Discriminatory Taxes, Secondary Effects, and Tiers of Scrutiny — part 4 in a series

Previously, I blogged the abstract, introduction, Part I, and Part II of my new article, Taxing Nudity: Discriminatory Taxes, Secondary Effects, and Tiers of Scrutiny, which has just been published in the Journal of Free Speech Law. It’s based on my work with the Georgia Association of Club Executives v. Riley case, where we challenged a Georgia tax on adult entertainment establishments on First Amendment/free speech grounds.

In this final post, I’ll give you Part III, “Taxation and the Necessity Inquiries”, which explains why the tax should fail under strict scrutiny and is vulnerable even under intermediate scrutiny. And I’ll end with the Conclusion. (The article obviously has a lot of footnotes — go to the article itself if you want to see those.)

*     *     *

III. Taxation and the Necessity Inquiries

A. Taxes Fail Strict Scrutiny

Once we determine that a tax is content-based and requires strict scrutiny, it must necessarily fail. Strict scrutiny requires that the government choose the least restrictive (i.e., least content-discriminatory) means of pursuing its goal. But the government can always pursue its goal—providing services to victims of sex crimes, or similar—by providing the same amount of money from general revenues. As the Supreme Court wrote in Minneapolis Star & Tribune Co. v. Minnesota Commissioner of Revenue:

The main interest asserted by Minnesota in this case is the raising of revenue. Of course that interest is critical to any government. Standing alone, however, it cannot justify the special treatment of the press, for an alternative means of achieving the same interest without raising concerns under the First Amendment is clearly available: the State could raise the revenue by taxing businesses generally, avoiding the censorial threat implicit in a tax that singles out the press.

“The same is true of a tax that differentiates between members of the press,” and the same is true of a tax that differentiates between erotic and non-erotic messages, or different types of content more generally.

Now perhaps I’m being too absolutist here—perhaps we shouldn’t take too literally the idea that we need to find the absolutely least discriminatory alternative; perhaps we should care at least a little bit whether those less discriminatory alternatives serve our purposes somewhat less well. Maybe we need to do somewhat more balancing—like intermediate scrutiny, but drawing the line in a more speech-protective way, for instance by requiring the government to tolerate a greater harm to its asserted interest when strict scrutiny applies than when intermediate scrutiny applies.

That’s not how the strict scrutiny cases state the test, so I’m inclined to say my absolutism is warranted here. Still, if one wants to be less absolutist, we’ll want to look closely at a variety of asserted government interests. But I’ll save that analysis for the next subsection, in which I argue that the tax is vulnerable even under intermediate scrutiny (which implies it would be even more vulnerable under strict scrutiny).

B. Taxes Are Suspect Under Intermediate Scrutiny

Suppose we instead evaluate such a tax under intermediate scrutiny. This might be because we’re considering a nudity tax instead of an erotic-expression tax; or maybe, even as to an erotic-expression tax, you’re not convinced by my strict-scrutiny argument presented above. Under intermediate scrutiny, a tax wouldn’t necessarily fail, but it would still be vulnerable.

One of the requirements of intermediate scrutiny is that the government’s interest must be “unrelated to the suppression of free speech.”

Another requirement is that the tax has to be “narrowly tailored to serve a significant governmental interest.” Both strict and intermediate scrutiny use similar “narrow tailoring” language when discussing the requisite means-ends relationship, but this is confusing; in the context of intermediate scrutiny, “narrow tailoring” doesn’t require the least-restrictive alternative characteristic of strict scrutiny. Still, under intermediate scrutiny, there must be a “reasonable fit” between the government interest and the chosen means. Any burden on speech must “promote[] a substantial government interest that would be achieved less effectively absent the [burden]”; the governmental action can’t “burden substantially more speech than is necessary to further the government’s legitimate interests.”

In this section, I examine a number of possible governmental interests that might be thought to support these taxes. I conclude that some possible government interests themselves are invalid—for instance, because they’re speech-suppressive or discriminatory. Other interests are legitimate, but a discriminatory tax doesn’t serve them any better than would a more neutral tax—e.g., any amount of revenue raised by a targeted tax could equally well be provided out of general revenues.

I identify one interest—the interest in combating secondary effects—that is both legitimate and could conceivably be pursued more effectively with a targeted tax. But the government must explicitly assert this interest and support it with empirical evidence that is squarely on point. Moreover, because any increase in effectiveness derives directly from the tax’s chilling effect on expression, courts should be very skeptical of an argument based on this interest.

Therefore, these taxes are likely unconstitutional under intermediate scrutiny.

1. Reducing the activity as governmental interest

To figure out whether the tax satisfies intermediate scrutiny, we have to start by carefully identifying the government’s interest, determining whether that interest is valid, and if it is, looking at whether it has the proper means-ends relationship (i.e., looking at the narrowness of the tailoring).

