Article

Externalities and Public Goods: Theory OR Society?


How much does the standard theory of externalities and public goods really say?

‘Progressives’ interested in making things better have often been attracted to the study of externalities and public goods. [1] The body of theory that these topics represent involves the recognition that a market system may require correctives in order to generate desirable results, and that there is a potential role for government in this process. It is perhaps unsurprising that this field of ideas has been the nub of disagreement, less in regard to theory (such technical matters as the description of efficient outcomes) than to policy (how best to advance the presumed aims, for instance, through decentralized ‘Coasean’ solutions vs. ‘Pigouvian’ taxes).

In this comment, we aim to present a critical perspective on this body of thinking, recognizing what it has achieved but also identifying respects in which it may be inadequate or tellingly silent. As elsewhere in this series of blog posts, we make special reference to the presentation of the standard ideas in the influential textbook by Mas-Colell, Whinston and Green (MWG), although we intend our comment to have more general application. We consider the two topics together as that is how they are presented in that text (although the connection is, as we shall discuss below, one that must be examined).

Externalities: the Problem of Definition

What’s in a name? That which we call a rose by any other name would smell as sweet.”

What is an externality? The concept is taken for granted by most economists but when one delves under the surface is surprisingly difficult to explain. The definition presented in MWG, following one of the established traditions, is that “An externality is present whenever the well-being of a consumer or the production possibilities of a firm are directly affected by the actions of another agent in an economy” (chapter 11, p.352). [2] The authors go on to elaborate that “When we say ‘directly,’ we mean to exclude any effects that are mediated by prices”, thus excluding what Jacob Viner (1931) referred to as pecuniary external economies. The aim is presumably to isolate those externalities that have productive or allocative “efficiency” effects by creating a gap between private and social costs or benefits of activities, and not include distributional effects associated with gains and losses from price changes  — in order to focus on the restoration of aggregate efficiency rather than on who is affected. This definition has the merit of focusing on the failure of market prices to accommodate all relevant information, but it suffers from deficiencies.  In particular, the distinction between pecuniary and non-pecuniary externalities is strained. One reason, emphasized by Greenwald and Stiglitz (1986) is that pecuniary externalities have aggregate welfare effects in the presence of other ‘imperfections’ such as asymmetric information or incomplete markets, leading to effects that are much the same as arising from “technological” externalities.  Another reason is that agents’ actions may not adequately take note of consequences experienced by others, and thus lead to non-optimal consequences, even in non-market systems in which prices do not explicitly figure - making a price-based definition not straightforwardly applicable.  

An alternative definition of an externality in terms of “uncompensated interdependencies” rather than in terms of effects not mediated by prices would not exclude the pecuniary case, since the effect of a transaction on prices faced by third parties might lead to resulting costs and benefits that were uncompensated. Moreover, the concept of an uncompensated interdependence is also too broad in another way, as it entails that any concern with others (whether philanthropic or psychopathic) could give rise to an “externality” according to this definition.  Only persons entirely unconcerned with others’ actions or well-being (“solipsists”) ensure the non-existence of externalities arising from inter-personal concern (see e.g. Shaikh (2012) and Sen (1977)). If ordinary humanness involves pervasively experiencing such “externalities” then they can hardly be viewed as deviations. 

The concept of externality appears then to be caught between the Scylla of excessive narrowness (externality as any ‘direct’ effect not mediated by the market) and the Charybdis of excessive breadth (externality as any uncompensated interdependence including those arising through pecuniary effects and interpersonal concern).

The authors of MWG note that a consequence of their preferred definition in terms of direct effects not mediated by prices is that “the presence of an externality is not merely a technological phenomenon but also a function of the set of markets in existence”. This recognition presents the thin edge of a rather larger wedge. In fact, the presence of an externality, however ‘real’ as a behavioral phenomenon involving a gap between costs or benefits perceived by individual agents and those experienced by the whole society, is in basic respects a consequence of the specific social structure and institutional organization of an economy and not only of prevailing technology. Whether externalities can be said to arise or not is a consequence, for instance, of ownership relations: In James Meade’s famous example of the beekeeper and the apple grower, one solution is to agglomerate the enterprises. Although this can happen through common ownership it can also happen more generally through mechanisms of shared governance that lead similarly to coordinated or unified decision-making.[3] Externalities are a failure of institutions, of which ‘missing markets’ are a special case.[4] The concept of the allocation that would be chosen by a maximizing “social planner” (and replicated by the market under first-best conditions) allows standard economics to evade the discussion of these detailed aspects of the structure of an economy by resorting to an utterly unrealistic abstraction which offers little or no guidance.

