The Social Cost Frontier

In Terror in the Balance, Eric Posner and Adrian Vermeule introduce a “tradeoff thesis” to explain how courts balance between security and liberty in times of criss. To illustrate this tradeoff, the authors produced a “security-liberty frontier”  (Figure 1 on page 2) that is similar to a Pareto frontier (like the Guns and Butter curve from Econ 101).

On the X-Axis is Liberty. On the Y-Axis is Security–the ability of the state to protect the people–in the context of their book, from terrorism. On the frontier (the curve) “any increase in security will require a decrease in liberty, and vice versa.” This curve rejects any claims that “liberty is priceless” or “security at all costs.”

Posner and Vermeule use this curve to illustrate how the relationship between security and liberty can be used to “maximize the aggregate welfare of the population”–“[b]oth security and liberty are valuable goods that contribute to individual well-being or welfare. Neither good can simply be maximized without regard to the other.” The authors note that the frontier by itself “conveys no information about where the optimal tradeoff point lies.” Rather that point “depends entirely on the values or preferences of the people in the relevant society.” Effectively, the frontier “represents a constraint on the opportunities available to governments.” The shape of the Frontier is not static–“balance between security and liberty is constantly readjusted as circumstances change”

A valuable contribution of the Frontier model is that it allows people to recognize that “security and liberty are comparable” and “can make judgments about the relative worth, to them, of increases (decreases) in security that produce a concomitant decrease (increase) in liberty.” However, this is not to say that all increases in liberty will result in a decrease in security, or that all increases in security will result in a decrease in liberty. An important aspect of this model is the recognition that certain government policies can reside below the curve (Points Q, Q*, R, and R*), where liberty and security are not directly related.

The Posner/Vermuele frontier is helpful in order to illustrate the relationship between liberty, social costs, and security I discussed in The Constitutionality of Social Cost (which is now in print in the Harvard Journal of Law & Public Policy). I adapted this frontier, with a few alterations, in what I call the Social Cost Frontier. I re-labeled the X-Axis as individual liberty, rather than liberty. This focuses on what is implicit in Posner and Vermuele’s analysis–liberty inures to the benefits of individuals, while security inures to the benefit of society, at large (and these two factors are often at odds). I re-labeled the Y-Axis more broadly as Collective Safety–security is really a subset of collective safety. The red curve represents social cost (and corresponds to a second axis on the right, labeled social cost).

In The Constitutionality of Social Cost I defined the social cost as a negative externality that could result from the exercise of individual liberty (such as harm that could result from using a firearm for ill will). Effectively, the more liberty a person has, the more risk there is for the individual to cause social cost to society as a whole. In future works, I aim to broaden this definition to also include social cost that can result from the exercise of government power. Effectively, the more power the government has, the greater the risk of the state of causing social costs to individuals (and their liberties). The red social cost curve, represented by a parabola, illustrates this relationship.

This frontier helps to explain judicial review and the relationship between the tiers of scrutiny and social cost. Scrutiny is effectively a question of how much deference the courts will give–but deference to whom? Generally we think of rational basis scrutiny as the Court deferring to the determinations of the majority and exhibiting skepticism towards the individual’s liberty interest. Strict scrutiny, on the other hand, employs skepticism to the determinations of the majority, and deference to the individuals’ liberty interest. These determinations are linked with considerations of social cost.

High social costs from an individual that can harm society nudges the courts to show deference to society (majority rule) viz strict scrutiny. High social costs from government power that can harm individual liberty nudges courts to show deference to individuals (individual liberty) viz rational basis. In this sense, strict scrutiny and rational basis are mirror images of each other, reflected along the social cost curve. These mirror images are illustrated by the two clouds on the figure.

In the first cloud, which represents, strict scrutiny, courts will defer to individual liberty and be skeptical of the state’s power to provide for collective safety and minimize potential social costs from the state; as a result the law in question will likely not be upheld. Looking at the curve at this point, we see that the individual liberty interest is quite low, but the state’s power to provide for collective safety is quite high. Additionally, the social cost of the government’s power to the individual is also quite high (represented by the high point of the parabola). In these cases, where the social cost is too high, the protection of individual liberty is too low, and the power of the state is too high, courts are skeptical of the government’s power, and defer to the liberty interest. This is the essence of strict scrutiny, visualized.

In the second cloud, which represents rational basis scrutiny, courts will defer to the state’s power, and be skeptical of the liberty interest in order to minimize social costs from the individual; as a result, the law in question will likely be upheld. At this point, the social cost of the individual’s liberty can be quite high (represented by the high point on the parabola). In these cases, where the social cost is too high, the protection of the state’s safety is too low, and the liberty interest is too strong, courts are skeptical of that liberty interest, and defer to the majorities. This is the essence of rational basis, visualized.

