Public Goods beyond Markets

ABSTRACT The aim of this article is to contribute to a clearer depiction of public goods. The theoretical conundrums of the standard approach are highlighted before discussing a broader view of public goods that includes various types of merit good. It is suggested that there is no definite technical criterion to decide which type of good should be deemed public: a decision concerning societal priorities is needed. The article therefore discusses who is to decide, and how, these priorities and indicates that voting does not lead to a univocal decision unless appropriate decision criteria are depicted. Public goods should allow people to choose how to organize their society but, in so far as they are not provided, people may be unable to adequately choose the very public goods they need. Elucidation of this vicious circle may provide for a more rigorous classification of public goods and suggest a pragmatic way to bypass it.


Introduction
Recent policy in most of the Western world has focused on reducing government action in the economy. This has occurred because at the macroeconomic level public expenditure is claimed not to be an appropriate means to stabilize or enhance growth. It has occurred also because at the microeconomic level public regulation and/or the (direct or indirect) provision of goods is supposed to involve inefficiency. 1 As far as the microeconomic aspect of the question is concerned, generally public discourse takes it for granted that markets are efficient, and that the very notion of efficiency is clearly defined. The type of goods (including services) that government is supposed to provide involves a slightly more slippery issue. It is generally acknowledged that public goods that are typical of the minimal state-e.g. defence and the preservation of the legal order-should be provided by government. As for other goods, they may be deemed important for growth or improving equity, but the problem is that either they are inadequately provided or they involve costs that clash with budget constraints. Under these circumstances, most policy discussions are centred on the relative efficiency of the actors-public or private-who are supposed to provide the public good. Political discussions, therefore, tend to be reminiscent of the so-called equity-efficiency trade-off whereby the combination of these two policy goals is a matter of value judgement.
Although this type of argument tends to be commonplace, my view is that it is not grounded in a rigorous theoretical discussion. The problem is that the demand for these goods is most often associated with the pursuit of ethical goals whereas the dominant economic approach-the New Welfare Economics-considers ethical and technical issues as conceptually independent of one another. This is precisely where the problem lies. Indeed, a major tenet of this article is that what public goods are does not depend on their technical characteristics alone but on the ends that a society pursues. What is wrong with public discourse, therefore, is less a lack of technical details and more a belief that the issue is basically technical rather than ethical.
The aim of this article is not to discuss public discourse per se, but to investigate the theoretical backing for the claim that government should or should not provide certain goods. These will be referred to generically as public goods.
In general, the peculiarity of public goods is that, although they are beneficial, it is up to government to provide them because private actors have no incentive to do so. This broad depiction raises the following issues. First, who are these goods beneficial for: does it have to be everybody or can it be just a section of the community? Second, how are benefits to be assessed: should a Pareto criterion hold or can we claim that the benefits they provide to someone may be at the expense of someone else? Third, should the benefits be associated with individuals or can they transcend them? Fourth, how much of these goods should government provide?
Depending on the answers to these questions, our list of public goods may range from those normally associated with a minimal state (e.g. a legal system that protects property rights) to those provided by the welfare state (e.g. education for everybody) and beyond (e.g. a manufacturing structure that avoids environmental degradation). On the other hand, depending on how we conceive of the economy, those same goods may perform different tasks. Thus, the legal system that protects property rights and is conventionally considered to benefit everybody may be viewed as a means to defend those who have property from those who do not have it (Ver Eecke 1998). This interdependence between the definition of a public good and the economy to which it refers implies that a clearer view of how these goods can and should be conceived can emerge only as we proceed.
Given these premises, it is appropriate to situate the ensuing argument within the general discussion about public goods. The next section therefore surveys possible strategies to investigate public goods, distinguishing them by their theoretical and methodological premises. Since the aim is to discuss public goods less as a theoretical conundrum for students in public economics than as a major issue for economic policy, the focus is on social and societal relevance rather than on internal consistency issues. Following this survey, the standard approach is outlined to highlight the theoretical difficulties that it faces. The standard definition of a public good is rather restrictive. To take account of goods that are-or could be-provided by the government, the discussion is extended to a broader approach that includes various types of merit good, i.e. goods whose benefits for the community cannot be traced back to individual benefits alone.
There are legitimate reasons for this more extensive perspective, but the discussion suggests that there is no definite technical criterion to decide which type of good to include or omit, instead suggesting that a decision concerning societal priorities is required. The discussion therefore moves on to who is to decide and how, and points out that-owing to macro effects, historical change and non-converging interests-the expression of preferences through voting does not lead to a univocal decision about what to do. A broader perspective is required that considers decision criteria. Following a brief outline of what these may be, it is argued that public goods are supposed to allow people to choose how to organize the society in which they participate. At the same time, in so far as these goods are not provided, people may be unable to adequately choose.
This vicious circle may have no strictly logical solution, but its elucidation is important for two reasons. First, it provides the elements for a more rigorous classification of public goods in terms of meta-preferences. Second, it suggests a pragmatic approach to bypass the vicious circle. The main point of the argument is that the solution is not based on some type of economic technicality: it is a political issue. To support this claim, the discussion highlights that, once we relax some restrictive assumptions of conventional economics, the degrees of freedom for political action increase.
Concluding remarks briefly consider how to understand prevalent contemporary public policies.
The article discusses public goods in relation to contributions that are most often centred on neoclassical theory. It avoids stressing the latter's internal inconsistencies and the implausibility of its assumptions. Rather, the focus is the plausibility of a pricecentred coordination of economic activity when public goods are considered. 2 It is contended that even those who are willing to accept a neoclassical approach must acknowledge that, given the prerequisites to function properly, an economy such as the one in which we live is hardly conceivable as static and that the degrees of freedom concerning change are much greater than is conventionally believed. I posit that, if these conclusions are reasonable for scholars in the neoclassical tradition, they should be a fortiori plausible for non-mainstream approaches.

