime that imposes such huge burdens on expression for the purpose of avoiding negligible costs fails the tailoring test. The threat of astronomical damages deters risk-averse speakers from speaking on issues of public importance, even when they have taken reasonable precautions against falsity.201 And the “actual malice” rule is best understood as a remedy for such poor tailoring. Purposeful and reckless lies are likely to be especially costly, since they are often intended to injure. Thus, limiting liability to just those cases allows heavy constitutional burdens only in exchange for presumptively large reductions in cost. Moreover, when speakers already know that their speech will be inaccurate, they can correct it before publication, avoiding liability without much constitutional burden.

This explanation of the “actual malice” rule is borne out by the Court’s subsequent decision in Gertz v. Robert Welch, Inc.202 There, the Court held that defamation liability may be imposed without “actual malice” if the plaintiff is not a public official or figure.203 But in such cases, the plaintiff “may recover only such damages as are sufficient to compensate him for actual injury.”204 That is, the damages must be well-tailored, with the constitutional burden mimicking actual social costs. Tailored damages mitigate concerns about “media self-censorship” by risk-averse speakers.205 Yet, as in Sullivan, the rule leaves open the possibility of particularly high constitutional burdens—punitive damages—for statements made with “actual malice.” This, the Court reiterated, was because such statements are more likely to constitute particularly harmful “reprehensible conduct.”206

Thus, Sullivan’s apparently categorical “actual malice” rule is best understood as enshrining a particular judgment about how to ensure proportionality in a species of speech-regulating law. Rather than perform the proportionality analysis in every new defamation case involving a public official, the Court did it once and applied the result going forward. Blocher calls this process “common law categoricalism.”207 But as with all standards that have calcified into rules, changed circumstances could precipitate a swing back in the other direction, toward case-specific proportionality review.208

C. A Social-Cost Picture of Individual Rights

If legitimacy and tailoring drive the constitutional analysis—as they do in these two doctrinal areas209 —what do rights amount to? Is a constitutional right worth anything at all if it allows the core benefits it protects to be bartered away for reductions in social cost?

Under the view advanced here, individual constitutional rights function to limit and then channel tradeoffs between constitutional benefits and other social goods. This provides substantial protections for the constitutional benefits. First, the legitimacy requirement imposes categorical restrictions on the goals at which regulation may aim. What constitutes a social cost, as opposed to a benefit the constitution protects, is a contestable—and historically contested—question.210 The first layer of the legitimacy analysis resolves such contests and flatly bans any law that wrongly treats a constitutional good as a social ill.211

The second layer of the legitimacy analysis places an additional thumb on the scale for constitutionally protected benefits. Even when a law targets an effect that is sometimes legitimately regulable, such laws still burden protected benefits. Depending on the burden’s weight, the legitimacy requirement will ask about the importance of the regulated cost. Particularly large or odious burdens demand particularly important costs.212 A total ban on some important species of protected activity may trigger a heightened legitimacy requirement.213 So may a law that singles out protected activity in a suspicious way.214

This prevents governments from imposing odious constitutional burdens to reduce qualitatively trivial, but quantitatively substantial, costs. Regulators presumably could not, for example, ban all radio broadcasting, even if it could be proved that the electromagnetic waves gave everyone occasional mild headaches. The social cost of headaches, even if large in the aggregate, is not sufficiently important to warrant such an odious First Amendment burden.215

In addition to categorical restrictions on what regulators may treat as a cost to be regulated, constitutional rights limit how those costs can be traded off against protected benefits. The tailoring requirement ensures that laws do not sacrifice lots of protected benefit in exchange for small reductions in social costs—even very important ones. The requisite proportionality varies depending on the law under review. When the legitimacy analysis is heightened, so too is the tailoring requirement strengthened. Strict scrutiny thus requires “narrow” proportionality,216 with very little excess constitutional burden.217 But a law that imposes a less serious aggregate burden on protected benefits may survive if it is merely “substantially” tailored.218 This again places a thumb on the scale for protected benefits, giving lawmakers more berth when they regulate with a light constitutional touch.

Here, a puzzle arises. What is the right unit of analysis for the proportionality inquiry? Consider, for example, the limited firearms bans—such as possession by the mentally ill—that Heller endorsed.219 If the right unit were the individual, such laws would surely be unconstitutional. Statistically speaking, they certainly apply to at least one person who, in the end, poses little or no threat of gun violence. Alternatively, imagine a ban on another group of people, each of whom derives fifty units in protected benefits from owning a firearm. Assume it is obvious to legislators and the courts that half of them will impose one hundred units in gun-related social costs, while the other half will impose zero units. Then, across the whole population, the average constitutional burden of fifty units precisely matches the average reduction in social costs. But this nevertheless seems like a proportionality problem, since the law is plainly overinclusive. The answer, then, must lie somewhere in the middle. Proportionality must be analyzed across populations affected by a given set of regulatory requirements. But it must also require—with varying strictness—that lawmakers draw their regulatory categories so as to minimize errors within them. That this principle remains unclear in the case law reflects a general lack of quantitative sophistication with which regulations of protected activity are promulgated and litigated.

In addition to the primary proportionality requirement, tailoring analysis sometimes imposes other rights-protective side constraints. For example, strict scrutiny dooms even well-proportioned laws if another regulation would achieve the same reduction in social cost while imposing a smaller constitutional burden.220 That this requirement usually applies only under higher levels of scrutiny could reflect a judgment about institutional competencies. Legislators may generally be better positioned than judges to assess the relative cost-reducing potential of different regulations. But when the constitutional stakes are high, judges feel empowered to second-guess those assessments.

In sum, under this Article’s model, constitutional rights protecting individual activity are neither absolute nor trivial. Regulators may impose constitutional burdens for the sake of other societal goals. But the Constitution limits both what tradeoffs they may make and how. Occasionally, as explored below, the Constitution may even limit the maximum tradeoff permissible, including for well-tailored regulations of important social costs.221 This constraint is much less pervasive in current doctrine than those discussed above, but there are reasons—including distributional justice—to advocate its expansion.222

IV. Why Pigouvian Taxes Outperform Traditional Regulations

Now, finally, we have the tools to assess the constitutional advantages of Pigouvian taxes over traditional regulation. As discussed above, in at least two key areas of conflict between regulation and constitutional protection—guns and speech—a prime difficulty has been tailoring.223 This Part shows how, in such cases, Pigouvian taxation can overcome tailoring problems and enable regulation of the social costs lawmakers care about most. It also illustrates a limiting case. Pigouvian taxes are unlikely to resolve the legitimacy-based conflict at the heart of the debate over regulating abortion.

A. The Inherent Advantages of Pigouvian Taxes over Traditional Regulation

Why are traditional regulations so difficult to adequately tailor? And why should we expect Pigouvian taxes to fare better? There are two primary reasons, and they mirror the reasons that economists prefer Pigouvian taxes to traditional command-and-control rules.

First, recall that, to impose a command-and-control rule, lawmakers must estimate both the benefits and the costs of the regulated activity.224 For purposes of pure efficiency, the lawmaker should try to evaluate all benefits and costs. But what matters for constitutional tailoring is proportionality between constitutionally protected benefits and the regulation’s targeted social costs. Underestimate protected benefits or overestimate targeted costs, and you end up with a constitutional tailoring problem.

Not every estimation error will be fatal. As discussed above, the Constitution’s tailoring requirements vary by context.225 Laws that only lightly impinge on protected benefits need not be as tightly proportioned as those that more heavily impinge.226 But the need for two estimates presents two opportunities to go astray.

This problem is especially acute when regulators do not know what benefits the Constitution protects. For example, the Supreme Court has so far been silent about whether the Second Amendment protects benefits from, say, hunting, in addition to self-defense.227 In the face of such uncertainty, how should lawmakers tailor regulations on hunting rifles? The only option is to make a guess about protected benefits, enact a law based on that guess, and wait to see if the courts strike it down.

Pigouvian regulators, by contrast, need not estimate how much private constitutionally protected benefit people derive from engaging in a given activity. To set the right Pigouvian tax, they need to estimate only the marginal targeted social costs the activity generates, and set the tax at that amount.228 If properly sized, the constitutional burden of a Pigouvian tax is not merely proportional to its targeted social costs; the two are identical. This presents just one opportunity to go wrong. Moreover, it does not require regulators to know everything that the Constitution does protect, only that it doesn’t protect their particular targeted costs. And existing case law supplies much information about the latter.229

Regulators promulgating Pigouvian taxes might sometimes misestimate costs. And such errors might sometimes produce tailoring problems. However, between estimating targeted social costs and estimating constitutionally protected benefits, the former is less likely to go wrong. Regulators are often most concerned with highly visible harms—things like physical injuries and monetary losses. As detailed below, existing legal and social scientific tools provide well-established and constitutionally acceptable—if metaphysically imperfect—methods for measuring these familiar costs.230 And such costs are at least easier to measure than, say, the intangible constitutionally protected benefit of freely speaking one’s mind. Moreover, if a social cost turns out to be difficult to estimate, Pigouvian regulators can simply ignore it. They can measure what is measurable and set the tax rate accordingly. This may result in less-than-perfect efficiency, but that is not a constitutional problem.231 Command-and-control regulators have no such luxury when it comes to estimating protected benefits. The proportionality inquiry does not let them simply ignore their law’s effect on some benefit simply because it is difficult to quantify.

Even more problematic for command-and-control regulators is transmuting cost-benefit estimates into legal rules. There are really only ever two choices: ban an activity, or allow it.232 To be sure, not all bans are as far-reaching as, for example, the near-total ban on handguns at issue in Heller. They can be much more targeted—applying only to certain people, at certain times, engaging in protected activity in a certain manner. A background check for gun purchases, for example, operates as a ban on ownership for those who fail the check. Such microbans must, by practical necessity, apply to categories of people or activity: “no gun sales to people under 18 years of age,” “no protests on the highway,” and so on.233

The task, then, is this: To impose proportionate, well-tailored command-and-control rules, lawmakers must identify coherent categories of activity for which social costs likely outweigh protected benefits. But what if costs are diffuse and hard to predict, such that identifying such categories is difficult or impossible? All gun ownership carries the risk of an accidental shooting. And every accidental shooting is extremely costly. But they happen to only a small fraction of gun owners. Thus, the average per capita expected cost from accidental shootings is likely quite low—too low to justify any serious broad-based restriction. Nor is it clear that lawmakers have any good way of predicting accidents. Without that predictive information, they will not be able to identify any significant categories of gun owners for whom expected costs exceed benefits, which would enable substantial regulation.

Sometimes, small categories of particularly costly activity can be identified and banned via a command-and-control rule. But such bans are likely to be quite narrow, thus leaving the vast majority of cost-imposing activity unregulated. For example, assault weapons may be particularly dangerous, and thus ban-able. But even if so, they are used in only a miniscule percentage of homicides and suicides.234 Thus, even a perfectly enforced assault-weapons ban with no substitution of guns could, by its own terms, reduce gun-related costs by only about 2%.235 The rest of the costs remain completely untouched by regulation.

Additional rules, even if similarly narrow, might aggregate to greater effect. But still, their maximum scope remains limited. Consider that federal law already makes it illegal for anyone previously convicted of any felony to possess a gun.236 This includes the 52.3% of felons who committed nonviolent drug, property, and other crimes.237 Yet even if this broad prohibition were perfectly enforced, it would prevent gun possession for only about 30%–40% of people eventually convicted of homicide.238 In turn, homicides account for at most about 38% of annual gun deaths; gun suicide rates are nearly twice as high.239 Such scant regulatory coverage, likely touching only a small fraction of costly activity, may be the best we can do using command-and-control rules. That is because lawmakers are running out of identifiable categories of people whose gun use could be constitutionally restricted. For example, people between the ages of 18 and 21 commit their fair share of gun crimes.240 But, as Professor Michael Dorf has recently argued, a law banning gun ownership for everyone under 21 would be unlikely to survive constitutional tailoring analysis.241

Pigouvian regulators do not face this problem. Pigouvian taxes are not bans, or even micro-bans. Under them, anyone who determines that the protected benefits from an activity exceed social costs is free to engage in that activity. And the regulatory regime remains well tailored so long as the Pigouvian tax rate approximates targeted social costs. As a result, lawmakers are not limited to regulating only those tiny slivers of activity where costs identifiably exceed protected benefits. Even in cases where costs are widespread but individually unpredictable, the solution is straightforward: tax everyone on an expected-cost basis, and let people themselves decide how to act in response. Pigouvian taxes can therefore regulate up to 100%—not 2% or 15%—of cost-producing activity.

Moving on, Pigouvian taxes are dynamic enough to be well tailored even under tests requiring the least burdensome regulatory alternative. If a Pigouvian tax is well-keyed to actual social costs, it allows—indeed, incentivizes—taxed entities to find the least burdensome way to reduce those costs. Think, for example, of a smart gun that can be fired only by its owner.242 Such a precaution, if cheap and effective, could work significant reductions in social cost with small constitutional burdens. A command-and-control rule can reap this welfare-enhancing benefit only if a smart gun requirement is written explicitly into the law.243 But under a well-designed Pigouvian tax, taxed entities will seek out such precautions—or even invent new ones—whether or not the law explicitly contemplates them.244 So long as taxes follow costs,245 a Pigouvian tax will simulate the least burdensome command-and-control regime, whatever that would be.

Traditional regulations aside from command-and-control rules also suffer from tailoring problems that Pigouvian taxes can avoid. Tort regimes like the one overturned in Sullivan can impose constitutional burdens far in excess of reductions in social cost. This is because of shortcomings in institutional competence. Juries theoretically peg compensatory damages to social costs. But they can be biased. Their racial prejudices might, for example, produce immense awards for trivial false statements made by Black civil rights leaders.246 Or, as laypeople, they might be less well-equipped than experts to accurately evaluate social costs, even when acting in good faith.247 Pigouvian tax rates, by contrast, can be set by experts using established social scientific techniques designed to produce the best available estimates of social cost.248 Such expert calculations are certain to contain some inaccuracies. But they would be careful, systematic, and public, whereas jury awards are not. Moreover, even the strictest constitutional tailoring requirements do not require laws to estimate social costs with metaphysical perfection.249 If it were otherwise, there could be no extant examples of permissible regulations on protected activity. But many such examples exist,250 despite being based, as they are, on imperfect cost estimates.

B. Two Promising Cases: Guns and Fake News

Pigouvian taxes will generally fare better under the tailoring analysis than traditional regulations of constitutionally protected activities. But what about specific taxes? Here, we return to our hypothetical Pigouvian taxes on guns and fake news. This Section argues that a well-crafted version of each would avoid the pitfalls of its forebears and thus be upheld as constitutional.

Begin with the dual-layered legitimacy test.251 The important point here is that both the fake news and the gun taxes would target costs whose legitimacy the Supreme Court has already approved. In its case law evaluating the taxes’ traditional analogues, the Court has identified some costs as truly distinct from constitutionally protected benefits. And it has opined on the importance of those costs. In both cases, we should expect the social costs at which our Pigouvian taxes are aimed to easily pass the legitimacy threshold.

The Pigouvian tax on guns would target social costs from accidental and intentional shootings. More specifically, the tax would be imposed in an amount equal to losses—like injury, death, and lost wages—resulting from crimes, accidents, and suicides. These are surely the kinds of harms the Heller majority had in mind when it approved the regulation of “handgun violence in this country.”252 At least, they are the kinds of harms targeted by the many longstanding regulations on gun ownership that Heller explicitly endorsed.253 Laws barring gun ownership by the mentally ill, the “carrying of dangerous and unusual weapons,”254 and even garden-variety tort law all attempt to avert such costs.