When taxes are involved, the governmental interest that most obviously springs to mind is the interest in reducing the activity: policymakers often suggest taxing undesirable activity and subsidizing desirable activity. Ideally, one can design an optimal system of deterrent taxes, where the amount of the tax is precisely equal to the amount of the social harm from the activity; this would optimally deter the harmful activity by making bad actors internalize its social cost. And even in a non-ideal world, one can just levy taxes of some amount in order to reduce the amount of the activity by some amount; the precise amount might not be optimal, and there might not be any serious attempt to quantify the amount of social harm (perhaps because, in morals contexts, harms might be subtle and diffuse), but it might still be better than nothing. This is the sort of thinking that goes by the name of “polluter pays” in environmental contexts, “sin taxes” in morals-regulation contexts, and “Pigouvian taxes” in public-finance contexts.

But—common and praiseworthy though this thinking may be elsewhere—this can’t be the governmental interest when burdens on expressive activity are concerned. To say that the government has an interest in reducing erotic dancing is just to say that the government is relying on the tax’s “chilling effect” on expressive conduct. But if the First Amendment means anything, it must be that a chilling effect on constitutionally protected expression can’t itself be the governmental interest; i.e., it can’t be a positive in a First Amendment case.

Perhaps a tax or regulation might pass First Amendment scrutiny on the ground that, given an important enough interest, the chilling effect is justified by, or outweighed by, some other benefits. But the chilling effect can’t itself be the benefit. Such a “sin tax” rationale would run afoul of intermediate scrutiny’s separate requirement that the state interest be unrelated to the suppression of speech.

One could argue that the interest is not in reducing erotic dancing as such, but in reducing the combination of erotic dancing with alcohol. But, as I’ve already shown above, the doctrine doesn’t support any different First Amendment ana­lysis when alcohol is involved. If your interest is to reduce erotic dancing with alcohol but not to reduce other kinds of dancing with alcohol, then your problem is with erotic dancing—which means that you’re still claiming the chilling effect on speech as a positive.

If your view is that erotic dancing plus alcohol uniquely contributes to some social problem, then combating that social problem (i.e., a secondary effect) might be a good governmental interest—and conceivably, the chilling effect might be justified as an acceptable way to pursue that interest. But if that’s the case, you’re not using the chilling effect—reducing speech, either by itself or with alcohol—as the governmental interest itself, and so this possibility has to investigated separately. (I discuss that possible governmental interest below.)

2. Revenue-raising as governmental interest

Once we eliminate reducing the activity as a legitimate governmental interest, another possible interest could be raising revenue. But this doesn’t work, for the same reason it didn’t work in the strict scrutiny analysis above.

Raising revenue is obviously a legitimate interest. But under intermediate scrutiny’s narrow tailoring standard, a tax that burdens expression fails, because the State could achieve that goal equally well, with less burden on expressive conduct, by merely applying a more broad-based tax to raise the same amount of revenue—e.g., the state’s ordinary income taxes, sales taxes, property taxes, or the like. Money is money, so any way of raising a particular amount serves the revenue-raising goal as well as any other; a targeted tax is thus unnecessary to the government’s legitimate goal of raising revenue.

Am I importing strict scrutiny’s least-restrictive alternative analysis into intermediate scrutiny? Not at all. Under strict scrutiny, when a less-restrictive alternative is available, the government has to choose it, even when that’s harmful to its (compelling) interest. Intermediate scrutiny doesn’t require such sacrifices to the governmental interest. As I’ve quoted above, there just needs to be a “reasonable fit”; the burden must “promote[] a substantial government interest that would be achieved less effectively absent the [burden]”; and so on. But what this means is that if, hypothetically, one could identify a less-restrictive alternative that promoted the government’s interest exactly as well as the challenged burden, at the same cost and with the same effectiveness, then the burden must fail intermediate scrutiny. The Supreme Court stated as much in explaining the concept of “reasonable fit” in City of Cincinnati v. Discovery Network, Inc.:

We reject the city’s argument that . . . our consideration of alternative, less drastic measures by which the city could effectuate its interests . . . violates [the doctrine] that regulations on commercial speech are not subject to “least-restrictive-means” analysis. . . . A regulation need not be absolutely the least severe that will achieve the desired end, . . . but if there are numerous and obvious less-burdensome alternatives to the restriction on commercial speech, that is certainly a relevant consideration in determining whether the “fit” between ends and means is reasonable.

The availability of a broad-based tax that does just as well in pursuing the revenue-raising goal thus means that this interest cannot support intermediate scrutiny any more than it can support strict scrutiny.

