Manfred Tietzel and Christian Müller

One of the key aspects in James Buchanan’s work is the question of how far governmental action should legitimately reach. In this respect his ideas often stand in conflict with those of the orthodox public finance school of thought as, for instance, with Richard Musgrave’s. According to Buchanan’s classification, Musgrave represents the insider-Harvard vision of socio-political reality whereas he himself stands for the outsider-Chicago-Virginia-public choice school of thought (Buchanan, 1989, p. 291). Both scholars’ ideas concerning government intervention appear to be mutually exclusive (see also Hansjürgens, 1999). Buchanan emphasized that all governmental action should aim at fulfilling the interests of only the individuals concerned. He stressed ‘that in the conceptual derivation of the origins of the state ..., there is no resort to any source of value external to the expressed preferences of individuals who join together in political community.’ Consequently, he states that ‘the state does not exist as an organic entity independent of the individuals in the polity. The state does not act as such, and it cannot seek its own ends or objectives’ (Brennan/Buchanan, 1985, p. 22).

In sharp contrast to this Musgrave (1956/57; 1959; 1987; 1990), by means of his notion of ‘merit goods’, tried to justify even those kinds of governmental interventions which intervene into allegedly distorted individual preferences. In the case of so-called ‘(de-) merit goods,’ paternalistic governments try to induce or even enforce a certain market allocation of private or club goods for which demand is regarded as being ‘insufficiently low’ or ‘too high’. Here the state intervenes, for example, by (wholly or partly) subsidizing (or taxing) the market supply of goods (or bads) or by directly regulating prices to lower or higher than equilibrium levels. The prohibition of production, of marketing and of consumption of addictive drugs is an example of a ‘demerit good’; the provision of subsidized opera performances or of ‘free’ school lunches can be seen as examples of ‘merit goods’ which supposedly satisfy wants ‘society’ as a whole considers to be desirable. Although the merit concept as such has remained remarkably confused to this day, its undisputed core is that the ‘satisfaction of merit wants, by its very nature, involves interference with consumer preferences’ (Musgrave, 1959, pp. 13-14).

Notwithstanding these fundamental differences in the underlying normative approaches of Buchanan and Musgrave, it has been suggested that Musgrave’s concept of merit goods can be justified by means of a constitutional approach (see Head 1988; Priddat, 1993). The purpose of this paper is to examine the correctness of this assertion. First, we try to demonstrate the welfare effects connected with the provision of merit goods. In a second step we shall put forward a coherent definition of the notion ‘merit good’¹ which, surprisingly, is still under debate. We then proceed by analyzing the arguments usually put forward to justify their public provision and by confronting these arguments with the normative implications of the constitutional approach. We shall come to a negative answer in all key applications of the merit concept.

Figure (a) illustrates the welfare effects of the provision of merit goods. The abscissa shows the consumption units of a given commodity x_i, say, the number of visits of a museum. The government is assumed to consider this good to be one that satisfies ‘merit wants’². The ordinate represents units of another good x_j which can be thought of as a basket of goods or ‘all other commodities’. Let x_i and x_j be substitutable, and let x_j be not inferior. In the following partial analysis, the welfare effects of governmental intervention into the market allocation of good x_i will be analyzed. The income, substitution and welfare effects on x_j will be neglected.

The indifference curves I1 and I2 represent an individual’s preferences for both these goods at different budget constraints Z1 and Z2. Z2 prevails as a consequence of an increase in the relative price of good x_i. As a rational utility maximizer, the individual will demand the combinations A and B respectively, given the different budget sizes.

From that the individual’s demand curve D for x_i in figure (b) can be deduced. Given the supply curve S³, the household equilibrium is in G where it consumes x_i^G units at a price . The individual’s total expenses for good x_i amount to 0x_i^GG, and the individual’s (net) consumer surplus is GJ.

What are the effects of a government’s attempt to ‘correct’ this household’s preferences in order to induce an additional consumption of x_i^m - x_i^G?

Voluntarily, the household, at a price , will demand the quantity x_i^m which, however, is only supplied at a higher price . Thus, the overall expenses 0x_i^m for quantity x_i^m have to be subsidized with a total of F, since the household is only willing to pay 0x_i^mF.

Given the condition that the burden of financing this subsidy is not imposed on the consuming household but on the general taxpayer, the consumer surplus rises to FJ. As indicated in figure (a), subsidization results in a reduction of xi’s relative price. The budget line then revolves from Z₁ to Z3, thus leading to a new consumption optimum in C on a higher indifference curve I3.

We are now in the position to explain what is meant by the expression of a ‘correction’ of individual preferences by governmental intervention: the subsidy induces the household to choose a consumption pattern, as if it had preferences for goods x_i and x_j represented by indifference curves and . Suppose, these were the household’s true preferences. Its demand curve for x_i, then, would be Dm, and it would voluntarily consume quantity x_i^m at a price . This situation is represented by point C' in figure (a) where the budget constraint Z2 is tangent to the indifference curve . Having such a preference structure, the household would receive the same consumer surplus as in the case of a paternalistic intervention (FJ = R), but it would have to bear the full cost of 0x_i^m. As a result, the subsidized supply of x_i^m will cause an inefficient allocation of goods. On the one hand, the household’s consumer surplus rises by FG; on the other hand, however, the subsidy F exceeds this gain in consumer surplus by FG. In their role as taxpayers the economic subjects stand to lose more than what they gain as subsidized consumers.