Disentangling the public goods issue
To conceptualize public goods three analytical strategies are possible. The first one is to rely on the conventional approach. From a neoclassical perspective, public goods are a case of 'market failure'. While coordination of economic activities should occur through relative prices, 'market failures' prejudge allocative efficiency. Thus, the problem is how to reconcile price-centred coordination with any public action initiated in response to those purported failures. While the issue has a strictly technical dimension-e.g. how an information-transmitting mechanism such as relative prices can deal with incomplete information-it also has an ideological dimension. This emerges particularly when merit goods, and the ethical judgements they involve, enter the picture. More specifically, the very fact that someone may think ethically, rather than merely process data to pursue self-interest, begs for a reassessment of the conventional conception of individuals: whether they merely pursue self-interest or are also able to choose according to some value system. 3 This reassessment, in turn, has important implications for the relation between individual and collective choices. As we shall see, this involves either neglecting a range of important issues or facing world view aporias. 4 The second strategy is to redefine and extend the conventional concepts by drawing on factual experience and on insights from other disciplines. This may be a useful point of departure but the issue it raises is whether it qualifies and supplements the conventional view 5 rather than advocates an alternative conceptual framework. 6 More specifically, the attempt to introduce new concepts within a compact theory such as the neoclassical one tends to raise more problems than it solves. For instance, as mentioned above, Ver Eecke (1998) points out that even public goods generally associated with the minimal state or neoliberal views favour someone at the expense of someone else, which means that they involve ethical judgements and, for this reason, must be conceived of as merit goods.
He adds that these goods should be provided in so far as they are justified by the wishes of citizens and consumers. The justification, he contends, is that the 'first thing that Western citizens as economic actors wish is a free market' (ibid, p. 140). If we agree for the sake of argument, the problem lies in the term 'free market'. If it was defined by reference to neoclassical theory, its compatibility with justice and human dignity is, at the very least, subject to specific conditions concerning distribution and how 'market failures' are dealt with. To meet these conditions-e.g. through redistribution-we might have to go beyond the minimal state and conceive of merit goods that clash with some 'freedom' of the market. The main point here, is that, by merely juxtaposing ethics with the conventional approach, Ver Eecke fails to acknowledge the need for an assessment of the implicit ethical judgements that underlie the 'free' functioning of a neoclassical market.
Similar difficulties appear in Malkin and Wildavsky's (1991) work. They claim that public goods are a social construct, which cannot be defined in strictly technical terms, but at the same time they are adamant about the existence of an objective positive economics. It is not clear whether they consider that attempts to provide better definitions of public goods are a matter of whim, which would be an oversimplification of reality. They do fail to consider that, in so far as the notion of public good is a social construct, the same may be said about the concepts that underlie economic thought, so that different views of what the economy is and how it functions may coexist. In the same manner as Ver Eecke's treatment of ethics, Malkin and Wildavsky juxtapose insights from theories of cognition with the conventional view of economics. The result is a failure to reconcile their relativistic approach to the knowledge of public goods with an absolutist approach to the knowledge of the economy.
The third strategy eschews the conventional approach in that it does not assume that price-centred coordination of the economy, or the individualism underlying it, exists a 3 Different views exist, ranging from the extreme conception of individuals as knaves (Farrant and Paganelli 2005) to Colm's (1956, p. 409) claim that, [a] more useful concept, I believe, is that [each voter] has an idea of what the government should do for him but also for others and the nation or even the world as a whole, and possibly for future generations. 4 A detailed survey of recent attempts to overcome these aporias is provided in Florenzano (2010). 5 According to Mann (2006), for instance, 'The hypothesis is that the current utilitarian framework for justifying market interventions is not sufficiently developed' (p. 511). 6 Desmarais-Tremblay (2017) provides a detailed historical survey of how the discussion on merit goods impinges on these two views.
priori. Based on this strategy, Altvater grounds his discussion of public goods on the distinction between human security-that is, 'freedom from fear and freedom from want' (2004, p. 47)-and commercial security, i.e. 'the security to do business in a troublefree environment' (p. 52). These two types of security reflect two assessment criteria: an ethical one concerning the quality of human life and a strictly economic one concerning how a capitalist economy is historically structured. The distinctive aspect of this approach, relative to the first two strategies, is that the former criterion is not viewed as a constraint on the economy but as a policy guideline of the same standing. From this perspective, Altvater further points out that commodities are unlikely to achieve human security while public goods can. Similar to Ver Eecke, he points out that, in so doing, public goods 'can be good for some but 'no good' for others' (ibid, p. 47) but he does not pretend to make these reflections consistent with the compact framework of conventional pricecentred coordination. Colm (1960) adopts a similar strategic stance when he debates the characteristics of the public interest-thus what the government is supposed to do-not only as a compensation for market failures but also as the only way to pursue the ultimate goals of a society. His approach focuses on what is deemed a relevant issue, quite independently of its implications for conventional theory.
The problem with the third strategy is that, since no general framework is assumed that circumscribes the notion of public good, it is likely to raise more issues than it can deal with. More specifically, in so far as we consider public goods in relation to particular aspects of the economy, reference to the latter as a whole is unnecessary. It is an altogether different issue if these goods affect the economy as a whole and involve an assessment of the general goals of society. What appears to be a drawback, however, may turn out to be conducive to a more insightful and theoretically fruitful conclusion precisely because it forsakes not only the conceptual straightjacket of conventional theory but also the pretence of being an all-encompassing account of the economy.
Let us consider the implications of these strategies for the specific public policy issue of this article. The first two approaches assume that the economy is based on price-centred coordination. Their problem is to deal with circumstances that prejudge that coordination. In the first case it is a market failure, a shortcoming that is internal to the coordinating mechanism itself. In the second case it is ethics or cognition, that is, exogenous interferences. The theoretical issue is to reconcile these shortcomings with an as efficient as possible price-centred coordination. The third strategy proceeds in the opposite direction. It assumes that there are social and societal issues that require some type of policy. It investigates the kind of rules that are required to set up an economy that provides for a solution to these issues. 7 From a methodological perspective, we are confronted with two different conceptions of the economy. 8 The first two strategies rely on a closed-systems view whereby, given a range of conditions, the economy can function independently of any interaction with 7 Altvater's question is how to achieve freedom from fear and want in a capitalist economy whose main goal is to ensure profit. Similarly, Colm's view reflects the dominant post-Second World War stance whereby public policy needs to organize and direct the economy in order to achieve social and societal priorities. 8 The term 'economy' is referred to here as the set of relations that allow for the material reproduction of society. the society it is a part of. The third strategy is based on an open-systems perspective whereby how the economy functions depends on the society it is a part of. 9 The two views lead to different approaches to policy. In the first case, the economy has its own rules and the issue is to remove, or make up for, the shortcomings that prejudge them. Difficult as this task may be, it is up to the economist to devise a solution or a set of solutions that the policy-maker can choose from. In the second case, the economy is basically organized according to rules that reflect historical circumstances and societal priorities. Since the latter are potentially countless, there is a corresponding infinity of economic arrangements. Economists can use their theoretical toolbox to identify possible inconsistencies, but the identification of a desired arrangement is not strictly economic: it also depends on society and on how it interacts with the economy.
