# Political Philosophy and Welfare Economics **URL:** **By** Tyler Cowen **Published:** September 1, 1987 --- “Nonconsequentialist or agent-relative theories claim that the rightness or wrongness of an action results from moral constraints we place on the behavior of agents.” Ever since the pioneering work of Robert Nozick and John Rawls, the field of overlap between political economy and political philosophy has been one of the fastest growing and most exciting areas in the social sciences. Many of the world’s leading economists and philosophers have exchanged ideas on the nature of the good society and its justification. Before the initiation of these debates in the early 1970s both political philosophy and welfare economics were in a relatively stagnant state. The period between Sidgwick and Rawls yielded little of significance in political philosophy, perhaps due to the influence of positivism. Likewise, the “new welfare economics” dating from the 1930s had developed into a super-refined mathematical structure that was largely irrelevant to many interesting normative questions. Developments in the 1970s and 1980s can be interpreted as a rediscovery of the aims of political philosophy as it was developed in the post-Renaissance period by such figures as Hobbes, Locke, Rousseau, and Kant. Normative inquiry should be directed at obtaining useful knowledge about the structure and nature of available alternatives, rather than the analysis of linguistic propositions or refining the art of model-building for its own sake. The novelty in recent work is that old questions in political philosophy are combined with new methods, specifically rational choice theory and economic science. These tools allow a more advanced and clearly defined analysis of political theories. Rawls, for instance, used an integration of choice theory and economics in his _Theory of Justice_ (Cambridge: Harvard University Press, 1971) to help determine what rational individuals would choose from behind a veil of ignorance. Robert Nozick’s _Anarchy, State and Utopia_ (New York: Basic Books, 1974) also uses economic analysis to outline the evolution of the minimal nightwatchman state as well as to puncture holes in alternate theories. For an introduction to the entire field, the reader is advised to start with two collections of essays, _The Limits of Utilitarianism_, edited by Harlan Miller and William Williams (Minneapolis: University of Minnesota Press, 1982) and _Utilitarianism and Beyond_, edited by Amartya Sen and Bernard Williams (Cambridge: Cambridge University Press, 1982). Both books are based on conference proceedings and provide an overview of many of the critical issues in this field. Nozick and Rawls are examples of philosophers who have been influenced by economics. Such influences have also run from political philosophy to economics. Many economists, such as Sen, have been inspired by political philosophy to demand that welfare economics compare different conceptions of well-being, as well as examine important normative questions such as the nature of rights, the importance of nonutilitarian considerations, possible conflicts between rights theories and utilitarianism, endogenous preferences, and optimal population theory. I shall examine a number of topics in the areas of overlap between political philosophy, rational choice theory, and welfare economics. ## Consequentialism Perhaps the single most important question in moral philosophy is whether different states of affairs ought to be ranked on the basis of outcome (their consequences) or whether they should be evaluated according to “agent-relative” considerations. Under consequentialism, outcomes are evaluated on the basis of their objective consequences apart from the intention of any specific agent. Nonconsequentialist theories claim that we can (or should) only ask whether an action is good or bad based on the formal characteristics of the agent’s choice. Utilitarianism is the best-known form of consequentialism, as it evaluates outcomes on the basis of total utility. Consequentialism, however, is broader than utilitarianism and may encompass rights considerations. A rights-consequentialist theory, for example, would assert that we ought to prefer the outcome that contains the smallest number of rights violations. The important feature for the ranking is the consequence (how many rights were violated?) and not what need be done to minimize the number of rights violations. In contrast, nonconsequentialist or agent-relative theories claim that the rightness or wrongness of an action results from moral constraints we place on the behavior of agents. These moral constraints prescribe rightness or wrongness regardless of whether the agent’s action may lead to good or bad consequences. These consequences may include the violation of process values, which imply that certain ways of treating others have intrinsic value. For example, a consequentialist theory (under _ceteris paribus_ conditions) might recommend murdering one innocent man to prevent a madman from murdering five other innocent people. An agent-relative approach, on the other hand, would not call for summing up either consequences or the violation of process values from an agent-neutral point of view. It is wrong for an individual to kill an innocent person, regardless of how good the consequences of such an act may be or even if it entails a maximization of certain process values (for example, nonaggression). Rights theories have traditionally been agent-relative, but this is quite different from relativism. Instead, they consist of objectively valid rules which stipulate that individuals must respect the