This implies two things. First, the Pigouvian tax targets costs distinct from “individual self-defense,” which is “‘the central component’ of the Second Amendment[’s]” protected benefits.255 To carefully avoid illegitimately targeting that protected benefit, the Pigouvian tax should exclude from its rate calculation any costs resulting from justifiable self-defense. Second, costs from injury and death are sufficiently important to justify a right-sized tax. Heller endorsed those costs as important enough to justify odious burdens like total criminal bans for whole categories of people.256 All the more, then, should the Court endorse a tax—which is not a ban, and certainly not a criminal penalty—in service of the same goals.

One can run a similar analysis on the targeted harms of a Pigouvian tax on fake news. That tax would be designed to incorporate certain harms from demonstrably false speech—mostly online—posing as real news. Specifically, the targeted harms would fall into two categories: damage to reputation and harm from unwarranted reliance on such lies. As with the gun tax, these are social costs that the Supreme Court has already approved as targets of traditional regulation. Reputational harm from false speech is the classic social cost that defamation laws target. And in its cases following Sullivan, the Supreme Court has repeatedly emphasized that such harms are legitimately regulable.257 Harm to reputations from false claims of fact is thus distinct from the protected benefit of being able to freely express “pernicious [ ] opinion[s]” and other “ideas.”258 Moreover, such costs are sufficiently important to warrant imposing monetary payments in the form of defamation damages.259 There is little reason to think that the imposition of monetary payments in the form of Pigouvian taxes should be any different.

The same goes for social costs from unwarranted reliance on false information. These are the classic harms targeted by laws penalizing fraud. Even though false speech receives some First Amendment protection,260 the “legally cognizable harm associated with [fraud]” justifies its regulation.261 In a fraud, one party lies to another to induce an action, the other acts in reliance on the lie, and the latter is worse off for it.262 The main act that fake news has historically attempted to influence is voting.263 But when fake news induces people to buy products or financial instruments, a Pigouvian tax could target those fraudulently induced choices, too.264 Such harms are at least important enough to justify laws imposing tort damages in the amount of the harm.265 So too, then, are they important enough to justify a Pigouvian tax imposing similar economic costs.

This brings us to tailoring. As noted above, a right-sized Pigouvian tax on protected activity is inherently well-tailored.266 The detriment it imposes on constitutionally protected benefits is not merely proportional to, but rather identical to, the targeted social cost. Thus, the baseline assumption should be that a properly sized267 Pigouvian tax will withstand scrutiny under any version of the tailoring test. That assumption is bolstered when one considers how the Pigouvian taxes proposed here would avoid the specific tailoring pitfalls that felled their traditional analogues.

Begin this time with the tax on fake news. Recall that both the fee scheme in Forsyth and the defamation regime in Sullivan were struck down for poor tailoring.268 In both cases, the source of inadequate tailoring was institutional incompetence. The administrator in Forsyth set the permit fees not by carefully estimating each demonstration’s regulable costs, but rather by “whim.”269 The jurors in Sullivan allowed their racial animus to infect their estimation of costs, imposing extraordinarily high damages on pro-civil-rights criticism of the police.270

A Pigouvian tax on fake news would avoid this pitfall by vesting highly competent experts with the authority to measure social costs and thereby set tax rates. They would rely on well-established empirical methods.271 And unlike Sullivan’s jurors or Forsyth’s administrator, their rationales would be public, stable, and directly reviewable by courts.272 The goal would be social cost metrics that approached the most accurate available anywhere. Indeed, because the costs targeted by the fake news tax—and the gun tax—are familiar in law and social science, existing empirical methods are robust.273 Such careful estimates would avoid the wild disproportionality between constitutional burden and social cost that pervaded Sullivan and Forsyth.

That leaves the problem of risk aversion. Sullivan might be read as worrying that even accurate damage awards, if sized to unpredictably large harms from apparently minor misstatements, could chill valuable speech. The Court solved this problem by imposing an “actual malice” standard, allowing liability only for those who, in advance, know—or should know—of a serious risk.274 Pigouvian taxes can do even better. Consider that, even under the “actual malice” standard, a known, but apparently trivial, false statement can still produce unexpected and massive damages. When the defamed turns out to be an eggshell victim, risk-induced chilling of speech can still arise. But a Pigouvian tax on costs that vary unpredictably case-to-case can be levied in the amount of average expected cost, smoothing eggshell-victim effects.275 Then, costs per instance are both moderate and predictable in advance.

Moreover, there is little risk that a Pigouvian tax limited to fake news would ensnare diligently reported stories. The line between fake news—utter fabrications presented as accurate—and genuine news need not be blurry. PolitiFact, for example, keeps a running scorecard of misleading news stories.276 Of the stories it has evaluated, a whopping 94% fall into its two most blatant categories of inaccuracy—“False” and “Pants on Fire.”277 A “False” story might, for example, edit a famous writer’s tweet to favor Trump, rather than Clinton, by blatantly swapping their two names.278 A “Pants on Fire” story might opine on the strictures of the Constitution’s Twenty-Eighth Amendment, though only twenty-seven amendments exist.279 These are not subtle lies. They could be regulated without much fear of Orwellian governmental overreach.280 And given such egregious falsehoods’ enormous share of the fake news pie, a Pigouvian tax focused only on them could still have strong effects. A fake news tax, narrowed in this way, would in fact approximate the effect of Sullivan’s “actual malice” requirement. Only easily identified lies would incur a tax.

There is comparatively little to add about the tailoring of a Pigouvian tax on guns. The law at issue in Heller281 essentially forbade everyone from carrying any handgun of any kind.282 The tailoring problem there was the same problem with all broad bans. They indiscriminately destroy constitutionally protected benefits, banning all protected activity, irrespective of reductions in social costs. And like all right-sized Pigouvian taxes, the gun tax would incentivize people to refrain from protected activity only when, by their own lights, costs exceed benefits.

Guns and fake news thus represent exemplary areas where Pigouvian taxes can succeed, though traditional regulations have not. They can target costs that are already known to be legitimately regulable. And they would avoid the tailoring problems that beleaguer their traditional regulatory analogues.

C. Measuring Social Costs

The previous sections have argued that right-sized Pigouvian taxes can effectively reduce the social costs that regulators care about while overcoming constitutional tailoring problems. Both of these claims depend on lawmakers’ ability to adequately measure the targeted social costs from protected activity. If they cannot do so, and the taxes are set significantly too low or too high, then either efficiency or constitutional problems can result.

In many important cases—including our promising taxes on guns and fake news—there should be little concern that the constitutionally mandated cost measurements will prove infeasible. For our model taxes on guns and fake news, the targeted social costs are fairly familiar. Our long-established frameworks for estimating them will almost certainly pass constitutional muster.

The Pigouvian tax on guns would aim to reduce social costs from shootings in the form of bodily injury and death.283 Bodily harm and death are among the most often quantified harms in existence. Not only social scientists, but also legal actors—like juries, legislators, agency heads, and even judges—are constantly trying to measure them. Indeed, they often must do so to resolve even the most basic tort suit or to design simple safety regulations. The result is a panoply of well-established and oft-employed models for converting maimings and murders into dollar figures.

Modern estimates of the value of lost life or limb are rooted in statistical estimates of peoples’ willingness to pay to avoid such fates.284 The basic method is to observe how labor and product prices vary with risk, controlling for as many other factors as possible.285 Since the 1980s, such estimates have become extremely sophisticated. They rely on increasingly rich and accurate datasets.286 They differentiate between ways people might be injured or killed.287 And increasingly, such estimates are sensitive to the imperfections in human risk assessment and decision-making.288 Despite differences in methodologies, these measurements substantially converge. Today, the value of a statistical life in the United States hovers roughly in the $12 million range.289

Valuing injury costs associated with gun ownership requires another inferential step. Regulators need to know the relationship between gun ownership and injuries. For example, does increasing gun ownership increase the murder or suicide rate? Or, in the absence of guns, would the same murders and suicides be committed by other means? Here, too, there are longstanding social science frameworks for evaluating cause and effect. Most commonly, researchers use sophisticated statistical analysis to observe how, controlling for other factors, rates of gun injury and death vary with rates of gun ownership.290

These analyses are not merely theoretical. In 2006, social scientists at Duke and Georgetown Universities estimated the annual marginal social cost per gun-owning household to be $600.291 This estimate is conservative, using low-end figures for the value of a statistical life and the effect of gun ownership on injuries.292 A less conservative number would approach $1,800 annually.293 Both figures include only costs from intentional shootings.294 Other researchers have attempted to quantify the effects of gun ownership on other outcomes, like suicide295 and accidental deaths.296 Thus, a conservative Pigouvian tax on gun ownership might easily exceed $1,000 per household annually.297 We should expect more research of this type in the near future. For the first time in over 20 years, Congress has recently appropriated federal funding to study the causes of gun violence.298

The same kind of logic would apply to estimating the social costs of fake news. As with bodily injury, harms associated with damaged reputation—lost revenues, job opportunities, etc.—can be evaluated using traditional tort frameworks.299 What about the costs that occur when fake news stories bamboozle people into making bad decisions? If those decisions are commercial, then other tools from tort—this time, fraud—will again be instructive.300

However, as described above, much fake news is designed to influence voting.301 How could Pigouvian regulators estimate the social cost of a fraudulently induced vote? This is again a two-step procedure. First, regulators would need to determine the likelihood that any given instance of fake news affects a vote. Already, academics have begun to study how such lies spread online and the extent to which they affect electoral outcomes.302 Then comes the question of how much each fraudulently induced vote is worth, from a social-cost perspective. Measures of voters’ willingness to pay might be hard to come by. Not often are citizens offered the opportunity to pay for precautions against being tricked into voting against their true interests.

Social scientists have, however, begun to measure campaigns’ willingness to pay for a flipped vote, in terms of legitimate advertising expenditures.303 Bad actors may prefer to influence votes with fake news instead of legitimate political advertising because the former is comparatively cheap and easy. The social cost of fake news can thus be understood as the difference between the cost of flipping a vote with a lie and the cost of doing it legitimately. This, too, mirrors a longstanding tort mechanism for regulating antisocial activity: disgorgement, which strips wrongdoers of their ill-gotten gains.304

What if regulators wished to use Pigouvian taxes to target costs from protected activity that are less familiar than those targeted by our sample taxes?305 As the preceding paragraphs suggest, different costs will invite different measurement frameworks. And there is no guarantee that every conceivable cost will admit of reasonable estimation. Yet there is good reason to believe that many will. We live in the legal epoch of cost-benefit analysis (CBA). Since the Reagan era, administrative agencies have been required to quantify both the benefits and social costs of major regulations.306 In doing so, they often make unusual estimations, like the cost of a dead fish307 or, alternatively, of a lost day of recreational fishing.308 Today, agencies are doing this kind of thing all the time. Between 2010 and 2013 alone, they issued 106 rules requiring CBA, resulting in dozens upon dozens of occasionally complex cost estimates.309 True, agency CBAs sometimes fall short, with agencies declining to issue quantitative figures.310 But it is quite rare for them to decline because they believe costs or benefits are unquantifiable “as a matter of principle.”311 And even when they do, they may often be wrong. As Professors Jonathan Masur and Eric Posner have argued, “in all of these cases” between 2010 and 2013, the effects, in fact, “could and should [be] calculated.”312

The real constitutional question, then, is not whether some method will exist for measuring a given targeted cost. One almost always will. The question instead is which methods are, constitutionally speaking, good enough. Long-established frameworks for estimating the familiar costs associated with our taxes on guns and fake news should certainly suffice. When diligently followed, such tools provide a basis for calculating—for example—accurate tort damages, which the Constitution blesses in various contexts.313

Beyond these well-trodden methods, the question is more open-ended, largely because the Court’s proportionality analyses have historically been quantitatively unsophisticated.314 In the cases described above, the Court approved many regulations as well-tailored based on little more than gut instinct about the magnitude and distribution of social costs.315 Rule-of-thumb intuition is apparently good enough to find proportionality in a flat ban on gun ownership by all previously convicted felons, despite strong evidence that many of them pose no elevated threat of violence.316 We thus ought to expect most good-faith empirical analyses to likewise suffice.317 We might think so even if people could reasonably disagree about methodology, and hence about ultimate cost estimates.318 In sum, if there are constitutional limits on good-faith empirical cost estimates, they are not yet well developed.319

So, regulators can constitutionally promulgate Pigouvian taxes on protected activity by estimating only targeted costs and relying on rigorous, but imperfect, good-faith measurements. But should they? This Article takes the normative position that lawmakers ought to try their best to regulate in light of all the effects of their rules. Myopic attention to only targeted costs and protected benefits may be constitutionally permissible, but it is not optimal.320

The biggest potential challenge for Pigouvian regulators designing all-things-considered, cost-benefit-justified taxes affecting protected activity is accounting for such activity’s positive externalities.321 Some First Amendment–protected speech, for example, likely produces benefits beyond those for the speaker. Art is thought to create such prosocial effects.322 Likewise, it is sometimes argued that widespread gun ownership acts as a deterrent to crime, such that even non–gun owners experience benefits from guns.323 If effects like these exist, and are large, taxes internalizing social costs but ignoring positive externalities might, on net, reduce social welfare.

Pigouvian taxes should, then, ideally be set to track net social effects of protected activities. And if a given activity’s positive externalities exceed its costs, the right Pigouvian rate is zero, or even negative.324 This is unlikely to be the case for our Pigouvian tax on guns or fake news. Existing social-scientific efforts to quantify gun costs already focus on net, not gross, effects.325 The dominant view is now that increasing rates of gun ownership lead to larger, not smaller, net costs.326 Unlike other kinds of speech, it seems unlikely that blatantly false news stories posing as the genuine article generate significant positive externalities.

Not every case will be so easy. What should regulators do if faced with plausible but hard-to-quantify positive externalities from protected activity? Should they assume positive externalities are infinite—swamping costs—and thus refrain from taxing? That approach imposes significant opportunity costs by refusing to tackle known harms. The opposite approach—assuming zero positive externalities and imposing a Pigouvian rate based on costs alone—incurs costs from lost positive externalities.

Masur and Posner—analyzing administrative rulemaking—argue that a better approach exists. Advocating a rough Bayesian method, they contend that lawmakers should rely on the best information they have—even if intuitive or inchoate—to assign values to social costs and positive externalities.327 The initial valuation should be updated as more information becomes available.328 This approach produces the regulations most likely to be optimal—even if error bars are large.329 Importantly, it avoids the nearly certain mistake of arbitrarily assuming positive externalities are either zero or infinite. If, in the end, this best-estimation approach leaves regulators suspecting a Pigouvian tax may be net welfare reducing, they should refrain from enacting it.330

It is worth reemphasizing here that the problem of positive externalities is mostly pragmatic, not constitutional. Regulators ought to try to account for them. But neither the constitutional principles discussed above nor the cases implementing them seem to suggest that such third-party effects are among the benefits the Constitution protects.331 We do not, for example, discount tort damages for accidental shootings, consistent with the possibility that widespread gun ownership deters some crimes. And a central metaphor for understanding free speech is as protecting “the market place of ideas.”332 A market incentivizes the creation of societal goods by allowing individuals to capture some—but not all—of the benefits from goods they create. Constitutional rules protecting people’s individual benefits from speaking, but not the positive externalities they generate, are consistent with this picture. Moreover, the Constitution enshrines additional, independent mechanisms to reward speakers for speech that society deems especially valuable. Copyright does this for art, patent for science, and elections do it for political speech.333

D. A Limiting Case: Abortion

The sections above showed how Pigouvian taxes could be used to successfully regulate harms from guns and fake news. This Section, in contrast, demonstrates the proposal’s limitations. In the case of abortion, Pigouvian taxation is unlikely to resolve existing tensions between regulation and constitutionality. That is, Pigouvian taxes would not allow pro-life regulators, who view the termination of a pregnancy as imposing extreme social costs, to heavily penalize abortions on that basis. Indeed, it is questionable whether Pigouvian taxation would allow regulators to impose significantly greater restrictions than they have already achieved using traditional tools. On the contrary, switching from current regulatory tools to Pigouvian taxes could lead to greater abortion access, not less.

