Cost-Benefit Analysis as an Expression of Liberal Neutrality

Joseph Heath

The past few decades have seen an expansion in the use of cost-benefit analysis as a tool for policy evaluation in the public sector. This slow, steady creep has been a source of consternation to many philosophers and political theorists, who are inclined to view cost-benefit analysis as simply a variant of utilitarianism, and consider utilitarianism to be completely unacceptable as a public philosophy. I attempt to show that this impression is misleading. Despite the fact that when construed narrowly, cost-benefit analysis does look a lot like utilitarianism, when seen in its broader context, in the way that it is applied, and the type of problems to which it is applied, it is better understood as an attempt by the state to avoid taking sides with respect to various controversial conceptions of the good.

In traditional cost-benefit analyses of public projects, every citizen’s willingness to pay for a project is given an equal weight. This is sometimes taken to imply that cost-benefit analysis is a democratic method for making public decisions, as opposed to, for example, political processes involving log-rolling and lobbying from interest groups. Politicians are frequently criticized for not putting enough emphasis on the cost-benefit analyses when making decisions. In this paper we discuss the extent to which using cost-benefit analysis to rank public projects is consistent with Dahl’s (1989) criteria for democratic decision-making. We find several fundamental conflicts, both when cost-benefit analysis is used to provide final answers about projects’ social desirability, and when used only as informational input to a political process. Our conclusions are illustrated using data from interviews with Norwegian politicians.

Public Administration and the Liberal State Joseph Heath [draft 24/09/14] Chapter 5. Cost-Benefit Analysis as an Expression of Liberal Neutrality The past few decades have seen a steady expansion in the use of cost-benefit analysis as a tool for policy evaluation in the public sector. In 2003, for instance, the Green Book was introduced in the United Kingdom, with its requirement that “all new policies, programmes and projects, whether revenue, capital or regulatory, should be subject to comprehensive but proportionate assessment, wherever it is practicable, so as best to promote the public interest... This is achieved through: identifying other possible approaches which may achieve similar results; wherever feasible, attributing monetary values to all impacts of any proposed policy, project and programme; and performing an assessment of the costs and benefits for relevant options.”1 The book as a whole is described, with admirable directness, as providing “binding guidance for departments and executive agencies.”2 And in the United States, where many progressive groups have historically refused to participate in costbenefits analysis (on the grounds that it is part of a broader anti-regulatory agenda), it is increasingly being argued that these groups need to learn to accept the approach, and use it to their advantage.3 This slow, steady creep has been a source of consternation to many philosophers and political theorists, who are inclined to view cost-benefit analysis as simply a variant of utilitarianism, and consider utilitarianism to be completely unacceptable as a public philosophy. This interpretation of cost-benefit analysis as a type of applied utilitarianism is, of course, exacerbated by the fact that many of its most prominent defenders are utilitarians (or moral consequentialists), and defend it on the grounds that it approximates what they take to be the correct moral view.4 Furthermore, there is the obvious fact that cost-benefit analysis has at its core a calculation that is consequentialist, aggregative, and appears to presuppose a welfarist theory of value – all positions that are associated with utilitarianism. Finally, and perhaps most damningly, cost-benefit analysis seems to exhibit what many people take to be the central flaw of utilitarianism, viz. that it does not respect the distinctness of 1 HM Treasury, The Green Book (2003), p. 1 (https://www.gov.uk/government/publications/the-green-book-appraisal-andevaluation-in-central-governent) accessed Sept. 22, 2014. 2 Green Book, p. v. 3 Richard L. Revesz and Michael A. Livermore, Retaking Rationality (Oxford: Oxford University Press, 2008) 4 For example, see Matthew D. Adler and Eric A. Posner, New Foundations of Cost-Benefit Analysis (Cambridge, MA: Harvard University Press, 2006), or Eric Schokkaert, “Cost-Benefit Analysis of Difficult Decisions,” Ethical Perspectives, 2 (1995): 71-84. persons. This is reflected in a variety of ways, including an apparent commitment to distributive neutrality (the view that, as long as the total quantity of welfare is constant, it is a matter of indifference from the standpoint of society how that welfare is distributed) as well as its willingness to place a valuation on human life (suggesting that, if one persons’ death would produce a sufficient compensating benefit to others, then society should be happy to let that person die). These observations do conspire to present what is, admittedly, a rather damning circumstantial case. The situation, however, is more complex than it may at first appear. The first and most important thing to appreciate is that cost-benefit analysis is not literally a decision procedure, which is to say, it is never applied “baldly” to any particular policy question. It is always embedded in a set of more complex institutional decision procedures, which impose a set of constraints that reflect essentially non-utilitarian concerns.5 My central objective in this paper will be to show that, because of this, costbenefit analysis as institutionalized in representative modern welfare states is better understood as reflecting a commitment to set of concerns that are shared very widely by political liberals, including a commitment to resolving collective action problems, to treating all citizens equally, and to having the state avoid taking sides on controversial questions of value. Because the introduction of CBA has been so controversial, its use has generated a bewildering array of objections, along with replies to the objections, replies to the replies, and so on, creating a body of literature that is very difficult to survey and assess. As my point of entry, therefore, rather than trying to present a synoptic overview, I will instead present what I think of as a “primrose path,” that starts with a fairly uncontroversial form of liberalism, but ends with a full-fledged commitment to comprehensive CBA of the sort mandated by the Green Book. I invite the reader to imagine a junior analyst, who shows up for work on the first day of her new job in the civil service, having not one speck of sympathy for utilitarianism, but affirming rather some form of deontological liberalism. Yet as she confronts a series of public policy questions, she finds herself accepting the reasonableness of CBA as a way of picking out the best policy. The process is not entirely one-sided – certain constraints get added to the CBA procedure along the way, in order to make it acceptable. What emerges in the end is a more complicated, but normatively more sophisticated, decision procedure, which I will try to articulate. 1. ‘Embedded” CBA It is important to recognize, from the very beginning, that CBA as institutionalized is very 5 A point emphasized the Lewis A. Kornhauser, “On Justifying Cost‐Benefit Analysis,” Journal of Legal Studies, 29:S2 (2000): 1037-1057 at 1053. different from the abstract decision procedure described in introductory public finance textbooks.6 (I will use the term “bald” CBA to refer to the textbook version, and “embedded” CBA to refer to the procedure within its broader institutional context.7) The overall impact of CBA, for instance, is strongly affected by the choice of problems to which it is applied. To take one, particularly obvious example, when “comprehensive” CBA was first introduced in the United States by the Reagan administration, it was applied to all new major regulatory initiatives, but was specifically not applied to deregulatory initiatives.8 So, for example, a proposal to “save drivers money by adding lead to gasoline” would have been exempt from scrutiny, because it could be achieved simply by eliminating various regulations on fuel additives. Yet if later one wanted to reintroduce the ban on leaded gasoline, a full-scale CBA would be required. Thus CBA was introduced with the not-so-subtle goal of making it harder to regulate than to deregulate. This example is intended only to show that the choice of problems to which one applies CBA can be very important. And while the decision to apply it asymmetrically, to regulation but not deregulation, reflects little more than an ideological hostility to government, there are many other forms of selective application that have a more principled basis.9 For example, governments do not apply cost-benefit analysis to programs that are purely, or even primarily, redistributive. It is not difficult to see why. Suppose one were to propose a new program called “taking money away from the rich and giving it to the poor.” Using a standard willingness-to-pay (WTP) willingness-to-accept (WTA) framework for measuring the benefits and costs (respectively), this program would be guaranteed to be at least CBA neutral. (The amount that a person should be willing to pay, to receive a transfer of $1000, should be exactly the same as the amount that a person should be willing to accept, in order to accept a loss of $1000, viz. $1000.) Because money is subject to diminishing utility, the UK Green Book also recommends imposing a set of distributional weights on the WTP/WTA amounts, stipulating that benefits to the lowest income quintile (“the poor”) be subject to a multiplier in the range of 1.9-2.0, while benefits to the upper quintile (”the rich”) be discounted, with a multiplier in the range of 0.4-0.5. Using the upper end of these values suggests that taking $1000 away from a rich person and giving it to a poor person produces a net benefit to society of $1500. Of course, the fact that “bald” CBA has this consequence should come as no surprise to anyone, 6 The one I happen to own provides a good example. See Harvery Rosen, Beverly George Dahlby, Roger Smith and Paul Boothe, Public Finance in Canada, 2nd edition (Toronto: McGraw-Hill, 2002). 