Externalities: Where From?

Conceived ahistorically, externalities are extant ‘market imperfections’ that arise due to technological and institutional facts giving rise to a mismatch between valuations used in private decision-making and in social evaluation. On this interpretation, the internalization of externalities through appropriate public policies or private actions leads to a once-and-for-all solution to the problem of restoring the perfection of the market. The only question of interest then becomes one of how to best restore this perfection, or the nearest ‘second-best’ approach to it.

In a contrasting dynamic view, externalities are not taken as mere data of the problem but as being potentially produced. In such a perspective, externalities are born as a result of a variety of actions on the part of agents, acting individually and collectively. For instance, because employing a “dirty” means of production may lower private costs, a firm may actively pursue opportunities to do so, by locating in jurisdictions that are insufficiently able to mobilize counter-power in order to resist polluting production, or by exercising lobbying effort to weaken the definition and enforcement of environmental protections.[5] From this perspective, the active creation of externalities (as well as the active resistance to their elimination) may be even seen as part and parcel of the rough-and-tumble nature of real competition (as distinguished from perfect competition) in which competitive efforts to lower the cost curve play as much of a role as does moving along a given cost curve to minimize costs. An account of how externalities are generated and sustained is needed to complete a theory of them, since in its absence they cannot be explained.[6] They are an inefficiency that ought to have been already eliminated by agents finding ways (whether in the form of Coasean bargaining or public policy implementation) to move to the efficient frontier. If externalities are pervasive in contemporary societies, as they are, then that requires explanation.[7]

Public Goods, Club Goods and the Commons: Naming is Power

A valuable contribution of the technical literature on goods with public good characteristics has been to provide an analytical template through which different kinds of goods could be classified and, thus, the specific problems of provisioning and distribution in the different cases better approached. In the case of such goods it is widely agreed that there is an externality aspect (benefits that are at least partially non-rivalrous) and an aspect relating to boundaries (excludability). The development of this analytical framework and associated classification system came late. There were early discussions of the problem of “collective” or public goods provision by Henry Sidgwick (in 1883) and John Stuart Mill (in 1848) who recognized the problem of diffuse and difficult to exclude benefits, using examples such as that of the lighthouse within the British tradition[8]. Much of the definition and analysis of specific cases developed only from the 1960s[9]. Whereas in pure forms, private goods are both excludable and rivalrous, and public goods are neither, club goods are excludable but non-rivalrous and the commons are non-excludable but potentially rivalrous. Efficiency conditions and available management strategies can be influenced both by the type of the good within this classification scheme and its more specific additional traits (for instance depletable but renewable goods such as fisheries stocks present different problems than ones that are non-depletable but expansible such as intellectual property). The identification of these different conceptual categories has been useful as it has provided a basis for understanding better the specific difficulties that bringing about a desirable patent of provision in each case presents. The taxonomy may be of considerable use to any analyst, even if specific behavioral assumptions, for example regarding motivation and the propensity for ‘free-riding’, as distinguished from its arithmetic, are not universally accepted (as we shall discuss further below).

Still, it is far from obvious that the work of constructing such a taxonomy is anything near complete. Public goods are treated by MWG as a special case of externalities. However, this is far from obvious. A beautiful sunset that generates non-excludable and non-rivalrous benefits is a public good but it would be strange to describe it as involving an externality since no particular agent’s activities are involved in its production. In contrast, if all the residents of a neighborhood read in the light reflected from a single well-lit house, that may be a case of a public-good it is also evidently a case of an externality, insofar as it involves the incidental benefits created for others by one person’s choices.

Public Goods between Utopia and Politics

Samuelson’s (1954) was by his own description preceded by others in providing an efficiency condition for public goods provision (Samuelson specifically mentions Sax, Wicksell, Lindahl and Musgrave in this regard) who had identified their “essential logic” and describes Bowen (1943)[10] as having made an “independent discovery” of the same result. Samuelson’s was not a discovery, but a new derivation and restatement, in a framework in which ordinal utility is employed and an arbitrarily large number of persons and goods, public and private, is allowed. Both Lindahl (1919) and Bowen (1943) identify the condition for optimal provision as being that the sum of marginal rates of substitution between a public and private good must equal their marginal rate of transformation. [As Kolm (2010), op cit, notes, this efficiency condition was also very well-understood, from as early as the 1830s, in the French tradition of public economics, where it was interpreted in terms of the maximization of the surplus generated from an investment project.].