One could dispute the view that a broad-based tax is less burdensome (and thus unproblematic from a First Amendment perspective) than a targeted tax, because a broad-based tax hits everyone while a targeted tax hits only a few. But if a broad-based tax were more burdensome for First Amendment purposes, we could challenge the income tax itself (or any general tax), which does, after all, fall on some expressive activities along with everything else. Could we use the First Amendment to force the income tax and other general taxes to exempt all expressive activities? But that seems intuitively wrong. Moreover, it is doctrinally incorrect. “[E]very civil and criminal remedy imposes some conceivable burden on First Amendment protected activities,” but First Amendment scrutiny doesn’t come into play unless a policy burdens “conduct with a significant expressive element.” Burning a flag or a draft card does have such an element, but earning income or selling goods doesn’t. A general income or sales tax burdens general activities with no significant expressive element, and thus raises no First Amendment concerns.

Could we do somewhat better than the mere revenue-raising interest? Perhaps the government interest might be raising a given amount of revenue while generating the least economic distortion. Public-finance economists have a lot to say about how to set differential tax rates on various goods or activities so as to raise money while minimizing economic distortions. The distortionary effect of a tax arises from people’s tendency to reduce the amount of the taxed activity or consumption of the taxed good. The ideal tax, from this perspective, is one that raises money without changing anyone’s behavior. This is of course an unattainable ideal, because all money is used for something. But the least distortionary tax tends to be one that is assessed on goods or activities that aren’t very responsive to price, i.e., goods or activities with “inelastic demand.” If a government followed that sort of economic advice, it’s conceivable that it might choose nude dancing as one of these price-inelastic activities.

But this motivation, while theoretically conceivable, is unlikely to be implicated here. In the first place, as a general public finance matter, broad-based taxes (like income taxes) are likely to be less economically distorting than targeted taxes on particular activities, because the tax base is much broader and so the same amount of revenue can be raised with a much smaller tax increase.

And in the second place—as a factual matter in this context—the taxes discussed here are levied uniquely on adult entertainment, with no apparent concern for minimizing economic distortions. On the contrary, the statutory context suggests that these taxes are overwhelmingly motivated by the connection between the activity and some social ill—i.e., the secondary effect. If we take the statutory context at face value, this suggests that the government’s true interest is in fighting the secondary effect (on which more below), not in raising money in efficient ways.

3. Fighting secondary effects as governmental interest

What about combating the secondary effect? Fighting child sex trafficking is obviously a legitimate interest (even a compelling one), and so are many other secondary effects that might be adduced in the adult-entertainment context. But the government still has to establish that the tax will reduce the secondary effect in the way required by intermediate scrutiny.

There are two main pathways for the tax to reduce the secondary effect. First, the tax money could be spent on activities that will reduce the amount of the secondary effect, like anti-trafficking enforcement; we could call this the “spending effect.” Second, even if there is no such spending, the tax can by itself reduce the amount of the secondary effect—by raising the price of the activity to consumers or by raising its cost to producers. We could call this a “deterrence effect,” but we should be more precise here. A regulation—say, a zoning ordinance—seeks to reduce the secondary effect, but doesn’t necessarily affect the total amount of the expressive activity. By contrast, if we ignore the spending effect, the only way a tax can reduce the secondary effect is by reducing the amount of speech. So we should more properly call this the “speech-deterrence effect.”

a. The spending effect

First, let’s just focus on the spending effect. That is, let’s assume that the government isn’t relying on any reduction in the expression at all; or, let’s assume that the tax is so small (or is structured in such a way) that it has no effect on the level of expression. Thus, suppose the only way that the tax alleviates the secondary effect is that the revenues will be spent on various relevant activities—like providing services to trafficked women or children, or funding sex-crime prosecutions.

Though alleviating the secondary effects is an adequate interest, merely raising money isn’t enough here; it has to be actually spent. Putting the money into the general fund with an expectation that the legislature will fund activities (like in Tennessee) isn’t good enough, because the government won’t be able to show the necessary linkage between the tax and the spending. At a minimum, the money raised has to somehow be directly tied to combating the secondary effect; one possibility would be to use a dedicated fund.

Maybe not 100% of the money raised needs to go into that fund—in Texas, some of the money raised is used to subsidize health-insurance premiums for poor people, and maybe this isn’t fatal. But at least there needs to be some direct relation between the money raised and the money spent on reducing the purported secondary effect.

If the government uses a fund, then the tax adds money to the revenue pool, and at least part of that money will be spent on activities that reduce the secondary effect. That seems promising—but in an intermediate scrutiny context, even that isn’t good enough. A tax that burdens speech not only must promote a governmental interest, but must promote it more effectively than would be possible without the burden on speech. As noted above, the governmental action can’t “burden substantially more speech than is necessary to further the government’s legitimate interests.”

And just as in the strict scrutiny context—and just as we saw above with the revenue-raising interest—the state could combat the secondary effect equally well, with less burden on expressive conduct, by merely applying a broad-based tax to raise the same amount. An obvious way of doing so would be to fund the relevant activities directly out of general revenues (whether or not this is done through specific “funds”) rather than from a targeted tax on adult entertainment establishments.