sphere of rights possessed by others. These rules, however, are adopted for reasons other than the fact that they may have good consequences. Such rules cannot be abandoned when they lead to bad consequences or even to a net increase in the number of rights violations (as in the madman example above). The classic defense of agent-relative theories can be found in Bernard Williams’s “A Critique of Utilitarianism,” (in _Utilitarianism: For and Against_, edited by Williams and J. Smart, \[Cambridge: Cambridge University Press, 1973\]). Williams argues that individual lives cannot have integrity or continuity unless we allow agents to give priority to their own values and plans. Consequentialist theories are not compatible with individuals having meaningful lives, since such theories force individuals to abandon their personal plans any time they are capable of producing good consequences for others. For instance, this might require individuals to devote their entire lives to charitable work. > Nonconsequentialist or agent-relative theories claim that the rightness or wrongness of an action results from moral constraints we place on the behavior of agents. The most powerful critique of agent-relative theories to date can be found in Sidney Scheffler’s _Rejection of Consequentialism_ (Oxford: Oxford University Press, 1982). Although Scheffler is sympathetic to agent-relative theories, he finds them hard to justify. If we assert that one agent ought not to violate the rights of others, this implies that rights violations are in some sense bad. Scheffler presents a lengthy argument that this tends to collapse into arguing that we should minimize rights violations, a form of consequentialism. A defense of agent-relative theories can be found in Sen’s article “Rights and Agency,” (_Philosophy and Public Affairs_ Winter 1982) and in his exchange with Donald Regan in the Spring 1983 issue of the same journal. In a collection of his essays, Sen also criticizes Paretian theory at length for its inability to accommodate nonutility and nonconsequentialist criteria (_Choice, Welfare and Measurement_ \[Cambridge: MIT Press, 1984\]). Y. K. Ng responds to Sen in the 1979 issue of _Economic Journal_ (“Welfarism: A Defence Against Sen’s Attack”). Another significant contribution to the debate between consequentialist and nonconsequentialist theories is provided by Derek Parfit’s brilliant _Reasons and Persons_ (Oxford: Oxford University Press, 1984). In the first section of this book, Parfit argues that consequentialist and nonconsequentialist views are not necessarily incompatible. To be coherent, both need to be formulated in a manner that is not collectively self-defeating. A simple example of a collectively self-defeating theory would be the stipulation that everyone seek his self-interest without constraint. Collectively applied, this would likely make everyone worse off, just as in the classical prisoners’ dilemma game. Parfit believes that if both consequentialist and nonconsequentialist theories were properly formulated in nonself-defeating fashion there would be no conflict. Parfit’s ingenious attempts do not succeed in producing such a theory, however, as is noted by several commentators in the Parfit symposium in the June 1986 issue of _Ethics_. The conflict between consequentialist and nonconsequentialist theories is a real one and cannot be ignored or explained away by arguing for a general harmony between the recommendations of both approaches. > I attempt to show that an individual’s adoption of a nonconsequentialist moral code can often be justified on consequentialist grounds even when we abstract from its effects on the well-being of other individuals. At the 1986 general meeting of the Mont Pelerin Society in St. Vincent, Italy, Roland Vaubel and Hannes Gissurarson debated whether the defense of a liberal order should be based on a consequentialist or an agent-relative theory. While Vaubel attempts to reconcile consequentialist and “procedural’ justifications of a free society, he insists that procedural considerations are grounded on a consequentialist foundation, and that, “To be viable, morals have to be consequentialist and be concerned with the well-being of a group.” Gissurarson raises the questions of “What consequences are to be considered in judging an ethical or political system?” and “Which consequences are to be considered beneficial?” While Gissurarson also sees no necessary contradiction between consequentialist and nonconsequentialist ethics, he perceives nonconsequentialist considerations underneath consequentialism itself, notably in the historical “experience of individuality or the acquisition of the disposition and ability to make choices.” For reasons that cannot be adequately set forth in this essay, my own leanings are toward consequentialism. Nonetheless, like Parfit, I believe that when consequentialism is properly formulated it possesses a considerable degree of harmony with many habits and tendencies that may appear irrational or bad when viewed in terms of the “common good.” In one of the chapters of my dissertation (“Dynamic Inconsistency and the Theory of Consumer Choice,” in _Essays in the Theory of Welfare Economics_, Harvard University, 1987), I attempt to show that an individual’s adoption of a nonconsequentialist moral code can often be justified on consequentialist grounds even when we abstract from its effects on the well-being of other individuals. ## Formulating Moral Codes A related body of literature examines whether it is possible to formulate a moral theory that is not collectively self-defeating. As evidenced by the work of Regan (_Utilitarianism and Cooperation_ \[Oxford: Oxford University Press, 1980\]) and of Parfit, this is a formidable task. While most of this discussion has been within a utilitarian framework, the structure of the arguments could apply to many other moral theories as well, including rights theories. Critics often charge, however, that such theories are in principle incapable of determining or defining when a rights violation has taken place. This challenges defenders of the “reasonable man” approach to provide a consistent definition of reasonableness. One problem in formulating a viable moral theory is the relevant “unit of choice.” For example, specifying that individuals should take account only of the marginal impact of their actions can generate situations where the combination of a large number of marginally insignificant actions produces a significantly bad result. Automobile pollution may be one example. Even altruists who maximized the welfare of others on the basis of the marginal increment of their actions could generate a result that would make everyone worse off. Increasing returns to scale in bad effects can make the sum of marginal products worse than their constituent elements. > From the Hayekian perspective, moral codes are the product of evolution, classic instances of phenomena that are ‘the result of human action but not of human design.’ On the other hand, specifying that individuals take account of the average product of their actions also encounters problems. It could imply a waste of resources in any situation where marginal and average products were not equal. Blood donations may have a high average product (the first set of donations are important) but if the value of my marginal donation is very low, should I necessarily give my blood? Regan has attempted to circumvent such problems in _Utilitarianism and Cooperation_ by developing a theory of rational cooperation. Such questions, along with the closely related issue of rule-versus act-utilitarianism, are debated in the October 1985 issue of _Ethics_ by Regan, Allan Gibbard, John Harsanyi, Jon Elster, and Brian Barry. A very different perspective on many of these problems is provided by the work by Friedrich Hayek (see, for instance, his three-volume set _Law, Legislation and Liberty_ \[Chicago: University of Chicago Press, 1973, 1976, 1979\]). Hayek has emphasized the impossibility of a fully rational and articulable moral code, a theme which will be further developed in his forthcoming book, _The Fatal Conceit_. The problems dealt with in the literature on formulating moral codes, discussed here, can be interpreted as a rigorous proof of Hayek’s claim. From the Hayekian perspective, moral codes are the product of evolution, classic instances of phenomena that are “the result of human action but not of human design.” Rather than being interpreted as impossibility theorems, some of the problems and paradoxes discussed above may simply be indications that some of the thinkers have misinterpreted the results they have generated. The analyses of the paradoxes and problems have been important and illuminating because we have obtained considerable insight into the structure of certain types of moral propositions. This sort of rigorous structural analysis, however, should not be expected to provide definitive answers to moral questions. The Hayekian tradition suggests that moral codes cannot and should not be given perfectly articulable, rationalist foundations. This would imply interpreting Hayek within the philosophical framework provided by Willard Quine (_Word and Object_ \[Cambridge: Harvard University Press, 1958\]) and Richard Rorty (_Philosophy and the Mirror of Nature_ \[Princeton: Princeton University Press, 1979\]). Moral codes are a set or “web” of propositions that possess a certain amount of internal coherence and external practicality. This web cannot be given rigorously derived foundations and should instead be judged on the basis of its usefulness and in accordance with certain of our moral intuitions. Such a Hayekian research program, however, remains in its infancy and requires considerable development before it could be considered a serious contender with the other approaches discussed above. Thomas Schelling’s _Strategy of Conflict_ (Cambridge: Harvard University Press, 1960) is a classic work in game theory whose connections with Hayekian themes have not been sufficiently developed. Schelling demonstrates how game-theoretic problems in the moral sphere require a spontaneous-order solution that cannot be fully articulated or even defined apart from the process of their emergence. Some of the relations between the work of Hayek, Schelling, and David Hume have been analyzed in Robert Sugden’s excellent book, _The Economics of Rights, Cooperation and Welfare_ (Oxford: Basil Blackwell, 1986.) ## Self-Management and Endogenous Preferences > Recognizing the endogeneity of individual preferences creates difficulties for both utilitarian and rights theories. Another set of interesting questions can be found in the literature that analyzes problems involving preference formation, self-management, and possible conflicts of interest among different preferences in the same person. The seminal essay in this area is Robert Strotz’s “Myopia and Inconsistency in Dynamic Utility Maximization” (_Review of Economic Studies_, 1955-56). It was not until the work of Elster and Schelling, however, that the implications of Strotz’s insights were fleshed out (Ulysses and the Sirens \[Cambridge: Cambridge University Press, 1979\] and _Choice and Consequence_ \[Cambridge: Harvard University Press, 1984\], respectively). Recognizing the endogeneity of individual preferences