An individual household will readily agree with subsidizing the supply of x_i, as long as its own share of the tax burden does not exceed its willingness to pay 0x_i^mF, with the rest of the whole cost being borne by the other taxpayers. The household thus attains a higher utility level on indifference curve I3. The subsidy lowers the price that it has to pay for x_i from to , and the budget line shifts from Z2 to Z3. Meritorious policies are easily enforceable if the tax burden necessary to finance the subsidy spreads over a large number of taxpayers and is individually imperceptible.

The situation is quite different, though, if the household is forced to consume quantity x_i^m at its full price. One might think of compulsory school attendance, the full costs of which are borne by the students themselves or by their parents. In this case, the household is led to allocate its budget on x_i and x_j in a way as represented by point C’. This, however, does not only pin down the individual on a utility level as indicated by I4, which is lower than the level attainable with his budget, but also induces him to choose an inefficient consumption bundle. At the same relative prices, the individual could realize the same utility level in C’’ by spending a smaller amount of money. The gain in consumer surplus from x_i, KFG, is smaller than the utility loss incurred by the additional expenses x_i^Gx_i^mG.

In such a situation, the affected households are most likely to resort to resistance and evasion strategies. When, for example, at the end of the 18th century school attendance was made mandatory in Prussia, teachers in the countryside complained about high rates of student absenteeism during seed and harvest times. From the point of view of families with children required to attend school, the opportunity costs of school attendance raised so dramatically during those times that they rather put up with a sanction than to do without their children’s help.

Given the preceding analysis, we propose to refer to only those governmental actions as ‘meritorious’ which induce an additional consumption of a good, be it by subsidy or by coercion. Individuals, then, act as if they had preferences different from those that they would reveal in an undistorted market. Where merit policies are at work, a government interferes into the market allocation of a good by inducing a consumption pattern different from the one voluntary exchange would have brought about. That is to say that the quantity demanded is no longer solely determined by the individuals’ true preference orders and incomes. Rather, it is chosen according to a fictitious ‘political’ preference order, consumers ‘ought’ to have in the government’s opinion. Market allocation is considered to be ‘too exclusive’ for merit goods and ‘too inclusive’ for demerit goods.

Given our definition, not only those goods the consumption of which is to be increased by subsidizing market prices are to be classified as merit goods as, for instance, is often the case with theatre tickets or low-cost housing (merit goods in a narrow sense). Our notion of merit goods also includes goods consumed compulsorily. Seat belts and protective helmets are well-known examples.

Analogously, ‘demerit goods’ are those the consumption of which should be lowered or kept at low levels according to some regulator’s political preferences. Excise taxation of tobacco or alcohol are pertinent examples. As we see it, the notion of ‘demerit goods’ also comprises goods which are subject to a general prohibition of consumption as, for instance, certain habit-forming drugs.

There can be no doubt that in reality politicians frequently attempt to alter allegedly distorted individual preferences in order to achieve some notion of social welfare. In societies based on less-than-unanimity decision rules, preference correcting ‘merit good’ policies are widespread phenomena. The mere recognition of this fact, though, does not imply that these policies can be justified from an individualistic perspective. This requires a normative justification of this particular type of governmental action.

Two question have to be clearly separated in this context: First, it has to be asked if, for the merit good issue in question, interventions into consumer preferences can be justified by reference to some ‘defect’ in the individual’s decision itself (condition 1). Most of the literature on merit good issues confines itself to this question. Given some justification of intervention, the means by which such an intervention should be carried out requires closer attention and, related to this, by which actor (condition 2). In the following two sections we first examine, whether meritorious interventions meet conditions 1 and 2. Second, we compare the merit good argument to a Buchanan-type constitutional justification of governmental action which, as will be seen, is much more appropriate for public policy evaluation.

a) The concept of merit goods as lauched by Musgrave is confined to the first question, namely the claim that individual decision-making is defective. The normative presuppositions put forward in favor of meritorious policies are, however, all but clear. On the one hand, Musgrave (1959, p. 14) takes on the ‘basic doctrine of consumer sovereignty’ as a starting point. This norm may be regarded as one instant of normative individualism according to which the individual is the best arbiter of his own needs (see Hamlin’s 1990 illuminating discussion of this principle). As James Buchanan (1986, p. 249) puts it, this principle ‘locates sources of value exclusively in individuals.’ Whatever the individuals concerned decide has to be taken as legitimate in a given public policy issue; value judgements ‘externally’ imposed on them do not bear any normative relevance (see, for example, Buchanan, 1977, p. 142).

On the other hand, according to Musgrave’s repeated assertion, the merit concept is meant to achieve more than an individualistic justification for governmental action. Musgrave concedes that a ‘position of extreme individualism could demand that all merit wants be disallowed, but this is not a sensible view’ (Musgrave, 1959, p. 13). In one of his more recent publications on the subject, Musgrave (1987, p. 452) stresses the fact that an ‘evaluation of a good (its merit or demerit) derives not simply from the norm of consumer sovereignty but involves an alternative norm.’ The postulate of normative individualism is the core of standard normative economics. Therefore, deviations from that premise have to be justified thoroughly. What can be a sufficient reason for forcing an individual with preferences represented by indifference curves I1 to I4 to behave as if his preference ordering was represented by indifference curves to instead? Musgrave neither puts forward any such sufficient reason nor describes an ‘alternative norm’ which could serve as an individualistic justification for such a kind of policy. For some, though by no means all, kinds of merit goods issues, he refers to what he calls ‘community values’ allegedly held by a society as such (Musgrave 1987, p. 452; 1990, pp. 208-209); in other examples his quotations of Rawls or Harsanyi, for instance, seem to indicate that it is a Kantian norm of universality which he has in mind (see Musgrave 1987, p. 453; 1990, pp. 208 and 210).