This is what makes targeted taxes fundamentally different from targeted regulations: a targeted regulation, like a zoning ordinance—if properly targeted, which can be established with good empirical evidence—can have a direct effect on whatever secondary effect is being addressed, because the regulation itself forces changes in behavior. A non-targeted regulation would be no more effective, or even less effective (surely all businesses can’t be zoned into an adult area of town). But a targeted tax doesn’t do any better than a non-targeted tax, unless we can also consider the reduction in the expression (i.e., the speech-deterrence effect).

So, as a governmental interest, the revenue-raising interest works no better here than it did for strict scrutiny, even if we tie it to the interest in fighting the secondary effect. The only way fighting the secondary effect can work as a governmental interest is if there is also a deterrence effect—which is the subject of the next subsection.

b. The speech-deterrence effect

i. May one rely on the speech-deterrence effect?

As to the speech-deterrence effect, we’ve already seen above that reducing the amount of speech can’t itself be the governmental interest. But as long as something else (like reducing the secondary effect) is the government interest, it might be permissible to take the reduction in speech into account in determining whether that interest is sufficiently furthered.

Why do I say “might be”? Because it depends on how one interprets Justice Kennedy’s concurrence in the judgment in Alameda Books. Justice Kennedy wrote:

[A] city may not regulate the secondary effects of speech by suppressing the speech itself. . . . Though the inference may be inexorable that a city could reduce secondary effects by reducing speech, this is not a permissible strategy. The purpose and effect of a zoning ordinance must be to reduce secondary effects and not to reduce speech.”

At first glance, one might think that of course, regulation of adult entertainment must reduce its amount, because doesn’t regulating an activity always burden that activity, and thus reduce its amount? Not so: a zoning regulation that places adult businesses together might actually promote competition among adult businesses and thus reduce their prices; or, if it makes law enforcement easier by putting adult businesses in one place, that might improve safety and attract more customers that way. The opposite might also be true: if having too many adult businesses in one area promotes crime, then a zoning ordinance that spaces out adult businesses might improve safety and thus attract customers. All that’s necessary here is to observe that (1) this is an intensely empirical question, and (2) regulation of adult entertainment needn’t entail speech suppression.

According to Justice Kennedy’s approach, then, the government needs to reduce secondary effect without reducing speech. As I’ve mentioned above, courts have disagreed on whether Justice Kennedy’s opinion is controlling. But if we do treat it as controlling, that would suggest that we should treat any speech-deterrence effect as a negative, or at least ignore the speech-deterrence effect in deciding on the constitutionality of these targeted taxes. Indeed, as Justice Kennedy wrote immediately afterward: “A city may not, for example, impose a content-based fee or tax. This is true even if the government purports to justify the fee by reference to secondary effects.” And if that’s the case, then the targeted tax would fare no better under intermediate scrutiny than under strict scrutiny.

This would apply more broadly than the tax context: the insistence on not reducing speech would mean that the government could never use Renton analysis to ban a certain type of entertainment outright. Presumably we wouldn’t be able to pretend that the secondary effects convert a facially content-discriminatory ban into a content-neutral one, and we’d have to use ordinary strict scrutiny analysis.

The Supreme Court hasn’t encountered such a claim so far, so we don’t know how whether it would follow Justice Kennedy’s approach. Let’s just observe that the tax would be invalid under that approach, and then explore what would happen if the Court doesn’t follow that approach, and if the speech-deterrence effect can be considered when the governmental interest is fighting a secondary effect.

In that case, the government can argue that a targeted tax would be more effective than a broad-based tax because the speech-deterrence effect can be added to the spending effect—that is, a targeted tax fights the social ill better than a general tax does, because the targeted tax has the extra advantage of making the amount of adult entertainment go down. But even then, the tax would not automatically be valid.

ii. The government must make the argument

First, the government must actually make this argument. This might sound obvious, but there are reasons why a government might make a different strategic decision. In Georgia, for instance, the government disclaimed any interest in reducing the speech as such, stressing that the tax was de minimis and left nude dancing establishments absolutely free to do whatever they were doing before—and even free to continue with their speech and avoid the tax as long as they stopped serving alcohol. The need to raise revenue to fight child sex trafficking—i.e., the spending effect—was its primary argument; the speech-deterrence effect was minimized or completely denied.

Why not forthrightly admit that a tax would beneficially reduce the amount of the activity? The government might want to rhetorically blunt the argument that the tax is merely motivated by dislike of adult entertainment—even if this is actually the case. It might want to at least pay lip service to First Amendment values, and avoid giving courts the impression that reducing the activity is the actual governmental interest—rather than a collateral consequence that contributes to fight­ing the secondary effect. The de minimis argument (which is in strong tension, if not downright inconsistent, with the speech-deterrence argument) might be used to convince the court that the constitutional challenge is entirely frivolous; or it might be used to support a “greater power includes the lesser” argument (a type of argument I address below).

