The Online Library of Liberty

A project of Liberty Fund, Inc.

• Introduction
• Part One: the Needless State (chapters 2, 3, 4 Are Not Available Online For Copyright Reasons)
• 1: Who Gave Us Order? On Exclusion, Enforcement, and Its Wherewithal *
• 5: Empirical Evidence *
• Part Two: Redistribution (chapter 7 Is Not Available Online For Copyright Reasons)
• 6: A Stocktaking of Perversities *
• 8: Disjunction, Conjunction *
• Part Three: Justice (chapters 10, 11 Are Not Available Online For Copyright Reasons)
• 9: Justice As Something Else *
• 12: Slicing the Cake Nobody Baked *
• Part Four: Socialism
• 13: Ownership, Agency, Socialism *
• 14: Market Socialism: “this Square Circle” *
• Part Five: Freedom
• 15: Right, Wrong, and Economics *
• 16: The Paretian Liberal, His Liberties and His Contracts *
• 17: The Bitter Medicine of Freedom *
• Works Cited

16

The Paretian Liberal, His Liberties and His Contracts*

WITH HARTMUT KLIEMT

1.

Introduction

The debate about the putative impossibility of a Paretian liberal has been going on since 1970. Looking back, the impression is one of a mixture of clear formal argument and often confused interpretations. Confusion arose, in particular, from a failure to distinguish conceptionally between the relevantly different phenomena of “liberties” and “rights.” As we hope to show subsequently, the alleged paradoxes of liberalism lose their paradoxical character if one realizes that liberties differ from rights in the following way: We are at liberty to do something if we are under no constraint or obligation1 to act otherwise, we have a right only insofar as others have certain obligations towards us to act in ways demanded by us.

If person A has the liberty to decide whether to wear a green or a red dress and if person B has the same liberty to choose which dress suits her, B, neither of the two has a right to demand that a certain dress be worn by the other. Correspondingly, failing specific evidence to the contrary, neither of the two has any obligation to wear either kind of dress, nor is either of the two under an obligation to choose one color rather than another, even if their choices are not agreeable to each other. Each is at liberty to choose how to act. Individuals may, however, be willing to trade their respective liberties of choosing the color of their own dresses and thus to create rights and obligations.

Assume that a mutually agreeable trade confers on one person the right to choose the other person’s color of dress, green or red. As a result of contracting, the latter is under an obligation to wear a dress of the color specified by the former, i.e., the right’s holder. Assume also that the holder of the right has retained her liberty to choose the color of her own dress. Then, after the first individual has traded away her liberty, the second individual as holder of the right will be entitled to choose a state of affairs or to make a social choice. She may choose the color both of her own dress and that of the other. Therefore she has full control over which state of a set of social states—each defined by a combination of the colors of the two ladies’ dresses—will be chosen.

It is impossible, though, that two individuals should have full control over the same pair of states of affairs. If person A has the right to choose one from a pair of social states, then person B cannot have a right to choose with respect to the same pair. Both cannot simultaneously have a right to decide which combination of dress colors of two individuals will form the state of the world. Nor could one have the right to choose either of two social states which both specify the colors of both dresses so long as the other still retains the liberty of choosing her dress.

Subsequently we shall illustrate our claim that the alleged paradox of liberalism loses its bite if one makes the distinction between how liberties and rights function. In a first step we shall present that distinction in a somewhat more formal manner (2). If the alleged liberal paradox should rest on such an obvious confusion as we claim, it must be explained how it could emerge and be taken seriously at all. After proposing our account of that matter (3) we try to present a more traditional and, as we feel, more adequate liberal view of the role of liberties, rights, and Paretian values (4). Some concluding remarks follow (5).

2.

The Non-Paradoxical Paradox of Liberalism

2.1.

FEASIBLE, PRE-EMPTED, AND ADMISSIBLE CHOICES

We take all social states rendered possible by nature as the given feasible set. A subset of the feasible set (e.g., reading lewd books, or buying them tax-free) cannot be chosen because of collective prohibitions (“do not read lewd books”) or collective commands (“pay a pornography tax”). This then is the pre-empted subset. Its complement is the admissible subset, which includes everything that is feasible and not prohibited. (For our purposes, we may ignore the possibility of choosing alternatives in violation of prohibitions and commands.)

Prohibitions and commands are by their general nature collective choices (made for a collectivity either by a dictator or by a sub-collectivity or even the whole collectivity), leaving the choice between residual alternatives, if there are any left, to individuals. Evidently, there may be no residual. Short of this, the collectivity may choose not to choose, and to restrict its own domain of choice by a substantive meta-rule (constitutional provision), which specifies what is put into the public domain of collective or political decision-making and what shall be decided non-politically by individuals in their several capacities. (A procedural constitutional rule, as distinct from a substantive one, instead of delineating private and public domain lays down how a collective choice from a domain of alternatives is to be reached—e.g., by aggregating votes.)

The preceding way of dividing the feasible set into public and private treats collective choices as basic. Therefore, on the most fundamental level of decision-making, individual rights and liberties cannot impose any constraints on the collective choice of the proper realm of collective as opposed to private decision-making. We need a kind of Archimedean point preceding any collective decision if on that level constraints on collective choice are assumed to exist. Without some initial exogenous division between pre-empted and admissible, there may be no liberties to start with.