The sections that follow will pursue this third strategy to link these concepts to a general view of society's ultimate goals. The critiques in the general discussion, however, are not presented as proof that the alternative approaches are inevitably wrong. They point out specific issues that, while suggesting that the third strategy is more insightful, aim to engage scholars who follow different strategies. 10

Public goods: The standard approach
The standard definition of public good is straightforward: a public good is non-rivalrous and non-excludable. The policy issue is how much of it the government should provide. The problem lies in non-rivalry. In order to consume a private good, individuals are obliged to buy it, thereby providing information about the benefit they gain from the good. When they face a public good, they can benefit from the provision of the good independently of whether they paid for it or not.
Optimal provision of the good requires that, given a range of appropriate circumstances, the total cost must be equal to the sum of the prices that individuals would be willing to pay for it. Since they can benefit from provision of the good even if they do not pay for it, it is convenient for them not to provide the information and to act as free riders. This precludes the determination of the amount of good that should be (optimally) produced. To make up for this drawback, a government can base its decision on a social welfare function: Samuelson (1954, p. 389) clearly states that: the failure of market catallactics in no way denies the following truth: given sufficient knowledge the optimal decisions can always be found by scanning over all the attainable states of the world and selecting the one which according to the postulated ethical welfare function is best. The solution 'exists'; the problem is how to 'find' it.
In this case, however, what the individuals are unwilling to provide-information concerning their preferences-remains unavailable to the government. Individuals can avoid paying for the good so long as the good is non-excludable. Excludability can, therefore, turn out to be a partial solution to the problem that public goods raise. In so far as it is possible to devise some technological (e.g. TV decoders) or legal (e.g. patents) form of excludability, a public good may become a club good. This prevents consumers from withholding information and allows a price to be fixed. Excludability may, therefore, allow a 9 See Dow (1996, p. 14), Chick (2004), Chick and Dow (2005) and Lawson (1997). 10 A more detailed account of this methodological stance is provided in Dow (1990). return to market-based allocation. Unfortunately, even if all other assumptions that allow for optimality hold, this is not enough to achieve an efficient outcome. As Musgrave and Musgrave (2003) state, 'Exclusion may be possible-for example, by charging a toll to cross a bridge. But the charge will be inefficient if the bridge is underused, because in that case the bridge's benefits are nonrival' (p. xiii).
How relevant are these issues in practice? One way to deal with this question is to identify what goods meet the above definition. Traditionally, the public goods par excellence are national defence and law enforcement. Other goods are non-rival, however, in that they are characterized by consumption externalities: computer software, infrastructures, network goods such as utilities or the internet, and payment systems ranging from credit/debit cards to money. In many instances, they are provided by private agents, but this happens in much the same way as for the bridge in the example above.
Leaving aside specific goods, if the production of any output involves environmental degradation, through the depletion of natural resources or the production of waste, basically that output affects other people. Apparently, this issue has to do less with public goods than with externalities. It is nonetheless possible to argue that services that reduce pollution or make up in any other way for the depletion of the environment are public goods. Furthermore, if the deterioration is not the same for all goods, (the composition of) output as such turns out to be a public good. The issue is that, if interdependent utility functions characterize a great deal of economic activity, social accounting should not be restricted to individual utility.
Summing up, the standard theory of public goods is centred on non-rivalry. A great many goods and the very composition of output fit this feature. Non-rival goods may or may not be excludable. When they are not, the provision issue emerges completely: there is no actual way to decide the type and quantity to provide optimally. When they are excludable, optimality is not achieved anyway because the goods are priced as if they were rivalrous. Under these circumstances, the criteria to decide who should provide a good-whether the government or private agents-and how much of it should be provided are rather shaky. The problem is that the efficiency criterion is not clear. If market failures exist, then public policy pursues a 'second best', which is a rather difficult goal to define in practice. It is not only that, as Markovits (1998) argues, 'many law professors and an increasing number of economists ignore Second-Best Theory because they like clear bottom lines and correctly perceive that TBLE [thirdbest-allocatively-efficient] analysis will often lead to conclusions that are fact-dependent and, on that account, contestable ' (p. 8). What is most likely is that more than one solution can be envisaged to make up for a market imperfection, which leads us back to the knowledge requirements referred to by Samuelson.
The absence of generally accepted guidelines, compounded by the possible ineffectiveness of governments in providing adequate solutions, leads to a debate centred on personal sensitiveness, which boils down to the degree of optimism concerning governments. Thus, it is not surprising that, in a discussion with Musgrave, Buchanan summarizes their divergences in very simple terms: My differences with Musgrave can be summarized along a unidimensional spectrum, with neither of us located at the extremes. He places considerably more relative weight on the 'good' that collective action can do, whereas I place more relative weight on the 'bad' that unconstrained collective action might do. (Buchanan and Musgrave 2000, p. 108)

Public goods: A broader approach
Public action often relates to a rather broad variety of goods (and services), including not only national defence and law enforcement but also health, education, pensions, basic goods such as clean water, a sewage system and housing, and basic scientific research. It may consist of providing the goods as such, in subsidizing their purchase or in regulating the legal features of the market where private agents provide these goods. Whatever the specific features of this action, it has to do with goods that are particularly important for the standard of living of a community: in some cases, one might argue that they involve lexicographic preferences. Decisions to enhance or reduce public action may, thus, dramatically affect the community. Some of the above goods apparently do not meet the criteria set forth by Samuelson even though, as a matter of fact, they are often provided by governments. Samuelson (1954), who originally argued that government expenditure was required only because of the non-rivalrous nature of public goods, subsequently acknowledged that other circumstances could account for the provision of goods by government. Aside from a range of circumstances that are generally included among so-called market failures, he mentioned that government might provide goods such as education and hospitals to make up for tax and transfer policies that fail to achieve a desired redistribution. 11 The Musgraves provided a different approach to this issue by introducing the concept of merit good: There exists a community interest as such, an interest which is attributable to the community as a whole and which does not involve a 'mere' addition, vertical or horizontal, of individual interests. This community interest then is said to give rise to communal wants, wants which are generated by and pertain to the welfare of the group as a whole. (Musgrave and Musgrave 1989, pp. 54-55; see also Musgrave 2008) Merit goods-which satisfy these wants-are like the public goods discussed so far in that they affect many individuals. They differ in that the benefits they provide can seldom be measured in terms of individual utility. With this type of good, Samuelson's solution is inadequate on strictly formal grounds, independently of its actual practicability.