Begin with the constitutional law of abortion. As with guns and speech, regulations affecting abortion access live or die by the legitimacy-tailoring framework. However, unlike with guns or speech, pro-life lawmakers’ primary obstacle to achieving their regulatory goals has been legitimacy, not tailoring.

In Roe, the Supreme Court applied strict scrutiny to Texas’s criminal abortion ban, requiring that it serve “a compelling state interest” and be “tailored to” that interest.334 Texas asserted two statutory goals: protecting prenatal “potential life” and women’s health.335 The Court determined that both interests were “important and legitimate.”336 But the legitimacy hurdle for strict scrutiny is high. A merely “important” interest is not good enough.

The Roe Court held that, for most of a pregnancy, the state’s asserted interests could not justify its ban. It is only “as the woman approaches term . . . [that] each [interest] becomes ‘compelling.’”337 In other words, the Court held that the costs of terminating a fetus that could meaningfully live outside its mother’s body are categorically different from those of terminating a previability pregnancy.338 A total abortion ban—or a similarly burdensome regulation—thus passes the legitimacy test only if limited to postviability procedures.339

Thus, the legitimacy requirement, not tailoring, clashes directly with pro-life lawmakers’ core regulatory vision. If one’s view is that the termination of a previability fetus is akin to the loss of a postnatal human life, then the only satisfying regulations of abortion will be highly burdensome. And any such restriction is extremely likely to trigger Roe’s stringent legitimacy requirement, demanding a “compelling” cost. Certainly, pro-life regulators believe that terminating a fetus creates just such a cost. Whatever the ontological truth, substantive constitutional doctrine dictates otherwise.

Pigouvian taxation is no help here. Suppose that, instead of Texas’s criminal ban, lawmakers wished to impose a Pigouvian tax on abortions, either at the locus of the patient or the provider. If one views the termination of a fetus as akin to the killing of a person, the right-sized Pigouvian rate is the same for both. Current estimates of the social cost of a lost-life converge around $12 million.340 Compare a $12 million tax with Texas’s maximum penalty for providing a requested abortion—five years’ imprisonment.341 Quite likely, if the latter penalty triggered strict scrutiny’s exacting legitimacy requirement, so would the former. In either case, the law fails for lack of a “compelling” social cost.

It is true that some command-and-control regulations of abortion have been struck down because of tailoring, not legitimacy, problems.342 Certain putative health-and-safety measures substantially restrict abortion access, while providing a “virtual absence of any health benefit.”343 Yet a “virtual absence of . . . benefit” suggests that lawmakers are not really trying to improve safety. Rather, so-called TRAP344 laws can be used as sub silentio attempts to impose such heavy regulatory burdens on abortion providers that they are forced to close.345 That they have been struck down demonstrates how the tailoring requirement can help prevent lawmakers from accomplishing by subterfuge what they could not do in the open.346 These setbacks notwithstanding, TRAP laws appear to have seen significant success. As of 2017, there were five states with only one abortion provider each and another eight with two or three.347

Should lawmakers impose Pigouvian taxes instead of otherwise-constitutional, health-and-safety rules? Not if their primary goal is to reduce the number of abortions. If the tax accurately tracks social cost, it incentivizes healthcare providers to avoid risks as efficiently as possible.348 If providers discover a cheaper precaution than the command-and-control rule would have required, the result will be lower costs for abortions, and thus more, not fewer, procedures.

Finally, one might wonder whether pro-life lawmakers could moderate their views of the social costs associated with abortions, and thereby impose Pigouvian taxes consistent with existing precedent. After all, Roe does say that protecting “potential life” constitutes an “important and legitimate,” though not “compelling,” regulatory goal.349 Perhaps, then, a moderate tax keyed to this “important” cost, whatever it may be, would withstand the more relaxed legitimacy inquiry promulgated in Planned Parenthood of Southeastern Pennsylvania v. Casey.350

The problem here is that, while the Supreme Court has been adamant that previability “potential life” is not like actual life, it has scrupulously refused to decide what “potential life” is. This is a hard question on which society does not agree. And the case law offers little guidance. Gonzales v. Carhart351 is the Supreme Court case most directly addressing the issue. It held that governments may, under the auspices of protecting potential life, ban surgical abortions that might particularly “shock[ ]” the public and “coarsen society.”352 But this just confuses things more. On this view, the value of potential life has nothing to do with the fetus itself. The identified costs are instead to postnatal humans. Other holdings further muddy the water. Consider that, in every case upholding a regulation targeting lost potential life, the Court has understood the regulation to impose only de minimis burdens on abortion access.353 This leaves open the possibility that, whatever the cost of a lost potential life, it may be quite small. A permissible Pigouvian tax on abortions, then, would have to be commensurately de minimis.

Without a coherent picture of what lost potential life is, Pigouvian taxes targeting this cost are highly speculative. Lawmakers could take a guess at what the Court means when it approves such costs as important. But they would be quite likely to guess wrong. And, in doing so, they might well identify a set of concerns that failed, under the legitimacy test, to justify their tax’s burden on constitutionally protected benefits. This is especially likely if the regulatory burden is substantial. Then, regulators would find themselves exactly where they are today; not with a tailoring problem, solvable via Pigouvian taxation, but with an intractable legitimacy problem.

These points are generalizable. There are other non-abortion-related regulatory projects for which Pigouvian taxation would be of little use. For example, some legislators view hateful speech directed at vulnerable minority populations as inherently costly. But the Supreme Court has repeatedly held that the ability to espouse hateful racial views is among the benefits the First Amendment protects.354 Thus, a Pigouvian tax on hate speech qua speech would fail at the first stage of the legitimacy test.355 Mirroring its refusal to elucidate potential life, the Court has not yet clarified whether the Second Amendment protects benefits associated with hunting.356 An animal-rights-focused legislator who wished to tax hunting would thus be saddled with similar uncertainties to those arising from any attempt to tax losses of potential life.

V. Practical and Normative Challenges

So far, this Article has argued for Pigouvian taxation as a method for effectively and constitutionally regulating perceived social costs from certain protected activities. But there are challenges, both practical and normative. This Part takes those up.

A. Practical Challenges

1. Manipulation and judicial review.

The previous Part argued that methods exist for regulators, acting in good faith, to measure with constitutional sufficiency the social costs targeted by Pigouvian taxes. But what about bad faith? Gun-rights advocates, for example, doubtless worry that left-leaning lawmakers promulgating a firearms tax would inflate—intentionally or unconsciously—social-cost estimates, taxing guns out of existence. Ought we worry here that endorsing “the power to tax” will, in effect, grant the “power to destroy”?357

There are three reasons for cautious optimism. First, recent administrative law scholarship suggests that cost estimates may not be as easily manipulated as it might appear.358 The Trump administration’s executive agencies have blatantly attempted to manipulate their CBAs in an effort to roll back Obama-era regulations.359 Despite this, they often succeeded only in “massag[ing] some of the numbers at the margin,” such that they “could not bring [themselves] to argue that the [rollbacks] were cost-justified.”360 In other instances, where “the Trump EPA did gin up phony CBAs, [ ] their phoniness is plain to anyone who cares to examine them.”361

This leads to the second point: The judiciary can and should review lawmakers’ social-cost estimates when scrutinizing regulations affecting constitutionally protected activity—Pigouvian or otherwise. Precisely how exacting such review should be is a difficult question, turning on the balance between agency expertise and judicial neutrality.362 At the very least, judicial review can and does reject laws when lawmakers have made no attempt to estimate targeted costs.363 Nor should we worry that courts are incompetent to disapprove estimates whose “phoniness [is] plain.”364 In fact, the mere threat of moderate review may have a significant restraining effect, including on regulators otherwise engaged in motivated reasoning.365

Third and finally, movement on the margin toward judicial demands for and review of cost-benefit estimates would improve constitutional analysis as compared with the current system. The doctrinal analysis above shows that courts already make judgments about whether regulations’ targeted social costs are proportionate to their burdens on constitutionally protected benefits.366 But they usually do so based on hunches, rules of thumb, and heuristics.367 Such reasoning likely admits even more manipulation than would be possible if lawmakers had to show their work. Consider again Heller’s unreflective endorsement of firearms bans for felons, based presumably on an intuition that such individuals are particularly likely to commit violent crimes.368 That intuition turns out to be false as to certain easily identified subcategories of convicted felons—including those over fifty.369 Current doctrine, then, endorses an unjustified total deprivation of Second Amendment rights for an entire category of older convicts—a disproportionate number of whom are Black and Brown.370 Pigouvian regulation based on explicit cost estimates could therefore help increase constitutionally justifiable regulation of protected activity while simultaneously decreasing unjustified deprivations of constitutional rights.

2. Whom to tax, and how.

Assume Pigouvian regulators have identified and measured a set of legitimately regulable social costs associated with a protected activity. Whom should they tax? And should every taxed entity pay the same amount? A gun tax, for example, could be imposed directly on gun owners or indirectly on sellers. It could be charged on an annual basis or only at the point of sale. It could be the same for every person and gun or vary with a number of factors. The values guiding such choices should be efficiency and administrability.

First, on efficiency, the key point is that Pigouvian taxes should be designed as well as possible to track actual social costs.371 Consider again the classic example of the polluting factory. Assume it generates ten units of pollution per widget. Regulators may be tempted to simply set the tax rate using this ratio and indefinitely charge on a per-widget basis. But doing so would create no incentive for technological improvements reducing per-widget pollution.372 Instead, regulators might install sensors on the factory’s smokestacks that measure actual pollutant output, thereby rewarding the factory when its per-widget output falls. If the smokestack option is too costly—because of technology, fraud, or other factors—regulators could tax using the per-widget estimate but update the rate as technology improves.373

The hypothetical taxes on guns and fake news could employ either design, or a mix of the two. Regulators could annually determine, using random sampling, the number of fake-news views or interactions on Twitter or Facebook. They could tax those companies on a per-view basis, perhaps varying the rate by category of fake news story.374 This is a smokestack sensor–style design, so long as the costs of each view are, in expectation, fairly uniform. The same would go for a gun tax that, starting on a specified date, charged gun manufacturers some share of the actual costs caused by their own guns sold after that date.375

The gun tax might alternatively be levied, at least in part, on an estimation basis. A tax on individual firearm owners could be charged in expectation, rather than after each person’s actual costs were tallied. But to adequately track social costs, such taxes should be sensitive to variations in expected cost between categories of gun owners. Existing research on the costs of gun ownership has already begun to account for such variation, quantifying cost differences between localities.376 Similar analysis could be performed to determine, for example, the marginal effects of precautions like trigger locks or gun safety courses on social costs. As new technology—like “smart guns”377 —emerges, its effects should be incorporated, too.

The end result of this estimation-based design is that gun owners would be sorted into a number of risk tranches. Those tranches would be charged rates consistent with lawmakers’ best estimates of expected cost.378 Certainly, such a system would fail to perfectly mirror actual costs. Some gun owners might have private information at the time of taxation that they intend to cause more harm than others similarly situated.379 But often no one, including owners themselves, will be able to predict intra-tranche variations. Few people expect to eventually kill themselves or for their child to accidentally discharge their weapon. In these situations, Pigouvian regulators need not predict the truly unpredictable. Instead, individuals’ cost estimates would equal average expected costs. Then, a Pigouvian tax in the amount of average cost is well-tailored and induces efficient behavior.

Some scholars have argued that accurately predicting cost variations is difficult—so difficult as to present a serious efficiency problem for Pigouvian taxation generally.380 But insofar as this is a problem, it is one for all modes of regulation, not just Pigouvian taxes.381 Moreover, Pigouvian taxes require regulators to make fewer estimates than alternative tools would.382 Thus, there is reason to be confident—or at least as confident as regulators can ever be—that carefully designed Pigouvian taxes would constitute responsible public policy.383

To pick from the options for whom to tax, and how, lawmakers should rely on considerations of administrability and efficiency. Ideally, taxes should be levied on entities that are easy to monitor and enforce against, and which are well-situated to take cost-justified precautions.

Sometimes, these considerations will converge, pointing to a single best locus of taxation. Consider the Pigouvian tax on fake news. It would be prohibitively costly to directly tax every individual or group that produced a fake news story. Even if regulators tried, the worst offenders, like foreign agents,384 would likely be the least compliant. But most fake news is spread online, via social media.385 The tax should thus be imposed on companies like Twitter and Facebook for two reasons. First, relatively few entities would need to be taxed, and they would be likely to comply with the law. Second, social media companies are well-positioned to develop solutions—including technological ones—for identifying and eradicating fake news stories. Already, small tech companies are developing machine learning techniques for detecting fake news with a high degree of accuracy.386 And Facebook has announced its intention to label, but not remove, fake news stories on its platform.387 It is no stretch to suppose that, spurred by a Pigouvian incentive, such companies, which employ the world’s best engineers, could develop even better solutions.

In other cases, administrability and efficiency considerations could suggest taxing more than one entity. A tax on gun manufacturers might, for example, incentivize the invention of safer, more advanced firearms. But because guns are durable, millions already exist on the secondary market.388 Thus, part of the Pigouvian tax should be designed to reach existing owners. This could be achieved in multiple ways. A government might impose an annual tax directly on households that own firearms or a sales tax every time a gun changed hands. Alternatively, it might tax a consumable complement of firearms, like ammunition.389 The optimal design might do all of these things in some proportion.

What about avoidance? Some scholars have suggested that, especially for guns, individuals who cause the most social cost are precisely those most likely to evade regulation.390 Responsible owners will pay their taxes, but criminals will not.391 As a preliminary point, this objection applies principally to social costs arising from criminal shootings. It is less salient regarding accidental shootings and suicides. And annual gun suicides alone outnumber gun homicides nearly two to one.392 Moreover, it relies on a picture of people who commit gun crimes as possessing a deep-seated criminal mentality: the television gangster whose business is violence and whose murders are long premeditated. But if most homicides instead reflect some combination of coincidence, inflamed passion, and tragic circumstance, the objection weakens. Then, eventual criminal actors’ preferences at the time of taxation—and thus the time of decision about gun ownership—are likely to be quite average. If most eventual perpetrators of homicide are ordinary people who live in dangerous surroundings, acquire cheap guns for protection, and have no standing commitments to homicide, then, at the time of taxation, they ought to respond to incentives just like the rest of us.