Interestingly, and specifically unlike Samuelson, both Bowen and Lindahl viewed the problem of provision as having to be approached through political economy. Lindahl is concerned with the political equilibrium between two different classes possessing different views in regard to the net marginal benefit of tax-financed public good provision. Bowen is concerned with interpreting the aggregate consequences of multiple voters casting their ballots in accordance with the valuations they place on public goods. In this respect, Samuelson’s description of Lindahl as having “not fully appreciated” that it is “in the selfish interest of each person… to pretend to have less interest in a given collective consumption activity than he really has”, and his consequent insistence that “taxing according to a benefit theory of taxation can not at all solve the computational problem in the decentralized manner possible [for private goods]” is rather beside the point, since Lindahl is from the first concerned with a rather different problem. Whereas Samuelson lampoons the “utopian” scheme of “Scandinavian consensus”, Lindahl is in fact concerned with a rather more hard-headed picture of Scandinavian “political equilibrium” in which competing interest-groups (one might think here of the apex business and workers organizations actually present in Scandinavia) face each other in a “repeated game” public goods provisioning context. The idea of transcending the ostensibly inexorable free-rider problem does not seem entirely “utopian” once put in this way, and Samuelson’s looks like an almost willful misreading of Lindahl’s argument, that squeezes out its original political content, which is what gave it credibility. This ‘sanitized’ depiction has unfortunately deeply affected the subsequent interpretation of the latter’s ideas.[11]

Public Goods between Model and Reality

The identification of the free-rider problem in collective action (and the associated problem of truthful revelation of preference) is an important contribution of the technical theory of public goods, and has led to important insights in applied contexts (for instance as to how and why collective action might be expected to vary according to the type of good, the number of participants, the extent of conflicting interests, and other factors). [12] However, the relevance of the theory is questionable from at least two distinct perspectives, one internal and one external to it.

First, the theory can be indeterminate, due to the possible presence of multiple equilibria in public goods provisioning models, for instance due to strategic complementarities giving rise to outcomes dependent on expectations. Even with identical agents, there can be asymmetrical efforts to provide public goods, determining for example upon prevailing mutual expectations (on which, see Cornes and Sandler (1995), op cit).[13] As a result it is necessary to call upon social or cultural norms , institutional factors etc. to arrive at predictions, even within the presumed self-interested behavioral frame of the standard analysis. If the possibility of publicly spirited behavior is considered, including that which may be conditioned on similar behavior by others, a still larger role for such determinants is introduced.

Second, the empirical record of the theory is mixed at best. Experimental studies of public goods provisioning find a wide range of results depending on many distinct factors including the population of experimental subjects (for an early survey see Ledyard (1995)). As is now well known, and made most famous by the work of Elinor Ostrom and her collaborators the extent of actual collective action in practical situations depends greatly upon institutional factors, and cooperation often appears to evolve in response to the specific locally experienced imperatives and difficulties, although it takes different forms in distinct social contexts. On the margins of the world economy and society, the addressing of collective action problems is often a matter of survival, and not surprisingly solutions are sometimes found, despite enormous pressures. In a variety of contexts, including the international political economy and business cartels, collective action problems are often at least partially overcome. The body of theory on public goods, although offering potential insights, is often undermined by a propensity to turn theoretical deduction into presumed fact.

Much of the empirical work on collective action has adopted the framework of the very theory it often undermines, becoming as a consequence a discourse concerning the conditions for the successful application of the theory rather than how best to understand the world. A fuller and more inductive approach might begin with comparative studies in society and history to arrive at a productive taxonomy and at such inferences as are warranted as to when and where collective action best works, complementing this with experimental findings. The role of deductive reasoning in such a picture is not to define the language game but to complement it, enabling the construction of analytical narratives. The interior (or –emic) as well as exterior (or –etic) perspectives on when and why collective action arises to a greater or lesser extent can both play crucial and complementary roles in such an approach to understanding. The goal must be to go beyond the empty formula that collective action fails (except when it doesn’t). Instead we must understand how, and when societies can be more or less able to promote the common good, drawing on the entire range of concepts that help us in doing so, freed from mental shackles that prevent our looking for them.