It may be reasonable that governments provide the broad variety of goods listed above. On strictly economic grounds, output growth is likely to be greater if workers are healthy and educated, if they can rely on the advanced technology that can only be achieved through basic scientific research and if they do not have to worry about basic living conditions for their households, including elderly people. It is, nonetheless, unlikely that they can deal with these issues by themselves, owing to incomplete information and an intertemporal preference that prejudges their understanding of the longer-term consequences of their choices.

11
He also referred to what we might label as 'extra-economic' situations where 'people … do not regard the results from spontaneous market action as optimal' (Samuelson 1955, p. 356) and 'government expenditure … in areas where government probably ought not to operate from almost anyone's viewpoint' (ibid). Hence, his suggestion that an exhaustive understanding of these issues required not only further economic enquiry but also 'contributions from neighboring disciplines' (ibid).
In the case of scientific research, apart from the public good features of scientific and technological knowledge (Arrow 1962), individuals may be unable to predict the outcomes-and gains-to which their investment will lead. The issue, here, is not that information is incomplete but that it is altogether absent: by definition, research consists of dealing with the unknown.
More generally, incomplete information may lead individuals to underestimate the future; that is, it may affect intertemporal preferences. 'Inappropriate' intertemporal preferences, in turn, may lead individuals to underestimate the information they require. Thus, what is at issue is not only the absence of informational inputs but also the inability to adequately appreciate and, consequently, process those inputs that are available. Truly, governments may not be a fully-fledged solution to these problems: they do not know individual utilities and the intertemporal preferences of citizenship. It is nonetheless the case that, while individuals focus on their specific goals and intertemporal preferences, governments are organized to transcend individual goals and individuals' lifetimes.
The type of good listed above is not the only one that involves a possible public action. Natural resources (oceans, lakes, forests, glaciers, coastal reserves, wildlife), the national heritage, (clean) air, financial stability and peace, amongst others, are generally believed to be significant for society. The services required to preserve them may be considered public merit goods. From a conventional perspective, this type of good raises several problems. Growth is generally viewed as a positive outcome for all but it may affect the natural (Boulding 1966) and cultural (Pasolini 1976) environment. Financial innovation and the armaments industry may favour output in some sort of short run, but their long-term consequences may be disastrous (Eisenhower 1961;Minsky 1982). What is at issue, here, is not only a lack of information but two contrasting goals. The conventional way to deal with it is centred on the existence of a trade-off. The issue is how to assess it. Should we decide according to existing prices, even though they are inefficient prices, i.e. prices that reflect market failures, thereby precluding choices that lead to the efficient allocation of resources? Should we assign property rights to goods that are not private goods since they affect the utility of society as a whole?
Similar issues arise also when public merit goods are provided for redistributive reasons. This is the case with, for instance, public schools and a national health system: their purpose is to ensure that everybody has access to these services, independently of the income or wealth of the beneficiaries. Here, too, the conventional approach is to deal with the issue in terms of a trade-off, assuming you can only base your choice on the existing set of relative prices. While this may seem reasonable from a pragmatic point of view, it is not a rigorous way to frame the issue. Along with the above considerations about the inefficiency of prices when they reflect market failures, there is no reason to assume a priori that original endowments are socially desirable.
It is beyond doubt that, whatever we choose under these circumstances, a positive outcome for everybody is unlikely to occur. Someone is bound to suffer the consequences of the restrictions on output that follow anti-pollution laws or of the reduction in relative income or wealth consequent on any redistributive action. Musgrave and Musgrave (1989, pp. 57-58) acknowledge this issue: Individuals as members of their society may feel obliged to share certain costs (e.g., for maintaining the Lincoln Memorial) or to accept certain priorities in the use of their own funds because this is called for as a matter of respect for community values.
This statement points out that community values may prevail over individual preferences. When and how this should occur is an important problem. Before discussing this issue in greater detail, it should be stressed that community values need not focus on individual benefits alone. For many Americans, making sure that the Lincoln Memorial-or the Statue of Liberty-is in proper condition is important not only because these are nice statues to look at: they represent symbols of a national identity. Allowing everybody to look at them is considered a way to reinforce not only one's own but everybody else's sense of belonging.
There is another aspect of public merit goods that deserves attention. Action in favour of the community does not merely reflect existing community values. It also feeds back to their creation and evolution: Consider the city of London: to the extent that in our work, travel and play we participate in the irreducibly common actions of, say, crowds, queues, rush-hour, Sunday afternoon on the London South Bank, and so on (ad infinitum), so we help to engender an extraordinarily complex and obviously irreducibly common good-one entirely unavailable to any of the individuals concerned acting separately. (Deneulin and Townsend 2007, p. 11) Whatever our thoughts about the quality of life that such an environment leads to, it is beyond doubt that we generate that environment by interacting with others. In turn, that environment becomes part of us and shapes the way we are. Thus, the decision to provide Londoners with facilities such as public transport eventually affects how and where they interact, thereby determining what London-and being a Londoner-means to them. In this sense, a community cannot be reduced to the sum of its members: it is not a mere collectivity. All this implies that whether a good is public or not depends not only on the benefits it may provide for individuals but also on how it feeds back to the community as a whole.
Notwithstanding Deneulin and Townsend's (2007) specific reference to a common good, what this leads to is that community values need not be considered exogenous. Their creation and change may be part of policy goals that are centred on the quality of life, i.e. on something that encompasses strictly economic welfare but cannot be reduced to it. Viewed from this perspective, merit goods involve much more than the incomplete information discussed above. They beg the question: what kind of community do we want? To some extent, this is not enough, however. In so far as we all belong to different communities-religious, political, professional, national, local, etc.-decisions concerning merit goods relate to the broader community that-in one way or another -encompasses them all.
Summing up, even if we remain within the bounds of a conventional economics approach, it is difficult not to deal with a range of goods that, just like standard public goods, benefit more than one individual but, differently from them, reflect goals that are irreducible to individuals. This raises two issues. The first and original one is what goods and how much of them should be provided. The second one is how to deal with two different types of choice criteria: that of individuals and that of a community.

Whose decision?