7 Cass Sunstein, having started as an academic advocate of CBA, then accepted an administrative position where he was involved in assessing CBAs for a broad range of agencies in the U.S. federal government, has written a number of interesting articles that attempt describe the differences between textbook CBA and CBA as practiced. 8 Revesz and Livermore, Retaking Rationality. 9 Cass Sunstein provides several examples, “The Cost-Benefit State,” Chicago Working Papers in Law and Economics. since this sort of egalitarianism is a well-known implication of utilitarianism (when applied to a world in which money, or consumption generally, is subject to diminishing returns). So the fact that CBA is not applied to this sort of a problem (e.g. “how high should Old Age Security payments be?”), should serve as the first hint that, underlying CBA, there is not actually a commitment to utilitarianism. When one looks at the problems that CBA is applied to, what one finds is that they are almost exclusively instances of market failure (or, more generally, collective action problems). In the United States, Executive Order 12866, issued by President Clinton, makes this reasonably explicit with respect to regulations: “Federal agencies should promulgate only such regulations as are required by law, are necessary to interpret the law, or are made necessary by compelling public need, such as material failures of private markets to protect or improve the health and safety of the public, the environment, or the well-being of the American people. In deciding whether and how to regulate, agencies should assess all costs and benefits of available regulatory alternatives, including the alternative of not regulating.”10 Despite a somewhat complex history of subsequent amendment, the impact of this order has been to put the burden squarely upon agencies proposing new regulations to specify the market failure that they are responding to.11 Thus the goal of a regulation cannot simply be to increase social welfare, it must be to solve a collective action problem that private parties are unable to resolve through voluntary contracting. This suggests that the underlying rationale for CBA is actually Paretian (i.e. aimed at producing Pareto improvements), rather than utilitarian. Given the importance of this distinction, it is probably worth dwelling on it for a moment, in order to make sure that it is not subject to any misunderstanding. An intuitive way of representing the difference between Paretianism and utilitarianism is to show how they resolve a two-person distribution problem. Figure 1 represents a set of possible social states, with the status quo corresponding to a particular allocation of utility between two individuals. The line U shows the set of other social states that contain the same aggregate utility level as the status quo.12 Utilitarianism therefore ranks any point to the north-east of this line as superior to the status quo. The Pareto principle, by contrast, ranks a particular social state as superior only if it makes at least one person better off, and no one worse off. Thus it ranks every social state to the north-east of only the point representing the status quo – including those strictly north and strictly east – as superior. Informally, this is the set of win-win (or at least no-lose) outcomes. Changes that make one person better off but 10 http://www.epa.gov/fedrgstr/eo/eo12866.htm 11 Sunstein, “The Real World of Cost-benefit Analysis: Thirty-Six Questions (and Almost as Many Answers),” Columbia Law Review, 114 (2014): 167-211 at 191 & 193. 12 For purposes of illustration, it is helpful to set aside the issue of interpersonal comparability in the utility scales. Disagreement about what one person's utility is “worth,” in terms of the other's, concerns the slope of U, but not the basic form of the social welfare function. leave the other worse off are Pareto-noncomparable, and are not ranked by the Pareto principle. Thus it has nothing to say about the more controversial set of proposals, to the north-west and the south-east of the status quo. Utilitarianism, by contrast, makes the controversial claim that it is acceptable to harm one person, to the benefit of another, so long at the magnitude of the benefit is greater than or equal to that of the harm. The Pareto principle, by contrast, is about as attractive and uncontroversial as a normative principle can get (although it achieves this through the somewhat dubious means of simply saying nothing about the set of cases most likely to arouse controversy). Pareto (x' ≥ x,y' ≥ y) Person y status quo U (x+y=k) Person x Figure 1. Utilitarianism vs. Paretianism In textbook treatments, the way that CBA is usually presented to students is not by starting with the moral case for utilitarianism. Typically the argument starts with the Pareto principle, which is often treated as either self-evident or self-recommending. The argument then moves along quickly to the observation that, under real-world conditions, we are unlikely to find many pure Pareto improvements. Making an omelette always seems to require breaking at least one egg. (For example, no matter how magnificent and useful a particular infrastructure project would be, there always seems to be one landowner who refuses to sell at any price.) We do not want it to be the case that a harm to a single person is able to hold up a vast project, which could benefit many others. So we need some way of articulating the idea that the benefits to the many should, on some occasions, outweigh the harm to the few. The Kaldor-Hicks principle is then proposed as an alternative criterion – that if the gains to the winner are great enough that they could be used to compensate the loser, then the proposal is ranked better than the status quo. This is sometimes known at the “potential Pareto” principle, because whenever it is satisfied, it means that you could, in principle, carry out the compensation, which would then make the outcome an actual Pareto improvement. There is often some mild sleight of hand involved in the presentation of this argument. The way that the Kaldor-Hicks principle is presented often makes it sound like only a slight modification of the Pareto principle. Once it is clear, however, that compensation need not actually be paid, it is obvious that the principle is practically equivalent to full-blown utilitarianism (they diverge only in cases where, even though a particular change would increase aggregate utility, compensation is not possible because of differing attitudes towards the medium in which compensation would be paid). This observation is what has led critics, such as Amartya Sen, to argue that the Kaldor-Hicks principle is either redundant or unacceptable.13 If the intention is to actually pay the compensation, then there is no need for a “potential” Pareto principle, all one needs is the Pareto principle, because after the compensation has been paid an actual Pareto improvement will have been achieved. So in order for the principle to have any purpose, there must not be any intention to pay compensation. But if that is the case, then the principle is subject to all of the traditional objections to utilitarianism. In particular, it seems difficult to justify imposing a loss on someone just by saying “well someone else is gaining more than you are losing.” The major claim I want to make in this paper is that, even though bald CBA uses the KaldorHicks criterion, embedded CBA does not, but is instead used in the pursuit of actual Pareto achievements. Primarily this is achieved though the choice of problems to which it is applied. Formally, this can be represented by subjecting the basic calculus to two “screening” procedures, one on the input and the other on the output side.14 On the input side, policy proposals that are not essentially a response to collective action problems (e.g. questions about the progressivity of the tax code, changes to the criminal law), are set aside as unsuitable for CBA. Once the CBA is conducted (and, as we shall see, an enormous number of further normative judgments go into determining how it should be conducted), then the results are weighed against other considerations. Most obviously, in cases where there would be significant costs imposed upon some people, the question of whether they will actually be compensated may become important. If it is impossible to compensate them (e.g. because they cannot 13 Amartya Sen, “The Discipline of Cost-Benefit Analysis,” Journal of Legal Studies, 29 (2000): 931-952 at 947. 