b) As it turns out, the concept of merit goods is of little help to determine circumstances under which an overruling of individual preferences can be justified (condition 1). In addition to this, the approach does not offer any guidance as to what kind of regulation can count as meritorious and who should be the regulator (condition 2). No way is offered to select among alternative means of intervention such as provision of information, subsidization or compulsion. Analogously the demand for demerit goods can as well be reduced by means of sales or production taxes as by prohibitions of their production, trade or use. This may well be taken for an invitation to abuse governmental power, as Musgrave himself concedes.⁴ Since no restrictions on governmental action are offered in the concept of merit goods, even a tyranny of a democratic majority cannot be ruled out as a possibility.

a) One possible way of reconciling meritorious policies with normative individualism might be seen in an extension of the range of the individualistic postulate (see Brennan/Lomasky, 1983; Head, 1988; Walsh, 1990). From the point of view of this wider interpretation, narrow individualism, which is implied by Musgrave’s approach, is equivalent to consumer sovereignty. The set to of indifference curves, then, necessarily represents ‘external’, non-individualistic values which should overrule individual preferences.

Although it is recognized that interventions on the background of this more general kind of normative individualism are irreconcilable with consumer sovereignty, the opinion is being held nonetheless that such interventions are compatible with a position which is individualistic in the wider sense. to , under these auspices, do not necessarily reflect ‘external’ values, but, under certain circumstances, may be regarded as an individual’s own ‘enlightened’ preferences. Unless an individual’s process of decision making were in some way defective, he would, in the light of these preferences, agree with the meritorious intervention.

In the last analysis this argument equals a constitutional interpretation of merit wants. In constitutional economics the outcome of a collective decision making process is regarded as fair or justified if a hypothetical consensus of all members of the decision making body can be assumed. This assumption is met if none has to bear (avoidable) costs imposed upon him by others without an adequate compensation. Otherwise the person affected would have a reason to veto the decision. This is equivalent to saying that a consensus on a possible Pareto improvement is conceivable under the condition that all negative externalities are internalized. In a situation like the well-known prisoners’ dilemma, all individuals could gain in ‘generalizing’ their own behavior and in agreeing on a mutual contract in which everyone promises to abstain from free riding. The assumption of a ‘veil of ignorance/uncertainty’ is a well-known metaphor for this ‘generality principle’ (see Buchanan/Congleton, 1998). According to this idealized premise of contractarianism, in order to reach impartial and unanimous decisions, one should imaginatively leap into a constitutional decision-making situation in which everyone is ignorant of his future place in society and of his class or status. By denial of all natural and social differences, every individual, by assumption, is able to recognize his ‘constitutional interest’ (Vanberg/Buchanan, 1988) in a mutual internalization of all external costs.

A key condition for a viable social contract, however, is that everyone abides by his constitutional promises. Constitutional economists regard it as a legitimate role of government to enforce contractual reciprocity. In their view, an intervention is individually justified if ‘constitutional’ interests are properly represented by indifference curves to . Preference-correcting merit policies, though irreconcilable with consumer sovereignty, may meet the criterion of an extended normative individualism if the individuals concerned can be expected to agree according to their constitutional interests.

Whereas constitutional economics usually deals with problems of interindividual competition for scarce resources, attempts have recently been made to apply the approach to what has been dubbed a ‘constitutional economics of temptations’ (see Koboldt, 1995) which Buchanan (1990, pp. 3-4) called for some years ago. This branch of research, sometimes labeled as ‘egonomics’ (Schelling, 1978a), focuses on addiction phenomena and other kinds of temptation situations. This kind of decision-making defects is sometimes modeled as ‘intrapersonal prisoners’ dilemmas’ (see Elster, 1985, pp. 254-55; Kavka, 1991, 1993; Moreh, 1993). In such situations, by acts of ‘self-paternalism’, individuals may wish to commit themselves to certain restrictions on their own behavior. ‘Personal constitutions’ (Koboldt, 1995, p. 15), as which these rules can be thought of, are intended to prevent them from making myopic and ill-considered decisions. A well-known illustrative case for such ‘weakness of will’ phenomena is that of Ulysses who once let himself be bound to the mast of his ship in order to be able to withstand the Sirens’ singing (see Elster, 1979).

This view suggests that a constitutional ‘as if’-test of rules justification is an evaluative criterion of governmental intervention preferable to the concept of merit goods. Moreover, the constitutional approach provides a unified framework for evaluating from an individualistic perspective collective decisions of kinds as different as the provision of public goods, the redistribution of wealth or the shaping of seemingly irrational behavior. However, whenever a constitutional justification can be given, the concept of merit goods is simply redundant.