The symmetry between the two ends is illustrated by the model–striking down a law and deferring to individuals is the opposite of upholding a law and deferring to the state. In this sense, the Court can be both a majoritarian and counter-majoritarian/libertarian institution depending on the level fo social costs.

Both strict scrutiny, and rational basis, attempt to guard the Social Cost Frontier, and keep the law below the curve, in the orange-shaded region. The cases at the margin–along the curve–are often the toughest to decide. The Supreme Court, as an institution, has assumed the role of policing these cases, and making sure that border is seldom reached by nudging cases down.

In this orange-shaded region, liberty and safety can be varied without affecting each other. Considering the Posner/Vermeule curve, a change from policy Q* to Q results in an increase in individual liberty without a concomitant decrease in security. Likewise, a change from policy R* to R results in an increase in security without a concomitant decrease in liberty. With these policies, liberty can be increased without decreasing safety, and safety can be increased without decreasing liberty. As the authors noted, “[o]f course, not every issue of security policy presents such a tradeoff. At certain levels or in certain domains, security and liberty can be complements as well as substitutes. Liberty cannot be enjoyed without security, and security is not worth enjoying without liberty.”

In anticipation of comments, Vermuele does a very thorough job of responding to a number of criticisms (I expect to receive) with respect to plotting liberty and security on a graph in this paper. Most saliently, the frontier, has its limitations–it is “not an empirical claim about security policy,” nor does it aim to “evaluate, on the merits, whether” certain policies are optimal, but rather serves as only an “expository device.” For example, Posner and Vermuele conceded that the concept of individual liberty is “more difficult to determine,” so they “basically have to ignore it.” Viewed in this limited fashion, and putting aside objections about representation of these items on a graph, the explanatory features of this model are quite illustrative of how courts approach judicial review.

Cross-Posted at

You may also like...

13 Responses

  1. Ken Rhodes says:

    Interesting model, Josh. In the graphic representation it depends on “measurement scales” for Liberty and Security, which may be a complicated problem in itself. But the model concept seems like sound math/economics.

    I have to wonder about the region below the frontier. Where liberty and security do not have to trade off, it may still be rational for government to fail to move higher on the Security measure, primarily because of dollar cost. But why in the world would we fail to move farther to the right on the Liberty measure?

    Having written that sentence, I suppose the answer is obvious, though distressing — there are many among us who are not willing to accord liberties to others to behave in ways they find distasteful.

  2. Josh Blackman says:

    Thanks for the comment. As for the scales for Liberty and Security, as well a Social Cost (the red axis), I am not deciding it now. Getting bogged down in precise numbers vitiates the value of this expository model.

    I think your instincts about increasing liberty may be about right, and that directly ties in with how Judges defer to majorities rather than deferring to liberty. I hope to elaborate on that in future works. This initial topic will be more descriptive, rather than normative.

  3. A.J. Sutter says:

    Sorry, but there is a fundamental fallacy here. The curve doesn’t prove anything — it’s just a metaphor for the authors’ thesis that “security and liberty are comparable” and that one “can make judgments about the relative worth of increases (decreases) in security that produce a concomitant decrease (increase) in liberty.” I.e., the conclusion precedes the diagram. Where is the evidence that the curve exists, that it is differentiable, that the “frontier” is relevant, or that it has the shape posited, or that “security” and “liberty” have any useful meaning when each is projected into one dimension? This is, so to speak, an object lesson in reification — and in misleading reification, at that.

    You are also positing a judiciary that makes rational choices about “social cost” rather than decisions that have a political context, and decision-makers who are embedded in that context. John Kenneth Galbraith already pointed out in several works (e.g. The Affluent Society, A History of Economics, The Economics of Innocent Fraud) the convenient fiction in neoclassical economics that political power doesn’t exist. Your importation of the metaphor of costs — and your attempted explanation of judicial decisions on that basis — does economics one better: you’re trying to take the politics out of politics.

  4. Josh Blackman says:

    Thanks for the reply. I recognize that you reject all the premises on which this model is based. Without accepting any of my assumptions, even for the purposes of this thought experiment, I’m not sure what else to reply, other than with this classic quote from The Matrix, in response to your comment of “Where is the evidence that this curve exists?”

    Spoon boy: Do not try and bend the spoon. That’s impossible. Instead… only try to realize the truth.
    Neo: What truth?
    Spoon boy: There is no spoon.
    Neo: There is no spoon?
    Spoon boy: Then you’ll see, that it is not the spoon that bends, it is only yourself.

    There is no curve.

  5. Shag from Brookline says:

    Does this in the post:

    “Both strict scrutiny, and rational basis, attempt to guard the Social Cost Frontier, and keep the law below the curve, in the orange-shaded region. The cases at the margin–along the curve–are often the toughest to decide. The Supreme Court, as an institution, has assumed the role of policing these cases, and making sure that border is seldom reached by nudging cases down.”

    confirm in comment 4 that:

    “There is no curve.”?