As for committed tax avoiders, evasion is a potential issue for any imaginable gun control regime, including total bans. Thus, any substantive firearms regulation would need to be accompanied by measures designed to reduce evasion. Pigouvian taxes, in particular, are amenable to such complementary measures. Rules requiring seller licensing and universal background checks, while unlikely on their own to substantially curb gun deaths, would improve tax enforcement.393 So could high penalties for failure to pay the Pigouvian gun tax. Indeed, we should expect such penalties to actively incentivize compliance in a way not possible with traditional regulation. Under a ban, the regulated person has only two high-cost options: forgo a gun and live without protection or obtain one and suffer high expected regulatory penalties. But a Pigouvian scheme offers a third, lower-cost option: pay a moderate tax and obtain protection. Evading the tax incurs higher expected regulatory costs but confers no marginal protection benefit. Thus, the rational choice for most—even those who would evade a ban—will be to pay the tax.394

3. Tort liability and other equivalent designs.

Some readers may wonder why we need Pigouvian taxes when we have tort liability. The two regimes share many features. Perhaps most importantly, both are, theoretically, sized to match targeted social costs. They therefore ought to enjoy the same advantages, both in effectiveness and constitutionality. But, as currently implemented, tort systems fall short on both scores.

As for effectiveness, litigation is slow and expensive. In the U.S. system, parties usually bear their own costs, such that it makes no financial sense to pursue a claim with an expected value smaller than the cost of litigation.395 Small claims can sometimes be aggregated in class actions. But substantive procedural rules often bar classwide litigation by individuals who were harmed in nonidentical ways.396 Even if those procedural rules were modified, calculating actual damages as to every class member in every case would remain a long and costly process.397 Moreover, even if tort claims were cheap and easy to bring, judgment-proofness would often limit recoveries. All these factors conspire to produce substantial underdeterrence, leaving social costs well above the optimal level.

Tort regimes can also run afoul of constitutional strictures. Consider the constitutional deficiencies the Sullivan Court identified in Alabama’s defamation law.398 Juries can wildly overestimate damages, resulting in poor constitutional tailoring.399 Furthermore, even accurate damage assessments for eggshell plaintiffs can be large, perhaps burdening the risk-averse as they engage in protected activity.400

The problems above could be resolved by reforming liability systems to look more like Pigouvian taxes. Instead of costly litigation resolved by unpredictable juries, plaintiffs might submit grievances to administrative adjudicators. Those bodies could operate under streamlined procedural and evidentiary rules specialized to the kinds of claims they resolved. They could be staffed by experts who relied on standard methods for computing damages.

These reforms would help to solve the problems of litigation expense and inaccurate damages awards. But they would leave untouched problems of judgment-proofness and risk aversion. Both might be solved by mandating liability insurance.401 With mandatory insurance in place, costs are imposed in the amount of predicted harm, rather than in the amount of actual harm. This would largely eliminate the judgment-proofness problem and smooth out unpredictable variations in damages, reducing risk. In such a system, sophisticated insurers, anticipating predictable damages awards imposed by experts, would likely settle the vast majority of cases. This would further reduce litigation costs.

A tort system thus reformed would compare well with Pigouvian taxes, both in terms of efficiency and constitutionality. This is largely because, in the end, the two systems are quite similar. In both, government experts measure social costs and standardize prices for costly activity. The main difference between the two regimes would be that, when instances of harm need to be predicted ex ante, insurers’ market incentives might cause them to outperform governments. On the other hand, insurers might not be willing to cover everyone, such that the highest-risk individuals would be excluded from the ex ante payment system. Considerations like these should rightly inform policy design choices.

One can imagine other variations on regulatory design that share the important features elucidated here. Cap-and-trade systems can be designed to mimic Pigouvian taxes. The trick is to increase the supply of permits for an activity until their price mimics the optimal Pigouvian tax rate.402 Insofar as individual jurisdictions prefer such variations on the core proposal presented here, this Article raises no objection.

B. Normative Challenges

The foregoing parts argue that Pigouvian taxes on protected activities would, at least in some important cases, be both constitutional and effective at reducing legitimate social costs. The upshot is that, if we wanted to, we could regulate constitutionally protected activities with Pigouvian taxes. But should we? Such regulation raises important normative concerns. This Section explores them. Along the way, it shows how Pigouvian thinking illuminates normative problems for any scheme of regulating constitutionally protected activity. Such thinking likewise suggests avenues for doctrinal reform that would improve protections—especially for the poor—under any regulatory regime.

1. Distributional effects.

Pigouvian taxes on constitutionally protected activity would have distributional effects. Money has diminishing marginal utility.403 Thus, every dollar of tax liability for owning a gun or speaking online or obtaining an abortion hurts the poor more than the rich. At some price, people of sufficiently limited means simply will not be able to engage in the protected activity at all. Imagine if the properly sized, constitutionally adequate Pigouvian tax on gun ownership turned out to be many thousands of dollars per year. Then, low-income individuals and families would be functionally barred from partaking of any of the benefits the Second Amendment guarantees. Middle-class families might still be able to own a firearm for self-defense, perhaps by making other sacrifices. Billionaires could own whole arsenals and never consciously feel the tax’s effect. In such a world, the rich have nearly unlimited access to the benefits the Constitution protects, while the poor have none.

Consider, however, that every valid regulation of constitutionally protected activity, including traditional ones, raises the cost of the activity. Sometimes these costs are explicit. Zoning restrictions on theaters make protected speech more expensive, yet they can be constitutionally imposed.404 Constitutionally valid tort penalties for negligent shootings or health regulations for abortion providers likewise raise the price tag for engaging in protected activity.

Other times, regulations raise the price of protected activity implicitly by increasing the time or effort required. Such hidden price hikes can have the same disparate effect on the poor, whose nonmonetary and monetary resources are both strained.405 Recall that Casey actually allows some regulations that impose a burden on obtaining an abortion.406 If, for example, a well-tailored safety regulation causes some clinics to close, that increases the travel necessary for some women to obtain abortions. Low-income women, with less free time, less flexible jobs, and less support at home, will be the most affected. This is not a mere theoretical possibility. There are a dozen states that presently have three or fewer abortion clinics.407 One can tell the same story about permitting requirements for demonstrations, waiting periods for gun purchases, and numerous other laws.

The distributional problem is thus pervasive. Understanding it tells us something about regulation via Pigouvian tax. But it tells us much the same about all regulation of protected activity. All such regulations involve tradeoffs between protected benefits and social costs. And the detriment to protected benefits functions as an increase in the price of obtaining that benefit. Increases in price—implicit or explicit, monetary or in-kind—hurt the poor more than the rich.

Thinking of constitutional regulation in economic terms—as Pigouvian taxes require—brings this problem to the fore. It allows us to see clearly how people of different means are disparately impacted by the tradeoffs that our constitutional law already makes. But, hopefully, it also helps us to think more clearly about the potential solutions.

To begin, the Constitution already contains some measures for mitigating the distributional problem, either from traditional or Pigouvian regulations. The legitimacy requirement functions as a first line of defense against regulations that could swallow protected benefits whole. Imagine that a protected activity had some widespread but trivial effect on a huge number of people—for example, the above-posited mild headaches from radio waves.408 Lawmakers might choose to regulate that effect, and the resulting implicit price increase could be substantial, even from a well-tailored law. But if that were the law’s effect, a heightened legitimacy requirement would apply.409 Then, the Constitution would demand at least a moderately important regulatory goal, and preventing mild headaches would presumably not suffice.410 This protection is more substantial than it might first seem, given that essentially any effect of activity might constitute a social cost in some regulator’s mind.411 The legitimacy requirement thus heads off innumerable significant distributional burdens.

Legitimacy tests, however, cannot solve all distributional problems. It is not difficult to imagine regulations substantially raising the costs of protected activity while satisfying both the legitimacy and tailoring requirements. Some social costs from protected activities are both large and able to pass a high legitimacy threshold. The major harms from gun ownership may fall into this category. As discussed above, the average marginal social cost of one household with a gun may be over $1,000 per year.412

Furthermore, consider that, as a formal matter, the courts generally review only the laws before them for constitutionality. Heightened legitimacy and tailoring requirements are thus triggered only when the law under review is particularly odious or burdensome on protected benefits. It is therefore possible to imagine a mosaic of laws, enacted over time, each of which is moderately tailored to a moderately important social cost. The laws might pass constitutional muster individually, but if considered together, they would trigger strict scrutiny and be struck down. In fact, that very thing may be happening now in states with just a few abortion providers. Those legislatures may have enacted a series of health-related regulations, each of which passes a low legitimacy and tailoring threshold on its own. But if enacted and examined together, the laws would have triggered more stringent review, and they would have been struck down.413

Existing constitutional rules protecting individual activity thus allow at least some burdensome regulations that will strongly impact the poor. What is there to say about this? There are two options, and the Supreme Court has seemed to say both, sometimes in a single case. One response is to say that the situation is just too bad. The Constitution does not guarantee the right to partake of protected benefits cheaply, and it certainly does not require society to subsidize those benefits, including by absorbing associated costs. The Court said something similar in City of Renton v. Playtime Theatres, Inc.,414 denying that there was a constitutional right to engage in protected speech “at bargain prices.”415

The other response is to say that the Constitution guarantees everyone some minimum quantum of protected benefits, regardless of wealth. Put another way, maybe the constitution places a cap on the maximum burden a regulation may impose, regardless of legitimacy and tailoring.416 This principle appears occasionally in the case law. Cases analyzing time, place, and manner restrictions on speech—including, again, City of Renton—have required such laws to leave available sufficient “alternative avenues of communication.”417 A rule like that could be used more broadly to strike down laws that, while tailored to legitimate social costs, make constitutional benefits substantially unattainable. Moreover, whether alternative avenues remain available depends on, among other things, regulations other than the one under review.418 The rule might thus be used to target constellations of regulations that, taken together, put protected benefits out of many people’s reach.

In the end, the “no bargain prices” principle carried City of Renton. The Court upheld the regulation, despite the “alternative avenues” rule, on the theory that that “[t]he inquiry for First Amendment purposes is not concerned with economic impact.”419 But it is hard to see why. As just discussed, all regulations impose costs. It should not matter to an alternative avenues–style test whether the laws foreclose alternatives directly, or indirectly via insurmountable increases in price or effort. Perhaps, then, the right reading of the “alternative avenues” test, as it exists today, is as guaranteeing an opportunity to speak not for everyone, but just for someone. So long as a regulation leaves space for some people to air their views, it survives review. If that is the rule, the “some people” in question will generally be the rich.

Nevertheless, we could update our constitutional rules to recognize that everyone, including the poor, has a right to some minimum quantum of each protected benefit. Courts could apply something like a beefed up version of the “alternative avenues” test across doctrinal areas. When reviewing a particular law, courts would examine the entire regulatory landscape. They would ask whether—regardless of the law’s legitimacy and tailoring—it left available to the poorest members of society some minimum ability to partake of constitutional goods. If not, the law would be invalidated, at least until legislators revised some portion of the relevant regulatory scheme to provide for the requisite minimums.

Implementing such a test would not be easy. It would require courts to decide, through the adversarial process, how innumerable laws interacted and affected the poor. It would also require them to determine, as to each constitutional right, what qualifies as an acceptable minimum quantum of benefits. This task, while difficult, is not entirely alien to constitutional law. The Sixth Amendment, for example, guarantees a right to an attorney in criminal cases.420 And the Supreme Court has held that, while this does not mean a free attorney in every case, governments must appoint counsel for those who cannot afford it.421 What constitutes minimally competent, and thus constitutionally sufficient, counsel is a complex and oft-litigated question.422 The Constitution similarly guarantees some minimum access to the courts, even for those who cannot pay ministerial fees.423

Such a minimum-benefits rule would essentially require lawmakers to adjust existing networks of regulations of protected activities to make them progressive. This could be accomplished, for example, with regulatory waivers for the indigent or targeted subsidies for low-income households.424

The analysis of distributional effects so far shows that they are an important problem for both Pigouvian and traditional regulations on constitutionally protected activity. But are they a bigger problem for one? The answer depends first on how bad things are now, under our regime of traditional rules. It depends second on the predicted marginal effect of substituting traditional rules with Pigouvian taxes.

There is some reason to think that the distributional problems under the current regime are significant. As discussed above, existing constitutional rules can deal with them only occasionally and indirectly. Thus unregulated, the distributional effects of current regulations may well be substantial.

If that is right, would a transition to Pigouvian regulation make the problem better, or worse? The problem might get better insofar as Pigouvian taxes can more efficiently regulate social costs than can traditional regulations. Recall that many laws affecting constitutional rights need maintain only a moderate proportionality between social costs avoided and constitutional benefits reduced. Therefore, many current laws probably impose bigger constitutional burdens than are necessary to achieve the desired reduction in costs. Replacing those with Pigouvian taxes would reduce constitutional burdens and, thus, price out fewer people from protected benefits.

On the other hand, this Article contends that a transition to Pigouvian taxation would enable regulation of previously unregulable activity. So long as those new regulations target social costs worth reducing, that is a good thing from a welfarist perspective. But more regulation means higher costs. Even if it reduces important harms, additional regulation initially exacerbates distributional effects.

It is possible, though, that moving toward regulation via Pigouvian tax would make our pervasive distributional problems publicly salient. Price increases from Pigouvian taxes are transparent, unlike the increases from current rules. And if the public, legislatures, or the courts began to take the problem seriously, perhaps rules would arise protecting minimum access to constitutional benefits. Such rules would also be easier to implement in a world where constitutional regulation was primarily Pigouvian. The regulatory costs from Pigouvian taxes are easy to determine and offset. Thus, a shift toward Pigouvian regulation might enable long-run reductions in distributional problems, as compared with our current baseline.

One final note on distribution: The distributional consequences of regulation apply far beyond the contexts discussed here. All regulations—even responsible food, environmental, or highway safety rules—can raise prices. And those price increases can put basic goods, both constitutional and otherwise, beyond the reach of the poorest citizens. The root of these most severe distributional problems, then, is poverty. Thus, the best way to remedy them is not to tinker with every law, making it marginally less burdensome on the very poor. It is to enact aggressive, freestanding redistributive programs that help create a society in which there are no very poor.

2. The expressive function of traditional regulations.

Another normative concern about Pigouvian taxes might be that they cannot perform all the functions of traditional regulations. In particular, one might wonder whether bans, permitting schemes, or other command-and-control rules have expressive content that Pigouvian taxes lack.425 There are several reasons to think that little expressive power would be lost by adopting the proposals advocated here.

This Article is about regulating activity that has traditionally been difficult to regulate. In this context, the choice is not between traditional regulation and Pigouvian taxes. Rather, it is between Pigouvian taxes and little or no regulation. Thus, even if Pigouvian taxes lack any expressive quality whatsoever, no expression is lost by using them if no other option exists.

What about situations where lawmakers could choose either Pigouvian taxation or traditional regulation? One major reason to care about the law’s expressive function is that it may influence behavior. It can do so by shaping social norms426 or by providing information relevant to decision-making.427 Perhaps a severe restriction on handguns causes society to look down on those who buy them, thus imposing a “social tax” on their purchase.428 Or maybe such a law conveys information to gun owners about the risks from firearms to themselves or others.

A Pigouvian tax can do both things. If social norms are implicit taxes, Pigouvian taxes are just explicit versions of the same thing. As for laws’ information-transmitting role, economists have long held that prices—which Pigouvian taxes raise—are a tool for aggregating and transmitting information.429 Raising a price by the Pigouvian rate simply incorporates information about social costs into that price. This effect can be made more salient by flagging the existence and purpose of a given Pigouvian tax.430

Finally, it is not clear what the law should express when regulating constitutionally protected activity. Owning a gun or getting an abortion is not like committing a murder, which everyone agrees is bad and should be condemned. Some approve of these protected activities; others do not. And the Constitution enshrines a right to engage in them, absent some important countervailing harm. Should the law then turn social norms against protected activity, on account of the harm it causes? Or should it turn them in favor of it, so that people may more readily partake of constitutional goods? The answer is not obvious. Perhaps then, in this realm, the law ought to be silent, or at least speak only cautiously.