One such potential determinant, exogenous to the present, is history, which has bequeathed social convention to the present. Convention rules out certain alternatives for being torts, in the broad and ancient sense of the word, that is offenses against person and property subject to retribution and restitution. The concept is not very sharp-edged but it captures quite well our common intuitions about respecting other individuals as persons who are entitled to make certain choices.—In any event, we must start from somewhere. We will therefore begin our discussion under the assumption that the admissible sub-set, i.e., the initial area of liberties, is exogenously determined.

2.2.

LIBERTIES, RIGHTS, AND OBLIGATIONS

Whether or not we accept that there can be any individual liberties and rights preceding any form of collective choice, the admissible subset of an individual’s feasible choices consists of liberties, rights, and obligations towards other individuals. The individual exercises a liberty when performing an admissible act A that does not violate another’s right. He exercises a right R when his doing so obliges another to perform an act bringing about a “state” r corresponding to R. Finally, he fulfills an obligation when performing an act bringing about r to which another is exercising a right R (for the determination of rights, cf. infra.).

A driver is free (has the liberty) to drive his motor car on the road in a manner that causes no tort or a high risk thereof to other users of the road. Every other driver has the same liberty, notwithstanding that the simultaneous use of their liberties by everyone would bring traffic on the road to a standstill. This is to say that the exercise of liberties may be incompatible. The exercise of one of a pair of incompatible liberties is not a violation of the other. It is an adverse externality. A liberty is only violated by a tort, an inadmissible act.

More specifically, consider again the example of two women each of whom is at liberty to choose the color of her dress. Each of the two, who for convenience are christened 1 and 2, may decide to wear a green, g_i, or a red, r_i, i = 1, 2, dress respectively. We take it that for both of the women each of the two decisions is admissible and neither of the women has a right limiting or controlling the choice of the other. Given these premises the ensuing interaction may be represented by the following game form:

All of the results represented in this game form are admissible. They emerge as individuals exercise their liberties. Exercising a liberty is equivalent to the choice of a row, in the case of player 1, or a column, in the case of player 2. Individuals’ liberties are to be identified with their strategy sets (rows or columns) in the game form rather than with the social states (cells) brought about by the joint exercise of their liberties.2

If only liberties to choose the color of one’s dress—but no rights with respect to another wearing one color or the other—exist, each individual is free to choose among the alternatives over which she has a liberty. The other individual has no legitimate complaint as far as this is concerned. Neither has either of the individuals, using her respective liberties—normatively speaking—any claim over the choices of the other. Each can choose her own actions within the realm of her liberties. Neither can choose a social state. Whatever comes out of their separate choices will be the social outcome.

On the other hand, imagine that lady 2 has given up her liberty to choose the color of her dress. She has accepted the obligation to comply with lady 1’s wishes as far as the color of her (2’s) dress is concerned. Lady 1 has acquired the right to choose a social state (from a set of social states). She is entitled to choose among whole states of affairs since she is at liberty to choose her own dress and has the right to impose the color of 2’s dress. Contrary to this case, individuals, in exercising merely their liberties, can never bring about a collective result single-handedly.3 Their liberties allow for the simultaneous exclusion of sets of results from the collective choice set but never for the choice or exclusion of a single alternative from a set of alternatives. Thus, minimal liberalism in Sen’s sense—that is, the capacity to choose one state of at least one pair of social states—is not implied by “game form liberalism” based on the assignment of liberties rather than rights. Therefore, contrary to Sen’s claims, his arguments do not apply to what might be called liberal individualism.

Essentially the same point has been made by James M. Buchanan twenty years ago (printed for the first time in this issue). Since it was strongly criticized in Buchanan’s original presentation it may be helpful to look at it in some more formal detail in the light of our basic conceptual distinction between liberties and rights.

2.3.

SOCIAL CHOICES BY EXCLUSION

In the above game form, so long as no rights exist, there is neither an individual choice nor a social choice of a cell. There is simply no choice of a cell. On the other hand, each person, in exercising her liberties, insures that the social state finally emergent must fall within the subset defined by her choice. Exercising their liberties individuals end up in a cell. But the cell is not chosen by any individual.

The liberties of individual 1 may be represented by the set of sets D₁ = {{(r₁,r₂), (r₁,g₂)}, {(g₁,r₂), (g₁,g₂)}} while the liberties of individual 2 may be represented by the set of sets D₂ = {{(r₁,r₂), (g₁,r₂)}, {(r₁,g₂), (g₁,g₂)}}. As can be checked immediately ∀ x ∈ D₁, ∀ y ∈ D₂ : x ∩ y ≠ ϕ. Thus individuals 1 and 2 can simultaneously exercise their liberties in any way they like without precluding the emergence of a well-defined collective result in a situation characterized by the above game form.

However, if we postulate rights rather than liberties there is no guarantee that within the realm of the normatively admissible a well-defined collective result exists. This may be illustrated by Alan Gibbard’s well-known example of Zubeida and Rehana (1974, also quoted in Sen 1976/1982a, 312–13) who are going to choose the color of their dresses. Each of the ladies can very well have the liberty to choose green or red. However, if Zubeida had the right both to choose between red and green, and to wear the same color as Rehana, Rehana would have an obligation to choose red when Zubeida chose red (and green when the latter chose green). Rehana could not have the liberty to choose her own color. This would be pre-empted by Zubeida’s right. One’s right would negate that of the other and, for that matter, the liberty of the other. Both women’s “rights” could not simultaneously stand. No two contradictory rights can both stand.