In his treatment of public goods, R. Musgrave provides pragmatic solutions to some of Samuelson's drawbacks. The concept of exclusion qualifies public goods, allowing for a distinction between these and similar goods (club goods), which need not be provided by the government. The concept of merit good accounts for the public provisioning of goods that Samuelson (1954) was unable to include in what he referred to as a 'pure theory of public expenditure'. These qualifications, however, raise important theoretical issues. As noted above, exclusion cannot solve the optimization problem: at the very least it leaves an externality. As for public merit goods, they are identified according to non-individualistic criteria (societal values), which are different from-and somewhat inconsistent with-the choice criterion based on individual preferences. This raises the issue of what criteria should be used to achieve efficiency and societal goals.
Viewed from a conventional perspective, public merit goods act as an exogenous interference with individual market choices. As Ver Eecke (1998, p. 137, original emphasis) points out: Public goods are political economic goods which the government provides with the intention of respecting the wishes of consumers. The consumers need help because public goods have technical (factual) characteristics that make it difficult for individuals to acquire them in an optimal way. Merit goods, on the other hand, are political economic goods which the government provides by a method or at a level which disregards the wishes of consumers.
As a result, merit goods may not only affect market performance but even endanger democracy. The latter issue is clearly stated in Musgrave and Musgrave (1989), where they point out that there is an inconsistency between the individualistic stance of conventional economics and the notion of community. Thus, although 'common values and concerns do exist in a cohesive society and their existence may place some limitations on the conventional doctrine of individual choice' (ibid, p. 58) it is the case that 'the concept of merit or demerit goods … must be viewed with caution because it may serve as a vehicle for totalitarian rule' (ibid).
As we extend the discussion from standard public goods to merit goods, it is difficult to minimize the importance of decisions concerning government provision. The problem is that, while an ideal, if practically useless, solution exists that reconciles the provision of standard public goods with the rationale of individual choice, merit goods centred on community values are irreducible to individual choice. While this suggests that there undoubtedly is scope for a choice criterion that transcends markets, the original problem-what that criterion should consist of-remains.
It is possible to play down the importance of these issues by arguing that these 'communal' merit goods are an exception and that their overall influence on market performance is not particularly important. R. Musgrave (in Buchanan and Musgrave 2000, p. 42) seems to follow this approach when stating that he does not 'wish to assign merit goods thus defined a major role'. If we look at the variety of goods listed in the previous section, it is rather difficult to agree with this remark. On the other hand, not everybody may agree that merit goods include all that variety. The fact is that, as Musgrave and Musgrave (1989) argue, what we depict as merit goods depends on collective values. Based on this premise, to assess how pervasive they are, we should focus on what people value. Sekera (2015, p. ii) contends that 'we need to understand how public goods and services originate through collective choice and collective payment'. More specifically, she states that, orthodox economic theory, and Samuelson's definition, ignore the reality that public goods derive from social as well as economic forces. In reality, public goods originate through collective choice (voting) and are funded by collective payment (taxes). (Sekera 2014, p. 3) Sekera's emphasis is not on the technical features of the goods but on how they are valued. In this sense, although she refers to public goods, it is merit goods she has in mind. The two terms will be used interchangeably in what follows.
According to Sekera's definition, what public goods are depends on what voters decide them to be. This definition raises two issues. First, do voters always choose appropriately? For this to occur, we should expect them to be adequately informed: they should have all the information they need, and it should not be biased, e.g. through advertisements. Furthermore, they should be able to process that information appropriately: they should have an adequate rationality (as might be associated with a computer's hardware) and the ability to situate the information in an appropriate interpretative framework (the analogy being, in this case, with a computer's software). Finally, they should have the time to think their choice over. 12 The second issue is more specific: is it enough for voters to choose, independently of any discriminating criterion? Based on Sekera's argument, if a majority were to vote against public schools on the grounds that if you want an education you should pay for it, that would mean that education, and the knowledge it involves, is not a public good. Similarly, public law enforcement would be denied its public good nature if it was banished in the United States on the grounds that the Second Amendment to the Constitution allows individuals to take care of themselves. Leaving aside these extreme cases, should government budget problems lead to a cut in public expenditure, a good that is not provided any more would cease to be qualified as a public good.
These examples suggest that collective values-and the goods they qualify-may not be as changeful as voting majorities. For instance, public goods may be based on fundamental social rights-such as those listed in the Universal Declaration of Human Rights (United Nations 2015)-which suggests that, quite independently of pricing issues, no trade-off should exist: they reflect lexicographic preferences. Alternatively, they may be based on a notion of welfare or well-being that, although culturally and historically specific, is nonetheless stable over time. Kaul (2001) follows a different approach. She depicts publicness in terms of participatory decision-making, fairness and inclusion. Inclusion consists of extending the potential access that people have to a good. This allows for the provision of socially relevant goods to broad sections of society. 13 It is not residually-determined but the result of a policy choice.
Kaul acknowledges that fairness can be defined in different ways. Her suggestion is to allow the desired notion to emerge from participatory decision-making. Let us, therefore, examine this concept.
The difference between mere voting and participatory decision-making is that, in the latter case, people can exchange their views, elaborate on them and possibly change their minds consequently. This does not exclude that, owing to bounded rationality, they can make wrong decisions, but it does favour pondering, compromise and greater awareness of common-as opposed to individual-values. In some instances, it may also be a means to enhance inclusion and empowerment.
Much like in the voting case above, the issue that deserves further discussion is how to frame choices and decision-making. While this process may be relatively easy when small communities are concerned, it must meet quite a few requirements as the size and the complexity of the community increase. Awareness of what is at stake involves not only information-important as this may be 14 -but also a toolbox to adequately interpret it, independently of what personal experience may suggest. More specifically, decisionmaking undoubtedly requires the ability to make decisions on benefits, constraints and costs. These are not straightforward, however. The macro-consequences of microdecisions may not match expectations, either because the whole is not the sum of its parts (as in the Thrift Paradox) or because expectations become unreliable over time, owing to economic and societal change, such as the one discussed in the London transport example. Furthermore, decisions occur in a contested social environment. They affect different people in different ways, feeding back to economic, social and power relations (Kalecki 1943). Owing to varied identities, they also affect single individuals in complex ways: In fact we all have many identities, given by our relationships with various groups to which we belong and of which we see ourselves part. As members of a family, a community, a class, an occupation group, a nation, etc., we have a complex perception of 'our interests'. So the point, … that the pursuit of what we see to be our interest is not the only possible motive for choice, has to be supplemented by the further point that even what we see to be our interest may have various 'social' components. (Sen 1986, p. 349) Desai's (2003) historical assessment of the role of public goods complements these general issues. He points out that, during the 18th and 19th centuries, they were provided because 'fear of the mob-of fast-growing urban populations-made European states provide such goods even while they preached the doctrine of laissez-faire and balanced budgets' (p. 68). In the 20th century, universal franchise and the golden age of Keynesianism initially led to an exceptional growth in their provision. With the inception of inflation and of the fiscal crisis of the state, however, voters chose governments that curbed public expenditure.