    Or is this an attempt at a “fast ball”?

  6. Josh Blackman says:

    Shag, it’s more of a slider, by keeping it down and away.

  7. Ken Rhodes says:

    In re: AJ’s objection, I would say “the curve doesn’t exist as a function in two-dimensional x,y coordinates, but that doesn’t mean it doesn’t exist. The curve is a set of points in n-space.”

    In my initial comment I mentioned that the graphical representation depends on measurement standards for “Security” and “Liberty,” which itself is a daunting challenge. Most likely, those two attributes of a society are not measurable on a one-dimensional scale.

    However, that does *not* negate the concept of a Pareto-optimal bound on the multi-dimensional attribute-space we live in, nor does it require that it be representable in a two-dimensional graph. The graph is merely a simple construct to illustrate the complicated world.

    The proof that there is a Pareto-optimal set of boundary points for the Security-Liberty attribute space is the simple fact that certain actions to increase security seem to occur only at the cost of decreasing liberty. You do *not* have to measure those two things on a single scale (dollar value, for example) to discuss Pareto efficiency. You merely have to recognize that for an individual, there will be tradeoffs that will be favorably received and others that will be rejected. The set of tradeoffs that would be considered “neutral” by a given individual represent the Pareto optimal boundary for that individual.

    The model is proposed by Josh not as the basis for a programming exercise, but as a construct to understand our world. Another challenging step in trying to *apply* the model, rather than merely using it to illustrate the situation, is to try to come up with a method of “averaging” the Pareto optimal sets of individuals. Many individuals would reject any set other than their own, and most would reject any points in a set that were far away from their own. Gaining acceptance of a “compromise set” of boundary points in a large group may be hopeless.

  8. Shag from Brookline says:

    Perhaps a “balk.”

  9. Josh Blackman says:

    Shag from Brookline (Massachusetts I suppose),
    On the topic of baseball, you wouldn’t happen to be a Red Sox fan, would you?

  10. A.J. Sutter says:

    Ken: To pick but one nit out of many available, a set of points in n-space is not necessarily a differentiable curve. If you read Pareto (§ 89 in the Appendice to his 1909 Manuel d’économie politique), you’ll see that his argument depends on differentiability. No one bothers to ask if this makes empirical sense. You can’t establish the continuity of a “Pareto boundary” by ascertaining an individual’s “neutrality” toward a finite number of tradeoff scenarios. You can only assume it (which is what Pareto, Slutsky, Samuelson & al. did in their theory of consumer preferences.) Pareto’s formulation also rests on the assumption that choices are well-ordered, which isn’t empirically justified. As for a common scale between “security” and “liberty” that would allow one to claim, as P/V do and as Josh endorses, that one can “maximize the aggregate welfare of the population” or of any other group, Pareto did indeed have one, “ophelimity”. On top of that you have the intersubjecvtivity issue, since laws aren’t made for individuals, etc., etc.

    As for Josh’s reply @#4: I’m not sure which role you’re assigning to which position, but as the midrash says, let your ears hear what your mouth is saying. If there is no curve, then what does it mean to say it’s “quite illustrative of how courts approach judicial review”?

    This isn’t merely a matter of accepting/rejecting assumptions like one might accept or reject the Parallel Postulate or the Axiom of Choice, because you’re attempting to apply this explanation to real-world events. At minimum, you’re claiming that certain factors (considerations of “social cost”), expressed in low dimensionality and ignoring qualitative differences, are sufficient to explain judicial decisions in a positive sense. You don’t attempt to justify your assumptions of low dimensionality and quantifiability, nor to establish why that’s a better explanation than alternatives, other than by reasoning from the curve.

    Obviously this is more than academic wheel-spinning for you, or a late-night beer-fuelled argument for the sake of fun. You take it seriously enough to publish about it. Your claim that it’s “an expository device” then raises the legitimate question of what’s the narrative you’re promoting. You seem to simultaneously claim and disclaim that courts do reason this way; if you mean to suggest that courts should reason this way, or some other way based on “social cost”, then the fallacies in the style of reasoning ought to be exposed. I admit that especially in that normative context the issue of accepting assumptions does come back into play: disagreement about assumptions is called politics. You can’t wish that away with Keanu Reeves-style faux-zen discourse, much as many might like to.

  11. It’s humbling to read A.J. rationally and persuasively explain what I only intuitively and inchoately think and feel.

  12. Shag from Brookline says:

    To put what Patrick said in baseball terms, A.J. Sutter did a reverse A.J. Burnett shaving cream pie in the face to Josh.

    MA, yes; Red Sox fan, no. Bob & Ray of Boston radio fame had a sports skit that Bob signed off with: “This is Steve Bosco rounding third and being thrown out at home.” Josh, you should have punted (to segue to football). (If you’re interested, I am not a Patriots fan either.)