14 See Rosemary Lowry and Martin Peterson, “Cost-Benefit Analysis and Non-Utilitarian Ethics,” Politics, Philosophy and Economics, 11 (2011): 258-279. be identified), then one might decide not to pursue the policy, and so the potential welfare gain would be screened out on the output side. Thus the major commitment at the heart of CBA can be expressed in the form of a principle of efficiency: 1. Pareto-improvement. The state should attempt to resolve collective action problems that private actors are either unable to resolve, or unable to resolve at reasonable cost, through voluntary contracting. This is combined with a second set of commitments, which are central to liberalism, and closely connected to the privileging of the Pareto standard. The first is a commitment to liberal neutrality, broadly construed. The idea is that when it comes to controversial questions of value, where there seems to be some sort of reasonable disagreement, the state will refrain from using its powers of coercion to the benefit of just one party. Suppose, for example, that 45% of the population worships Kronos, and wants to build a temple to the one true God (viz. Kronos). The other 55% of the population worships Demeter, and wants to build a temple to the one true God (viz. Demeter). The commitment to neutrality suggests that the state should not be in the business of determining which group is right, or using public funds to support one or another project. It might decide to avoid the entire question, by insisting that only private funds be used for temple construction. Or if there is a case for state involvement, it might decide to allocate funds to both projects with a 45/55 split. This may seem obvious, but it is worth observing that, prior to the dawn of liberalism, European states were very much involved in making judgments about these sorts of questions, and using state power to enforce them. Furthermore, the principle of state neutrality does not follow from the democratic principle (the worshippers of Demeter, having won a majority in a general election, might decide that it is their prerogative, as the majority, to use state funds to build just their temple), or from the concept of individual rights. It is a sui generis constraint that captures a great deal of what we mean when we describe a particular political regime as liberal. And as we shall see, a great deal of the resistance to CBA is a consequence of the difficulty many people have accepting the constraints imposed by the principle of liberal neutrality when it comes to their own, most deeply-held (yet controversial) values. It is, after all, much easier to say that the state should not take a particular set of values as a basis for policy when they are values that one considers woefully mistaken than it is when they are values that one considers correct: 2. Liberal neutrality. The state should try to avoid taking sides on controversial questions of value. There is a close connection between this principle of neutrality and the Pareto principle. The central attraction of a Pareto improvement is that it allows the state to bring about a state of affairs that is deemed better for all, each from his or her own perspective – which is to say, according to each individual’s own system of values. Pareto efficiency is therefore a paradigm instance of a normative principle that the state can act upon without taking a position on any first-order questions of value, which is to say, without violating neutrality. And yet if the state “does not have a view” on the correctness of certain questions, and still it needs to regulate in that area, what is it supposed to do? The answer is that, as much as possible, it should defer to citizens’ own judgments about these questions, and to the extent that it does so, it should assign their views and interests equal weight. This is very closely connected to the idea of neutrality. In the same way that the state should not just pick a winner among a contested set of values, neither should it privilege some particular subset by assigning it greater weight in its deliberations. Thus the third principle, which also expresses a very standard liberal commitment, is to a certain form of equality: 3. Citizen equality. In areas of reasonable disagreement, the state should assign equal weight to the values and interests of each citizen in deciding which policies to pursue. Despite the fact that the Kaldor-Hicks criterion at the heart of bald CBA is notorious for assigned no weight to distributive considerations, I will try to show that a commitment to citizen equality is an important feature of embedded CBA. In particular, while textbooks often describe a perfectly general decision procedure, CBA as institutionalized is done quite differently in different areas. If one looks at the way that life is treated, for instance, sometimes CBA assigns a value to “lives saved,” but in other cases it assigns a value to “life years saved.” The difference is quite significant, in that the latter may privilege the interests of the young over the old in a way that the former does not. The fact that the calculations are done differently in different domains is not just eclecticism, but constitutes a principled difference. It reflects that fact that, underlying the application of CBA is the commitment to a particular conception of equality among citizens, but that the way equality gets interpreted differs in different policy domains (e.g. in workplace safety compared to health care resource allocation). Each of these three principles has been formulated in a way that is intentionally loose (for example, through inclusion of a “reasonability” constraint). That is because my goal is not to provide an airtight formulation of the central tenets of liberalism. On the contrary, my goal is to show that the commitment to a vague form of liberalism, of the sort represented by these three principles, is sufficient to motivate the use of cost-benefit analysis in most if not all of the policy domains in which it is currently being applied. This is, I will argue, what explains its appeal – departures from CBA seem to violate one of these three principles, and are, to that extent, widely perceived to be unacceptable (or else motivated by political ideologies that are, at some level, illiberal). It is worth noting, perhaps just in passing, that none of these three principles seems to involve any commitment to consequentialism, Furthermore, all three could easily be justified in a nonconsequentialist fashion – for example, they all follow quite naturally from certain forms of contractualism. Naturally, they create a normative structure within which consequences will be relevant to the evaluation of a policy. A collective action problem, after all, is defined in terms of the consequences of individuals' maximizing choices. This does not mean, however, that the Pareto principle must be justified through appeal to these consequences (and nothing but these consequences, as the consequentialist would have it). Thus the framework that has been presented here is perfectly compatible with all but the most implausibly strict deontological moral views. This is just to accentuate my basic point, which is that there is a lot more daylight between CBA and utilitarianism than is commonly realized. My goal then in the next five sections will be to describe a “primrose path,” that might lead a person committed to these three principles to accept CBA, including some of its most controversial aspects, such as the practice of assigning a monetary value to lives lost. I do so by way of a series of examples, of policy scenarios where state officials are inclined to use CBA. Some of these are rather stylized, for purposes of exposition – in real life, they would not be subject to CBA, for a variety of reasons, including the fact that they fall below the “threshold of significance” (e.g. because a CBA is very expensive to carry out properly, the requirement in the U.S. is that only regulations with an impact over $100 million require a full-blown CBA). In the discussion that follows I will be ignoring these complications, choosing instead examples that bring out in the most transparent way possible the normative issues at stake in the decision to use CBA. 2. Provision of a public good Suppose that a municipality comes into possession of some land (perhaps it seizes a derelict property for non-payment of taxes), and must decide what to do with it. Two proposals are made for its use: first, the land could be cleared and converted to a neighbourhood park, or second, it could be sold to a developer. It is the responsibility of the planner to decide which constitutes the best use. Now if the choice were between selling it to a developer who wanted to build a shopping mall, and selling it to a developer who wanted to build a condominium complex, then there wouldn’t really be any difficulty. One could simply sell it to the higher bidder, knowing that in so doing, one would actually be ensuring that the land was being put to its best use. The underlying question is what “the people” want most. The market is the mechanism through which their desires are transmitted and aggregated. If there are a lot of people who really want to live in that neighbourhood, and there is not enough housing, relatively speaking, then they will be willing to pay a lot for the condominiums, making that venture the more profitable one. If there is instead a shortage of stores in the area, then the shopping mall will be the more profitable venture, and so the commercial real estate developer will be