Courts, lawmakers, and society as a whole have long recognized that constitutionally protected activity is not all upside. Certainly, some of the benefits reaped from such activities are vitally important—important enough to be safeguarded by the “supreme law of the land.”431 But there are harms, too, some of which are likewise of great importance to lawmakers and their constituents. And the very supreme law that protects the benefits has also often made those harms difficult or impossible to control. Pigouvian taxation offers a regulatory structure that can do what few others can. In important cases, Pigouvian taxes can effectively reduce the most important social costs from protected activity, while maintaining the Constitution’s delicate balance between regulation and rights. Pigouvian taxation thus represents a valuable new regulatory tool in areas historically fraught with constitutional pitfalls. Indeed, absent sweeping constitutional change, it may be the only means available for meaningfully attacking modern problems like rising gun death rates or the scourge of online fake news.

  • 1John Gramlich, What the Data Says About Gun Deaths in the U.S., Pew Rsch. Ctr. (Aug. 16, 2019),
  • 2See Hunt Allcott & Matthew Gentzkow, Social Media and Fake News in the 2016 Election, 31 J. Econ. Persps. 211, 212 (2017).
  • 3The reality and severity of harm from certain constitutionally protected activities—like terminating a pregnancy—is contested. While I have personal views on such questions, they are not part of this Article’s analysis. This Article is about the structural relationship between individual constitutional rights and regulatory design. I therefore take as granted the normative preferences of legislators from across the political spectrum, asking how those preferences can or cannot be enacted into policy. On occasion, substantive constitutional law will supply answers—or at least boundaries within which answers might fall—to normative questions. For example, as discussed below, the Constitution’s protection of abortion rights limits the range of legitimately regulable social costs. See infra Part IV.D. Those limits do matter to the analysis.
  • 4New York Times Co. v. Sullivan, 376 U.S. 254, 279–80 (1964).
  • 5District of Columbia v. Heller, 554 U.S. 570, 635 (2008); McDonald v. City of Chicago, 561 U.S. 742, 791 (2010). By “deadliest,” I mean “used in by far the most shootings.” See infra note 47 and accompanying text.
  • 6554 U.S. 570 (2008).
  • 7Cheyenne Haslett & Samantha Sergi, Here’s Where the 2020 Democrats Stand on Gun Control, ABC News (Feb. 7, 2020),
  • 8See infra notes 50–55.
  • 9See infra Part I.
  • 10See, e.g., Sullivan, 376 U.S. at 283; see also Joseph Blocher & Darrell A.H. Miller, What Is Gun Control? Direct Burdens, Incidental Burdens, and the Boundaries of the Second Amendment, 83 U. Chi. L. Rev. 295, 310–13, 334–35 (2016).
  • 11Jonathan S. Masur & Eric A. Posner, Toward a Pigouvian State, 164 U. Pa. L. Rev. 93, 95 (2015).
  • 12Id. at 115–16.
  • 13This is a reasonable first-cut estimate of the average annual marginal social cost per gun-owning household. See infra notes 291–97 and accompanying text. As described below, the best Pigouvian taxes would vary the rates charged, commensurate with variation in social costs imposed by different taxed entities. See infra Part V.A.2.
  • 14See infra Part V.A.2.
  • 15See infra Part III.
  • 16See infra Part III.
  • 17See infra Part III.
  • 18See infra Part IV.A.
  • 19See infra Part IV.A.
  • 20See infra notes 224–27 and accompanying text.
  • 21See infra notes 228–31 and accompanying text.
  • 22See infra notes 232–41 and accompanying text.
  • 23See infra Part IV.B.
  • 24See infra Part IV.D.
  • 25See infra Part IV.C.
  • 26See infra Part IV.C.
  • 27See infra Part V.B.1.
  • 28See infra Part V.B.1.
  • 29See infra Part V.B.1.
  • 30See infra Parts IV.B, IV.D.
  • 31Certainly, factors other than constitutionality, like politics, also matter.
  • 32U.S. Const. amends. V, XIV.
  • 33U.S. Const. amend. IV. The Constitution presents no obstacle to banning government discrimination or unreasonable searches. But the Constitution’s prohibition on those government activities sometimes makes other regulatory goals more difficult to achieve. Antidiscrimination rules might impede, for example, public universities’ diversity goals. And the warrant requirement makes policing more laborious. Here, Pigouvian taxes might play a role—though inverted from the general scheme of this Article. If the government discriminates or conducts unreasonable searches as it pursues its other legitimate goals, perhaps it should pay citizens for the harm it causes, instead of the other way around. This is not so different from the schemes imposed, at least in theory, by 42 U.S.C. § 1983 and Bivens v. Six Unknown Named Agents of Federal Bureau of Narcotics, 403 U.S. 388, 392 (1971). See infra Part V.A.3 (comparing Pigouvian taxation and tort liability). But, as currently implemented, those schemes are quite underinclusive, granting compensation for only a small fraction of violations. See, e.g., Ziglar v. Abbasi, 137 S. Ct. 1843, 1857 (2017) (cabining Bivens); Devon W. Carbado, Blue-on-Black Violence: A Provisional Model of Some of the Causes, 104 Geo. L.J. 1479, 1519–24 (2016) (discussing qualified immunity).
  • 34U.S. Const. arts. I–III.
  • 35U.S. Const. amend. I.
  • 36See Paul v. Virginia, 75 U.S. 168, 180 (1869) (describing the constitutional right of “free ingress into other States, and egress from them”).
  • 37See Harper v. Va. Bd. of Elections, 383 U.S. 663, 668 (1966) (ruling that the Virginia poll tax violated the Equal Protection Clause). Someone must pay to print the ballots.
  • 38See Boddie v. Connecticut, 401 U.S. 371, 374 (1971) (holding that a state may not deny someone access to the courts to dissolve their marriage based solely on inability to pay). Someone must pay for judges, clerks, court reporters, bailiffs, and others.
  • 39See infra Part III.
  • 40Consider, for example, that under Employment Division, Department of Human Resources of Oregon v. Smith, 494 U.S. 872, 878–79 (1990), the Free Exercise Clause—unlike the Free Speech Clause—requires no tailoring of a “valid and neutral law of general applicability.” Id. at 879 (quotation marks omitted) (quoting United States v. Lee, 455 U.S. 252, 263 n.3 (1982) (Stevens, J., concurring in the judgment)). But see Hosanna-Tabor Evangelical Lutheran Church & Sch. v. EEOC, 565 U.S. 171, 189–90 (2012) (holding that Smith does not apply to all neutral and generally applicable laws).
  • 41See Erin Grinshteyn & David Hemenway, Violent Death Rates: The U.S. Compared with Other High-Income OECD Countries, 2010, 129 Am. J. Med. 266, 268 (2016).
  • 42See Jason E. Goldstick, April Zeoli, Christina Mair & Rebecca M. Cunningham, US Firearm-Related Mortality: National, State, and Population Trends, 1999–2017, 38 Health Affs. 1646, 1647 exhibit 2 (2019).
  • 43See Grinshteyn & Hemenway, supra note 41, at 270.
  • 44See Goldstick, et al., supra note 42, at 1650–51.
  • 45See, e.g., Sripal Bangalore & Franz H. Messerli, Gun Ownership and Firearm-Related Deaths, 126 Am. J. Med. 873, 875 (2013) (all firearm-related deaths); Philip J. Cook & Jens Ludwig, The Social Costs of Gun Ownership, 90 J. Pub. Econ. 379, 387 (2006) (homicide); Mark Duggan, More Guns, More Crime, 109 J. Pol. Econ. 1086, 1088 (2001) (homicide); J. John Mann & Christina A. Michel, Prevention of Firearm Suicide in the United States: What Works and What Is Possible, 173 Am. J. Psychiatry 969, 970 (2016) (suicide).
  • 46See Heller, 554 U.S. at 574–75; McDonald v. City of Chicago, No. 08 C 3645, 2008 WL 5111112, at *1 (N.D. Ill. Dec. 4, 2008).
  • 47See Fed. Bureau of Investigation, 2016 Crime in the United States: Expanded Homicide Table 4, (homicide); see also Steven Stack & Ira Wasserman, Gender and Suicide Risk: The Role of Wound Site, 39 Suicide & Life-Threatening Behav. 13, 18 tbl.4 (2009) (suicide); R.E. Kohlmeier, C.A. McMahan & V.J.M. DiMaio, Suicide by Firearms, a 15-Year Experience, 22 Am. J. Forensic Med. & Pathology 337, 340 (2001) (suicide).
  • 48See Heller, 554 U.S. at 628–30; see also infra Part III.A.
  • 49Heller, 554 U.S. at 626–30.
  • 50See Haslett & Sergi, supra note 7.
  • 51Id.
  • 52See Christopher S. Koper, Daniel J. Woods & Jeffrey A. Roth, An Updated Assessment of the Federal Assault Weapons Ban: Impacts on Gun Markets and Gun Violence, 1994–2003, at 2 (July 2004),; Christopher S. Koper, William D. Johnson, Jordan L. Nichols, Ambrozine Ayers & Natalie Mullins, Criminal Use of Assault Weapons and High-Capacity Semiautomatic Firearms: An Updated Examination of Local and National Sources, 95 J. Urb. Health 313, 314 (2018).
  • 53See, e.g., Lois K. Lee, Eric W. Fleegler, Caitlin Farrell, Elorm Avakame, Saranya Srinivasan, David Hemenway & Michael C. Monuteaux, Firearm Laws and Firearm Homicides, 177 J. Am. Med. Ass’n Internal Med. 106, 117 (2017).
  • 54See Haslett & Sergi, supra note 7.
  • 55See Effects of Background Checks on Violent Crime, RAND Corp. (Mar. 2, 2018),
  • 56See Allcott & Gentzkow, supra note 2, at 212.
  • 57Id.
  • 58Angie Drobnic Holan, 2016 Lie of the Year: Fake News, PolitiFact (Dec. 13, 2016),
  • 59Id. Note that Trump rally attendees did chant the similarly racist slogan, “send her back,” directed at four non-White Congresswomen. See Joseph Zeballos-Roig & Walt Hickey, Over One-Third of Republican Primary Voters Supported the Racist Chants at Trump’s North Carolina Rally, Bus. Insider (Aug. 2, 2019),
  • 60See Richard Lardner & Deb Riechmann, Intel Officials Detail How Russian Cyberattacks Sought to Interfere with U.S. Elections, PBS News Hour (June 21, 2017),
  • 61See Richard Gunther, Paul A. Beck & Erik C. Nisbet, “Fake News” and the Defection of 2012 Obama Voters in the 2016 Presidential Election, 61 Electoral Stud., Oct. 2019, at 6.
  • 62Id.
  • 63See Bobby Chesney & Danielle Citron, Deep Fakes: A Looming Challenge for Privacy, Democracy, and National Security, 107 Calif. L. Rev. 1753, 1772–73 (2019).
  • 64Id.
  • 65567 U.S. 709 (2012).
  • 66376 U.S. 254 (1964).
  • 67See Cass R. Sunstein, Falsehoods and the First Amendment, 33 Harv. J.L. & Tech. 388, 389 (2020).
  • 68Id. at 389.
  • 69See, e.g., Philip M. Napoli, What If More Speech Is No Longer the Solution? First Amendment Theory Meets Fake News and the Filter Bubble, 70 Fed. Commc’ns L.J. 55, 97 (2018); Daniela C. Manzi, Managing the Misinformation Marketplace: The First Amendment and the Fight Against Fake News, 87 Fordham L. Rev. 2623, 2645–49 (2019) (critiquing existing First Amendment doctrine because it prevents lawmakers from directly addressing the problem of fake news). But see Chesney & Citron, supra note 65, at 1791 (arguing that First Amendment precedent “leav[es] considerable room for carefully tailored prohibitions of certain harmful deep fakes”).
  • 70See Manzi, supra note 69, at 49 (collecting such proposals and discussing their likely ineffectiveness).
  • 71410 U.S. 113 (1973).
  • 72Jon Swaine, The Anti-Abortion Conservative Quietly Guiding Trump’s Supreme Court Pick, The Guardian (July 6, 2018),
  • 73Scott Lemieux, Yes, Roe Really Is in Trouble, Vox (May 15, 2019),; Devin Dwyer, Liberals Alarmed for Roe v. Wade As Supreme Court Conservatives Overturn 40-Year-Old Precedent, ABC News (May 13, 2019),
  • 74See, e.g., Whole Woman’s Health v. Hellerstedt, 136 S. Ct. 2292, 2311 (2016).
  • 75Denise Lavoie & Carl Willis, Abortion-Rights Groups Challenge TRAP Laws That Restrict Providers, Limit Women’s Choices, ABC7 WJLA (May 16, 2019),
  • 76See, e.g., Whole Woman’s Health, 136 S. Ct. at 2318.
  • 77The name “Pigouvian” (also rendered “Pigovian”) refers to Arthur Pigou, the British economist who first advocated corrective taxes to reduce industrial pollution. See generally Arthur Cecil Pigou, The Economics of Welfare (1920).
  • 78Donatella Baiardi & Mario Menegatti, Pigouvian Tax, Abatement Policies and Uncertainty on the Environment, 103 J. Econ. 221, 222 (2011).
  • 79Masur & Posner, supra note 11, at 95.
  • 80Id. at 96.
  • 81Id. at 101.
  • 82Id.
  • 83Id. at 101–02.
  • 84Since legislators often fail to show their work, it can be difficult to tell how closely certain taxes track the true Pigouvian rate. Masur & Posner, supra note 11, at 104–08.
  • 85One should expect such results, insofar as people make rational choices based on changes in price. People do not always do this, but they appear able to account for such changes, especially when taxes are made salient at the point of transaction. See Raj Chetty, Adam Looney & Kory Kroft, Salience and Taxation: Theory and Evidence, 99 Am. Econ. Rev. 1145, 1170–71 (2009).
  • 86Bruce Pasfield & Elise Paeffgen, How to Enforce a Carbon Tax: Lessons from the Montreal Protocol and the U.S. Experience with the Ozone Depleting Chemicals Tax, 14 Vt. J. Env’t L. 389, 393, 395–96 (2013).
  • 87Id. at 397–98.
  • 88See J. Andrew Hoerner, Taxing Pollution, in Ozone Protection in the United States: Elements of Success 39, 39 (Elizabeth Cook ed., 1996) (describing how the tax and related policies enabled the United States to exceed its targets); Internal Revenue Serv., Ozone Depleting Chemicals (ODC) Excise Tax Audit Techniques Guide 5 (2007) (describing the success of the tax and related policies at raising prices and reducing consumption).
  • 89Some of the social costs associated with consuming sugar are rightly considered internalities, not externalities. For a discussion of how internalities fit into this Article’s analysis, see infra note 321.
  • 90Anna H. Grummon, Benjamin B. Lockwood, Dmitry Taubinsky & Hunt Allcott, Designing Better Sugary Drink Taxes, 365 Science 989, 989 (2019).
  • 91Id.
  • 92Tatiana Homonoff, Lee-Sien Kao, Doug Palmer & Christina Seybolt, Skipping the Bag: Assessing the Impact of Chicago’s Tax on Disposable Bags 3–4 (2018),
  • 93Id. at 4.
  • 94Tatiana A. Homonoff, Can Small Incentives Have Large Effects? The Impact of Taxes Versus Bonuses on Disposable Bag Use, 10 Am. Econ. J.: Econ. Pol’y 177, 190 (2018).
  • 95See Frank Convery, Simon McDonnell & Susana Ferreira, The Most Popular Tax in Europe? Lessons from the Irish Plastic Bags Levy, 38 Env’t & Res. Econ. 1, 7–10 (2007); Rebecca L. Taylor & Sofia B. Villas-Boas, Bans vs. Fees: Disposable Carryout Bag Policies and Bag Usage, 38 Applied Econ. Persps. & Pol’y 351, 371 (2016).
  • 96While a few scholars have considered Pigouvian taxes on guns from the perspective of public policy, none has made this Article’s main argument: that Pigouvian taxes have a significant constitutional advantage over other regulatory designs. Professor Victor Fleischer, in a generalized argument against Pigouvian taxation, has written that setting the correct tax on guns presents difficult information problems. Victor Fleischer, Curb Your Enthusiasm for Pigouvian Taxes, 68 Vand. L. Rev. 1673, 1677–81 (2019); see also infra Part IV.C. Professor Samuel D. Brunson argues for something he calls a Pigouvian tax on guns. See Samuel D. Brunson, Paying for Gun Violence, 104 Minn. L. Rev. 605, 607 (2019). But one could question whether that approach is Pigouvian at all, insofar as Brunson’s “proposal is not intended to change individuals’ behavior.” Id. at 623. Professors Brian Galle and Murat Mungan have briefly advocated a true Pigouvian tax on firearm ownership for reasons of sound public policy. Brian Galle & Murat Mungan, Predictable Punishments, 11 U.C. Irvine L. Rev. 337, 380 (2020).
  • 97While a few scholars have considered Pigouvian taxes on guns from the perspective of public policy, none has
  • 98made this Article’s main argument: that Pigouvian taxes have a s
  • 99gnificant constitutional advantage o
  • 100er other regulatory designs. Professor Victor Fleischer, in a generalized
  • 101rgumen
  • 102against Pigouvian taxation, has written that setting the correct tax on guns presents difficult information problems. Victor Fleischer, Curb Your Enthusiasm for Pigouvian Taxes, 68 Vand. L. Rev. 1673, 1677–81 (2019); see also infra Part IV.C. Professor Samuel D. Brunson argues for something he calls a Pigouvian tax on guns. See Samuel D. Brunson, Paying for Gun Violence, 104 Minn. L. Rev. 605, 607 (2019). But one could question whether that approach is Pi
  • 103ouvian at all, insofar as Brunson’s “proposal is not intended to change individuals’ behavior.” Id. at 623. Professo
  • 104s Brian Galle and Murat Mungan have briefly advocated a true Pigouvian tax on firearm ownership for reasons of sound public policy. Brian Galle & Murat Mungan, Predictable Punishments, 11 U.C. Irvine L. Rev. 337, 380 (2020).