But where does this argument leave us? Is it correct, as Head⁵ seemingly suggests, that a constitutional justification can be given for any kind of a policy which Musgrave takes to be ‘meritorious’? We do not share this opinion. In what follows, we shall argue that a constitutional justification cannot be derived for each and every instance of a merit good or a demerit bad, that has been put forward. As we see it, a contractarian justification is not some one-way means of legitimization of governmental action but also one of its limitation. At the fictitious constitutional level, rational individuals behind a veil of ignorance/uncertainty would more probably agree to assign certain ‘insurance tasks’ to the government: They would solve the old Hobbesian problem of social order by founding a state in order to insure themselves against assaults on their lives and possessions. Moreover, they would be disposed to establish democratic checks and balances in order to protect themselves from political despotism (see Overbye, 1996). And since no individual can know in advance whether he will be poor at the post-constitutional level, each of them would rationally vote for the provision of some minimum level of social security. Quite in the same vein and already two centuries ago, the Hobbesian-spirited ‘insurance theory of taxation’ considered the state as a mere ‘insurance company for political and social accidents’ and regarded tax shares as periodically collected ‘insurance premiums’ necessary to finance its activities (see Lindahl, 1919, chap. 2, § 1, and Mann, 1937, pp. 105 and 214, for an overview).

The assignment of the role of an insurance to the state, however, does not confer an unlimited license on politicians to intervene into individual decision-making for whatever ‘insurance motives’. To every right there is a corresponding obligation which generates costs for the ones obliged. Behind a constitutional veil of uncertainty no one can predict whether, at the post-constitutional level, he will be a golfer in his lifetime. Yet this does not imply that, from a constitutional perspective, government should provide anyone with golf rackets and caddies. In other words, the constitutional insurance analogy can only justify a minimum standard of social security (see Murphy, 1977, pp. 240-41). The definition of ‘rights’ will become meaningless if society, for reasons of equality before the law (e.g. if swimmers claim the same rights as golfers do, and tennis players insist on being treated in the same way as the swimmers are), simply cannot afford to enact them. In section 3 of this chapter we shall discuss a couple of illustrative examples which, we believe, show that most of Musgrave’s examples of merit goods simply exceed these justifiable (and affordable) minimum standards.

b) In addition to this, we intend to show that the constitutional argument developed so far is a necessary, yet not a sufficient condition for governmental intervention. What is to be demonstrated is that, once a necessity to intervene has been established (condition 1), an individualistic theory has to indicate which kind of governmental action can be seen as an appropriate means of enforcing the legitimate intervention, and, by implication, who ought to be entrusted with the enforcement (condition 2). To state a need for intervention does not imply the ability of the state to do so.

Usually, in constitutional economics as well as in the theory of merit goods, condition 2 is passed over in silence. In a prisoner’s dilemma game, for example, to which constitutional economists frequently refer as a model of a situation of constitutional choice, there is one unique situation Pareto-superior to the equilibrium. In reality, however, there exists almost always a whole menu of equivalent rules that could make some persons better off without making others worse off. Imagine a number of citizens who, behind a veil of ignorance/uncertainty, share an interest in some minimum degree of redistribution in the post-constitutional society. There are many different alternative ways in which they could realize a Pareto-move as, for instance, the establishment of a governmental insurance monopoly, the introduction of a compulsory insurance or the institutionalization of in-kind-redistributions.

As it turns out, condition 1 focuses on a ‘conflict of interests’ (see Vanberg/Buchanan, 1989) in constitutional choice the solution of which presupposes a constitutional interest in leaving the initial state of nature common to all individuals. In contrast to this, condition 2 requires the solution of a ‘theory conflict’, i.e. to find an answer to the question of how, and also by whom, the legitimate intervention should be carried out.

We believe that normative individualism gives clear answers to both questions: Insistence upon respecting individual freedom of choice is obviously a consequence of a postulate one might label as the principle of minimal coercion. To put it in an operational way, this principle reads: ‘As much coercion as necessary, yet as little coercion as possible!’ The variable to be maximized is ‘freedom’ which is achieved if there is a choice among ‘genuine’ alternatives (see Vanberg, 1982, pp. 56, de Jasay, 1991, pp. 17). It would be grossly misleading to call a choice a ‘free’ one if the utility difference between the first best and the second best alternative is maximal. Both options, then, reasonably cannot be regarded as ‘genuine’ alternatives. A thief who imposes on his victim a ‘choice’ between ‘Your money or your life!’ factually does not leave him a choice. It is only among ‘positive’ alternatives that one can choose freely. A situation can be regarded as one in which one is not just ‘free to choose’ but also ‘choosing freely’, the greater the number of alternatives among which can be chosen and the smaller the utility differences between the first best and the second best options.

Minimization of governmental coercion requires that, from a set of at least two alternative interventions, which could equally serve to heal a given decision-making defect, one should choose the one which least manipulates the number of alternatives and the resulting individual net welfare. This implies, for instance, that problems which can be solved successfully by private action cannot be solved by governmental intervention in a legitimate way since the latter necessarily involves compulsion.

As far as the concept of merit goods is concerned Musgrave was reluctant to come forth with a thoroughgoing justification. He rested content with an enumeration of cases which he regarded as illustrative examples of justified government interventions. As a first set of cases he considered what he called ‘pathological cases’ of irrationality and rational ignorance (Musgrave, 1987, p. 452) From a constitutional perspective, these examples can be seen as uncontroversial (see, for a more detailed constitutional analysis, Tietzel/Müller, 1998, pp. 105-109) and will, therefore, be neglected in what follows. Three further types of meritorious interventions, which Musgrave regards as crucial for his approach, deserve closer inspection. These types consist, first, in redistribution in kind, second, in so-called ‘community preferences’ and, third, in situations in which individuals suffer from a ‘weakness of will’ (Musgrave, 1987, pp. 452-53; Musgrave/Musgrave, 1989, pp. 56-57).