willing to bid more for the land. It is noticeable, however, that there are never any developers lined up to buy land and convert them to public parks. The reasons for this are obvious – they could never get their money back, because they couldn’t charge admission, or if they did, very few people would come. Thus much of the benefit associated with the construction of a public park take the form of positive externalities (or as we say in the internet age, the “value stream” created be the park is difficult to “monetize”). Markets will therefore systematically underproduce these goods (this is the “market failure”). We refer to them as public goods in the informal sense, of being relatively non-rival and non-excludable. It is worth dwelling for a moment on the nature of the market failure in this case, because it is not actually all that difficult to exclude people from a park, one need only build a fence around it and charge admission. Many people have “backyards,” which are essentially private parks (open green space in the city for recreational use), condominiums have grounds, which often function as “parks” for building residents, and some neighbourhoods even have gated parks, where only residents in adjoining houses have the key. In Paris one can find children’s playgrounds in public parks where one must pay an admission charge to get in. So it is perfectly possible to provide parks and park-like amenities as a private or as a club good.15 What is important to observe is that in all of these cases, what is being provided is a different type of good. This is an observation made most famously by Fred Hirsch, who observed that having to pay for certain goods may change their character, in a way that amounts to quality degradation. The most compelling example he gives of this is sexual intercourse, which most people are inclined to regard as a different good when paid for with cash than when supplied through systems of informal 15 On club goods, see James Buchanan, “An Economic Theory of Clubs,” Economica, 32 (1965): 1-14. reciprocity. Similarly, the changes that must often be made to a good in order to make it excludable, such as building a wall or a fence around a piece of land, change the character of the good. Thus a public park, like the public beach or a public festival, is a distinct type of good, where part of the enjoyment that people get from its consumption arises precisely from the fact that entry and exit is free and open to all. Because of this, provision of such a good will often generate a collective action problem. Voluntary contribution will not generate the optimal level of provision, because of free rider problems (and the fact that it is impossible to identify some potential users – such as tourists, or people who live far away but occasionally pass through the neighbourhood). Thus the “desire” that people have for a park will not be transmitted faithfully into market demand for a park. Even actual willingness to pay for a park through a contributory arrangement may fail to generate market demand. The fact that the good generates this problem is in one sense quite accidental. The good itself has no special moral property, such that it must be provided by the government; it’s just an ordinary good that people like to consume, which happens to be very difficult to organize the provision of through any institution other than government. Thus there is no reason to think that there is a morally correct level of provision, other than what the people themselves want to consume. After all, there’s no point providing parks if people are not going to use them (just as there is no point lowering peoples’ taxes, if what they would really like to buy is more parks). And if some strange technological intervention were to make the underlying market failure go away (or if some burst of publicspiritedness made voluntary financing by “passing the hat” feasible), there would be no special objection to allowing market forces (i.e. supply and demand) to determine the level of supply. Thus what the state needs to engage in, when deciding whether to build the park, is essentially a “market-simulating” exercise. It needs to produce the outcome that the market would have produced, had the market failure not occurred. (Again, this is not because the market is special, but because the market is a general mechanism for channeling resources to their best employment, and in this particular case, the state shares that same objective.) Of course, we know how much people want the condominium project, because this takes the form of actual market demand, which is what generates the price that the developer is willing to pay. What we do not know is how much people want the park, because it does not show up as market demand. The only way to get at it is to try to ascertain a hypothetical willingness to pay (WTP). This can be done using a variety of methods, the main ones being the stated preference method (do a telephone survey of the neighbourhood, asking people what they would be willing to pay to have a new public park) and the indirect valuation (or “revealed preference”) method (do a comparative study, to look at what impact public parks have on property values in the surrounding neighbourhood – this provides a sense of how much people are actually willing to pay for access to parks). Once this is done, we need to compare one value to the other, to see which project is more desirable. But this is, of course, just what it means to do a cost-benefit analysis, since the “cost” of any one project is just its opportunity cost, which is to say, the benefit that is foregone when a decision is made to not do any of the alternative projects. Thus the cost of building a park is the fact that the land is then not used for private development, which is reflected in the price of the land, which constitutes the revenue that the state will not be receiving if it decides to build the park. Thus the focus on “cost” in CBA is potentially misleading: when performing a cost-benefit analysis, what one is really doing is comparing the benefits of a particular project to the benefits of the presumptively second-best use of the resources, in order to see which is greater. Thus it could be called “benefit-benefit comparison” or “relative benefit assessment,” which would in some ways be more confusing, but would have the advantage of emphasizing the fact that the comparison of costs and benefits in a CBA is not the same thing as the balancing of “gains to the winners” and “losses to the losers” that one would see in a utilitarian calculus. I think it can be shown that making any decision in this case other than the CBA-justified one would violate one of the three principles. First of all, the basic rationale for state action is to resolve a collective action problem among potential users of the park. The case is somewhat complicated by the fact that the park is what Evan Davis calls a “minority public good,” viz. it will not be used by everyone. This appears to give the project a Kaldor-Hicks quality, where some people are being asked to forgo a benefit that they want (e.g. whatever they would buy with their tax savings, if the land were sold to a developer), in order that others may enjoy having a park. This is why, Davis argues, it is important that the state provide a balanced mix of such goods, with the hope that in general and over the long term, everyone will benefit to a reasonable degree. (Not everyone benefits equally from snow removal and pothole repairs, since some people do not drive; not everyone benefits equally from mosquito control, since some people do not enjoy sitting outdoors; and so on.) This, combined with Tiebout sorting in the municipal context (people can move to neighbourhoods that provide the balance of amenities most closely matched to their tastes), creates a general presumption that these state activities are Pareto-improving. The next question then concerns the level of provision of these public goods. Here what is important to the justification of CBA is the idea that there is no correct level of provision independent of what people happen to want. This follows from neutrality and equality, because any deviation from it would, in effect, involve taking sides on the question of whose proposed use is “better” and assigning greater weight to the interests of that party. This sort of partiality is not unheard of. In the early 20th century, many urban planners were persuaded that parks would produce “moral improvement” in working classes – a judgment that was particularly influential in the decision to raze working class slums and replace them with residential towers surrounded by green space. This backfired spectacularly, as the green space remained largely unused, making it a magnet for crime. As a result, the sort of “parks are good for you” judgment that prevailed in the first half of the 20th century is now regarded as involving an impermissible partiality toward a particular set of values. Some people like trees and green spaces, some people don’t. Planners tend to share the values of those who do, but it is a clear violation of liberal neutrality to use that as a basis for decision. No matter how much planners may like parks, they must ultimately defer to patterns of use. 2.1 The commodification objection Perhaps the most global and all-encompassing critique of CBA is the argument that it “commodifies” public goods, and therefore represents a fundamentally wrong-headed way of thinking about goods that are, as Elizabeth Anderson puts, “not properly