    ling for more categorical analysis in abortion-rights doctrine); see also Genevieve Lakier, Reed v Town of Gilbert, Arizona, and the Rise of the Anti-Classificatory First Amendment, 2016 Sup. Ct. Rev. 233, 234–35 (2017) (analyzing an attempt to impose a categorical test for content neutrality in First Amendment cases).

  • 105See Ronald Dworkin, Taking Rights Seriously, at xi (1977).
  • 106See infra Part III.B.3; notes 188–92 and accompanying text; note 342; see also Fallon, supra note 102, at 109–11, 113–16.
  • 107See infra note 138 and accompanying text. It is easy to see how a philosophy like Professor David Strauss’s, with its emphasis on policy tradeoffs and common law change, could accept legitimacy and tailoring as emergent tests for constitutionality. See David A. Strauss, Do We Have a Living Constitution?, 59 Drake L. Rev. 973, 977–84 (2011).
  • 108Jamal Greene, Rights as Trumps?, 132 Harv. L. Rev. 28, 34 (2018). Greene rightly observes that present-day constitutional practice also includes categorical considerations, the resolution of which can stunt or foreclose proportionality analysis. Id. at 38–56. Such categorical considerations fall under the legitimacy prong of the framework developed here.
  • 109Jackson, supra note 103, at 3104–05.
  • 110Greene, supra note 108, at 109–10; Jackson, supra note 103, at 3106. Some originalist scholars would no doubt disagree. This Article, however, is concerned with understanding our present-day constitutional doctrine, originalist or otherwise. Moreover, some prominent originalists have recently argued that contemporary constitutional law is, in fact, rightly considered originalist. See generally William Baude, Is Originalism Our Law?, 115 Colum. L. Rev. 2349 (2015); William Baude & Stephen E. Sachs, Grounding Originalism, 113 Nw. U. L. Rev. 1455 (2019). If one accepts large swaths of extant individual-rights case law as correctly decided, it is difficult to see how one could simultaneously reject all models resembling the one advocated in this Article. Fallon, supra note 102, at 8.
  • 111Michael Coenen, Of Speech and Sanctions: Toward a Penalty-Sensitive Approach to the First Amendment, 112 Colum. L. Rev. 991, 1002–16 (2012).
  • 112Joseph Blocher, Categoricalism and Balancing in First and Second Amendment Analysis, 84 N.Y.U. L. Rev. 375, 413 (2009).
  • 113Josh Blackman, The Constitutionality of Social Cost, 34 Harv. J.L. & Pub. Pol’y 951, 1004–31 (2011) (arguing that, across individual rights, the Constitution balances social costs against protected benefits).
  • 114See, e.g., Jud Mathews & Alec Stone Sweet, All Things in Proportion? American Rights Review and the Problem of Balancing, 60 Emory L.J. 797, 813–33 (2010).
  • 115See Richard H. Fallon, Jr., Strict Judicial Scrutiny, 54 UCLA L. Rev. 1267, 1315–34 (2007); Fallon, supra note 102, at 47–66.
  • 116Fallon, supra note 102, at 48.
  • 117See id. at 47–59.
  • 118Id. at 48–54.
  • 119Id. at 54–59.
  • 120See infra Part III.C.
  • 121Fallon, supra note 102, at 60–66.
  • 122Note that Fallon’s book spends many pages mining the subtleties, variations, and contradictions in how these tests are applied. This Article lacks the space to be so expansive. Such subtleties and apparent contradictions exist. But I follow Fallon in contending that the only way to understand any significant portion of the case law is by invoking a framework like the one advocated here. Id. at 43–46.
  • 123Heller, 554 U.S. at 622.
  • 124Id. at 635.
  • 125See McDonald v. City of Chicago, 561 U.S. 742 (2010).
  • 126Heller, 554 U.S. at 634 (quoting id. at 689 (Breyer, J., dissenting)).
  • 127Id.
  • 128Mathews & Stone Sweet, supra note 114, at 802; Greene, supra note 108, at 85–96.
  • 129Heller, 554 U.S. at 626.
  • 130Id. at 636.
  • 131Id. at 632 (quoting Chapter 1501, in 5 The Colonial Laws of New York from the Year 1664 to the Revolution 244, 244–46 (Albany, James B. Lyon 1894)).
  • 132Id.
  • 133Id. at 633.
  • 134Heller, 554 U.S. at 626–27.
  • 135Id. at 627 n.26.
  • 136Id.
  • 137Blocher & Miller, supra note 10, at 310–13, 334–35.
  • 138Heller, 554 U.S. at 626–27.
  • 139Id. at 636.
  • 140See id. at 627–28.
  • 141Id. at 628 (quotation marks omitted).
  • 142Id.
  • 143Heller, 554 U.S. at 628.
  • 144See infra Part III.B.2.
  • 145See Sarah Herman Peck, Cong. Rsch. Serv., R44618, Post-Heller Second Amendment Jurisprudence 15–17 & nn.140–57 (2019), (describing the circuit courts’ use of traditional tiers of scrutiny and collecting numerous cases); Eric Ruben & Joseph Blocher, From Theory to Doctrine: An Empirical Analysis of the Right to Bear Arms After Heller, 67 Duke L.J. 1433, 1486–96 (2018). Note that, as Professors Ruben and Blocher describe, courts may often apply this logic without explicitly flagging that they are doing so. Id. at 1487.
  • 146505 U.S. 123 (1992).
  • 147Id. at 126.
  • 148Id. at 130 (explaining that the law “must be narrowly tailored to serve a significant governmental interest”).
  • 149Id. at 134 (quotation marks omitted) (quoting Ark. Writers’ Project, Inc. v. Ragland, 481 U.S. 221, 230 (1987)).
  • 150Id. at 134–35; see also Texas v. Johnson, 491 U.S. 397, 400, 404, 407 (1989).
  • 151Forsyth, 505 U.S. at 134.
  • 152395 U.S. 444 (1969).
  • 153Id. at 444. Brandenburg forbids penalizing speech because of listeners’ potential response unless the speech is “directed to inciting . . . imminent lawless action and is likely” to do so. Id. at 447. The “directed to” language heads off any potential for shutting down speech via a “heckler’s veto.” See Owen M. Fiss, Free Speech and Social Structure, 71 Iowa L. Rev. 1405, 1416–17 (1986). The Brandenburg rule currently applies only in the First Amendment context, but one could imagine it migrating elsewhere to head off similar moral hazards.

    For another example of a tax on speech that failed the legitimacy test, see Minneapolis Star & Tribune Co. v. Minnesota Commissioner of Revenue, 460 U.S. 575, 586 (1983). There, because the tax singled out the press, the legitimacy requirement was heightened, requiring a “compelling” governmental goal. Id. at 585. The state’s asserted goal—raising revenue—was not important enough. Id. at 586.

  • 154Forsyth, 505 U.S. at 133.
  • 155Id.
  • 156Id.
  • 157312 U.S. 569 (1941).
  • 158Forsyth, 505 U.S. at 136 (citing Cox, 312 U.S. 569).
  • 159Id. at 136 (quoting Cox, 312 U.S at 577).
  • 160Cox, 312 U.S. at 577.
  • 161Id. at 575–77.
  • 162See, e.g., Stonewall Union v. City of Columbus, 931 F.2d 1130, 1135 (6th Cir. 1991); Int’l Women’s Day March Planning Comm. v. City of San Antonio, 619 F.3d 346, 366 (5th Cir. 2010).
  • 163See Int’l Women’s Day March, 619 F.3d at 366; Sullivan v. City of Augusta, 511 F.3d 16, 35–36 (1st Cir. 2007); S. Or. Barter Fair v. Jackson City, 372 F.3d 1128, 1140–41 (9th Cir. 2004); Coal. for the Abolition of Marijuana Prohibition v. City of Atlanta, 219 F.3d 1301, 1322–24 (11th Cir. 2000); MacDonald v. Chi. Park Dist., 132 F.3d 355, 362–63 (7th Cir. 1997) (denying a preliminary injunction); Stonewall Union, 931 F.2d at 1134–36; cf. Nationalist Movement v. City of York, 481 F.3d 178, 184 (3d Cir. 2007) (“It is beyond peradventure that a city can establish a permit scheme whose goal is to ‘assure financial accountability for damage caused by’ an event.” (quoting Thomas v. Chi. Park Dist., 534 U.S. 316, 322 (2002))); Transp. Alts., Inc. v. City of New York, 340 F.3d 72, 79 (2d Cir. 2003) (noting that such fees can “clearly” be “consistent with the First Amendment”).
  • 164For an analysis of the Court’s difficulty distinguishing between content-neutral and content-based laws, see generally Lakier, supra note 104. If Professor Lakier is right, and the Roberts Court takes a broad view of what counts as content-based, the stricter legitimacy and tailoring standards described below will apply more often.
  • 165See Geoffrey R. Stone, Content-Neutral Restrictions, 54 U. Chi. L. Rev. 46, 47–54 (1987). As Professor Stone argues, multiple tiers of stringency may be embedded in intermediate scrutiny. Id. The truth might even be that there is a sliding scale of stringency between rational basis review and strict scrutiny.
  • 166See R.A.V. v. City of St. Paul, 505 U.S. 377, 386 (1992).
  • 167Stone, supra note 165, at 48–50; see also, e.g., City of Renton v. Playtime Theatres, Inc., 475 U.S. 41, 50 (1986) (“substantial”); United States v. O’Brien, 391 U.S. 367, 377 (1968) (“important”); Village of Schaumburg v. Citizens for a Better Env’t, 444 U.S. 620, 636 (1980) (“strong”).
  • 168Ward v. Rock Against Racism, 491 U.S. 781, 799 (1989); see also Stone, supra note 165, at 48–49.
  • 169Ward, 491 U.S. at 798.
  • 170City of Renton, 475 U.S. at 47–48 (emphasis omitted).
  • 171Id. at 52.
  • 172Ward, 491 U.S. at 796 (quoting Members of City Council of L.A. v. Taxpayers for Vincent, 466 U.S. 789, 806 (1984)).
  • 173Id. at 800.
  • 174Watchtower Bible & Tract Soc’y of N.Y., Inc. v. Village of Stratton, 536 U.S. 150, 168–69 (2002).
  • 175Id. at 164–65.
  • 176Id. at 165–68.
  • 177Id. at 165, 168–69.
  • 178Id. at 165.
  • 179McCullen v. Coakley, 573 U.S. 464, 496–97 (2014).
  • 180Id. at 490–94.
  • 181See Playboy, 529 U.S. at 813.
  • 182Id.
  • 183Id. (quoting Reno v. ACLU, 521 U.S. 844, 874 (1997)).
  • 184Id. at 826.
  • 185Id. at 819–21.
  • 186Playboy, 529 U.S. at 815.
  • 187Id.
  • 188Gerald Gunther, In Search of Evolving Doctrine on a Changing Court: A Model for a Newer Equal Protection, 86 Harv. L. Rev. 1, 8 (1972).
  • 189Adam Winkler, Fatal in Theory and Strict in Fact: An Empirical Analysis of Strict Scrutiny in the Federal Courts, 59 Vand. L. Rev. 793, 796, 844 (2006).
  • 190Holder v. Humanitarian Law Project, 561 U.S. 1, 28 (2010).
  • 191Id.
  • 192Id. at 30.
  • 193Sullivan, 376 U.S. at 279–80.
  • 194Id. Note that, for defamation not involving a public figure, the mens rea requirement is lower. See Gertz, 418 U.S. at 347–49.
  • 195See Gertz, 418 U.S. at 343 (affirming “the state interest in compensating injury to the reputation of [public and] private individuals”).
  • 196Sullivan, 376 U.S. at 256–57.
  • 197The paper misstated, for example, which patriotic song protestors sang and what precise formation the police stood in. Id. at 258–59.
  • 198Id. at 289.
  • 199Id. at 256, 262. The jury did not differentiate between actual and punitive damages. Id. at 284.
  • 200This inflation adjustment was calculated in April 2021 using U.S. Inflation Calculator,
  • 201Sullivan, 376 U.S. at 278–89 (citing John Stuart Mill, On Liberty 15 (Oxford 1947)).
  • 202418 U.S. 323 (1974).
  • 203Id. at 347–49.
  • 204Id. at 350.
  • 205Id.
  • 206Id. One might wonder why the Gertz rule should not apply in all cases, allowing recovery of actual damages upon a showing of negligence, even when the plaintiff is a public official. That would seem to allow more recovery for legitimate defamatory harm while maintaining tight proportionality in no-malice cases. Recall, however, that the jury in Sullivan did not differentiate, when imposing its colossal award, between actual and punitive damages. See supra note 199. This suggests that juries cannot always be trusted to accurately calculate actual damages. Perhaps the Court believed that they should not be trusted in high-constitutional-stakes cases involving public officials—like Sullivan’s clash over civil rights.