Referring to the preceding game form this situation can again be illustrated in a very simple way. Recall that the liberties in that situation were

D₁ = {{(r₁,r₂)}, {(r₁,g₂)}, {(g₁,r₂)}, {(g₁,g₂)}} and

D₂ = {{(r₁,r₂), (g₁,r₂), (r₁,g₂), (g₁,g₂)}} with

∀ x ∈ D₁, ∀ y ∈ D₂ : x ∩ y = x,

that is lady 2 has neither a liberty nor a right to choose.

Now, the latter construction may seem unfair to Gibbard. He does not assume the existence of a decision right over all pairs of alternatives for one individual. It may seem therefore that such a dictatorial competence over all alternatives is over-extending Gibbard’s use of the notion of a right. However, even under the most charitable interpretation of the approach a variant of the preceding argument would still apply.

Consider the following game tree in which player 2 is granted the “right” to decide between pairs of states of affairs contingent on the choice of the other. With this “contingent right” player 2 cannot require player 1 to choose in a specific way. As a second mover she can merely decide which of the social states will emerge after the first mover 1 has chosen her dress.

The corresponding decision “rights” then are

D₁ = {{(r₁,r₂), (r₁,g₂)}, {(g₁,r₂), (g₁,g₂)}}

D₂ = {((r₁,r₂), (r₁, g₂)), ((g₁,r₂), (g₁,g₂))}.

That is, the first can choose among sets while the second, contingent on the set chosen by the first, can choose among states of affairs. The decision rights do not let both choose among states of affairs. That is, they are not

D₁ = {((r₁,r₂), (g₁,r₂)), ((r₁,g₂), (g₁,g₂))}

D₂ = {((r₁,r₂), (r₁,g₂)), ((g₁,r₂), (g₁,g₂))}.

Thus, if “contingent rights” are construed appropriately not both individuals can hold “rights” such that an empty choice set emerges. The basic claim of those who think that there is a paradox of liberalism vanishes, since this claim amounts to nothing but the thesis that certain sets of axioms imply that an empty choice set emerges for some profile(s) of individual preferences.4 —It is obvious that the same argument holds good for the symmetric case in which 2 is the first mover.

Moreover, if the game form of the corresponding—“simultaneous move”—imperfect information game is presented in its extensive variant basically the same argument still applies. Informationally, both moves take place simultaneously. Since none of the players can have any knowledge of what the other chose, none can intentionally choose a social state. Each can make her own choice of an action but then must “wait” for the result that is going to emerge.

Of course, in a non-informational sense there may be a time sequence between the players’ moves. The second mover in time, though being ignorant of the choices of the first mover in time, may know that as a matter of fact by making her choice of a class she actually chooses between two states. But even if we would assume that this kind of a choice fully captures what we mean by a “choice of social states” it is clear that the argument that at most one player can be in a position to decide single-handedly between some pair of states of affairs still applies. For, the first mover is in the same position as before. Given the assumption about the time sequence in the “imperfect information tree” she must make her choice before the other player chooses and thus she can choose only between classes of states of affairs. For her this is not merely a matter of knowledge. From the point of view of the first mover the state of affairs will emerge only after the second mover has made her choices.5

In the case of the two girls choosing their dresses, Rehana can be normatively entitled to choose between two states of dressing only if she is entitled to require that Zubeida dresses the way Rehana chooses and Zubeida is obliged to comply. Thus, obviously, Zubeida cannot be at liberty to choose how she will dress if Rehana has a right to choose between a pair of completely specified social states. Thus, to reiterate, for entirely trivial reasons any of the individuals can choose a social state from a pair of social states only if she is—normatively speaking—in the position of a dictator entitled to determine all dimensions of the emerging state of the world.6 This is no paradox but rather follows immediately from the underlying construction of “rights to choose.”

To generalize, after recognizing the elementary distinction between rights, i.e., the choice of cells on the one hand, and liberties, i.e., the choice of columns or rows on the other, it is obvious that an individual i could virtually choose between two states of affairs—cells—only if all other individuals j ≠ i, from a set of individuals N, were under an obligation to choose according to her “orders.” Individual i must be normatively entitled to tell them how they must choose. They cannot have any liberty left to choose against i’s wishes. If they choose otherwise they violate an obligation towards i. Individual i is in the position of a puppet master who can lead all other individuals by the strings of their normative obligations to follow suit if she asks them to do their parts in picking a specific cell.

Obviously, the adherent of liberal individualism would have to reject such a construction. He does not feel that letting individuals take turns in playing the role of the puppet master expresses liberal values. The adherent of liberal individualism is primarily interested in what we in this paper call liberties. Rights, or what we choose to call such in this paper, are in his view a contingent consequence of liberties: a person A creates a right for person B by assuming an obligation to perform a particular act if B requires him to do so. B cannot have the right to this performance if A preserves his liberty to perform or not to perform the act. The free choice between preserving and surrendering liberties is a defining feature of the liberal creed, and of a liberal theory of rights.

We do not claim a monopoly of correct usage when we call one particular relation between persons and acts “liberty,” the other “right.” What we claim is that they are fundamentally different relations; calling them by the same name is to ignore the difference. If there is an excuse for doing so, it can only be the view that all such relations, i.e., both our liberties and our rights, are privileges conferred on individuals by collective social choice. However, even on this view they would be relevantly different, as a glance at the game form representation clearly reveals. What is puzzling, and needs explanation, is how so many eminent social choice theorists could fail to make the obvious distinction between the phenomena to which we refer as rights and liberties respectively and consequently could think that their collective choice concept of a right could capture intuitive individualist liberal notions of freedom of decision.

3.