In short, even if reducing the speech can contribute to an intermediate scrutiny argument based on fighting the secondary effect, the government must be willing to say so.

iii. There must be empirical evidence

Next, the government must have empirical evidence on the connection between reducing the speech and reducing the secondary effect.

It’s worth pausing to consider the role of empirical evidence in intermediate scrutiny cases. In this Article, I haven’t sharply distinguished between Renton secondary-effects scrutiny and O’Brien incidental-burdens scrutiny: both involve intermediate scrutiny, and both are broadly similar in how they state the test. But there are some subtle differences between the two doctrines.

Renton secondary-effects cases, such as Renton itself or Alameda Books, have hinged on the quality of the government’s empirical evidence tying the burdened expression with the secondary effect. The questions in those cases concerned whether Seattle’s experience was relevant to Renton or whether a study about concentration of establishments could be generalized to concentration of operations within a single establishment. But everyone assumed that some empirical evidence was necessary.

O’Brien cases, on the other hand, have sometimes had no empirical evidence at all. The O’Brien Court, for instance, did not seem particularly interested in whether a ban on destroying draft cards would pursue the government’s draft-related interests substantially more effectively than other policies. (On the other hand, O’Brien held that “assuring the continuing availability of issued Selective Service certificates,” not just the effectiveness of the draft generally, was itself the “substantial interest”: we probably don’t need empirical evidence to know that the destruction of draft cards harms that interest, or that regulations that allowed for draft card destruction would be worse for that interest than a ban on destruction.)

City of Erie, too, minimized the need for empirical evidence, noting the lack of empirical evidence in O’Brien. But at the same time, it engaged in a lengthy discussion of the adequacy of the government’s empirical evidence before upholding the city’s nudity regulation. The justices in City of Erie were split as to how much empirical evidence there needed to be that was specific to Erie’s own experience, but even the plurality considered it significant that there was some evidence.

There is probably less difference between these two strands of doctrine on this point that it seems. Even in Renton-land, governments are allowed to use imperfect studies that are “reasonably believed to be relevant” and may supplement imperfect studies with “common sense,” though they cannot get away with “shoddy data or reasoning.” And O’Brien-style cases rely in part on government officials’ “expert judgments” based on their “firsthand knowledge.” Both standards require an essentially factual inquiry into how well the regulation would further the government’s interest, whether the regulation is greater than essential or burdens substantially more speech than necessary—these are just different ways of expressing intermediate scrutiny’s “narrow tailoring” requirement. So one really can’t avoid some empirical analysis, even if that empirical analysis is fairly casual. “Empirical analysis” isn’t the same as “peer-reviewed quantitative studies.”

So really, the question is how casual an analysis a government should be able to get away with in the context of erotic-expression or nudity taxes. Justice Souter has remarked, in several First Amendment cases, that the standard is flexible. In his City of Erie dissent, he wrote: “It is not that common sense is always illegitimate in First Amendment demonstration. The need for independent proof varies with the point that has to be established, and zoning can be supported by common experience when there is no reason to question it.” At the same time, he warned that “we must be careful about substituting common assumptions for evidence.” And in his majority opinion in Nixon v. Shrink Missouri Government PAC, a campaign finance case, he wrote that “[t]he quantum of empirical evidence needed to satisfy heightened judicial scrutiny of legislative judgments will vary up or down with the novelty and plausibility of the justification raised.”

This is why actual empirical evidence is necessary in these sorts of cases: mere intuition is likely to be misleading here. And the empirical evidence must actually be on point; studies purporting to establish secondary effects are often sloppy and rely on casual “sex-industry-is-bad” generalizations. The next subsection goes into greater detail on this point.

iv. The empirical evidence must be on point

The government’s empirical evidence must not only be present but must actually substantiate the connection between reducing the speech and reducing the secondary effect. When governments try to present empirical evidence, they are not always careful in connecting the dots. For instance, in Georgia, the government presented evidence showing that child sex trafficking is a problem (a proposition that nobody disputed), and that this problem is connected, broadly speaking, to the adult industry (not specifically the adult nude dancing establishments that were challenging the tax). Much of the state’s evidence didn’t seek to show that the problem particularly stemmed from adult nude dancing establishments—as opposed to street prostitution or massage parlors, which (in part because of previous zoning) are often located near those establishments. Nor did it seek to show that reducing nude dancing establishments would alleviate the problem.