creates difficulties for both utilitarian and rights theories. Some of the problems for utilitarianism are pointed out in Elster’s “Sour Grapes — Utilitarianism and the Genesis of Wants,” an essay in the above-cited Sen and Williams collection, _Utilitarianism and Beyond_. Since social systems are a significant factor in shaping individual preferences, the ability of a system to satisfy the preferences it generates cannot be regarded as a decisive argument in favor of that system. We also need to ask if desirable preferences have been generated. > An effective critique of paternalism requires a theory of the spontaneous order of the self, just as Hayek has developed a theory of the spontaneous order of the market. The notion of “coercion” required by a rights theory is very difficult to define in any situation where conflicting impulses exist within the individual. What appears to be paternalism may actually be satisfying an individual’s higher-order preference. Many cigarette smokers, for instance, might consider themselves better off if smoking were to be outlawed, even if they know that they will feel “coerced” when the addictive element of their personality goes to light up a cigarette. This theme has been developed in the Schelling book. _Paternalism_ (Minneapolis: University of Minnesota Press, 1984), edited by Rolf Sartorius, also contains several relevant essays, both pro and con. Because of issues springing from the literature on self-management and endogenous preferences, I do not think that the “demonstrated preference” critique of paternalism (see Rothbard’s _Toward a Reconstruction of Utility and Welfare Economics_ \[New York: Center for Libertarian Studies, 1977\]) is sufficient, whether from a utilitarian or rights framework. Individuals may demonstrate conflicting and/or inconsistent preferences. Instead, an effective critique of paternalism requires a theory of the spontaneous order of the self, just as Hayek has developed a theory of the spontaneous order of the market. This critique would be based on the knowledge problem, that is, the inability of a central authority to have sufficient knowledge to effectively plan the course of an individual’s self-development. Making an explicit connection between the theory of spontaneous order and this form of government intervention would be an extension of the approach taken by the great classical liberal Wilhelm von Humboldt in his 1798 classic, _The Limits of State Action_ (translated and reprinted, Cambridge: Cambridge University Press, 1969). A similar approach is taken in an important essay by James M. Buchanan, “Natural and Artifactual Man” (_What Should Economists Do?_ \[Indianapolis: Liberty Press, 1979\]). ## Population Theory Most political philosophers would agree that part four of Parfit’s _Reasons and Persons_ has been both the beginning and the end of modern normative population theory. Parfit argues that we do not yet have a rational normative theory of population that avoids counterintuitive and repugnant conclusions. Existing rights theories are unable to handle the issue of future generations. Unborn individuals presumably do not have rights until they are born, but this implies that we can damage the interests of future generations without limit, provided that we do so before they are born (ruining the environment would be an example). Another difficulty is what Parfit calls the Non-Identity Problem. If our harmful actions also change the identities of future generations (through changing the procreation of present generations), we can even say that everyone is made better off by the policy. An irresponsible nuclear-waste disposal policy may result in the birth of a large number of crippled or deformed children. These same children, however, if given the choice would not object to such policies — with a different policy they would not even have had the opportunity to be born. Utilitarianism does not do much better — maximizing the sum of utilities can give rise to what Parfit labels the Repugnant Conclusion. This implies that the ideal society might contain very large numbers of people with the smallest possible imaginable unit of positive utility, that is, the smallest amount of utility at which life is still worth living. If the numbers of people are large enough, this world would receive a higher ranking than other more reasonable-sounding alternatives. Asserting that such large numbers are impossible is not a valid objection. Would we prefer them if they were possible? Since the principle of total utility can lead to this Repugnant Conclusion, Parfit suggests that the principle fails. Other forms of utilitarianism are examined by Parfit but they do not fare much better. Population dilemmas remain a puzzle. In a chapter (“Normative Population Theory”) of my dissertation I have attempted to formulate a version of the contractarian approach that would avoid the problems discussed above. Population alternatives can be evaluated by asking which would be preferred if an individual had to live out or experience every life in the world he chooses. This approach, however, encounters serious difficulties with aggregating different sets of preferences. > The resurgence of normative moral and political philosophy in recent decades owes much to the cross-fertilization of two hitherto independent fields of thought. The new political economy emerging from this development offers many exciting opportunities for classical liberal scholarship. _Tyler Cowen has just completed his doctorate in economics at Harvard University and is currently a postdoctoral fellow at the Center for the Study of Market Processes at George Mason University._