Rights as “Softeners” of Social Choice

Sen does not accept the Nozickean view that “rights” are simply constraints imposing restrictions on the realm of collective choice. As a genuine social choice theorist, Sen models individual choices as acts of participation in an overall social choice. He therefore tries to build “rights” into the collective choice mechanism itself: in translating individual orderings into a common social ordering, society must rank any alternatives over which individual i has a “right” as i ranks them, and any alternatives over which j has a “right,” as j ranks them.

Let us reconstruct what that could mean by transforming the previously discussed example of a game form into a very simple voting game. The game form was defined by the set of players k, k ∈ K = {1, 2} and the set of strategy profiles Z = {(z₁, z₂) | z_k ∈ {g_k, r_k}, k ∈ K} which at the same time determined the set of possible states of the world characterized by the possible combinations of green or red dress colors of the two individuals. Now, let

Z* := {((Z₁₁, Z₁₂), (Z₂₁, Z₂₂)) | Z_k1 ∈ {G₁, R₁}, Z_k2 ∈ {G₂, R₂}, k ∈ K },

where capital letters stand for individuals’ voting rather than for their dressing strategies. Thus “Z_kj” must now be read as individual k votes in favor of bringing about a state of the world in which individual j acts according to “z_j.” Note, though, that according to this construction j is no longer entitled to choose z_j. All choices are made collectively or socially since the state of the world is determined in a voting process. To put it slightly otherwise: when dressing, individuals are merely acting in the way corresponding to z_j but the choice of their act has been made for them on the level of voting. (Think of the collective body as a “puppet master” who is deciding by majority vote on the script for a “dressing performance.”)

Whenever there is no unanimity the obvious question is whose wishes should prevail. For instance, individual 1 might vote (G₁, R₂) and individual 2 votes (G₁, G₂), etc. An obvious way out is giving dictatorial competence to one individual. Accordingly the next matrix shows what it would mean that 2 has dictatorial competence. In this matrix, whatever 2 chooses “for the collectivity” (by casting his vote according to one of the four pairs of “capital letter alternatives” in the top row of the matrix) is executed as the social choice and individuals merely act as “puppets on a string” when bringing about the socially determined result (one of the lower case alternatives forming the “inner” sub-matrix). By wearing a dress of the correct color they execute collective commands issued by the dictator.

To avoid dictatorship, individuals must change the voting mechanism. Individual 1 should not merely participate as a “dummy.” His vote should have real weight. If the mechanism is “softened” so that every individual can determine one issue by making his vote effective for that issue we get the following matrix of the voting game:

If we reduce the latter matrix to the starred rows and columns by leaving out the duplicated results we get the game form Γ:

This game form looks strikingly similar to the one presented before. The fact that the collective choice mechanism is “softened” by granting individuals a decisive vote in determining the collective command to be executed by them should not deceive us into believing, though, that the voting rights protect the individuals’ liberty of dressing as they please. They do not. The formal “similarity” between the voting game form and the liberal game form conceals that the change from lower case to capital letters in denoting strategy choices makes all the difference in the world.7

As long as individuals chose “small letter actions” directly they were entitled to choose the color of their own dresses separately. They had liberties rather than participatory rights in a voting process. Whereas now they have a vote in a collective choice. They can choose to vote in a certain way and by this bring it about that the collectivity issues the command that they dress in their preferred way. Still, when dressing, they merely execute a collective command but do not choose how to dress. The action of dressing now amounts to the execution of a command rather than to exercising a liberty.8

“Rights,” in the sense Sen uses that term, are elements of a collective command structure. They serve the function of keeping collective choice reasonably close to what could be accepted under the value premises of Paretian welfare economics.9 Thus, even though he presents it as an attack on the Pareto principle, basically the same Paretian-Wicksellian aim of “softening” collective choice that was driving Buchanan and Tullock in their Calculus of Consent (1962) seems to be behind Sen’s enterprise.10

Since the game form of the voting game and the reduced liberal game form of the preceding example look almost identical, one might be tempted to conclude that Sen’s construction amounts to the same thing as the liberal game form. This similarity explains why so many people could think that the social choice theorists’ representation of liberties as participatory rights in a social choice mechanism could capture what liberal individualism is all about. However, it is clearly inadequate to reconstruct the intuitive notion of what we call liberties in this paper as special “voting rights.”11 Explicating the concept of a liberal “right” in terms of individual entitlements to make social choices for a collectivity does not capture adequate notions of “rights” or, for that matter, “liberties.”

4.

A Liberal View of the Liberal Constitution

We could be content to let it rest at that. Yet the adherent of the social choice approach may still insist that even if individuals are entitled to make their “private” choices within the scope of their admissible actions a collective result or social state will eventually emerge. Since “the rules of the game” are collectively determined—at least they can be collectively changed under some rule of rule change—society cannot avoid responsibility for collective results—at least the responsibility of not changing the rules. In this sense the collectivity acting as a whole or through its agents, may be regarded as being responsible for the initial delineation of liberties, of what kinds of contracts are going to be enforced, of what kinds of behavior will be treated as torts, and so on.

4.1.