regarded as mere commodities.”16 Now there is no doubt an element of truth to this, since when calculated a WTP value, then using it as the basis for a decision, we are in effect asking “what would the market do?” where “the market” in question is a hypothetical market in which all goods are available, and all costs and benefits are fully reflected in their prices. Yet if all that it meant to “commodify” something was to subject it to such acts of the imagination, then it is not clear where the harm would lie in “commodification.” After all, no one is talking about about turning public parks into actual commodities. The plan is still for it to be provided as a public park, free and open to all. Asking people “how much would you pay for a new park?” as a way of deciding whether to build it, or how large it should be, does not really change anything about the eventual status of the park as a public good. In fact a casual visitor to the park, sitting on a bench enjoying the afternoon sun, might be surprised to discover that the surrounding environment had been turned into a “mere commodity” by the fact that a CBA was performed prior to the decision to turn the land into a public park. This suggests that the commodification charge needs to be made a bit more carefully. When it comes to debates over, say, the buying and selling of transplant organs, the question is whether a particular good should become an actual commodity, that can be bought and sold on private markets. But when the government performs a CBA in order to determine how large a park should be (or, for example, to decide whether the public health service should provide a vaccination without charge to the 16 Elizabeth Anderson, Value in Ethics and Economics, p. 190. public), there is no question of actually commodifying the good in question.17 Just thinking about the good as if it were a commodity is not itself a harm, and does not violate any taboo against the buying and selling of it. In order to show that there is harm, those who make the criticism need to show that this way of thinking will distort decision-making, leading to incorrect levels of provision of the good (including perhaps, in some instances, failure to provide it at all). This is what Anderson tries to show, although there are significant problems with the exposition. She starts by providing a helpful definition of what she means by commodification: “A good is treated as a commodity if it is valued as an exclusively appropriated object of use and if market norms and relations govern its production, exchange and distribution.”18 From this definition, it seems clear that the park is not being commodified. But she then goes on to claim that cost-benefit analysis “measures people’s valuations of these goods in market transactions and hence, only as they are valued as privately appropriated, exclusively enjoyed goods. This assumes that the public nature of some instances of these goods is merely a technical fact about them and not itself a valued quality. The possibility that national parks... might be valued as shared goods does not enter into its evaluations.”19 It is difficult to know what to say here, other than to observe that all of the substantive allegations made are false. It is simply not true that when you call someone on the phone and say “how much would you pay to have a public park in your neighbourhood?” that you are asking them “how much would you pay to have a private park that only you could use in your neighbourhood?” (this would be equivalent to asking “how much would you pay for a larger backyard?”) Apart from the fact that you wouldn’t need to ask this (you could just look at how much people actually pay for larger backyards), what you are asking about specifically is the private value (i.e. the value to the individual) of a public amenity. It is not the case that CBA tries to estimate the value of public goods as if they were “privately appropriated” or “exclusively enjoyed.” The same is true of revealed preference methods, such as looking at property values in surrounding neighbourhoods. Even though you are looking at the value of private commodities, you are doing so as a way of estimating the value of public amenities. It is a well-known feature of the real estate market that houses in “nice” neighbourhoods sell for a lot more money than houses in “not-sonice” neighbourhoods. Many cities have a housing stock that is architecturally fairly homogenous, so it is possible to find what amounts to the “same house” in a number of different neighbourhoods. Because the “private good” (i.e. the actual house and yard) in each case is the same, the differences in price 17 Many critics of CBA are either confused on this point, or write in a way that encourages significant confusion. Anderson. Steven Kelman, for example, makes it sound as though CBA involves actually assigning prices to things, “Cost-Benefit Analysis: An Ethical Critique,” Regulation, 33 (Jan, 1981): 33-40 at 39. 18 Anderson, Value in Ethics and Economics, p. 193. 19 Ibid, p. 193. must be due to externalities, both positive and negative. With a good data set and sufficiently refined statistical techniques, this allows one to determine how much people value these externalities (which will include public goods, such as proximity to good schools and clean parks, the availability of transit, etc. as well as public bads, such crime, the noise from a nearby freeway, traffic congestion, etc.) This allows one to determine how much people really care about these various things (so that if it turns out that schools make a huge difference, while parks only modestly so, it might make sense to invest more money in schools, rather than parks). There is simply nothing commodifying about this. More generally, there is nothing paradoxical about the idea of eliciting people’s private valuation of a public amenity, or in thinking that the public valuation of public amenities should be a function of these private valuations. It should be mentioned that, when the numbers come back, the results sometimes can put pressure on one’s commitment to liberal neutrality. That is because the planners who do these studies all tend to belong to roughly the same social class, and therefore tend to have views on these questions that are both similar to one another and strongly held. For example, there is a marked tendency to think that people ought to value neighbourhood green space over proximity to a freeway. So when it turns out that people in certain neighbourhoods have the ‘wrong’ preferences, according to one’s class-specific values, there is a temptation to want to impugn the methodology of the study, or the validity of a decision procedure that assigns those ‘incorrect’ preferences the same weight as one’s own. It is important to observe, however, that the problem here lies not with the CBA; the source of the tension is actually the more fundamental liberal commitments that are embodied in CBA. Thus when a critic like Anderson suggests that a process of “deliberation” should be used instead of CBA, one suspects that this is motivated, at least in part, by the thought that such a process would result in everyone having preferences much closer to her own. (In other words, it is unlikely that she is promoting deliberation as a way of giving herself greater opportunity to discover the error of her own ways. It is more likely motivated by a desire to put a “thumb on the scales” in favour of certain perfectionist values, in a way that CBA does not permit.)20 2.2 The inequality objection There is a second objection that can be made, which is far more serious than the “commodification” claim. The example that I chose of a neighbourhood park was intended to neutralize it to some degree, but still it will have occurred to many people. The point of a CBA is basically to determine how much people, in the aggregrate, want something, and whether they want it enough to 20 This issue is discussed further in chapter 6. justify forgoing what they would have to forego, should the state choose to provide it. This sort of aggregative exercise requires interpersonal comparisons, and so any CBA must select some sort of a metric, to serve as a basis for these interpersonal comparisons. Standard CBA uses money. Now, given the rhetorical appeal of the commodification objection, it is important to emphasize that the reason for assigning a dollar value to costs and benefits has nothing to do with imposing “market norms,” and it certainly does not reflect any underlying commitment to actually selling the good in question. Money is used simply as a metric, or as a basis for comparing the benefits going to one person to the benefits going to some other person. But of course, money is a very poor metric, because it is not valued equally by all persons. Indeed, money is subject to diminishing utility, which means that it is more highly valued, the less of it one has. As a result, a poor person will be willing to give up less of it, in order to achieve some benefit, than a rich person. (Which is to say, the WTP of the wealthy will be greater than the WTP of the poor for benefits that are of the same magnitude, in terms of utility.) As a result, it has seemed obvious to many that WTP numbers are going to be skewed, to the benefit of the wealthy. This is an analytic consequence of the idea that money has diminishing marginal utility, as can be seen in Figure 2. A person with b amount of money (and thus at utility level 2) would be willing to pay c-b for a single-point gain in utility, while a person with d amount of money (and thus at utility level 4) would be willing to pay e-d for a single-point gain. Thus if one were to auction off a good that promised a single-point gain in utility, the wealthier person would be willing to pay much more for it. This means that if you do a CBA, eliciting WTP figures from people in different social classes, the amounts that you get are going to be distorted by the value that individuals place on money.21 Whether or not this is a problem, however, depends very much on the project being contemplated, as well as the class profile of those affected. It may not be a problem if one is contemplating a project that will affect the entire population, and what one is looking for is an average WTP. In this case, the high WTP values one gets by asking the rich will counterbalance the low WTP values that one gets from asking the poor, producing the average value. The difference becomes problematic, only when one is considering projects that will have a differential impact on the rich as the poor. In such cases, it is true that using an unadjusted WTP value will systematically privilege the interests of the wealthy. 