According to Musgrave a typical case of legitimate merit good policies is that of giving in kind:

‘An individual donor may choose to give in kind rather than in cash, because he or she considers certain uses by the recipient as meritorious. Or taxpayers may prefer social programs which provide in-kind aid, such as food stamps, or low-cost housing, over cash grants.’ (Musgrave/Musgrave, 1989, p. 57)

Musgrave suggests that a society may modify the primary market distribution of income by means of monetary tax-transfer schemes but also by providing everybody with a minimum endowment of in-kind transfers. ‘Goods separated out for non-market distribution might then be viewed as merit goods’ (Musgrave, 1987, p. 453).

In the following we shall take a closer look at two cases which fit this pattern and which, we believe, shed light on the problems involved in Musgrave’s argument. Suppose, first, that welfare aid is provided by in-kind units rather than by the more common monetary transfers. Musgrave (1987, p. 452) compares the role of the state to one of a benevolent donor who has a right to freely choose the way in which he offers his ‘gifts’. Yet, it is questionable whether in-kind transfers can also be considered legitimate from an individualistic perspective. Head (1988, p. 36), in his quasi-constitutional approach of multiple preferences, answers this question in the affirmative. We believe that this judgement is untenable.

As a contract theorist one would hardly oppose a redistribution by means of governmental coercion insofar as it is backed by constitutional ‘insurance motives’. One could, for this reason, regard in-kind transfers as fulfilling condition 1. Condition 2, however, is certainly violated by in-kind transfers. One reason for this is that Musgrave’s comparison of governmental redistribution to private gifts is mistaken. On the one hand, the donor does not give voluntarily in the usual sense of the word. If redistribution from the rich to the poor is in everyone’s constitutional interst, then the state as a donor has an obligation to provide the transfer and cannot evade it legitimately. On the other hand, the recipients have a right to receive the transfer. The objection raised by Andel (1984, p. 644) that the recipient could easily refuse the redistribution offer, should he dislike it, is invalid. This is because redistributive transfers are not an offer in the usual sense. In common usage, an offer is the supply of a good or service without an obligation. In contrast to this, the donor in this example has a constitutional obligation to make the ‘offer’; its rejection by the recipient, therefore, would amount to giving up a legal right. Moreover, the case under consideration is an instance of the aforementioned decisions, in which one does not choose freely. To reject a minimum in-kind-transfer of goods, which is just sufficient to survive, implies maximum opportunity costs. Owing to the large utility difference between accepting and refusing the offer, both options cannot count as ‘genuine’ alternatives. In the extreme, the ‘choice’ between acceptance or rejection of, say, food transfers can amount to a ‘choice’ between life and death.

As a second argument, in-kind redistributions are more costly than monetary payments. Food and clothing have to be bought, shipped, stored and distributed; in case of monetary transfers, mere money payments are sufficient. If, on the one hand, the in-kind-transfers are not to be reduced in the amount of this cost difference, the donors have to bear an equivalent increase in taxation. Compulsory contributions restrict a donor’s choice opportunities more than necessary; therefore, given the quest for minimal coercion, monetary transfers are clearly preferable to those in kind. If, alternatively, the level of transfers is to be cut back the recipients’ choice opportunities are restricted. In-kind transfers and monetary payments could no longer count as equivalent alternatives. Monetary transfers represent an insurance benefit which, at equal cost to the donors, is superior.

What is more, the recipients’ needs and tastes will normally differ. Given any pattern of in-kind distributions, there will probably be potential Pareto-moves attainable by exchange. As a consequence, in-kind transfers will usually be inefficient. Given constitutional commitment power, the decision-makers at the constitutional level would, therefore, rationally reject them. In addition to this, exchange is not costless. Such transaction costs reduce the choice opportunities of the recipients. Again, money payments are constitutionally preferable to in-kind redistributive programs. In order to avoid such a result, one might think of prohibiting the sale of in-kind transfers received. This, however, comes only at the avoidable cost of an increased degree of governmental coercion.

In sum, welfare aid by means of in-kind redistribution programs cannot be justified by way of contractarian reasoning. Can one conceive of any other sort of in-kind transfers, say, subsidizing opera performances, which do not suffer from the problems mentioned? Here, the lack of a constitutional legitimization turns out to be even more evident. Opera subsidies do not even take the hurdle of condition 1. Nobody would consider opera visits as being part of an essential minimum for life which alone is justifiable by contractarian reasoning. But even if we went so far as to postulate a general constitutional interest in opera visits, subsidization of music theatres would fail to meet condition 2. For an alternative is conceivable which involves less coercion than subsidizing opera performances, namely direct income transfers to the potential opera visitors. It may go the grain of paternalistic politicians, that direct monetary payments to the consumers leave it to the subsidized individuals themselves whether or not to visit an opera performance or to spend the money for different kinds of amusements. The contractarian principle of minimal coercion, however, does not allow for any other solution.

Other than in the preceding discussion, the indifference curves to depicted in the figure shall now be taken to stand for so-called ‘community preferences’. Musgrave regards such allegedly communal wants as another essential application of the merit concept. Regrettably, however, what is exactly denoted by the term ‘community preferences’ remains rather diffuse. He defines a ‘community value’ as ‘an interest which is attributable to the community as a whole and which does not involve a "mere" addition, horizontal or vertical, of individual interests’ (Musgrave/Musgrave, 1989, pp. 56-57). On the other hand, and somewhat surprisingly, Musgrave (1987, p. 452) explicitly excludes any relationship to some ‘organic’ theory of the state, an intuition that suggests itself.