FREEDOM OF CONTRACT

Sen thinks that there are certain decisions that are intrinsically private. These decisions should be left to the individuals in their private capacities. And, as far as this is concerned, he claims to be in good company since “. . . most social philosophies accept certain personal or group rights” (what the present paper insists on calling liberties). “The fact that unqualified use of the Pareto principle potentially threatens all such rights gives the conflict an extraordinarily wide scope.” (Sen 1976/1982a, 316) Indeed, as one could have guessed, the problem—if there is one—must go beyond lewd books, pink walls, sleeping on one’s belly, and other “personal things” (297). “If we believe [in unrestricted domain and almost any form of the Pareto principle] the society cannot permit even minimal liberalism. Society cannot let more than one individual be free to read what they like, sleep the way they prefer, dress as they care to, etc. irrespective of the preferences of others . . .” (Sen 1970a, 157; our italics). However, if there is unrestricted domain and P, Pareto optimality, and L, minimal liberalism, are the universal rules comprising the social choice mechanism, do they not apply to all pairs of alternatives in the critical preference configuration, regardless of their particular content? Why is the competence of L restricted to “personal” matters? And where do personal matters stop12 and “impersonal” ones begin? Are matters of livelihood, work, property “personal,” to be “protected” by L? If not, why not? The intended effect in Sen’s theory of distinguishing between what is under an individual’s control (that another may covet), and what he covets but can only get by giving up what he controls, is that subjecting the former to L (the dictates of freedom?) and both to P is capable of producing the impossibility result. The conflict is rooted in who controls what. At least in its formal logic it is not content-dependent. It would be arbitrary to make it so.

A substantive flaw of Sen’s thesis (though he is in good and numerous company), seems to lie in his attempt to discriminate between rights (and of course liberties) according to their content. There are “personal matters,” “a sphere of privacy,” “an area of autonomy” in which an individual is to be sovereign, “free to decide,” and the related preferences of others are “meddlesome,” intrusive. There are, presumably, other matters of which this is not true. But if the individual’s sphere of privacy, or area of autonomy, covers the set of his liberties and rights that must not be violated, has he any others that are not part of the set, and falling outside the protected area, may be violated?

If there are no liberties and rights that may be violated, so that no one can be made to do something against his will, which seems to be an inherent supposition of the “soft” social theory that uses Pareto-superiority as a criterion of “better,” then none is outside the “sphere of privacy” or “area of autonomy.” For what characterizes the latter is not that its content is particularly “private” (whatever that means, for aren’t all individual liberties and rights “private”?) but that it is the set of a person’s liberties and rights, over which he alone disposes. Expressions like “private sphere,” that have no very precise meaning if understood as a particular (“private”) class of objects of our options, are found to mean, more rigorously, the sum of an individual’s admissible actions. Their “area” or “sphere” is better defined, negatively, by what the rights of others, and tort law, leave over. And, from a liberal point of view the freedom of contracting away what is in one’s private sphere seems naturally included in the set of an individual’s admissible actions.

From this point of view it seems doubtful to envisage the Pareto principle as operating outside the “private sphere” of liberties and rights. The Pareto principle operates through the medium of liberties and rights, since individuals can only choose what they are, by virtue of their liberties and rights, free to choose.

This has some relevance for the real nature of the alleged conflict between P and L. Sen depicts it as one between the Pareto principle and “rights.” On a close look, it is a conflict between preserving some (any) liberty as dictated by L, and converting it into an obligation by selling others rights over it, as dictated by P, because the trade is mutually agreeable. But if L acts as an interdiction to trade certain liberties, can it be interpreted as “freedom to decide”?13 We think not. Still, even though it is arbitrary to refer to interdictions of trade as “protections of the freedom to decide” it may still be justified for some reason to interdict such trades. There can be indirect external effects of the trade of liberties that lead to Pareto-inferior results. That may hold true even with respect to such classical political liberties as “freedom of speech.” Even somebody who has no interest at all to make use of such liberties himself may have good reason to hope that others would make good use of them and thus may want to enforce an interdiction to trade away such liberties. On the level of constitutional choice individuals might therefore want to render inalienable certain of each other’s liberties and thus to restrict freedom of contract.

Of course, using traditional terminology one would speak of “inalienable rights” in this context. What is at issue here is not a mere quibble over words, though. It is rather the fundamental normative question whether the collectivity as a whole may or ought to interfere with the trade of liberties at all and if so in what way?

Forbidding certain contractual exchanges of liberties by making them inalienable is one thing; imposing trades on unwilling parties is another. The Pareto rule in the liberal paradox is dimly perceived by some as collective choice forcibly sacrificing liberties to get Pareto-improvements—the obverse of Rousseau’s “forcing people to be free.” P is thus confusedly interpreted as a social imperative to trade off a liberty “at a profit,” i.e., as an interdiction to preserve it. It is supposed to imply that “the guarantee of individual liberty [must be] revoked” (Sen 1976/1982a, 313).

This view seems quite strange indeed. For, if it were the case that a particular distribution of liberties and rights is an obstacle to Pareto-optimality, the obstacle would either be overcome by trade, i.e., voluntary conversions of some liberties into obligations (hence new rights for others) and voluntary interpersonal transfers of some existing rights, or not.14 If not, there must be obstacles stopping these mutually agreeable transactions. For all we know, there may be mutually acceptable means of removing such obstacles—we cannot prejudge that. But the means cannot possibly include the violation of “legal” liberties and rights, given that the parties would not want to be so violated—or so we may presume.

The freedom of contract is the engine of improving social states under “soft” social choice. A liberty can be contractually converted into an obligation, in exchange for value received (or to be received as of right). An employment contract, involving the conversion of certain liberties (to work or to play, to work for Jones or for Smith, etc.) into obligations to work as directed in exchange for rights to payments or other benefits, is a mundane example. More generally, one can regard every use of the freedom of contract as a renunciation or “consumption” of a liberty: for contracting parties, the acceptance of reciprocal obligations involves the abandonment of the pre-contract option they had to adopt a different course of action, a different commitment, a different allocation of their resources.