    As for why punitive damages should ever pass the tailoring test, there are two potential reasons. First, they might compensate plaintiffs for certain real harms that are hard to calculate on a line-item basis. Second, they might offset some underenforcement of defamation laws, and thus be well-tailored in expectation.

  • 207Blocher, supra note 112, at 430–31.
  • 208Imagine, for example, that a state imposed $100 million in statutory damages for all reckless misstatements, even if trivial. Surely this would merit a case-specific tailoring inquiry, Sullivan notwithstanding.
  • 209Their application in a third area is discussed in Part IV.D, infra.
  • 210See, e.g., R.A.V., 505 U.S. at 392 (examining whether racist speech may be treated as inherently costly).
  • 211See, e.g., id. at 392 (“The point of the First Amendment is that majority preferences [against certain ideas] must be expressed in some fashion other than silencing speech on the basis of its content.”). This comports with Greene’s observation that constitutional rights should operate categorically—as opposed to proportionally—when “courts must defend the very existence of individual rights against government bigotry, intolerance, or corruption.” Greene, supra note 108, at 128.
  • 212See Blocher, supra note 102, at 345.
  • 213See generally, e.g., Heller, 554 U.S. 570; Playboy, 529 U.S. 803; Roe, 410 U.S. 113.
  • 214See, e.g., Johnson, 491 U.S. at 411–12.
  • 215Such a ban might also fail the tailoring test. But that would depend on how much First Amendment benefit was lost, given the availability of alternatives like internet broadcasting.
  • 216Playboy, 529 U.S. at 813.
  • 217See Fallon, supra note 102, at 60–62. Note that the Court also sometimes strikes down laws for being underinclusive—regulating costs from protected activity but ignoring similar ones from nonprotected activity. Id. One can understand such underinclusivity as demonstrating regulations’ illegitimate goals. Blocher, supra note 102, at 369–70. This test, however, is tricky to apply, and the court does so inconsistently. See Fallon, supra note 102, at 60–62. For example, regulations on one category of costly speech do not necessarily evince illegitimate motives simply because they fail to restrict other similarly costly speech. See Geoffrey R. Stone, Content Regulation and the First Amendment, 25 Wm. & Mary L. Rev. 189, 202–03 (1983).
  • 218O’Brien, 391 U.S. at 377.
  • 219Heller, 554 U.S. at 626–27, 627 n.26.
  • 220Playboy, 529 U.S. at 813.
  • 221See infra notes 417–18 (discussing this principle in the rules governing time, place, and manner restrictions on speech).
  • 222See infra notes 417–18.
  • 223See supra Part III.A–B.
  • 224See supra Part II.A.
  • 225See supra Part III.C.
  • 226See supra Part III.C.
  • 227See Joseph Blocher, Hunting and the Second Amendment, 91 Notre Dame L. Rev. 133, 134–35, 134 n.5 (2015).
  • 228See supra Part II.A.
  • 229See supra Part III. In the cases discussed there, the Court has enumerated many legitimately regulable social costs, which are thus not protected.
  • 230See infra Part IV.C.
  • 231For more on this, see infra Part IV.C.
  • 232Command-and-control rules requiring precautions can also be understood this way. Then, the relevant ban is on precautionless activity. And the proportionality question asks whether the decision to forego precautions imposes, on the margin, more social cost than constitutionally protected benefit.
  • 233But see generally Anthony J. Casey & Anthony Niblett, The Death of Rules and Standards, 92 Ind. L.J. 1401 (2017) (positing that, in the future, artificial intelligence might deliver legal directives in the form of individualized, second-by-second commands).
  • 234Koper et al., supra note 52, at 2; Christopher S. Koper, William D. Johnson, Jordan L. Nichols, Ambrozine Ayers & Natalie Mullins, Criminal Use of Assault Weapons and High-Capacity Semiautomatic Firearms, 95 J. Urb. Health 313, 318 (2018) (homicides); see also Steven Stack & Ira Wasserman, Gender and Suicide Risk: The Role of Wound Site, 39 Suicide & Life-Threatening Behav. 13, 18 tbl.4, 19 (2009) (indicating that 13% of suicides involved rifles, of which assault rifles are a small subset).
  • 235Koper et al., supra note 52, at 314.
  • 23618 U.S. Code § 922(g)(1).
  • 237Incarcerated Felon Population by Type of Crime Committed, ProCon (Jan. 28, 2015),
  • 238Philip J. Cook, Jens Ludwig & Anthony A. Braga, Criminal Records of Homicide Offenders, 294 J. Am. Med. Ass’n 598 (2005).
  • 239Everytown, EveryStat, EveryStat, It is possible that many suicides are committed by people who illegally possess firearms. But I was able to locate no research to this effect. Moreover, it seems dubious, given our laws’ primary focus on banning possession by convicted felons. See, e.g., 18 U.S.C. § 922(d), (g). It is at least intuitive to predict that the correlation between prior felonies and homicides is much stronger than between prior felonies and suicide.
  • 240Vittes et al., supra note 236, at 28.
  • 241Michael C. Dorf, Do 18 Year Olds Have a Constitutional Right to Guns?, Take Care (Feb. 27, 2018),
  • 242See Michael Ram, Technology of Safety Devices for Firearms, 12 Loy. Consumer L. Rev. 218, 221 (2000).
  • 243See supra Part II.A.
  • 244See supra Part II.A.
  • 245See infra Part V.A.2.
  • 246Sullivan, 376 U.S. at 256.
  • 247See generally, e.g., Cass R. Sunstein, Reid Hastie, John W. Payne, David A. Schkade & W. Kip Viscusi, Punitive Damages: How Juries Decide (2002). See also Michael J. Saks, Lisa A. Hollinger, Roselle L. Wissler, David Lee Evans & Allen J. Hart, Reducing Variability in Civil Jury Awards, 21 Law & Hum. Behav. 243, 244–45 (1997) (surveying the literature and finding that, while jury awards roughly track actual losses, “both systematic and random errors make the [jury] process less than satisfactory”).
  • 248See infra notes 313–19 and accompanying text.
  • 249See infra notes 313–19 and accompanying text.
  • 250See supra Part III.A–B.
  • 251Discussed in Part III.C, supra.
  • 252Heller, 554 U.S. at 636.
  • 253Id. at 626–27, 627 n.26.
  • 254Id. at 627 (quotation marks omitted).
  • 255McDonald v. City of Chicago, 561 U.S. 742, 767 (2010) (emphasis omitted) (quoting Heller, 554 U.S. at 599).
  • 256Heller, 554 U.S. at 626–27.
  • 257See, e.g., Gertz, 418 U.S. at 343 (acknowledging “the state interest in compensating injury to the reputation of [public and] private individuals”).
  • 258Id. at 339–40.
  • 259See Sullivan, 376 U.S. at 277–78. Recall that, insofar as tort payments raised a constitutional concern, it was because of their potential for extreme ill-tailoring. See supra Part III.B.3.
  • 260Alvarez, 567 U.S. at 721–22.
  • 261Id. at 719.
  • 262See Restatement (Second) of Torts § 525 (Am. Law Inst. 1977).
  • 263See supra Part I.
  • 264See Fredrik Vold, Will Bitcoin SV Repeat Pump & Dump of 2019?, Cryptonews (Jan. 14, 2020), (describing a fake-news-fueled pump-and-dump scam for cryptocurrency).
  • 265Alvarez, 567 U.S. at 718–19. Moreover, some fraudulent speech can be punished criminally. Id. at 720.
  • 266See supra Part IV.A.
  • 267For a discussion on how social costs can be measured—and taxes thus properly sized—see Part IV.C, infra.
  • 268See supra Part III.B.1, 3.
  • 269Forsyth, 505 U.S. at 133.
  • 270Sullivan, 376 U.S. at 294 (Black, J., concurring).
  • 271See infra Part IV.C.
  • 272See infra Part V.A.1.
  • 273See infra Part IV.C.
  • 274See supra Part III.B.3.
  • 275See infra Part V.A.2. This would mimic full insurance. See infra Part V.A.3. As with insurance, average prices could vary categorically. If, for example, it turned out that fake political stories systematically impose higher costs than fake stories about celebrities, then those two categories of fake news—not individual instances of each—would incur different tax rates.
  • 276See Fake News, PolitiFact,
  • 277Id.
  • 278Samantha Putterman, No, Andy Borowitz Didn’t Make Disparaging Statement About Clinton Voters, PolitiFact (Dec. 13, 2019),
  • 279Samantha Putterman, Viral Meme Makes Up New Constitutional Amendment, PoltiFact (Dec. 11, 2019),
  • 280Consider also that existing governmental entities—securities regulators, defamation juries, consumer protection bureaus, etc.—routinely regulate truth and falsity, often in much closer cases.
  • 281Similar to the one at issue in McDonald.
  • 282Heller, 554 U.S. at 574–75.
  • 283See supra Part III.A.
  • 284See generally, e.g., Thomas J. Kniesner & W. Kip Viscusi, The Value of a Statistical Life, Oxford Rsch. Encyclopedia Econ. & Fin. (2019),
  • 285Id. at “Estimating the VSL”.
  • 286Id. at “Representative Estimates”.
  • 287Id. at “How You Die”.
  • 288Id. at “How Markets May Be More Complex”.
  • 289Kniesner & Viscusi, supra note 284, at “Representative Estimates”.
  • 290See generally, e.g., Cook & Ludwig, supra note 45; Duggan, supra note 45.
  • 291Cook & Ludwig, supra note 45, at 390. This is an average marginal cost figure. The article also estimates variations in marginal cost according to geography. Id. at 389.
  • 292Id. at 390.
  • 293Id.
  • 294Id. at 390 n.15.
  • 295See generally, e.g., Mark Duggan, Guns and Suicide, in Evaluating Gun Policy: Effects on Crime and Violence 41, (Jens Ludwig & Philip J. Cook eds., 2003) (estimating the effect of gun ownership on suicide rates).
  • 296See generally, e.g., Phillip B. Levine & Robin McKnight, Firearms and Accidental Deaths: Evidence from the Aftermath of the Sandy Hook School Shooting, 358 Science 1324 (2017).
  • 297Cook & Ludwig, supra note 45, at 390.
  • 298Jessie Hellmann, Congress Reaches Deal to Fund Gun Violence Research for First Time in Decades, The Hill (Dec. 16, 2019),
  • 299See Restatement (Second) of Torts §§ 621, 623 (Am. L. Inst. 1977).
  • 300See Restatement (Second) of Torts § 549 (Am. L. Inst. 1977).
  • 301See supra Part I.
  • 302See, e.g., Soroush Vosoughi, Deb Roy & Sinan Aral, The Spread of True and False News Online, 359 Science 1146, 1150 (2018); Allcott & Gentzkow, supra note 2; Gunther et al., supra note 61, at 2–6.
  • 303See generally, e.g., Brett R. Gordon & Wesley R. Hartmann, Advertising Effects in Presidential Elections, 32 Mktg. Sci. 19 (2013); Jonathan Nagler & Jan Leighley, Presidential Campaign Expenditures: Evidence on Allocations and Effects, 73 Pub. Choice 319 (1992).
  • 304See Restatement (Third) of Restitution § 13 cmt. h (Am. L. Inst. 2011). See also generally Robert Cooter & Ariel Porat, Disgorgement Damages for Accidents, 44 J. Legal Stud. 249 (2015).
  • 305Assume that the costs in question pass the legitimacy threshold.
  • 306Exec. Order No. 12291, 3 C.F.R 127 (1982) (revoked 1993).
  • 307U.S. Env’t Prot. Agency, Economic and Benefits Analysis for the Final Section 316(b) Phase II Existing Facilities Rule, D1-3.2 (2004).
  • 308U.S. Env’t Prot. Agency, Economic Analysis for the National Emission Standards for Hazardous Air Pollutants for Source Category: Pulp and Paper Production; Effluent Limitations Guidelines, Pretreatment Standards, and New Source Performance Standards: Pulp, Paper, and Paperboard Category—Phase 1, 9-16 (1997).
  • 309See Jonathan Masur & Eric A. Posner, Unquantified Benefits and the Problem of Regulation Under Uncertainty, 102 Cornell L. Rev. 87, 100–01 (2016).
  • 310Id.
  • 311Id. at 104.
  • 312Id.
  • 313See, e.g., Sullivan, 376 U.S. at 277–78; Alvarez, 567 U.S. at 719; Blocher & Miller, supra note 10, at 310–13, 334–35.
  • 314See, e.g., Gonzales v. Carhart, 550 U.S. 124, 162–64 (2007) (avoiding engagement with competing empirical claims); City of Renton, 475 U.S. at 51 (“The First Amendment does not require a city . . . to conduct new studies or produce [new, city-specific] evidence . . . so long as whatever evidence the city relies upon is reasonably believed to be relevant.”).
  • 315See supra Part III.A–B.
  • 316Heller, 554 U.S. at 626–27. Empirical research suggests, for example, that previously convicted felons over fifty pose little threat. See The Effects of Aging on Recidivism Among Federal Offenders, U.S. Sent’g Comm’n, 22–27 (2017), Moreover, many felonies do not involve any violence in the first place. See, e.g., 18 U.S.C. § 1341 (mail fraud).
  • 317Bad-faith empirical analyses are discussed in Part V.A.1, infra.
  • 318Such disagreements arise everywhere, including in our most well-established cost estimates. See Kniesner & Viscusi, supra note 284, at “Representative Estimates” (describing variation among estimates of the value of a statistical life).
  • 319See infra Part V.A.1 for more on judicial review of cost estimates.
  • 320Masur & Posner, supra note 309, at 119–21.
  • 321As described above, Pigouvian regulators need not estimate private benefits to regulate efficiently. See supra Part II. And it is a weak argument against regulating that a Pigouvian tax is too small, having ignored some of an activity’s social costs. Such a tax improves on the margin over the status quo, and lawmakers can raise rates as they measure additional costs.

    So far, this Article has talked in terms of regulating “social costs,” not “negative externalities.” This is because some of the costs targeted by our hypothetical Pigouvian taxes are rightly considered internalities, not externalities. Suicides fall into this category. Pigouvian taxes can be efficient if they force an agent to account for such costs—accruing costs to the agent which she would otherwise ignore. See generally Hunt Allcott & Cass R. Sunstein, Regulating Internalities, 34 J. Pol. Analysis & Mgmt. 698 (2015).

    This is mostly a pragmatic point, not a constitutional one. The Supreme Court approves of paternalistic regulations on protected activity. See, e.g., Planned Parenthood of Se. Pa. v. Casey, 505 U.S. 833, 881 (1992). One might worry that imposing such regulations, including Pigouvian taxes, creates a kind of double penalty. The regulated party first pays a cost in expectation and then pays again if a bad outcome befalls her. Then, one would argue, constitutional burdens begin to outweigh targeted social costs. If this were a constitutional problem, it could again be avoided with Pigouvian regulation. Revenues from a Pigouvian tax can be redistributed to taxed entities—either in the amount of actual harm or on an average-cost basis. See Brian Murray & Nicholas Rivers, British Columbia’s Revenue-Neutral Carbon Tax: A Review of the Latest “Grand Experiment” in Environmental Policy, 86 Energy Pol’y 674, 677 (2015). Such a design preserves the full Pigouvian incentive.