According to Musgrave

‘... individuals, as members of the community, accept certain community values or preferences, even though their personal preferences might differ. Concern for maintenance of historical sites, respect for national holidays, regard for environment or for learning and the arts are cases in point. Such acceptance in turn may affect one’s choice of private goods or lead to budgetary support of public goods even though own preferences speak otherwise. By the same token, society may come to reject or penalize certain activities or products which are regarded as demerit goods. Restriction of drug use or of prostitution as offences to human dignity (quite apart from potentially costly externalities) may be seen to fit that pattern. Community values are thus taken to give rise to merit or demerit goods. ... Without resorting to the notion of an ‘organic community’, common values may be taken to reflect the outcome of a historical process of interaction among individuals, leading to the formation of common values or preferences which are transmitted thereafter’ (Musgrave, 1987, p. 452).

At closer inspection, it turns out that Musgrave mixes up several different aspects in this quotation. First, in his discussion of historical sites or regard for the environment he refers to pure public goods. Let us take the protection of historical monuments as an example. Behind a veil of ignorance, tearing down a historical building in order to replace it by a modern, more functional one, cannot be justified. In the Rawlsian variant of the veil-of-ignorance assumption, the ‘persons in the original position have no information as to which generation they belong’ (Rawls, 1971, p. 137). Since the monument cannot be reproduced, its demolition would amount to an illegitimate overuse by one generation at the expense of those to follow. In addition to this, protection of historical monuments has the usual properties of a public good. It makes no difference whether the property rights to the historical building are held privately. If a certain historical townscape is to be conserved for the future, private owners of the monuments to be protected are captured in a prisoners’ dilemma-type situation: On the one hand, constraints on the use of historical buildings are in everyone’s constitutional interest; on the other hand, every owner of a house can gain by building a modern and functional house. In equilibrium, general destruction of historical sites will prevail. As far as public goods are concerned, there is no need to resort to the imprecise notion of ‘community values’. Maintenance of historical sites is implied by the conventional public good paradigm.

Secondly, when Musgrave considers support for the arts as a merit good, the provision of which can be justified by ‘community values’, he refers to one special type of ‘redistribution in kind’. In that respect, what has been said above may be sufficient. Another aspect allegedly covered by ‘community values’, the use of drugs, will be treated in the next section in which problems caused by ‘weakness of will’ phenomena will be discussed.

The only one among the cases subsumed under the category of ‘community values’, which cannot be readily analyzed in terms of other theories, is the case of prostitution. Musgrave argues that prostitution can be conceived of as an ‘offence to human dignity (quite apart from potentially costly externalities)’ and, therefore, can be declared as ‘demeritorious’. Without a doubt, cases like this one are the classical domain of paternalistic government action. A majority of citizens makes use of the legal institutions of a non-unanimous democracy to impose their personal opinions of ‘good life’ on others. Musgrave’s list could be easily extended. Prohibitions of homosexual actions or of pornography – even if only adults are voluntarily involved and no third party is affected – fit this pattern.

It takes, indeed, an excursion into the metaphysics of ‘community values’ to justify interventions into the preferences of autonomous individuals. An individualistic-constitutional justification is not available, even if such policies are pursued with the best of intentions. Since in the above examples the set of indifference curves to can in no way be interpreted as reflecting the ‘enlightened’ preferences of the participants themselves, a constitutional justification would already fail to fulfil condition 1. Admittedly, individuals concerned often tolerate without a protest that the state decides in their place. One of the reasons for this may be found in the collective goods properties such a protest would have. Under certain circumstances even the voters themselves can be expected to vote against their own preferences and for paternalistic collective action. One reason for this could be the existence of a ‘veil of insignificance’ (Kliemt, 1986) which results from the lack of decisiveness from which an individual suffers in large-numbers elections (see Brennan/Lomasky, 1983; Head, 1988, pp. 30; Brennan, 1990). Whatever the reason may be, failure to object to governmental acts of paternalism is by no means equivalent to its unanimous acceptance.

As an interim result we can conclude that a constitutional legitimization can hardly be established, not even for those types of government actions which, according to Musgrave (1987, p. 453), go ‘to the heart of the merit concept’. ‘Redistribution in kind’ generally fails to meet the contractarian criterion of legitimacy. The same is true for the examples of interventions which supposedly are backed by ‘community values’. It remains to be seen if merit policies are tenable at least in the case of ‘weakness of will’ phenomena which Musgrave offers as another potential source of merit legitimization.

There is a multitude of situations in which people succumb to a temptation without taking into account their long-term ‘enlightened’ interests. Preference theory models this problem of a potential ‘weakness of will’ on the assumption that a person may have multiple sets of preferences that may be in conflict with each other. Given these divergent preference orders, very different actions may appear to be optimal.

Thaler and Shefrin (1981) model a conflict between short-run and long-run preferences by distinguishing a two-step hierarchy of given preference systems within a single person. An individual is thought of as consisting of a ‘planner’ and a ‘doer’. The planner is farsighted and represents the person’s ‘enlightened’ long-run interest; the doer, in contrast, is completely selfish and myopic. The planner within a motorcyclist, for instance, may wish to wear a helmet during every ride. This, on the other hand, bothers the doer who prefers to do without it. Each drug addict wishes to get ‘clean’; in the short run, however, when the symptoms of drug withdrawal become overwhelming, he will nonetheless succumb to his addiction. By the same token, an individual may be aware of the advantages of appropriate provisions for his old age; as long as the person is young his short-term preferences may tell him to make different expenses that appear more pressing at that point in time. As a consequence, necessary investments will tend to be suboptimal or, in the extreme case, not carried out at all.