21 See Anderson, where objection is incorrectly put. Issue is not that the wealthy actually have more money, and therefore have greater ability to pay – since no one is going to make them pay anyhow. The issue is that declining utility affects valuation, skewing the answers given. utility a b c d e money Figure 2. Diminishing marginal utility of money increases willingness-to-pay To see this, one need only consider a case in which a large municipality found itself with funds to purchase land for a new park, and had to decide which neighbourhood to build it in. If it decided to find out where a park was most needed by surveying residents from one end of town to the other, asking people how much they would be willing to pay for a park, the consequence of this would almost certainly be a recommendation that the park be built in a wealthy neighbourhood, simply because the residents there would have the greatest willingness to pay for a park. But this is not because they value the park more, it is actually because they value money less (i.e. the marginal dollar). Thus if one were to sketch out the social welfare function reflected in bald CBA, it is easy to see that it is anti-egalitarian (or that it endorses the perverse principle of St. Matthew, “to him that hath shall be given, to him that hath not shall be taken away”). Figure 3 represents these social welfare functions as indifference curves (where every point along the curve is considered just as good as any other, from the standpoint of social welfare). The utilitarian indifference curve U, just as in Figure 1, is a straight line, indicated that as long as the sum of individual satisfaction is the same, the utilitarian is indifferent with respect to distribution. CBA with a monetary metric, by contrast, generates the convex curve A, where the more unequal the two parties are, the more the richer one is willing to pay, relative to the poorer one, and so the greater the priority assigned to his interests. By contrast, as Amartya Sen has observed, it is generally thought that a plausible theory of justice should be concave, as in the prioritarian indifference curve P.22 Such as SWF tolerates deviations from equality, but as the level of inequality increases, the greater the benefits to the better-off must be in order to justify any further deviation. A P Person y U Person x Figure 3. Prioritarian and anti-prioritarian social indifference curves Now some people have argued that the A curve is just fine, since this is how the market distributes most other goods.23 Cass Sunstein, for instance, has observed that if the city were to build the park in a poor neighbourhood, even though the WTP there was quite low, it would essentially be forcing the residents to accept a good that they themselves would willingly give up in exchange for money at market prices (if, for example, a developer were to come along and offer to buy the land). So this starts to look a bit like forcing people to consume a good that they don’t want. Nevertheless, there are many who feel that, given the range of undeserved inequalities in the distribution of wealth, the 22 Amartya Sen, “The Discipline of Cost-Benefit Analysis,” p. 938. 23 Cass Sunstein, “Cognition and Cost-Benefit Analysis,” p. 34 n. 76. See also Kip Viscusi, “Risk Equity,” Journal of Legal Studies, 29 (2000): 843-871 at 857-8. state is under an obligation to do more in the way of promoting equality than to mimic the pattern of distribution favoured by private markets. This is reflected in the fact that, at the “back end,” goods are typically paid for by a tax system that incorporates a measure of progressivity. And at the “front end,” goods are often – although not always – distributed in ways that are intended to be independent of either ability or willingness to pay. Ideally all transactions could be organized in this way, but practical difficulties in implementing such a scheme leave us no choice but to defer to the market pattern of allocation in many cases. But this represents nothing more than a concession to practicality. In cases where the state is providing a good directly, there is no reason to privilege the market pattern.24 As a result, it is not uncommon for planners to adjust the WTP values in a CBA, in order to reflect the income level of respondents. If there is reason to think that one proposal is likely to benefit the poor disproportionately, while another is more likely to benefit the rich, then one can simply impose a multiplier on the reported WTP values, reflecting one’s best estimate of the diminishing marginal utility of money. This has the effect of “unbending” the A curve, so that it more closely approximates the utilitarian curve U. One can even give the CBA an egalitarian bias, by increasing the weights until the curve more closely resembles P. (For instance, one could look at the progressivity of the income tax system, infer from it the shape of the implicit SWF, then derive from it a set of multipliers that will give the CBA the same shape.) None of this is technically difficult, and as mentioned earlier, the U.K. government Green Book recommends that it be done whenever appropriate and feasible, using multipliers in the ranges set out in Table 1. Quintile Weight bottom 1.9-2.0 2nd 1.3-1.4 3rd 0.9-1.0 4th 0.7-0.8 top 0.4-0.5 Table 1. Distributive weights suggested by U.K. Green Book 24 For more extensive discussion, see Joseph Heath, “Efficiency as the Implicit Morality of the Market,” in Morality, Competition and the Firm (New York: Oxford University Press, 2014). Now the use of such multipliers does raise a number of delicate issues. My point is merely to show that CBA is a very flexible tool, which can be made more or less egalitarian. Furthermore, it is worth observing that it is very difficult to find anyone willing to endorse bald CBA, with its convex SWF, across the board. Even Sunstein, who endorses strict market simulation in the standard run of cases, draws the line when it comes to the VSL (or “value of a statistical life”) measure. This will be dealt with more thoroughly in section 4, but for now suffice it to observe that if one uses WTP to price the risk of death associated with various activities, then one will find that the poor are willing to pay less than the rich are for a reduction of risk of given magnitude. This implies that the VSL for a poor person should accordingly be lower than for a rich person. If taken seriously, this would mean that employers need not spend as much on workplace safety if their workers are poor, that the state need not spend as much money on road safety in poor neighbourhoods, and so on. To avoid this outcome, a standard VSL, the same for every person, is used. One way of thinking about these sorts of adjustments in the way that the CBA is performed is to imagine that another value, viz. equality, is trumping the concern for aggregate social welfare that is articulated by the CBA. This is, I believe, not the best way of thinking about it. On the contrary, the concern for equality is not trumping the concern for welfare; the concern for equality is being built into the CBA. A better way of thinking about it is to see that every CBA has built into it a particular conception of equality – some more persuasive than others. That having been said, there is no question that a concern for equality can raise some difficult informational challenges. For example, it is important that if one is using distributive weights on one side of a calculation, one must use them on the other as well. This can be difficult if, as is often the case, one has a market valuation of the costs but not the benefits. It becomes particularly important if there is asymmetry between the two, e.g. if the poor would be the major users of the park, but the rich would benefit most from a condominium development. In this case, it is not sufficient merely to ask about income when surveying for WTP values, and then adjust the valuations of the benefits up if the respondents are low income. One must also adjust the market valuation of the costs down, if one has reason to believe that the primary beneficiaries of the alternative use would be high income. The latter is, needless to say, very difficult to do. That having been said, if the problems raised by the use of a monetary WTP measure are significant enough, there may be the option of abandoning it in favour of some other metric. In this