Of course, some liberties cannot advantageously be converted into obligations-cum-rights, because they have no exchange value. Many of Sen’s illustrations of “minimal liberalism” have this character: whether I read naughty books or not, sleep on my back or my belly, have pink walls or white, is not only (as he stresses) my strictly personal business, but (pace both Sen and his critics) it is difficult to see anyone else making it his business to the extent of compensating me for allowing it to become his business. Our reciprocal preferences simply do not make room for potential gains from trade. These liberties of mine may never be worth as much to anyone else as they are to me. They are destined to remain my liberties.

The preceding line of argument does not restrict “collective choice” or the state to a completely passive role as far as contracting is concerned. Where the structure of trade is not self-enforcing the question of contract enforcement typically arises. In particular one may ask whether and when the state should act as an enforcer of freely chosen contracts. This may be an issue of constitutional choice.

4.2.

ENFORCEMENT OF CONTRACTS

It is a commonplace that an unexecuted contract is a “game” of prisoners’ dilemma. If potential gains from trade fail to be realized (the contract is not concluded, or concluded but not executed), we may say that the game was solved in a Pareto-inferior manner. Consider the matrix below

Like every other potential contract, the interaction we are considering can be reduced, in a first approximation, to one of two ideal types. One is the non-cooperative game, where credible commitments are ruled out. In this setting dominated strategies should never be chosen and thus both players should use their non-dominated strategies. In a more psychological vein we could elaborate on this in the following way: Whether i chooses p or q, the dominant strategy of j is to choose r. Even if he offered to contribute s, a rational i would have to assume that j rationally will default and in fact do r. Given his correct perception of j’s best strategy, i has no hope of qs being “available,” hence no hope that he could bring that result about by his own contribution and thus no reason to contribute q to the joint result. He must opt for p if only for the “maximin” reason of escaping qr. The rational solution of this game is therefore pr, as in the simple one-shot prisoners’ dilemma.

The other ideal type is cooperative: i offers q conditional on j producing s. The equilibrium solution is qs (which will satisfy P), if the contract providing for i performing q, and j performing s, is binding, or rather believed to be so. Other things being equal, the latter will be the case if it is “enforceable.”

However, the binary alternative “commitments are/are not enforceable” is too crude even for a first approximation. A broad continuum of varying degrees of subjectively perceived credibility—in turn a function of enforceability—would serve better. But no continuum could be stretched to accommodate some of the cases that Sen puts in the foreground. How could Prude’s promise to read even the lewdest passages of the lewd book in the privacy of his study, or Jack’s promise to sleep on his back behind the closed door of his bedroom, be credible to any degree to someone who had to pay for this promise with a promise of his own?

Clearly, such undertakings cannot form either side of an arm’s-length transaction. They might be credible as between persons linked by ties of affection and trust; but then they would not normally take the form of trades, commitments fulfilled for a consideration. Promises to feel, to think, or to believe something, promises to perform unwitnessed acts leaving no trace, are worth no consideration, since it is impossible to monitor, prove, or disprove their performance; and where there is no consideration, there is no contract. Sen knows this perfectly well, and puts it beautifully when having the gentle policeman call on Prude to inquire about his reading the good book (Sen 1982b; 1986, 227–a), though it is the very raison d’être of such contracts, rather than their dubious or socially objectionable enforcement, that he should have questioned. Why, then, did he pose the conflict between keeping a liberty and selling it in a Pareto-improving contract, in terms of objects that simply cannot be contracted for?—so that the question of the Pareto-improving solution cannot even arise? L will then prevail every time, as there is no contest with P. “How do you sell your freedom of thought?” is not, in this context, a mere rhetorical question.

It is obvious here that it may be unnecessary to protect such liberties against being traded away. For those who want these liberties to prevail the best constitutional policy may simply be following a maxim of “hands off.” However, liberties and rights that enter into reciprocal preferences, and are sensible objects of arm’s-length exchanges, may pose a genuine problem. The question that we ought to pursue a little further is whether contracting should be facilitated or not by public enforcement.

The standard means of making the cooperative solution of the prisoners’ dilemma available to the parties is to refer to the historically accurate fact that in our type of civilization most contracts that suffer from no formal vices are enforced by the political authority. The effect of believing this is to stabilize the qs solution against the temptations of the default strategy that is dominant yet Pareto-inferior. Thus are people, so to speak, forced to be better off.

Can one, however, still describe the resulting qs solution as satisfying P? For it might be objected that qs is Pareto-optimal only if it is freely chosen, but not if it is weighed down by coercion (however latent); the two are not commensurate, nor is a freely chosen pr commensurate with the coerced qs. To defeat this objection, it would have to be argued that the coercion needed to transform qs into an available option is already allowed for in both individuals’ preference orderings. It is not qs they prefer to pr, but “qs cum coercion to deter default.”

Sen is anxious to establish (1986, 225–27) that the parties may not even wish to negotiate a contract (for qs) because their non-utility reasons in favor of preserving their relevant liberty outweigh the extra utility they would gain by converting it into an obligation. If utility is used in a narrow sense, that leaves room for non-utility reasons to induce choices; this is plainly something one is free to assume. The impossibility in that case is resolved by an assumption that makes L counter-preferentially stronger than P; the parties will conform to it, and the choice dictated by L will be the social choice. If, however, preference is to be taken broadly to encompass everything that influences choice, and “preferred” is to mean the choice waiting to be made if given the chance, counter-preferential choice is beyond the pale of theory; qs then yields a surplus of the entity, whether we call it utility or something else, that is supposed to motivate choice, and we are not free to assume that the parties have no wish to seek it.