Res Publica: of, for and by the People

How people act is a question of their understanding of themselves and not merely of their instrumental reasoning regarding incentives: the standard theory of externalities and public goods is in this perspective a particular if very important case. The extent to which public goods are provided depends on who we see as part of “ourselves” and what we see as “ours”. It is thus influenced greatly by the psychological – the degree of identification of individuals with others and with shared endeavors —- as well as the cognitive — norms of public reasoning or action that may provide a reason to go beyond narrowly self-interested instrumentalities and to encompass conceptions of enlightened self-interest or of un-self interested moral action. These aspects of personal identification and reasoning are in turn inextricably tied to the social ground of common experience. How persons conceive of and indeed govern themselves (unavoidably involved both in what private goods they pursue and in whether and how they make voluntary contributions to the public good) thus enter quick on the heels of the social nature of public goods.

The classical phrase, res publica, not surprisingly refers both to goods held in common and to the activities of the community in and through the government that it constitutes and by which it is constituted. Questions of what we have in common lead quickly to questions of politics, through which interests are adjudicated (as Lindahl and Bowen understood) and identities also defined.

The choice among distinct modes of internalizing externalities or of addressing the free-rider problem in the provision of public goods, even when it leads to ostensibly “market-based” solutions, can involve a ‘meta-level’ governance decision to implement such a solution (one may think for instance of the implementation of a tradable permits regime instead of a tax as such a case).[14] Not surprisingly, questions of distribution and level of provision figure simultaneously in relation to such governance dilemmas, as recognized explicitly by early analysts such as Lindahl. This is no bad thing, as to separate the conditions for efficient provision from those of what is desirable in society, if we do not make the cloud cuckoo land assumption of lump sum tax and transfer mechanisms, is to abstract fatally what ultimately matters. This is true both for evaluative and empirical reasons. It is now well understood from experimental studies and widespread observations that perceptions of fairness and justice play an important role in determining contributions to public purposes, which can be ratcheted up or down depending on what people think of prevailing norms and arrangements. The purely privatist comparison of marginal benefits with marginal costs is far from sufficient to understand who does what.

The necessity to consider together questions of ‘efficiency’ and ‘equity’ is nowhere more apparent than in relation to perhaps the most wicked problem of our age — climate change. Here, the aversion of a public bad, in relation to which which benefits and burdens fall unequally and differently, is the form taken by the public good [on the question of the level and distribution of public bads, see the important work of the late Ulrich Beck]. The avoidance of climate change provides an example of a global public good, which in its case is of concern not only to the entire world population but both to present and future generations. Level and distribution must necessarily be ‘thought’ together. The application of proposed fixes (such as carbon taxes and tradable permits) which have derived their justification from standard economic theory, have run aground on the shoals of differing visions, political conflict and institutional inadequacy. The deeper origins of the problem in the contemporary pattern of economic growth, and the need for alternatives is hardly addressed. Both because the problem is forbidding in its complexity and because it is urgent, we must attend not to the best but to the good. Is the theory adequate, or must we transcend it? Time will tell.


[1] Sanjay G. Reddy; [email protected] ; The New School for Social Research, New York. This is one of a series of blog posts on which Raphaele Chappe and I have jointly collaborated throughout, although they are each primarily authored by one of us. I am deeply grateful to Tanweer Akram, Andre Burgstaller, Raphaele Chappe, David S. Grewal and Julia Harrington Reddy for their most helpful suggestions. I would like to thanks Ibrahim Shikaki for his helpful assistance and Anwar Shaikh for a useful remark.

[2] The Oxford English Dictionary defines an externality, in its usage in economics, as “A side-effect or consequence (of an industrial or commercial activity) which affects other parties without this being reflected in the cost of the goods or services involved; a social cost or benefit.” [Interestingly, this is reported as a “draft addition” which was only added in 1993 and the first example of usage that is offered is from 1957]. In contrast, Webster’s Dictionary defines it as “a secondary or unintended consequence (pollution and other externalities of manufacturing)”. The concept of “external economy” as used by Marshall and Viner or “incidental uncharged services or disservices” as used by Pigou were perhaps more precise since they highlight the specifically economic character of the external effect produced by the decision-maker. Although associated with the analysis of externalities, Pigou was centrally concerned with examples such as the congestion and degradation of roads which appear to be more in line with public goods analysis. We discuss the history of discussions of public goods separately below.

[3] See also Papandreou (1998) and Papandreou (2000) for a detailed discussion of the conceptual difficulties involved in presenting a definition of the concept of an externality and arguing the case for a theory of externalities centred on “gaps” or “incongruence” between physical and institutional descriptions of an economy.