In his account of why constituencies changed their attitude, Desai points out how economic and social change eventually led to a more differentiated demand for public goods: As countries prospered and citizens became richer, their preferences became more sophisticated. They were also able to express their multiple identities in a political way. So instead of a homogeneous electorate with single peaked preferences, there are citizens groups organized 14 Kaul points out, in this respect, that: Reality seems to be ahead of our thinking on this point. Just consider the many so-called 'watches' that have emerged in the areas of human rights, social development and the environment. Increasingly, we also see impact assessments, notably on such issues as 'trade and human development' or 'trade and the environment'. If these types of studies present sufficiently disaggregated measurements, people themselves would be able to decide whether and up to what point they consider a particular PG [public good] as a fair deal. (2001, p. 17) along age, gender, and ethnic lines. Their demands for public goods are fragmented by the quality and variety of goods they want. (p. 72) Since social fragmentation leads to demand fragmentation, different aggregations of interests turn out to be possible according to what coalitions prevail. 15 Under these conditions, participatory decision-making may be a rather difficult matter since what must be deliberated is not only what is most appropriate for the community-or even for one's self-but also how to achieve something through log-rolling and other forms of 'political market' transaction. The very notion of fairness that Kaul believes should emerge from participatory decision-making is likely to fade away or, at best, be very fuzzy. In this respect, Desai's conclusion that 'it seems the time has come for the public to be much more directly involved in formulating preferences, promoting just political bargaining, and contributing to the production of national and global public goods, nationally and internationally' (p. 74) may be encouraging but it begs the question: how can the public be more directly involved? Although it is not possible to exhaustively answer, we return to this in the following section.
Summing up, in the absence of a strictly technical, price-centred, solution to the question of what (broadly defined) public goods to provide, it is reasonable to suggest that the decision should be left to those who are directly involved. This suggestion, however, should deal with a range of issues that deliberation involves, and that people usually are not informed about: macro-effects of micro-behaviour, historical change and non-converging interests. Neglect of this issue seems to be based on the implicit assumption that there is a unique preference ordering, which allows for unequivocal decisions in an undisputable economic context. One would expect that economists and other experts, together with political leaders, must somehow have a role in this deliberating process. The primary issue, however, is what decision criteria should inform deliberation.

Decision criteria
It was previously pointed out that, as we extend the range of goods that governments can provide and thereby include different types of merit good, two problems emerge. One is that, given the irreducibility of merit goods to market preferences, some alternative criterion must be sought. The other is that, in so far as such a criterion is discretionary, it may become arbitrary.
From a market-centred perspective, these two problems are straightforward. The conventional view contends that prices provide the information that is required to allocate available resources. Market failures may require some public action, but this action is nonetheless constrained by the requirements of price-based resource allocation. If we choose to provide goods independently of these requirements, we forsake such a criterion and end up with no allocation rule to abide by. Economic welfare does not depend on relative prices any more but on the discretionary value judgements of the policy-maker.
The problem here is that, when market failures abound and cannot be eliminated, there is no way to achieve the conditions that underlie the first fundamental theorem of welfare economics. Under these conditions, there is no unique policy solution that achieves efficiency and a pragmatic approach is called for: at the very least, policy-makers must 15 Desai's conclusions are reinforced by the existence of vested interests that have all to gain from such fragmentation, as when business shifts its private costs onto the community, thereby turning them into social costs (Kapp 1971).
choose among different 'second best' solutions. The practical indeterminacy of this situation remains, however, to the point that New Institutionalists dismiss the whole problem on the grounds that any such attempt is just blackboard economics (Coase 1988) and that the real issue is simply to achieve 'relative efficiency' (Williamson 1993). The fact of the matter, however, is that once you acknowledge that actual markets do not comply with the requirements set out by conventional theory, it is unclear what the efficiency these scholars discuss is supposed to be 'relative' to (Stiglitz 1993). Ultimately, reliance on markets merely depends on a value judgement.
If market failures are downplayed, it is easier to argue that actual markets are not too distant from theory. Thus, R. Musgrave's reliance on neoclassical theory's relative prices as an allocation criterion is reinforced by his minimization of merit goods that deal with distribution and 'ethical preferences' and his total neglect of their cumulative effects on societal change. In fact, the problem with these goods is that they reflect what Sen (1982) terms meta-preferences. If preferences clash with meta-preferences, and decisions are made according to the latter, prices may be 'wrong' in that they do not reflect scarcity and optimization. If prices are 'wrong', choices are misplaced, and the rationale of relative prices is undermined.
How does this relate more specifically to the depiction of public goods? First, once we acknowledge that coordination of economic activity requires some choice criterion other than the one centred on individual preferences, what is this going to be? This begs the question: what economy does it pursue? This issue may be addressed in two ways. The traditional one is to consider exogenous action as an interference with existing relative prices, whereby the issue is to assess whether the resulting allocation is going to be more or less efficient: a typical case is Okun's (1975) bucket metaphor, whereby the pursuit of equity reduces the very availability of resources required to achieve that goal, much like carrying money from the rich to the poor with a leaking bucket.
The alternative approach is to recall that the traditionally basic assumption in neoclassical theory is that preferences, endowments and technology are given. Another assumption, which is less emphasized, is that no market transaction is possible unless there is an underlying legal system that ensures how it is actualized. As Bromley (2006, p. 120) states: Recall that public policy is collective action in restraint, liberation, and expansion of individual action. Notice, therefore, that public policy is nothing but a modification of the institutional structure of an economy that redefines choice sets (fields of action) for individuals. 16 This means that, given a range of conditions, the distribution that follows resource allocation may be consistent with marginal productivity and relative scarcity but the desirability of that distribution and of that allocation-including the composition of output and the technology deployed to produce it-depends on choices concerning endowments and the legal-economic nexus.