In all these cases a person acts as if he had two sets of preferences, I1 to I4 and to , only one of which is in charge at a single point in time. In other words, the welfare effects of a single action, such as putting on a helmet or consuming drugs, may be very different at different points in time. The person in question, then, has a ‘problem of self-command, or self-management’ (Schelling, 1984, p. 87) which ‘consists in the fact that he does not act in accordance with his constitutional interests (defined over the whole range of decision sequences)’ (Koboldt, 1995, p. 13). What appears to be utility-maximizing consumption in the short run, is inferior in the long run.

Therefore, the farsighted planner may wish to exercise control over the doer’s choices. By way of ‘self paternalism’, a person, according to the planner’s intentions, could choose to impose rules on himself that limit the set of choices the doer faces. In contrast to the institutions considered so far, these rules consist of constraints that are to prevent the person from self-damages (‘intrapersonal external costs’). If a government enforces these constraints by coercive measures, according to Musgrave (1987, p. 453), the notion of merit policy applies.

One could imagine that governmental intervention brings about just the decision required by the planner’s enlightened self-interest. For instance, the wearing of helmets could be made mandatory or drug consumption could be prohibited. Then, just as a benevolent, omniscient psychiatrist, the government would simply help people to protect themselves against a temporary lack of willpower.⁶

According to Head (1988, pp. 15-16), the government can legitimately use taxes, subsidies and any other regulatory or coercive measures in order to solve such individual self-management problems. On the basis of the constitutional approach put forward here, one will not deny the existence of a constitutional interest in the control of self-management problems. Therefore, our first condition can be assumed to be satisfied. This is simply due to the definition of long-term preferences as a person‘s own interests. As far as the policy recommendations are concerned, which may be derived from the diagnosis of a defect in an individual decision, we are much more skeptical than Head is. If coercion is to be minimized, additional information on the dangerous consequences of one’s actions should be preferred to government interventions in order to help a person overcome the internal prisoners’ dilemma in which he is trapped. If, for instance, friends or relatives repeatedly fail to persuade a motorcyclist to wear a helmet, he should be assumed to have good reasons for not doing so; from an individualistic point of view, these reasons have to be accepted unconditionally. Occasionally, the objection is voiced that, under such a laisser-faire regime, other policy holders of the driver’s health insurance would have to bear the financial consequences of a potential injury. However, this objection is not well-founded. The insurance companies concerned could simply exempt such cases from their liability. An analogous externality argument could be used in order to prohibit motorcycling, driving a car or cycling entirely. In extreme form, such an argument would immediately lead into totalitarianism.

Given the discretionary power with which the merit concept vests the state, it does not come as a surprise that similar risky activities with a high risk of danger are often regulated in very different ways. In many countries, for example, motorcyclists have to wear helmets even during very short and safe rides; at the same time, mountain climbers, despite the extreme risks they often face, are not subjected to any similar rules of conduct.⁷ By the same token, one might ask why governments usually content themselves with information campaigns in order to prevent the spreading of a disease as serious as AIDS although the risks of unprotected sex may well be as harmful as those of motorcycling without helmets. From such a point of view, the compulsory use of condoms is as legitimate as the compulsory use of protective helmets. The merit concept is unable to exclude absurd consequences such as these since it does not provide clear-cut normative criteria for choosing among alternative measures to re-shape preferences.

As far as demerit policies are concerned, government intervention faces additional problems. As a rule, general information on potential injuries is entirely sufficient in order to declare the use of head protection helmets or condoms as being ‘meritorious’. In contrast to this, in order to counter addiction phenomena properly, a government needs detailed information on which persons are addicted to what extent to alcohol or drugs. Even in one’s own circle of acquaintances it is often difficult enough to discover that a close friend or colleague is suffering from alcoholism or drug addiction. Diagnosing a lack of willpower in unknown and anonymous persons from a distance simply amounts to a ‘pretence of knowledge’. Friends and relatives are certainly equipped with more information and have a motivation than some government official to persuade a person with multiple sets of preferences to act in accordance with his own long-run interests. On top of this, discriminatory government measures will hardly be reconcilable with the fundamental principle of equal treatment before the law.

Therefore, the only way to solve weakness-of-will phenomena by collective action is to apply demerit policies in a uniform, non-discriminatory way. One might, for instance, think of introducing an excise tax affecting equally all consumers of a demerit good or of prohibiting the consumption of the good entirely. However, such interventions are necessarily inaccurate since it is a characteristic feature of addiction phenomena that only a small fraction of a population is seriously concerned. To tax or to punish the consumption of a drug does not just affect addicts as addressees. As a negative externality, it also restricts the freedom to choose of individuals who are able to deal reasonably with the demerit good.

A demerit tax on the consumption of a demerit good should ideally be designed as a kind of ‘intrapersonal Pigou-tax’ in order to internalize the ‘intrapersonal external costs’ involved. However, such excise taxes face about the same problems as environmental taxes do: They operate accurately only in a few exceptional cases. To individuals who are not addicted and who wish to consume the demerit good in a rational way, governmental sanctioning amounts to an illegitimate compulsion. Addicts, in contrast, will (and can) hardly be deterred from consumption by governmental intervention; for them, the tax or (the expected costs of) a punishment will always be too low. Both kinds of sanctions will serve their purposes poorly. The low elasticity of demand for socially accepted drugs such as tobacco or alcohol is a temptation for revenue-maximizing politicians to tax these goods heavily. From a contractarian perspective, however, the taxation of goods with a low elasticity of demand is to be rejected (see Brennan/Buchanan, 1980, chap. 4).