regard, it is important to keep in mind that there is nothing intrinsically significant about money (pace proponents of the “commodification” critique), it is merely be used as a basis of establishing interpersonal comparisons. If, however, one can find some other basis of interpersonal comparison, which is not subject to the same distortions, then one may well be better off using that. In health care resource allocation, for example, life expectancy is used as the basic metric, rather than WTP. Similarly, municipal planners may choose to use “commute time” as a metric of comparison for transit planning. There may be an element of stipulation in this – for example, even if an extra year of life expectancy may not be of equal value to all persons, the state may nevertheless be justified in treating it as though it were. These metrics all have their flaws and limitations, the point is simply that if they are able to provide the basis for a more compelling assessment of the relative benefits of a set of policy options, then there is no barrier either in principle or in practice in using them. Finally, it is important when considering the possibly inegalitarian features of CBA not to commit the “nirvana fallacy,” of comparing it against some entirely idealized decision procedure, particularly not one that makes use of impossible-to-obtain information. There are many examples in the United States of statutes that specifically prohibit the use of cost-benefit information in determining the way that they will be administered. Yet it is not clear that these operate to the benefit of the poor. On the contrary, the political system, the judicial system, as well as the informal will-formation mechanisms of civil society are all substantially biased against the poor. As a result, it is not clear that the poor will fare any better under a “democratic” decision-making process than under a technocratic CBA. Kip Viscusi, for example, as observed that the distribution of “superfund” environmental cleanup resources in the United States – a well-known example of non-CBA justified spending – basically mirrors the distribution of “pork barrel” spending. By contrast, there can be little doubt that the imposition of CBA with a standard valuation of life has had equalizing consequences. So while I am not aware of any systematic investigation of the distributive consequences of CBA versus other decision procedures, it is important to remain realistic about the challenges involves in designing an alternative to CBA that will not exhibit many of the same defects that one was hoping to correct. 3. Imposing a regulation Suppose that our municipality has recently made a number of zoning changes, to allow mixeduse development in the downtown core. Slowly people are beginning to migrate back to the city from the suburbans. Several warehouse conversion projects are underway, being turned into condominiums, but also mixed-use commercial-industrial-residential complexes. This is successful, not just at attracting new residents, but also new businesses – bars and restaurants, but also small-scale manufacturing. Unfortunately, with success comes conflict. The residents are beginning to complain about a number of things, including noise levels. They want new regulations, constraining the amount of noise that businesses can produce (and at what hours). In particular they want manufacturers to install noise-abatement technology (such as sound-proofing insulation). Unlike the case of a public park, which is not so difficult to imagine being a Paretoimprovement, regulation seems to have much more of a win-lose structure. Any noise regulation that is introduced (assuming that there is currently none) is clearly going to impose a cost on business, while generating a benefit for residents. The question that CBA can be used to answer is whether the benefit to residents is large enough to justify the cost imposed on business. This seems very much like a utilitarian calculation – and indeed, it is often described that way (e.g. “regulations seek to counteract externalities by restricting behavior in a way that imposes harm on an individual basis but yields net societal benefits.”25) These appearances, however, are misleading. The purpose of CBA, when examining a regulation, is actually to determine where the Pareto efficient outcome lies (and thus, whether the status quo arose only through market failure). This is the point that Ronald Coase’s celebrated paper, “The Problem of Social Cost,” sought to establish. His argument, however, is on several points quite subtle and so merits repetition. One of Coase’s ambitions was to criticize the assumption, made by Alfred Pigou, that merely because there was an externality being produced in a particular market, that the resulting outcome would be inefficient.26 Coase showed that market outcomes may still be efficient, even in the presence of externalities, depending on how important those externalities are. He illustrated this with a classic example, of a railroad track that passes through a farmer’s field. The faster the trains travel, the more sparks they throw off, and the more likely they are to set the farmer’s field ablaze. Under the status quo, the sparks are a negative externality, the cost of which is manifest in the form of crops lost to fire. There are two solutions to the problem: either the railroad could run the trains slower, or the farmers could increase the set-back between the railbed and the land where their crops are grown. Naturally the farmers prefer that the trains run slower, while the railroad prefers that the set-back be increased. Coase used this example to make one obvious point, which no reader has failed to grasp, and two less obvious ones, which are sometimes overlooked. The obvious point is that one need not necessarily impose a regulation in order to resolve this problem, all it would take is a clear assignment of rights. Furthermore, there is one sense in which it doesn’t really matter which party is given the rights, because the two sides can be expected to negotiate to the efficient outcome regardless of their 25 Colin Camerer, Samuel Issacharoff, George Loewenstein, Ted O'Donoghue and Matthew Rabin, “Regulation for Conservatives: Behavioral Economics and the Case for 'Asymmetric Paternalism',” University of Pennsylvania Law Review, 151 (2003): 1211-1254. 26 Pigou’s view is quite intuitive, and still very widely held. Elizabeth Anderson, for example, assumes it when she says “Markets do not produce efficient outcomes when market transactions impose welfare changes or ‘externalities’ on third parties,” p. 192. The correct claims is that market may not produce efficient outcomes when there are externalities, not that they do not. starting positions. Suppose, for example, that the farmers would suffer a loss of $80 from increasing the setback, while the railroad would suffer a loss of $100 by slowing down its trains. If one gives the railroad the right to run its trains as fast as it likes, then it will continue to do so, and farmers will have to absorb the $80 loss. But if instead one gives farmers the right to have their fields protected from sparks, it actually won’t change the outcome – the railroads will still run the trains fast, it is just that the railroad will be forced to negotiate with the farmers and pay them $80 to increase the setback. Either way the parties will negotiate to the efficient outcome. The difference between the two outcomes is purely one of distribution – which party has to bear the $80 loss. If the right is given to the railroad, the farmers bear the loss, if the right is given to the farmers, then the railroad has to bear it (by paying the farmers). This separation of the distribution issue from the efficiency issue is the first rather counterintuitive point that Coase makes.27 The second point is that the mere presence of a negative externality (in this case) is not necessarily inefficient, it all depends on how much the externality costs the person who suffers it, and how much it benefits the person producing it. In the case of the railroad and the farmers, the status quo ante, with the production of the sparks, is the efficient outcome – as witnessed by the fact that, even if the farmers are given the right to block production of the externality, they would choose not to do so, by selling the right to the railroads. So while the presence of the externality creates what many people are inclined to regard as a distributive injustice (i.e. the farmers are given no compensation for their lost farmland or ruined crops), it was not actually producing an inefficiency. Now there are many circumstances in which externalities are being produced, but the parties will not be able to negotiate a solution (typically because of transaction costs, understood broadly). Thus the state cannot just assign rights and let the parties decide, it must actually impose an outcome. In order to do so, it should be guided by some conception of what the parties would have decided, had they been able to negotiate freely. First and foremost, it must decide whether the parties would actually have negotiated a limit on the externality (and thus, whether the state should even regulate at all), and then, if it determines that they would have agreed to reduce it, what level of output they would have settled on. This is precisely what a CBA does. The important point is that this calculation is 27 This is one of the points that separates the way many economists think about social questions from more common-sense moral