It is sometimes difficult to distinguish these two approaches in practical policy decisions because neoclassical theory assumes away historical continuity. There never is a moment when endowments, preferences, technology and the legal system are determined: they are taken for granted. It is nevertheless the case that these prerequisites must be assessed and possibly changed-e.g. by initiating or repealing goods such as compulsory public schooling, technological research or a specific labour code-according to some exogenous choice 16 See also Schmid (1987), Bromley (1989), Commons (1990) and Samuels, Medema, and Schmid (1997). criterion. From a strictly neoclassical point of view, efficiency cannot be determined before they are selected. It is nevertheless the case that different sets of prerequisites lead to different and mutually incomparable allocations, i.e. different and incomparable sets of relative prices. The conclusion this leads to is that, even in the context of a neoclassical framework, decisions concerning public goods depend on ethical judgements that are a logical prerequisite for relative prices.

A potentially vicious circle
Moving beyond this static framework, whatever public policy-and public goods-we choose to provide, the outcome of our choice feeds back to society. For instance, the decommodification of a 'fictitious commodity' (Polanyi 1944) such as labour power, through the provision of public goods that stabilize employment and real income, may be deemed lexicographically superior to the provision of goods via the allocation of non-fictitious commodities. In turn, the former may affect the way in which the latter is achieved. This potential economic change extends the main question: what societynot just what economy-does the choice criterion pursue?
To appreciate this issue, let us go back to the discussion about participatory decision-making as a means to choose. One condition for it to be effective is that people must be in a condition to participate. Access to goods and services may be important but, as Sen (1999) suggests, with his example of the Indian woman who is unable to conceive of a different social role from the one she was brought up with, suggests, the internalization of gender or of social subordination may prevent people from appreciating what change is possible. Under these circumstances, the participatory goal will not be achieved fully. Another condition for participatory decision-making to be effective is that demand fragmentation should not hinder collective decisions. As Desai pointed out, this would lead to a situation whereby political bargaining prevails over participatory decision-making, citizens are distracted from their ultimate ends by short-term goals and governments reflect petty interests rather than societal priorities.
Such a situation may be aggravated by the asymmetry between governments and individuals, concerning the potential changes in the economy: by definition, governments are able to supervise the economy whereas individuals are more likely to focus on their specific situation. If the former fail to foresee and to pursue something better than petty interests, the latter are likely to accentuate their focus on their sectional concerns. Given the complexity of the world they live in and the uncertainty it involves, people may be unable to know and assess the technical details of single choices and may be unwilling to give up their short-run goals and their vested interests. The fact is, that the knowledge and power gaps between citizens and policy-makers inevitably lead to a paternalistic bent. Policy-makers may exercise discretion towards a segmentation of interests or towards their aggregation in a general view of the public interest.
The above issues point to what appears to be a potentially vicious circle: if people are not free to choose, they cannot effectively choose how to be free. There seems to be no rigorous way out of this, so that what Desai described may confirm the fragility of participatory decision-making, R. Musgrave's fear of a totalitarian turn or the nihilism of conservative authors whereby 'much of the moral rot that we observe, in both private and public behaviour, is traceable to the exaggerated size of the public sector relative to the total economy' (Buchanan, in Buchanan and Musgrave 2000, p. 217). Important as these worries may be, the fact is that they arise within an economy thatas the above discussion suggests-does not, and cannot, work if it is coordinated only by prices and by rules that are functional to price coordination. While this prejudges any conclusion that the above risks can be avoided by merely relying on markets, it also raises the question of whether an alternative to the vicious circle exists.
It is certainly the case that the sceptical view of government policy tends to be overstated by the claim that voters are blind to anything but immediate gain and that, consequently, politicians merely pursue their personal interests. To some extent, this claim implicitly denies that an ethical judgement may exist that identifies a merit good.
The above claim also understates Sen's argument that individuals have many identities, which are not always mutually consistent. It neglects that society consists not only of individuals but also of intermediate bodies-trade unions, consumer associations, religious groups, minority groups, small political parties, civil rights activists, etc.-that reflect these identities. These bodies may uphold values and pursue goals that need not be mutually consistent and, considering their variety, do not allow for a trivial 'I give that you may give' political market. If we acknowledge that this type of individual and organization exist, it is easier to see that policy-makers are obliged to encompass these different values and goals within a general policy perspective if they are to be made mutually consistent. At the very least, they must couple their segmentation-prone policies with the enhancement of an ideological consensus whereby 'there is no (better) alternative'.
This leads us to a second issue that the claim understates. It is that people do not merely process information nor do they passively adapt to circumstances. People who participate in a range of activities around London eventually get to understand what it means to be a Londoner. They make sense of the complex world they live in by learning from their personal experience and by interacting with others. People who go to look at the Lincoln Memorial face a value of national identity, which they may like or dislike but must come to terms with. They not only try to understand how the world they live in is but also how it should be. In other terms, independently of whether they are aware of it, they formulate meta-preferences.
These considerations provide theoretical support to Colm's (1960, pp. 301-230) view that, many people fail to recognize the function of leadership in a democracy because they confuse leaders and dictators. The nature of the public interest, as perceived by millions of voters, must be articulated by political leaders, even though the leaders' interpretations cannot become reality except by vote of these millions.
Colm's view and the considerations that precede it suggest that two sets of conditions possibly provide a way out of the vicious circle. The first one is that some of the neoclassical assumptions concerning individuals-basically, what goals they pursue and how they pursue them-must be substituted with more extensive ones, so that individual behaviour is subject to more degrees of freedom than is conventionally assumed. Second, precisely because individuals do not decide mechanically-i.e. by merely reacting to incentives in relation to preferences-what they choose to do depends on a myriad of circumstances related to their varied nature.
According to Colm, political leaders interpret what voters perceive to be the public interest. Perception does not mean fully-fledged knowledge or even awareness.
Interpretation, in turn, is not a mechanical process. What Colm suggests is that, rather than combining meta-preferences according to some algorithm or in any other way deducing aggregate meta-preferences, political leaders have an active role. It consists of 'conceiving, in the light of individual belief, judgment, or circumstance' 17 what the public interest may be, given the 'quick, acute, and intuitive cognition' 18 signalled by voters. In other words, political leaders do not process information. They try to make sense of what people communicate. They may succeed in this learning process just as they may fail. 19 The outcome is not some kind of collective (meta-)preferences. Much like Tinbergen's (1956) view, whereby attempts to draw a social utility function from individual preferences are too problematic, what Colm suggests is that policy is grounded in the (meta-)preference functions of policy-makers.