This surplus yielded by contract performance can be indifferently identified as one of three things: it is the reward for bearing default risk, it is a resource available for arrangements to deter defaults, or it is a resource for buying insurance against it. Nothing permits us to assert and no good argument favors the supposition that insurance can only be bought from the political authority (which would justify its taxing power as an alternative way of collecting premiums), or that it will be bought at all. The economist would expect to find a tendency for the contracting party to be indifferent, with respect to his marginal contract, between carrying the risk and insuring it. He would also expect the mix between risks assumed and premia to be such as to help bring about this equilibrium.

Coping with default risk does not necessarily, or only, mean providing the wherewithal for an enforcement mechanism, whether a do-it-yourself or a bought-in variety. It may also mean modulating the very need for enforcement by adapting the terms of contracts to the desired level of risk. Half-executory contracts are, cet. par., riskier than either “spot” or fully executory, “forward” ones. Simultaneous performances, each fully contingent on the other, have a self-enforcing property. Refusing to enter into half-executory contracts with certain parties under certain circumstances is tantamount to paying for reduced default risk by forgoing uncertain gains. Avoiding to deal with unknown parties in cases where performance is hard to define and easy to contest is another obvious way of acting directly on the level of risk, rather than dealing with a given level of it. A multitude of adjustment, protective, and risk-avoidance devices, positive incentives for reputation-building in the reliable discharge of obligations, and the many informal extra-judicial sanctions of default, constitute a net that upholds contracts. It may be stronger or weaker, and more or less finely meshed. It is costly to knot and to maintain. Part of the cost is intangible if not altogether conjectural, since it consists of forgone advantages, missed dealings, and contracts entered into that would pass for sub-optimal in a world without default risk.

There is an obvious kinship between the costs that, if incurred, help enforce contracts, provide substitutes for enforcement, and mitigate the consequences of its inadequacies, and two other famous classes of costs: those incurred to secure property rights, i.e., “exclusion costs,” and those that are entailed in their transfer from less to more highly valued uses, i.e., “transactions costs.” All three classes are admittedly hard to define, elusive, all too often the result of imputation verging on tautology. They are, so to speak, obstacles that are invisible to the spectator, who only sees the horse that balks but not the fence that made it balk.

Unfortunately, however, the older, and supposedly better understood, pre-Coase and pre-Demsetz cost categories, such as production costs and transport costs, are similarly tainted by imputation and metaphysics. Yet, tainted or not, both science and life need concepts and categories of cost, and nothing more “objective” is likely to serve any better than the ones we have. The relatively new-fangled and somewhat shadowy triad of exclusion, transactions, and enforcement costs15 goes some way towards explaining why asset markets discriminate, some goods become public and others private, many negative externalities are tolerated, and why some ostensibly Pareto-superior moves do not take place.

A commonsense resolution of the alleged paradox of the Paretian liberal is implicit in these considerations and is ready to be read off. If a choice mechanism combines two contingently contradictory rules—as, in Sen’s construction, L interdicting the negotiation of rights and liberties, and P mandating them—a meta-rule can “socially” justify the individual choices that are necessarily made in violation of one rule or the other. It is hard to think of a more neutral, less discretionary meta-rule than the submission of possible rival outcomes, rival social states obeying rival rules, to the test of costs. Costs are grassroots arguments against an outcome. As near as one can tell, they determine whether the game of the Paretian liberal is solved by contract, or by the failure to contract. Both make perfect sense, given the “argument against.” This is, it seems, as it should be; for why should we expect a uniform issue?

5.

Concluding Remark

A right in Sen’s framework amounts to being in a position to choose at least between two cells of the matrix of a game form. Sen’s frequent claim, that his minimal liberalism as entitlement to choose between at least one pair of states of affairs is implied by such concepts as for instance Gibbard’s “issue liberalism,” is correct. But, as we have shown, it is incorrect that the entitlement to choose between classes of social states, i.e., having a liberty, has the same implication. Having a liberty does definitely not imply the right to choose between at least two social states (i.e., liberal individualism as reconstructed here does not imply minimal liberalism in Sen’s sense).

If this is true, the paradox of liberalism is no paradox at all. The impossibility results, though formally correct, do not capture the essence of liberal individualism since such a view of the world is based on a fundamental distinction between liberties and rights. Still, Sen’s arguments as well as the general discussion of the alleged paradox of liberalism raise important and interesting issues of inalienability of liberties, rights, and enforcement of contracts in a free society. Even though the first three sections of our paper were critical of Sen and even though in section 4 we outlined a vision of the mutually compatible roles of liberties, rights, and Paretian policies that quite contradicts Sen’s views, it is a great accomplishment of Sen’s to put these issues again where they belong: at center stage of modern welfare economics.

[* ]By Anthony de Jasay and Hartmut Kliemt. Reprinted with permission from Analyse und Kritik, 18 (1996): 126–47.

The authors owe a particular debt to James M. Buchanan for his detailed comments and constructive criticism. One of the authors has also benefited from discussing some of the issues raised here with Amartya Sen. Friedrich Breyer, who does not agree with the thrust of the paper, nevertheless sent us some useful suggestions for “corrections of errors.” The usual disclaimers apply with added force.