[4] The “non-convexities” in production and in consumption that are deemed to result from the presence of externalities (notably discussed by Starrett (1972)) give rise to difficulty in establishing the existence as well as the efficiency of market equilibrium, but this is logical consequence, not definition, nor cause. As we have argued in other blog posts in this series, this “discovery” might be seen as an internal consequence of a particular choice of language game rather than a finding about the empirical world as such.

[5] Some of the public distemper at the infamous ‘Summers memo’ at the World Bank may have been related to the idea that even statically ‘Pareto improving’ trades are likely to become part of a real-world process in which the poor and politically disorganized are sought out actively as a dumping ground for waste, instead of addressing the problem of waste production and treatment at its origin. The environmental justice movement is concerned with such systematic and ongoing shifting of burdens, often as a result of endemically imperfect governance arrangements. There is no doubt that Summers’ reasoning was, from the standpoint of premises of the standard theory, indeed ‘impeccable’. An interesting example of the idea that an economic system can create incentives for the active creation of externalities in the form of environmental damage is provided by the work of Bagwell and Staiger (e.g. 2001) on the perverse incentives generated by the current rules of the WTO system to do so as a means of creating competitive advantage for national firms, by placing a ceiling on tariffs but no floor on environmental or social standards.

[6] A parallel might be seen in the failure to explain the presence of “comparative advantage. On this, see e.g. Unger (2007).

[7] The reasons for the emergence of externalities may differ in capitalist societies and in others. The environmental damage done by large-scale industrial production in existing socialist and indeed in pre- and non-capitalist societies serves to testify that un-internalized externalities are not the monopoly of capitalist societies. See also discussions of the “tragedy of the commons” in environmental history and other areas.

[8] As elaborated by Kolm (2010), the French public economics tradition was very preoccupied with public goods provision, concerned as it was with the financing and justification of bridges, canals, roads and other public infrastructure projects, and went rather further than the British school did in the same period in the identification of formal efficiency conditions to employ in project evaluation.

[9] For an earlier survey of these developments see Cornes and Sandler (1996). The case of club goods provides an example with many important economic applications. Whereas Sandler and Tshcirhart (1980) see Buchanan (1965) as being “seminal” and Olson (1965) as also being “influential” they identify the origins of the theory in applied problems addressed by A. C. Pigou and Frank Knight in the 1920s (toll charges to address congestion of public roads) and by Charles Tiebout (municipalities viewed as clubs) and Jack Wiseman (public utility cost-sharing) in the 1950s. As noted by Sandler and Tschirhart this literature developed rapidly thereafter to consider in particular the conditions under which a club structure could lead to more or less efficient provision. The possibility of establishing clubs may provide a partial (second-best) solution to public goods under-provisioning. On the other hand, if it leads to more widespread ‘enclosures’ of the commons it may have destructive effects, especially over time. This may, for instance, occur if stringent demarcation and protection of ‘intellectual property rights’ harms innovation.

[10] It is interesting to note that Bowen uses as a “social” (i.e. public) good the example of public education, which has subsequently often been treated as a good with public good characteristics as opposed to a pure public good.

[11] The difficulty of relying on truthful revelation of private valuations in public goods provisioning appears to have been long understood in the French tradition of public economics, in particular by Jules Dupuit. On this see Kolm (op cit) and Tubaro (2006). On Dupuit more generally and his approach to project assessment, see the chapter by Manuela Mosca in Faccarello (2002). The extensive modern literature on implementation of a scheme for truthful revelation of valuations has despite its technical sophistication not greatly helped to overcome the problem.

[12] See e.g. Olson (1965), Olson (1984), Bates (1981, 2014).

[13] In public goods models with continuous variation in costs of provision and contribution levels and in which all goods are normal, there is a single Nash equilibrium, but multiple equilibria can occur if either of these conditions is violated. See Bergstrom, Blume and Varian (1986), Cornes and Sandler (1996, op cit), and Offerman (1997).

[14] A very interesting example is provided by the canonical public goods example of the lighthouse. As acknowledged by Coase (1974) and elaborated by later writers many early lighthouses were not only financed by port charges imposed on ships assumed to have benefitted from their use but this financing was sometimes remitted to private entrepreneurs on the basis of royal “patents” which provided them the right to levy tolls. It is evident that this was a “public-private” institutional solution rather than a “market” solution as such.

Share your perspective