Policy-makers who believe that the best way to do their job is to focus on the public interest may find a way out of the above-mentioned vicious circle by acknowledging the manifold characteristics of actual economic actors. More specifically, they may take account of the distinction between preferences and meta-preferences. Policy-makersas well as political groups and committed citizens-may try to aggregate interests and formulate meta-preferences that represent what they interpret as the (ultimate) ends-the socio-economic priorities-of a community rather than merely comply with short-term or sectionally-specific vested interests. 20 It is most likely that, faced with such ends, people are going to be better able to appreciate what the general direction should be. In other words, it is possible to switch policy goals from the short-run emphasis on commodities to a long-run care for well-being. 21 Viewed from this perspective, policy does not just provide for the efficient and equitable operation of an economy. It directs it towards some desirable pattern of societal evolution. It provides a general view of what the key issues under discussion are, a cognitive framework that works as a meta-coordinating device (Ramazzotti 2007). Obviously, this reliance on meta-preferences implies that policymakers transcend the 'obsolete market mentality' (Polanyi 1947) of conventional scholars and their reliance on preferences alone.
Let us briefly consider what this implies for public goods in practice. Given the above discussion, a proper definition requires a focus on what type of ultimate ends are pursued. For instance, if we focus on individual interests alone, we can conceive of them as something that an individual would buy, if she only could. Consequently, 'health care' would presumably consist of the provision of pharmaceuticals, medical services and medical infrastructures to cure diseases. Similarly, 'education' would consist of the transmission of knowledge required to become more productive and to earn more. 'Public transport' would probably not be provided since users would simply buy tickets for what would boil down to a private service.

17
Definition of 'interpret' in the Merriam-Webster online dictionary (www.merriam-webster.com/dictionary/interpret; with adaptations). 18 Definition of 'perception' in the Merriam-Webster online dictionary (www.merriam-webster.com/dictionary/perception). 19 Dow's (1990Dow's ( , 1996 'Babylonian mode of thought' provides an outline of how this may occur and the uncertainty that it involves. 20 'The perspective of practical reason implies that politicians have an obligation to … develop an idea for progress and feasible improvement in the society conforming to the common good and ask for a fiduciary relationship with citizens in it' (Mastromatteo and Solari 2014, p. 98). 21 An example might be the adaptation to a specific institutional context of Nussbaum's (2006) 10 'central human capabilities' or of the human rights listed in United Nations (2015).
An alternative perspective-focused on the community as a whole-would consider benefits that individuals may not even envisage but that are important for the community. Thus, 'health care' might include early diagnosis but also prevention of diseases through changes in working conditions, in consumption of polluting goods, etc. 'Education' might include, and possibly favour, the preparation of conscious citizens. 'Public transport' might aim at reducing the pollution and traffic congestion that private transport causes. It might even be provided at a loss for the firm that administers it, should this loss be more than compensated by the reduction in pollution-related social costs.
In terms of public policy, the difference between the two perspectives is twofold. First, the former takes the status quo for granted-since no single individual would ever be able to change it-while the latter may envisage changes to the status quo. Second, the former uses private (market) costs as the basic point of reference whereas the latter uses social accounting, however difficult and demanding this may be. In terms of societal values, the difference is that the former focuses on a society characterized by isolated economic agents whereas the latter focuses on 'developing and enhancing the capacity of citizens to share power and hence to collaboratively govern themselves' (Brown 2005, p. 59).

Concluding remarks
The aim of this article was to discuss the theoretical grounding for the claim that the public sector should (or should not) provide public goods, whether directly or indirectly. Its focus was less on the internal consistency of theories than on their relevance for public policy. Following a brief survey of the analytical strategies one can adopt to deal with public goods, the goods that can be labelled public were outlined and these illuminated doubts concerning the uniqueness of such a concept. Consequently, the discussion considered the problems that the standard approach-based on the price-centred coordination of the economy-faces when it must deal with different types of public good.
Two main issues emerged. First, in so far as these goods imply market failures, it is logically inconsistent to acknowledge that price coordination 'fails' but, nevertheless, claim that it can remain the crucial criterion-albeit supplemented by institutional devicesto coordinate the economy. Second, since prices do not provide the required information, the question is: what alternative choice criteria exist and how do they affect the identification of public goods?
A reasonable alternative to individual actors choosing through prices may be a democratic process. A range of issues emerge, however, that shed doubt on the unqualified reliability of this solution. They include the irreducibility of individuals to a single preference ordering, the action of vested interests, a contested social environment and the everchanging nature of the economy. Given the complexity that all these elements lead to, one might rely on economists and other experts to outline a course of action. It has been argued, however, that, in the absence of an a priori technical criterion, choosing a course of action and deciding what goods to provide depends on a community's ultimate ends. This, in turn, begs the question of how to assess these ends. In so far as democraticand possibly participatory-decision-making is important, this implies that political leaders and other aggregators of interests, as well as the people involved, should focus less on people's preferences than on their meta-preferences and, as far as public goods are concerned, on the criteria that these meta-preferences imply.
The discussion of how the provision of different types of public good may affect the allocation of resources pointed out that public goods may affect relative prices, the overall economic structure and the organization of society. These effects feed back into communal values, which underlie the very choice criteria applying to public goods. The conclusion is that public goods are a component in a cumulative process of historical change. Public goods reflect but also affect people's capacity to take part in decisions concerning the society they live in. Ultimately, they affect whether people are free to choose. This raises a freedom-related vicious circle: freedom may be effectively achieved only through freedom. Alternatively, the absence of freedom may preclude the achievement of freedom.
A possibly pragmatic and progressive way to bypass the above vicious circle is to acknowledge that politicians can identify a public interest that takes account of the ultimate ends-rather than the sectional interests-of the people involved. Viewed from a slightly different perspective, it means they can formulate meta-preferences-what the public interest is-rather than preferences (what they have to gain, here and now). Whether they care to do so or not is a matter of political choice, not the implication of economic premises. Whether they succeed in doing so is a matter of political sensitivity.
An open-systems perspective such as the one followed in this article hardly allows for a once and for all claim of what policy should consist of. Historical change, complex interaction within the economy and between the economy and its societal environment point to the need for a pragmatic approach to economic policy. The discussion suggested a twofold guideline for such an approach: a reliance on a democratic-rather than pricecentred-choice of societal priorities; and the provision of public goods that increase the capability to take part in democratic deliberation processes. It is obviously possible to choose different value judgements and formulate different guidelines. Neoliberalism is an interesting example, in this respect. Its goal to reduce government spending, especially the provision of public merit goods, cannot be based on efficiency. Consequently, either the proponents of this policy are acting on false premises or they are pursuing ultimate ends that require fewer, rather than more, public goods. In the latter case, the relation that exists between public goods and participatory decision-making suggests that the ultimate ends of a neoliberal society are far from democratic.