  • 322See generally, e.g., Jeffrey L. Harrison, A Positive Externalities Approach to Copyright Law: Theory and Application, 13 J. Intell. Prop. L. 1 (2005).
  • 323See generally, e.g., John Lott, Jr., More Guns, Less Crime (1998) (arguing that gun ownership creates a net reduction in crime).
  • 324A negative Pigouvian tax is a Pigouvian subsidy. See Robin Boadway, Challenges for Second-Best Analysis, in Optimal Tax Theory to Tax Policy: Retrospective and Prospective Views 185, 206 (2012). For more on the choice between Pigouvian taxes and subsidies, see generally Brian Galle, The Tragedy of the Carrots: Economics and Politics in the Choice of Price Instruments, 64 Stan. L. Rev. 797 (2012).
  • 325See, e.g., Cook & Ludwig, supra note 45, at 387, 390.
  • 326See, e.g., Duggan, supra note 45, at 1104–05; Cook & Ludwig, supra note 45, at 390; Ian Ayres & John J. Donohue III, Shooting Down the “More Guns, Less Crime” Hypothesis, 55 Stan. L. Rev. 1193, 1285 (2003).
  • 327Masur & Posner, supra note 309, at 119–21.
  • 328Id. at 120.
  • 329Id. at 122.
  • 330Jonathan Masur & Eric A. Posner, Cost-Benefit Analysis and the Judicial Role, 85 U. Chi. L. Rev. 935, 945 (2018). Consider a Pigouvian subsidy. Indeed, in some cases, it might be feasible to design both a tax keyed to costs and a subsidy keyed to positive externalities. This is more complicated than doing just one or the other, but it creates incentives for people to improve the ratio of costs to benefits, producing less of the former and more of the latter.
  • 331See supra Part III.
  • 332United States v. Rumely, 345 U.S. 41, 56 (1953) (Douglas, J., concurring); see also generally Mill, supra note 201.
  • 333U.S. Const. art. I, § 8, cl. 8; art. I, § 4; amend. 12.
  • 334Roe, 410 U.S. at 155, 165.
  • 335Id. at 163.
  • 336Id. at 162.
  • 337Id. at 162–63.
  • 338Id. at 163–64.
  • 339Roe, 410 U.S. at 164–65.
  • 340See supra Part IV.C.
  • 341See Roe, 410 U.S. at 117 n.1.
  • 342See generally Whole Woman’s Health v. Hellerstedt, 136 S. Ct. 2292 (2016). Because the laws at issue there were less burdensome than those in Roe, Whole Woman’s Health was subject to the moderate tailoring-legitimacy standard represented in Planned Parenthood of Southeastern Pennsylvania v. Casey, 505 U.S. 833 (1992). In June Medical Services L.L.C. v. Russo, 140 S. Ct. 2013 (2020), Chief Justice John Roberts’s concurrence disparaged proportionality analysis in the constitutional law of abortion. Id. at 2134–35 (Roberts, C.J., concurring). But he advocated traditional intermediate scrutiny as the correct standard, which, as shown above, includes a proportionality inquiry. See supra Part III.B.2.
  • 343June Medical, 149 S. Ct. at 2112 (quoting Whole Woman’s Health, 136 S. Ct. at 2313).
  • 344Targeted Regulation of Abortion Providers. See Marshall H. Medoff, State Abortion Politics and TRAP Abortion Laws, 33 J. Women, Pol. & Pol’y 239, 240–41 (2012).
  • 345Id.
  • 346See Fallon, supra note 102, at 142–45. Of course, not every regulation of abortion providers is a TRAP law. Consider, for example, a handwashing requirement for surgical procedures.
  • 347Rachel K. Jones, Elizabeth Witwer & Jenna Jerman, Abortion Incidence and Service Availability in the United States, 2017, Guttmacher Inst. (2017), Some of the regulations restricting supply may be unconstitutional for the same reason as were those in Whole Woman’s Health. But many may be valid, imposing serious distributional effects.
  • 348See supra Part II.
  • 349Roe, 410 U.S. at 162.
  • 350505 U.S. 833, 876 (1992). The so-called undue burden test applies moderate legitimacy and tailoring requirements to laws imposing lighter constitutional burdens than the criminal ban at issue in Roe.
  • 351550 U.S. 124 (2007).
  • 352Id. at 160, 157.
  • 353Carhart, 550 U.S. at 166–67; Casey, 112 U.S. at 882–84.
  • 354See, e.g., Matal v. Tam, 137 S. Ct. 1744, 1764 (2017) (plurality opinion) (noting that the First Amendment “protect[s] the freedom to express ‘the thought that we hate’” (quoting United States v. Schwimmer, 279 U.S. 644, 655 (1929) (Holmes, J., dissenting))); id. at 1769 (Kennedy, J., concurring) (“A law that can be directed against speech found offensive to some portion of the public can be turned against minority and dissenting views to the detriment of all.”).
  • 355When mere speech evolves into a legitimate threat of bodily harm, the story changes. See Virginia v. Black, 538 U.S. 343, 359 (2003) (“[T]he First Amendment also permits a State to ban a ‘true threat.’” (quoting Watt v. United States, 394 U.S. 705, 708 (1969))).
  • 356See generally Heller, 554 U.S. 570; McDonald, 561 U.S. 742.
  • 357McCulloch v. Maryland, 17 U.S. 316, 431 (1819).
  • 358See Jonathan S. Masur & Eric A. Posner, Chevronizing Around Cost-Benefit Analysis, 70 Duke L.J. 1109, 1131–32 (2021).
  • 359Id. at 1114–36.
  • 360Id. at 1136.
  • 361Id.
  • 362Masur & Posner, supra note 330 at 939–40.
  • 363Cf. Minneapolis Star & Trib. Co. v. Minn. Comm’r of Revenue, 460 U.S. 575, 586–88 (1983) (striking down a tax because regulators refused to even specify, much less attempt to quantify, targeted costs).
  • 364Masur & Posner, supra note 358358, at 1136.
  • 365Id.
  • 366See supra Part III.
  • 367See supra Part III. Note how little of the analysis in these cases involves actual quantitative estimates. On occasion, when litigants have presented opposing empirical analyses of a law’s effects, the Court has taken pains to avoid resolving the issue. See, e.g., Carhart, 550 U.S. at 164.
  • 368Heller, 554 U.S. at 626–27, 627 n.26.
  • 369See The Effects of Aging on Recidivism Among Federal Offenders, supra note 316, at 22–27; see also, e.g., 18 U.S.C. § 1341 (mail fraud).
  • 370Darren Wheelock, Collateral Consequences and Racial Inequity: Felon Status Restrictions as a System of Disadvantage, 21 J. Contemp. Crim. Just. 82, 83 (2005).
  • 371This point also has a constitutional dimension since, as discussed above, proportionality is best understood as requiring at least modest efforts to reduce errors in allocating constitutional burden. See supra Part III.C.
  • 372See supra notes 83–85.
  • 373See Louis Kaplow & Steven Shavell, On the Superiority of Corrective Taxes to Quantity Regulation, 4 Am. L. & Econ. Rev. 1, 4–5 (2002) (discussing Pigouvian rate updates).
  • 374Some may object that § 230 of the Communications Act of 1934, passed into law as part of the Communications Decency Act of 1996, would forbid taxing social media companies. That problem is largely outside the scope of this Article, which is concerned with constitutional strictures. Moreover, § 230 is federal legislation, and it could easily be modified by new federal legislation imposing a fake-news tax. In 2018, Congress modified the law to allow liability for websites facilitating sex trafficking. See Pub. L. No. 115-164, § 3(a), 132 Stat. 1253 (codified at 18 U.S.C. § 2421A). And there is bipartisan appetite for regulating fake news on social media. See Li Zhou, Nancy Scola & Ashley Gold, Senators to Facebook, Google, Twitter: Wake Up to Russian Threat, Politico (Nov. 1, 2017), And if states were to attempt the regulation, it is not totally clear that § 230, which forbids certain “cause[s] of action . . . be[ing] brought,” would apply to a tax scheme. 47 U.S.C. § 230(e)(3).
  • 375Most of the points discussed above regarding 47 U.S.C. § 230 apply also to 15 U.S.C. § 7903 (governing civil liability for gun manufacturers). See supra note 374.
  • 376Cook & Ludwig, supra note 45, at 389–90.
  • 377See Cassandra K. Crifasi, Jayne K. O’Dwyer, Emma E. McGinty, Daniel W. Webster & Colleen L. Barry, Desirability of Personalized Guns Among Current Gun Owners, 57 Am. J. Preventive Med. 191, 193 (2019) (finding that while 79% of current gun owners support such weapons, few are willing to pay a premium for them). Such research suggests that a Pigouvian tax, which would make cost-justified precautions cheaper than the alternative, could have a significant effect on adoption.
  • 378Insofar as readers are wondering whether intra-tranche variation would raise constitutional concerns, the answer is almost certainly no. The Supreme Court endorses command-and-control rules governing categories of people for whom harm generally exceeds benefits, even though, for some individuals, the reverse is surely true. See supra Part III.C.
  • 379This would be the case when someone buys a gun with the specific intent of shooting someone else. Such cases exist, but they may be much less common than cases where people buy a gun with legitimate intentions and only later decide to use it for a crime.
  • 380See, e.g., Fleischer, supra note 96, at 1673; Brunson supra note 96, at 607 (arguing that these problems make it unlikely that a Pigouvian tax on guns would have more than a “margin[al]” or “incidental” effect on gun violence).
  • 381Masur & Posner, supra note 11, at 138.
  • 382See supra Part II.A.
  • 383See supra Part IV.C.
  • 384See supra Part I.
  • 385See supra Part I.
  • 386See Indre Deksnyte, How AI Can Create and Detect Fake News, Forbes (Sept. 12, 2019),
  • 387See Casey Newton, Facebook Will Label False Posts More Clearly as Part of an Effort to Prevent 2020 Election Interference, The Verge (Oct. 21, 2019), Facebook may believe that labeling is better than removal because users are likely to encounter fake stories elsewhere, too. But this equilibrium results from a collective action problem, wherein Facebook is regulating falsities and others are not. Regulation overcomes that problem. Moreover, recent research suggests that offering corrective information to people who have been affected by fake news makes them even less likely to believe the truth. See generally John M. Carey, Victoria Chi, D.J. Flynn, Brendan Nyhan & Thomas Zeitzoff, The Effects of Corrective Information About Disease Epidemics and Outbreaks, 6 Sci. Advances 7449 (2020).
  • 388See D. Kirk Davidson, Selling Sin: The Marketing of Socially Unacceptable Products 57 (2003).
  • 389See, e.g., Brunson, supra note 96, at 635.
  • 390See Fleischer, supra note 96, at 1677–78, n.17; Brunson, supra note 96, at 607.
  • 391See Fleischer, supra note 96, at 1677–78.
  • 392See Dylan Matthews, There are More Gun Suicides than Gun Homicides in America, Vox (Nov. 14, 2018),
  • 393See, e.g., Amy Sherman, 3 Things to Know About the ‘Gun Show Loophole’, PolitiFact (Jan. 7, 2016), (describing deficiencies in licensing regimes).
  • 394One can certainly imagine considerations to the contrary. For example, if taxed guns also had to be registered by serial number, perpetrators of gun crime might be easier to identify. This cost, however, applies only to people who expect at tax time to need their gun for a crime. Even then, it operates only as a marginal increase in total expected costs. It will deter tax compliance only if that marginal increase exceeds the expected enforcement costs of tax noncompliance.
  • 395See, e.g., Jonathan R. Macey & Geoffrey P. Miller, The Plaintiffs’ Attorney’s Role in Class Action and Derivative Litigation: Economic Analysis and Recommendations for Reform, 58 U. Chi. L. Rev. 1, 8–11 (1991); In re Rhone-Poulenc Rorer, Inc., 51 F.3d 1293, 1299 (7th Cir. 1995) (“In most class actions—and those the ones in which the rationale for the procedure is most compelling—individual suits are infeasible because the claim of each class member is tiny relative to the expense of litigation.”).
  • 396See Amchem Prods., Inc. v. Windsor, 521 U.S. 591, 609–10 (1997). But see generally Peter N. Salib, Intelligent Class Actions, 100 Tex. L. Rev. (forthcoming) (on file with author) (arguing that the use of artificial intelligence in class litigation could overcome this hurdle).
  • 397But see Salib, supra note 396 (manuscript at 40) (explaining how artificial intelligence could be used to cheaply calculate individual damages).
  • 398See supra Part III.B.3.
  • 399See supra Part III.B.3.
  • 400See supra Part III.B.3.
  • 401See, e.g., Galle & Mungan, supra note 96, at 380 (proposing such a scheme for guns).
  • 402Kaplow & Shavell, supra note 374, at 13.
  • 403See Abba P. Lerner, The Economics of Control 26–27 (1944).
  • 404See City of Renton v. Playtime Theaters, Inc., 475 U.S. 41, 54–55 (1986).
  • 405Consider that money can buy time, as when the wealthy outsource time-consuming activities like childcare and housekeeping.
  • 406See supra Part IV.D.
  • 407See supra note 347.
  • 408See supra Part III.C.
  • 409See supra Part III.C.
  • 410As outlined above, the legitimacy requirement is fundamentally qualitative. Preventing a murder is more legitimate than preventing millions and millions of paper cuts, even if it turns out that the aggregate amount society would pay for both is equal. See supra Part III.C.
  • 411See supra notes 210–11.
  • 412See supra Part IV.C.
  • 413Here, unlike in Whole Woman’s Health v. Hellerstedt, 136 S. Ct. 2292 (2016), constitutional strictures may be failing to bar legislators from doing by subterfuge what they could not accomplish in the open. See supra Part IV.D.
  • 414475 U.S. 41 (1986).
  • 415Id. at 54.
  • 416See generally Blocher, supra note 102.
  • 417City of Renton, 475 U.S. at 47. Consider also, from another constitutional context, Griffin v. Illinois, 351 U.S. 12, 19–20 (1956), which outlawed certain court fees being imposed on indigent parties.
  • 418In City of Renton, for example, the total regulatory landscape left “some 520 acres, or more than five percent of the entire land area of Renton, open to use as adult theater sites.” 475 U.S. at 53. Total land available is a consequence not only of a zoning regulation under review, but also of a city’s other land use laws.
  • 419Id. at 54 (quoting Young v. Am. Movie Theaters, Inc., 427 U.S. 50, 78 (Powell, J., concurring)).
  • 420U.S. Const. amend. VI.
  • 421See Alabama v. Shelton, 535 U.S. 654, 661 (2002).
  • 422See, e.g., Strickland v. Washington, 466 U.S. 668, 699 (1984).
  • 423See generally Griffin, 351 U.S. 12.
  • 424Such reforms would cut both ways, politically speaking. The idea of abortion vouchers for low-income women would be anathema to many on the right, just as gun subsidies would be to many on the left.

    Note that cash subsidies would preserve Pigouvian effects in a way that in-kind benefits would not. Cash can be used for any purpose, not just engaging in a single constitutionally protected activity. Recipients of cash thus face the same incentives as everyone else in a Pigouvian system to refrain from activity when costs exceed benefits.

  • 425See generally Cass R. Sunstein, On the Expressive Function of Law, 144 U. Pa. L. Rev. 2021 (1996); Wibren van der Burg, The Expressive and Communicative Functions of Law, Especially with Regard to Moral Issues, 20 Law & Phil. 31 (2001); Richard H. McAdams, The Expressive Powers of Law (2017).
  • 426Sunstein, supra note 425, at 2045.
  • 427See McAdams, supra note 425, at 136–99.
  • 428Sunstein, supra note 425, at 2031 (quotation marks omitted).
  • 429See generally Friedrich A. Hayek, The Use of Knowledge in Society, 35 Am. Econ. Rev. 519 (1945).
  • 430See Chetty et al., supra note 85, at 1170–71.
  • 431U.S. Const. art. VI.