Every jurisdiction has its own empirics and its own legal regime, so it’s difficult to say anything generalizable here. But imagine a jurisdiction like Atlanta, where adult nude dancing establishments are licensed, regulated, and inspected; where minors cannot legally enter as patrons; and where minors cannot legally work as dancers—in fact, where dancers need to have permits issued by the local police department, which presumably is trained in recognizing fake IDs. Given the ease with which the police can inspect the permits of an establishment’s dancers, and given the severe consequences of being caught employing an underage dancer, intuition suggests that sex trafficking of minors will overwhelmingly take place in places other than these regulated venues—and, indeed, that reducing the availability of regulated options might tend to drive patrons to less legal (and more exploitative) parts of the industry.

These are of course empirical questions. When I referred above to “intuition,” it was not to establish an affirmative point, but to question casual assumptions in the other direction and to underscore the need for rigorous analysis. When the government submits its empirical studies to support its taxes or regulations, it needs to present not just data, but the sort of data that would answer the relevant questions. Generic arguments that boil down to “the sex industry is bad” or “the sex industry generally is associated with child sex trafficking” aren’t enough. The government needs to at least try to answer whether the tax would reduce the speech at the specific type of affected business and whether that speech reduction would reduce the overall secondary effect. The evidentiary burden isn’t overwhelming, but at least it needs to be on point.

v. Taking intermediate scrutiny seriously

Finally, even if the government explicitly relies on reductions in speech aimed at dealing with secondary effects, and even if it adequately supports this with empirical evidence that is on point, courts should be wary of accepting restrictions when there are alternatives that would not reduce the speech by as much.

This is not strict scrutiny, where the least-restrictive alternative is required; but First Amendment intermediate scrutiny requires that options that reduce speech at least be disfavored alternatives. As Justice Kennedy wrote in his concurrence in the judgment in Alameda Books:

The challenge is to correct the [secondary effect] while leaving the [speech], as far as possible, untouched. . . . A zoning measure can be consistent with the First Amendment if it is likely to cause a significant decrease in secondary effects and a trivial decrease in the quantity of speech.

Or, as the Court wrote in McCullen: “Where certain speech is associated with particular problems, silencing the speech is sometimes the path of least resistance. But by demanding a close fit between ends and means, the tailoring requirement prevents the government from too readily sacrific[ing] speech for efficiency.”

Does this merely restate the rule of intermediate scrutiny? Perhaps, but it does so with a rhetorical force that reminds us that the level of scrutiny is meaningful, and that deferring too readily to the claims of governments is inappropriate outside of rational basis cases.

4. Targeting wrongdoers as governmental interest

Let’s explore a more nuanced governmental interest. Perhaps the argument that we should use taxes to reduce undesirable expression is no good, because the chilling effect can’t itself be the governmental interest in a free-speech context. And perhaps the revenue-raising interest—either by itself, or in conjunction with the need to combat the secondary effect—also isn’t good enough, because why couldn’t we raise revenue neutrally? But maybe we can creatively combine these and come up with an appropriate hybrid governmental interest that avoids those pitfalls.

Economists like the idea that all we should care about is the bottom-line quantity of an activity, the bottom-line amount of money, and traditional rich-vs.-poor distributional concerns. Under that view, there’s no point to a tax unless it either raises revenue, alters the amount of an activity (whether upward or downward), or redistributes wealth between the rich and the poor. Normal people, though, also care about fairness and the expressive function of the law; it might also matter to them whether the incidence of the tax is morally just. So, if it’s true that nude dancing in adult entertainment establishments causes or contributes to some negative secondary effect like child sex trafficking, one might think it’s appropriate (“it’s only fair”) to make those responsible parties bear the cost of fighting the problem, rather than insisting that the burden be borne equally by everyone else—people who have no connection to the problem. And one’s view that this is fair might well depend on the perceived moral appropriateness of the targeting—it wouldn’t matter if the tax did nothing to reduce the activity, or if revenue could be raised equally effectively with neutral alternatives, or if the net distributional effects of the tax were about zero.

Of course, this governmental interest would require establishing the causal connection between the parties targeted by the tax and the negative secondary effect—which, as we’ve seen, is often hotly disputed. But for now, let’s assume that the causal connection has been well established by good statistical studies. Would this governmental interest—merely imposing the tax on those responsible for the problem—be sufficient?

This governmental interest, while creative, still doesn’t work. To see why, we need to step back and consider what it means to say that adult entertainment establishments are “responsible” for secondary effects. We also need to consider the basic theory of antidiscrimination norms.

What does it mean to pin responsibility for a secondary effect like child sex trafficking on adult entertainment establishments? Obviously, it doesn’t mean that all such establishments participate in child sex trafficking. If some do, then everyone agrees that those establishments are legitimate subjects of enforcement (as well as whatever owners, managers, employees, etc., are specifically responsible).

What, then, does it mean to say that the industry as a whole bears responsibility—and should pay through an industry-wide tax? It means that even the establishments that don’t participate in the problem should also pay. And on what theory do we lump the non-participants in with the participants? Based on empirical evidence (which we’re assuming valid for now) that, due to statistical associations and correlations, identifies the industry as a whole as being associated with the problem. Of course, if a particular industry has a number of bad actors, the presence of those bad actors will be reflected in any collective statistics or correlations based on the industry as a whole. So this comes down to the appropriateness of making innocent participants pay for the acts of guilty participants.