[1. ]We use “obligation” as the negative corollary of another’s right. It is owed to the right-holder. A “duty” is not necessarily owed to anyone; however, if I owe a duty to someone, I do not do so as a matter of his right. He may, of course, have a non-enforceable moral claim to it. It seems best to preserve a distinction between the consequences of legal claims (and call them obligations) and the commands of morals (and call them duties). It makes good sense to say it is your duty to fulfil your obligation. “You have an obligation to do your duty,” if it means anything, means something altogether different.

[2. ]We feel that Sugden 1985; 1993; 1994, and Gaertner, Pattanaik, and Suzumura 1992 are basically right when suggesting that game forms are the appropriate tool for analyzing the alleged liberal paradox. However, contrary to their views we think that the distinction between rows/columns and cells should be reflected in a terminological distinction between liberties and rights. Consequently, unlike the precedingly mentioned authors, we identify individuals’ strategy sets with liberties rather than with rights. This difference may seem merely terminological but in view of the fundamentally different roles of liberties and rights it is of some systematic importance, too, to make this distinction.

[3. ]As shall become clear below there can be at most one individual that could single-handedly choose among social states. If all other choices are made already, one individual can choose between social states by exercising his liberties.

[4. ]With respect to the issue of Pareto optimality, Thompson and Faith 1981 prove that changing the information conditions such that a hierarchy of decision rights leading to what they call “truly perfect information” emerges, implies Pareto efficiency in any game.

[5. ]Replying to Bernholz 1974, who protests the confusion between choosing entire social states and their individual “features,” Sen states: “Given the rest of the world, . . . Jack’s choice between sleeping on his back and . . . on his belly is a choice over two ‘social states.’” (Sen 1976; 1982a, 304; his italics). However, even if we grant that speaking of a choice of social states in a state of ignorance about what one is choosing is meaningful the argument that at most one individual can do what Sen assumes still applies. One should not confuse hypothetical considerations that treat the choices of others as given—in that sense all can simultaneously treat the choices of all others as hypothetically fixed—with the choices of all others actually being made and fixed. Sen’s concept of a right to choose assumes the latter rather than the former!

[6. ]We shall henceforth neglect the special case of a last mover who as a matter of fact is making the “last choice” in a sequence of choices. Obviously our basic argument that at most one individual can be in the position to choose between states of affairs would apply in that case as well.

[7. ]Ignoring the distinction between lower case and capital letters in examples like the foregoing ones may provide an answer to Gibbard’s query in 1982, 597f.: “These liberal paradoxes carry, with them, an air of sophistry: they must in some way be creating problems that do not really exist. . . . To talk about paradoxes, then, is to explore the role of one kind of mathematics in thought about social norms and organization. What is it about the mathematical apparatus of social choice theory that apparently so misapplies to questions of liberty?”

[8. ]Even if individual liberties were to be viewed as ultimately chosen in a collective act of constitutional choice they would be different from participatory voting rights and, for that matter, obligations to behave according to collective commands.

[9. ]Bringing the Pareto principle into play on top of such “rights” as Sen does amounts to pursuing the same aim in two different ways. And, from this point of view, it is not surprising that inconsistency emerges.

[10. ]As far as the latter enterprise is concerned game theoretic analyses like Breyer and Gardner 1980 that focus on Pareto-dominated equilibria in the presence of “rights” may be most fruitful.

[11. ]In any event, if we use the construction of special voting rights in the way proposed here, the choice set will not be empty and thus the paradox is avoided.

[12. ]We find no place in Sen where he would seek to define the area of privacy or “personal matter,” but his examples suggest that he sees it as fairly narrow. Yet this may be doing him an injustice: for his objective, of course, is to show that even a puny area cannot be spared by the invasive Pareto principle. But then a larger area can a fortiori not be spared.

[13. ]Jesuitically, we may say that an interdiction to trade preserves freedom, in that once you have traded an object away, you are no longer free to decide what should happen to it. It is possible (though we think unlikely) that Sen means his “minimal libertarianism” to be freedom-protecting in this sense: we are only free to choose until we do choose, and lose a liberty irrevocably if we choose irrevocably. While Sen’s own position on this has at least a certain casuistic merit, its more widespread popular interpretation, where liberties are supposed to be suppressed by force in the name of Pareto-optimality, seems incomprehensible. Thus, one of his critics thunders: “It is, then, undeniable [sic] that if we propose a criterion for a good state of affairs like Pareto-optimality, then farewell legal rights” (Barry 1986, 94).

[14. ]Herbert L. Hart, discussing legal powers that some scholars call “norms of competence,” quotes A. Ross’s observation in the latter’s On Law and Justice: “The norm of competence itself does not say that the competent person is obligated to exercise his competence” (Hart 1961, 238).

[15. ]In his “The Problem of Externality” Dahlman 1979, 217, treats enforcement costs as part of transactions costs, and attributes the same view to Coase 1960. He goes on to argue that enforcement costs, like every other transaction cost, are in reality information costs: “enforcement costs are incurred because there is lack of knowledge as to whether one (or both) of the parties involved in the agreement will violate his part of the bargain.” (218) This is circular reasoning. A party may keep his bargain if there is enforcement, and violate it if there is not. Apart from the ethical and historical ceteris paribus, the probability of violation is best captured as a decreasing function of the enforcement costs being incurred. To say that they would not have to be incurred if we knew that neither party was going to violate a bargain, is true enough but no less circular and no more helpful.