Externalities and the Limits of Pigovian Policies

ABSTRACTPigovian policy is developed in economic theory as an efficient resolution to externality problems. The use of this type of policy to resolve real-world externality problems, including climate change in the form of carbon taxes, assumes that the Pigovian policy result derived in theory holds in the real world. By examining the bridging conditions from theory to the real world, I argue that this assumption holds only in an ambiguously defined subset of externalities. It is thus unclear when Pigovian policy could be coherently applied, which is problematic given its widespread use in critical policy contexts.KEYWORDS: Carbon taxclimate policyexternalitymarket-based policyphilosophy of economics AcknowledgmentsI am grateful to Margaret Schabas, Alison Wylie, John Beatty, and Joseph Heath for their helpful comments and suggestions. I also benefitted from audiences at several conferences and seminars, including the Philosophy, Politics and Economics Research Seminar at Arizona State University and the Biennial Meeting of the Philosophy of Science Association.Disclosure statementNo potential conflict of interest was reported by the author(s).Notes1. Pigovian policy is so called after the economist Alfred Pigou (Citation1920/2017) who is often cited as the progenitor of externality theory in economics (Medema, Citation2020).2. This is distinct to the question of whether implementing a specific Pigovian policy is justified, which has been extensively examined by economists. There are several well-known reasons that implementing such a policy might not be justifiable in practice, including high administrative costs relative to the expected benefits of the policy and insufficient information on the part of the policymaker about individual preferences (see Baumol, Citation1972; Tol, Citation2018; Winston, Citation2006). In order to assess the justifiability of implementing a particular Pigovian policy response, however, it must first be established that such a policy response is at least coherent in the given context. This paper is focused on this more basic question.3. Pigovian policy features prominently in justifying and guiding government carbon tax policies (Rennert & Kingdon, Citation2019). It is also endorsed by many leading economists in an open letter published in the Wall Street Journal (Akerlof et al., Citation2019) as well as in the informal ‘Pigou Club’, which is a list of economists and policymakers who endorse Pigovian policies (Mankiw, Citation2009).4. It is important to note that this paper does not call into question the general use of a carbon tax to address climate change. Instead, it calls into question the use of a particular type of tax that is justified by externality theory and which calls for a specific carbon tax rate determined by estimates of the value of the externality generated by carbon dioxide emissions.5. The concept of an externality in economics, however, is notoriously difficult to precisely define. In the latter half of the twentieth century, prominent economists including Kenneth Arrow (Citation1969), James M. Buchanan and William C. Stubblebine (Citation1962), Ronald Coase (Citation1960/2013), and James Meade (Citation1952) developed competing accounts of the concept of an externality. For more on the history of market failure and the concept of an externality, see Nathalie Berta (Citation2017), Nathalie Berta and Elodie Bertrand (Citation2014), Maurice Lagueux (Citation2010), Alain Marciano (Citation2011), Steven Medema (Citation2009, Citation2014, Citation2020), and Andreas Papandreou (Citation1994, Citation2003).6. Following economic terminology, I reserve the term optimality to refer to Pareto efficiency or Pareto optimality, which is also sometimes called social optimality. An outcome is Pareto optimal if there is no possible reallocation of resources that would make at least one agent better off without making any other agent worse off.7. A potential Pareto improvement exists when resources could be reallocated such that at least one person would be made better off and no agent would be made worse off.8. Ronald Coase’s (Citation1960/2013) influential critique of the Pigovian approach to externalities holds instead that externalities persist because of high transaction costs. If the two relevant parties could negotiate a contract over the externality, an efficient solution would ensue. Nevertheless, climate change is a case where Pigovian analysis is the appropriate approach. David D. Friedman (Citation2000), for example, argues that while Coase offers the correct general analysis, Pigovian analysis of externality problems is correct ‘under special circumstances, situations in which transaction costs are high, so that transaction between parties can be safely ignored, and in which the agent deciding which party is to be held liable already knows who the lowest-cost avoider of the problem is’ (Friedman, Citation2000, p. 42). Climate change is a clear case in which transaction costs are high because it is not possible for the current generation (those generating emissions) to form contracts with future generations (those impacted by the emissions). It is also a case in which there is good reason to believe that the current generation is the least-cost avoider of the problem. This is because there are hard limits to climate adaptation which indicates that it cannot be the case that the cost for the future to adapt to unmitigated climate change is lower than the cost for current generation to both mitigate and adapt (Pörtner et al., Citation2022).9. This paper focuses on Pigovian taxes because this is the most common form of Pigovian climate policy; however, these other forms of Pigovian policy are isomorphic in theory, and therefore the argument presented here is not specific to taxes and also apply to cap-and-trade as well as subsidy policies. If policymakers preferred a cap-and-trade policy, for example, and thus aimed to create a market for emissions, they would use the quantity at the Pareto optimal level to determine where to set the cap, and thus how to assign property rights over carbon emissions (which, in theory, could be assigned to polluters or those affected by pollution). Hence, what is called Pigovian theory in economics incorporates Coase’s (Citation1960/2013) insight that externalities are reciprocal and as such can be addressed by assigning property rights to either party. For more on the isomorphism of these policies in contemporary theory, see Martin L. Weitzman (Citation1974). The efficacy and political feasibility of these various policies, however, are not necessarily isomorphic in the real world. Lonergan and Sawers (Citation2022), for example, argue that positive incentives (e.g. subsidies) are often more effective than negative incentives (e.g. taxes), in part, because they are more likely to be viewed favorably by taxpayers.10. Versions of this framework can be seen in Charles Plott (Citation1966), Kenneth Arrow (Citation1970), Agnar Sandmo (Citation1980), and Hal Varian (Citation1994), for example.11. Those who prefer to skip the mathematical language of the theory can safely focus solely on the prose in this section.12. Derived utility is vi(p, wi, h) = max ui(xi, h) subject to p • xi ≤ wi ≡ øi(p, h) + wi. Since prices are unaffected by the choice of h, the price vector is suppressed. Therefore, derived utility is expressed as øi(h). See Mas-Colell et al. (Citation1995, pp. 352–354) for more detail.13. Note that contrary to the insights of Coase (Citation1960/2013), the model under scrutiny in this paper assumes that an analysis of the costs of administering the policy is separable from an analysis of the value of the externality. Hence, the model assumes that the reallocation of the unpriced good is costless, with the expectation that the administration costs and efficacy of a particular policy are assessed at a later stage of the policy analysis. Given that this paper aims to develop an internal critique of externality theory, it focuses on the first step of this process.14. Note that preference satisfaction refers to a state in which the object of a preference is realized, not the feeling of satisfaction (D. M. Hausman, Citation2012).15. Note that this terminology references contemporary Pigovian policy, not the type of externality or policy discussed directly by Pigou (Citation1920/2017). Pigou’s conception of an externality aligns more closely with the general view of externalities as unpriced spillover effects and, furthermore, Pigou’s economic framework for analyzing externalities differs significantly from contemporary economics. The term ‘Pigovian’ refers to a perceived economic tradition arising out of Pigou’s work. See Banzhaf (Citation2020) and Medema (Citation2020) for more on this topic.16. In economic terms, this means there is a corner solution in the two-agent constrained optimization framework.17. The scenario in which the budget constraint determines whether an externality is merely an unpriced spillover effect on welfare or a Pigovian externality raises well-known issues with the justifiability of implementing a particular Pigovian policy. One might reasonably question whether it is appropriate for a public policy to respond to income-constrained preference satisfaction (rather than preference satisfaction alone, for example), and thus instantiate an outcome dependent on existing income inequality (for example, see Anthoff & Tol, Citation2010; Sagoff, Citation2008). This paper sets aside this issue because it aims to examine the coherence of a Pigovian policy rather than the justifiability of implementing such a policy in a specific context.18. That is, where there is an internal solution to the joint welfare maximizing problem.19. As explained in the previous section, economics tends to hold that the converse is also true: if an individual’s welfare is reduced by an unpriced action taken by another individual, then this individual must have a willingness to pay to decrease the activity (at some level of income and at some quantity of the harmful activity, and barring non-economic values) that expresses both their preference satisfaction and welfare.20. This interpretation of Pigovian externalities resembles Buchanan and Stubblebine’s (Citation1962) account of relevant and irrelevant externalities. According to Buchanan and Stubblebine, externalities are irrelevant if individuals fail to eliminate them through exchange. This is because the persistence of an externality indicates that individuals have determined that it is costlier to eliminate the externality than to let it persist. Therefore, they argue that the persistence of some (irrelevant) externalities is consistent with a Pareto optimal state of affairs. The occurrence of exchange over an unpriced activity thus distinguishes between relevant and irrelevant externalities for Buchanan and Stubblebine (Citation1962), which implies that there is never a need for a Pigovian policy intervention. This is an inadequate account of policy-relevant and irrelevant externalities in the case of climate change in which exchange over the externality in question is not possible. In contrast, according to the view presented in this paper, the presence of untapped gains from exchange (not the occurrence of an actual exchange) based on preferences that are informative of welfare gains distinguishes Pigovian externalities from mere unpriced spillover effects. This is an important distinction especially in the context of climate change because it captures the value of the externality despite the time lag between the production and effect of the emissions. For more on Buchanan’s views on externalities, see James M. Buchanan (Citation1962, Citation1969) and Alain Marciano (Citation2011).21. In the case where insufficient income causes the divergence between gains from exchange and welfare gains, we would instead interpret hypothetical gains from exchange as what individuals would be willing to pay for an unpriced activity if they had sufficiently unconstrained income.22. In practice, the measurability of hypothetical gains from exchange could be delivered by a proxy; for example, lost GDP per capita caused by flooding could be used as a proxy for how much individuals should be willing to pay for improved flood mitigation infrastructure. An understanding of the concept being proxied, however, is required in order to justify the choice of proxy. This paper focuses directly on these foundational concepts.Additional informationFundingThis paper draws on research supported by the Social Sciences and Humanities Research Council.