But this is precisely the sort of stereotyping that antidiscrimination norms—like the norm against content-discrimination—disfavor.

Am I arguing that an industry as a whole can never be held monetarily responsible for fighting a problem that it contributes to on average? Not at all. To address auto accidents, there’s probably nothing unconstitutional about imposing a surcharge on the auto industry to fund safety programs or medical care for auto accident victims. This is true even if some members of the industry don’t participate in the problem. Maybe Volvos are so safe for their drivers and for bystanders that they’re involved in literally no accidents—but we can still impose a tax based on the statistical association between auto accidents and the auto industry as a whole, unfair to Volvo though it is. And to address global warming, there’s probably nothing unconstitutional about taxing fossil fuel-intensive industries—even if some businesses in those industries don’t contribute to the problem because they use fossil fuels in ways that don’t cause air emissions, or because they plant enough trees to be carbon-neutral.

But that’s because we have no constitutional antidiscrimination norms relevant to auto design or fuel usage. On the other hand, we do have strong antidiscrimination norms based on various other factors: for instance, based on race and sex (in the Equal Protection Clause), based on state of origin (in the Dormant Commerce Clause and the Article IV Privileges and Immunities Clause), or based on religion or content/viewpoint of speech (in the First Amendment). Presumptively, a government can’t hold members of a particular race or sex responsible for the bad acts of some members; presumptively, a state government can’t impose restrictions on people from a particular state because some people from that state have done bad things; and, presumptively, a government can’t penalize a category of expression based on statistical associations driven by particular bad actors within that category.

Is this necessarily true? Note that I did write “presumptively” and “disfavor” above (not “necessarily” and “prohibit”). Even in the First Amendment context, an industry can be singled out for disfavored treatment based on statistical considerations, provided the government can satisfy the relevant level of scrutiny. But while such discrimination might be justifiable in some cases because it’s an appropriately tailored way of pursuing some other interest, it can’t count as the governmental interest itself. The interest itself must be nondiscriminatory.

This is generally true of antidiscrimination norms. These norms generally don’t establish per se rules against discrimination: sometimes we can draw distinctions based on race, sex, content, etc. But the purpose of an antidiscrimination norm is to make us start out with a presumption that the discrimination itself is undesirable. Hypothetically, if we all agree that it is possible to pursue some goal equally well without discrimination (i.e., that the discrimination doesn’t contribute to the goal in any way), an antidiscrimination norm says that we must do so. We always ask whether the discrimination is justified as an appropriately tailored way of pursuing some different—and nondiscriminatory—interest. Otherwise, it’s too easy: merely announce that the goal is discriminatory, and discrimination will always be an excellent way of pursuing that goal. Anyone can play that game.

The Supreme Court encountered a similar argument in Simon & Schuster. The “Son of Sam” law prevented convicted criminals from keeping money they made from depictions of their crimes. What was the governmental interest involved? Of course, “suppressing descriptions of crime out of solicitude for the sensibilities of readers” would have been an inherently invalid governmental interest (and even then, why single out just criminals who tell their own stories, and not writers who tell equally grisly stories?). Compensating victims is a fine governmental interest, but why just expropriate criminals’ assets related to depicting their own crimes, rather than all other assets they might have? Preventing criminals from profiting from their crimes is a fine interest too, but the statute in that case was way too broad for that interest, because it covered income from a book on any topic—as long as the author at some point in the book admitted to committing a crime.

But what about a more complex governmental interest—like “ensuring that criminals do not profit from storytelling about their crimes before their victims have a meaningful opportunity to be compensated for their injuries”? This is clever, because (unlike the others) it establishes a direct link between the governmental interest and the state’s chosen means. But, said the Court, that was precisely the problem:

The Board cannot explain why the State should have any greater interest in compensating victims from the proceeds of such “storytelling” than from any of the criminal’s other assets. Nor can the Board offer any justification for a distinction between this expressive activity and any other activity in connection with its interest in transferring the fruits of crime from criminals to their victims. Thus even if the State can be said to have an interest in classifying a criminal’s assets in this manner, that interest is hardly compelling. . . .

[T]he Board has taken the effect of the statute and posited that effect as the State’s interest. If accepted, this sort of circular defense can sidestep judicial review of almost any statute, because it makes all statutes look narrowly tailored. . . . [S]uch an argument eliminates the entire inquiry concerning the validity of content-based discriminations. Every content-based discrimination could be upheld by simply observing that the state is anxious to regulate the designated category of speech.