As high prices do not deter addicts from drug consumption, a higher rate of crimes such as smuggling or stealing are to be expected as unintended side effects of taxation or prohibition.⁸ Minimal coercion requires that government should not intervene without further preconditions.

If the degree of coercion is to be minimized governmental intervention cannot take on legitimacy as far as individual self-management problems are concerned. For, here, the weakness of will is only a temporary phenomenon. In periods when his long-run preferences prevail, the individual concerned is able to take precautions himself which commit his own behavior at future moments in time to the preferences which are in charge at the moment the rules are chosen. In order to provide for his old age, for example, an individual could sign a savings agreement or take out a life insurance policy the untimely termination of which would carry with it a penalty for breach of contract. In the market place, such a commitment device can even be acquired for free: One promises to pay an amount of money in case one undertakes or refrains from undertaking a pre-specified action; the other party to the contract will agree if the expected value of the money to be paid exceeds his transaction costs. If, nevertheless, one believes, as we do, that governmental retirement insurance is necessary, one has to give reasons other than a potential lack of willpower.⁹

A similar argument holds in the case of drug addiction. Nobody is in a permanent state of euphoria. In phases of sobriety the individual concerned can take steps aiming at enforcing his long-run preferences. An effective way to get clean from drugs could be a contractual commitment to the custody of a clinic which legally binds the doctors to hold their patient captive even when he asks to get out. In many countries, however, such self-imposed constraints cannot not be enforced legally; ‘Dr. Jekyll can ask to be locked up for his own good, but when Mr. Hyde says ‘let me out’ they have to let him out’ (Schelling, 1984, p. 96). A permission to write contracts unchangeable and not terminable forever, though in the individuals’ constitutional interests, may be incompatible with some community value held in the merit concept

From a constitutional perspective, one cannot entirely deny the existence of some constitutional interest in drug regulation. This is because under free-market allocation of drugs, it cannot be excluded with certainty that children and mentally weak persons come into possession of drugs. On the other hand, the current system of general prohibition of the use of drugs implies more coercion on more persons than necessary. One could conceive of solutions which would avoid most of the undesired side effects of a general ban on drug use and which would, at the same time, increase individual freedom. Comprehensive information on the potential effects of drug consumption should be provided. A monitored opening of the market for drugs, as suggested by Thomas Schelling, could be accompanied by an institutional separation of the provision of drugs from their consumption. If, during that delay, an individual’s constitutional interests come to blot out his short-run preferences, the consumption decision may be finally left untaken completely (see Schelling, 1984, p. 104).

The aim of this paper was to examine if, and to what extent, public policies Musgrave denotes as ‘(de-) meritorious’ can be justified by way of Buchanan-type constitutional reasoning. Our result was negative for all key applications of the merit concept although we did not put into question its implicit premise that political agents act as ‘benevolent guardians’ of their citizens. As widely expressed in the public choice literature, this is anything but self-evident.

‘In all’, Musgrave (1987, p. 453) sums up the discussion of his conception of merit goods, ‘it seems difficult to assign a unique meaning to the term.’ From a constitutional perspective, there is nothing to add to this conclusion. For what is defensible in the theory of merit goods is not novel and covered more convincingly by other theories; and what is novel in the theory of merit goods is indefensible from an individualistic point of view.

1. For reasons of simplicity, we shall simply use the term ‚merit good‘ to refer to both merit and demerit goods.
2. The case of a ‘demerit’ good, such as the prohibition of drugs consumption, can be easily reconstructed from the following discussion. Therefore, it will not be treated explicitly.
3. In order to be able to study the welfare effects in a more general framework, S is depicted with an elasticity of less than infinity, which would be appropriate in the case of perfect competition.
4. According to Musgrave (1959, p. 14), ‘the satisfaction of merit wants remains a precarious task. Interference with consumer choice may occur simply because a ruling group considers its particular set of mores superior and wishes to impose it on others. Such determination of wants rests on an authoritarian basis, not permissible in our normative model based upon a democratic society.’
5. Head (1988, p. 30) emphasizes that the ‘multiple preference approach ... provides an individualistic normative theory capable of covering the entire range of merit wants and generalized social wants’ (italics added).
6. Schelling (1978) examined the case of players in the American Hockey League who deliberately chose not to wear helmets although most of them explicitly claimed to favor a mandatory rule to do so. This case differs significantly from the self-management problem considered above. For, many players believed that helmets would cut their efficiency and put them at a disadvantage since they were handicapped by their head protection. Thus, players here are captured in an ordinary prisoners’ dilemma-incentive structure. If wearing helmets is made mandatory, all players achieve a joint cooperative situation in which no one has an advantage over the others, and at the same time there is less danger of being injured. Here, compulsion on the players helps to overcome a social, not, however, an intrapersonal dilemma. Thus, according to our above definition, this intervention would not fit the category of merit policies.
7. The explanation suggests itself that the control of motorcyclists is much easier to exercise and consequently more profitable than that of mountain climbers.
8. For an economic analysis of drug prohibition and the resulting crime effects see Pommerehne/Hart (1991) and Koboldt (1995, chap. 6).
9. One such reason may be that, without compulsory retirement insurance, each individual has a rational incentive to expect that his fellow citizens will not let him starve to death if he has failed to insure himself. A governmental intervention, for example in the shape of a general compulsory insurance, could be justified by a normal (individualistic) externality argument.