Just as with the “reducing the activity” interest discussed above, targeting supposed bad actors because of the content of their speech cannot be an acceptable governmental interest. There are limited circumstances in which a targeted content-based tax can be imposed in pursuit of some legitimate governmental goal, but targeting the speakers based on their content cannot be the goal itself.

C. The “Greater Power Includes the Lesser” Paradox

There is an apparent paradox here. My argument is that some taxes would be unconstitutional, even though some regulations with an equivalent effect—even much more burdensome regulations—could be constitutional under Renton, given the proper empirical support. Isn’t that a perverse result?

But this apparent paradox is just a form of the “greater power includes the lesser” argument. Just because the government can do something draconian (e.g., “prohibit the combination of nude dancing and alcohol”) doesn’t mean it can therefore do everything that’s milder.

These sorts of “greater power includes the lesser” arguments are generally disfavored in First Amendment law (as well as elsewhere in constitutional law). The Supreme Court did once endorse a form of this thinking in Posadas de Puerto Rico Associates v. Tourism Co. of Puerto Rico. But it repudiated that line of argument in 44 Liquormart: “the ‘greater-includes-the-lesser’ argument should be rejected for the additional and more important reason that it is inconsistent with both logic and well-settled doctrine.”

The reason why the “greater includes the lesser” argument doesn’t work in this line of doctrine has to do with the concept of necessity implicit in narrow tailoring. Strict scrutiny and intermediate scrutiny differ on how narrow the tailoring needs to be, but they both require some degree of tailoring, i.e., some degree of necessity, by which we mean some degree of correspondence between means and ends. A regulation or prohibition of nude dancing with alcohol might be constitutional, if a government can produce adequate evidence of secondary effects to pass the Renton test. This would be because, if there’s adequate evidence of secondary effects, a well-crafted regulation (perhaps even a prohibition) could directly alleviate that secondary effect. A targeted regulation (targeted to the specific problem) makes perfect sense in that context because it’s necessary in the way required by the doctrine—and the more targeted, the better.

But that doesn’t mean a targeted tax must be constitutional as long as its overall burden is less than the prohibition. Constitutionality, here, doesn’t depend on the size of the burden, but on whether the tax is necessary, i.e., whether it actually furthers the government’s interest with the requisite level of tailoring. If the government can show that the tax addresses the secondary effect better than other options, so much the better—but that’s the relevant inquiry, not whether the tax is less burdensome than some different regulations that would be upheld.


Even if we accept the secondary effects doctrine, and even if the Supreme Court ends up reaffirming the doctrine as an exception to the facial-discrimination approach that would otherwise apply, taxes on erotic expression or nudity are still constitutionally vulnerable. This is true if we apply strict scrutiny, but can also be true even if we use intermediate scrutiny.

This isn’t an argument for changing anything about U.S. Supreme Court caselaw in this area, which overwhelmingly deals with regulations, not taxes. Indeed, as far as I know, the only cases that contradict my thesis are Combs v. Texas Entertainment Ass’n, Inc., where the Texas Supreme Court upheld an erotic-expression tax under intermediate scrutiny, and Bushco v. Utah State Tax Commission, where the Utah Supreme Court upheld a nudity tax under intermediate scrutiny. And the argument in this Article is that those cases were wrongly decided under existing First Amendment doctrine.

I haven’t covered various other arguments one can make against these taxes. Particular statutes might be vulnerable to overbreadth arguments, because it’s not necessarily easy to define adult erotic entertainment in a way that is both principled and excludes highbrow art. That is, a statute might be overbroad because it violates the rights of theaters or concert venues that feature shows with nude or partially nude content or sexual themes, like the musical Hair or Madonna concerts. Or a statute might be discriminatorily applied so as to exempt “legitimate” theater on the grounds that “we know it when we see it.”

These are all intensely fact-based arguments, and they’ll play out differently in different jurisdictions, depending on the scope of the problems in that jurisdiction and the quality of the respective governments’ empirical studies. But the value of the arguments discussed in this Article is that erotic-expression taxes are unconstitutional even if the state’s empirical studies are excellent; and nudity taxes, while more solid, are still not guaranteed to be constitutional.

Even if one believes that adult entertainment establishments are clearly responsible for a particular problem, a tax that defines such establishments based on their content must be analyzed under strict scrutiny. And whatever the level of scrutiny, a tax that burdens a particular form of expression like nude dancing is unlikely to be narrowly tailored, even under the lower standard of intermediate scrutiny. Most of the possible governmental goals are either illegitimate, because they are themselves speech-suppressive or discriminatory; or they are legitimate but unhelpful, because a broad-based tax would be just as effective.

At most, the governmental interest in fighting the secondary effect could support the tax, but only to the extent that the government is willing to argue that the tax would deter the activity and if the government has good evidence specifically showing that decreases in the targeted activities would decrease that secondary effect. And even then, intermediate scrutiny demands that such arguments be disfavored unless the benefit of deterrence is substantial.

Related Articles

Back to top button