perspectives. For example, many economists are not overly concerned about theft, because that is merely redistributive. If someone sneaks into your house and takes something, the total wealth of the nation is neither increased nor deceased, a portion of it simply passes out of your hands into someone else’s. But if you go out and change your locks, or install a burglar alarm, in response to the theft, then that is a concern to the economist, because now real resources are being expended, not to increase to stock of wealth, but merely to ensure that a particular redistribution of that wealth does not occur. Thus the amount spent on theft-deterrence is a loss, from the standpoint of efficiency. independent of the distributive justice decision. For instance, if the state decided to look at the situation between the farmers and the railroad, with an eye to regulating the externality, a CBA would recommend against it, on the grounds that the cost of slowing down the trains ($100) exceeded the benefit ($80). It might, however, also decide that the farmers were owed some compensation by the railroad, but this would be on the basis of separate, distributive justice considerations, not the CBA. In the case of the condominium residents and their noisy neighbours, the logic is much the same. Suppose that in this case, the residents would be willing to pay $100 for a quieter environment, while it would only cost the manufacturers $80 to reduce the noise level of their operations. What this means is that the current state of affairs is inefficient – there is a mutually beneficial (i.e. Paretoimproving) transaction between the residents and the manufacturers that could be taking place, but is not, because of incompleteness in the system of property rights (i.e. the residents could pay the manufacturers $80 to reduce the noise level, leaving them with a welfare gain worth $20). Thus the current, inefficient state of affairs, in which the manufacturers make too much noise, persists only because of a market failure. This is precisely what the CBA shows when it comes back positive. So when the state imposes that outcome through regulation, it is realizing a Pareto improvement over the true status quo ante, in which it is not yet determined whether there will be noise or silence. The regulation, of course, creates only a potential Pareto improvement over the existing state of affairs, because it imposes the cost of noise abatement on the manufacturers without requiring any transfer from the residents. The question of whether or not to turn it into an actual Pareto improvement is a separate issue, which needs to be decided by considerations of distributive justice, such as whether the manufacturers were “entitled” to make the noise that they were making in the first place. In general, whenever the state regulates, it has the option of compensating parties, based on who “should” have had the rights in question. In the case of these classic pollution externalities it usually does not compensate, on the grounds that the rights would typically not have gone to the emitter, but rather to those affected. So the cost impact of the regulation is more like a seizure of ill-gotten gains. In other words, the regulation is motivated by (Pareto) efficiency concerns, and efficiency concerns alone. The subsequent decision, whether to compensate those who are adversely affected by it, is a separate, distributive justice decision. Put the two together, and it looks like a utilitarian decision, but it’s actually two separate decisions, one about efficiency, the other about distribution. If there is a loss imposed on one party, this is due to the distributive justice decision. Regulation w/out compensation ($500,$600) Regulation with compensation ($580,$520) residents ($500,$500) Status quo ($580,$500) No regulation manufacturers Figure 4. The decision to regulate Figure 4 sketches out the situation. It one takes the “no regulation” outcome as the status quo, then a move to regulation without compensation is a Kaldor-Hicks improvement, but not a Pareto improvement. Starting with the “no regulation” outcome begs the question, however, because it is not clear that the manufacturers should have been entitled to produce the noise – and absent the market failure that prevents negotiations with the renters, they might never have done so. Thus the status quo point from which all regulatory decisions should be made is the outcome in which neither party is exercising any of the rights that are up for grabs. From that point of departure, the outcome in which the manufacturers make noise and the outcome in which the renters enjoy silence represent Paretoimprovements – the task of the regulator is to pick one. And here one can see that the outcome in which manufacturers make noise is not on the Pareto frontier. So despite appearances, the state is really not doing anything very different when it imposes a regulation than what it does when it decides to build a park. In both cases it is trying to bring about an outcome that would have been brought about, had there not been incompleteness in the system of property rights. Provision of public goods is a response to the real market’s underproduction of positive externalities, regulation is typically a response to the market’s overproduction of negative externalities. 3.1 The rights objection One of the complaints often made against CBA is that it does not recognize, or assign any status to, the “rights” of individuals affected by a decision. Thus one might object to the very idea that residents should have to negotiate with manufacturers, regarding noise levels, on the grounds that they have a “right” to a good night’s sleep (or, more plausibly, that farmers should not have to negotiate with the railroad, on the ground that they have a “right” not to have their crops lit on fire). The issue becomes even more significant when the externality in question poses a danger to people’s health, or even their lives. For example, many people who object to the use of CBA in deciding on carbon abatement policies to address global climate change insist rather that the issue must be addressed within a “rights” framework.28 There are two responses to this. Implicit in the objection is the idea that rights function as “trumps,” in the way the Ronald Dworkin defined them, as strict deontic constraints (“rights are best understood as trumps over some background justification for political decisions that states a goal for the community as a whole”29). In some circumstances, rights do function this way, so that no matter how much welfare is at stake, the state cannot take action aimed at realizing it if doing so violates an individual’s rights. For example, prior to the Kelo v. City of New London decision in the United States, it was understood that the “power of eminent domain” could used be used to seize private property, or forcibly transfer title, if the property was to be put to a “public use” – such as construction of a road, or a park. So while the state might contemplate all sorts of development projects, only ones that involve permissible forms of land appropriation are worth doing a CBA on, the rest are simply “trumped” by the property rights of individual land-holders. This can be handled on the “input filter” side of the embedded CBA procedure, when it comes to selection of problems. The fact that CBA is done so broadly, however, is a reflection of the fact that very few rights really do function as “trumps” in this way, despite the assertions of some philosophers.30 For example, Simon Caney argues that CBA is inappropriate as a framework for thinking about climate change, and 28 Simon Caney,“Climate Change, Human Rights, and Moral Thresholds,” in Stephen Humphreys, ed. Climate Change and Human Rights (Cambridge: Cambridge University Press, 2009, Henry Shue. Henry Shue, “Bequeathing Hazards: Security Rights and Property Rights of Future Humans,” in Limits to Markets: Equity and the Global Environment, ed. by Mohammed Dore and Timothy Mount (Malden, Mass.: Blackwell Pubs., 1998) 29 See Ronald Dworkin, “Rights as Trumps,” in Jeremy Waldron (ed.), Theories of Rights (Oxford: Oxford University Press, 1984), pp. 153–67. 30 Steven Kelman writes that, “The notion of human rights involves the idea that people may make certain claims to be allowed to act in certain ways or to be treated in certain ways, even if the sum of benefits achieved thereby does not outweigh the sum of costs. It is this view that underlies the statement that ‘workers have a right to a safe and healthy work place’ and the expectation that OSHA’s decisions will reflect that judgment.” The first sentence constitutes a typical expression of the “rights as trumps” idea, but then the second sentence goes on to specify a right that could not possibly serve as a trump, simply because no workplace is every perfectly safe or healthy. For example, anyone who has to drive a vehicle as part of his/her job is thereby exposed to a non-negligible mortality risk. urges instead that we think about the problem within a rights framework. He then posits a set of very broad rights, which he goes on to claim that climate change violates. For example, “HR2, the human right to health: All persons have a human right that other people do not act so as to create serious threats to their health.”31 One need only reflect upon this, however, to see that such a right could only be “trumps” if one construes both “serious” and “health” in an extremely narrow way. For example, we are extremely hesitant to quarantine people with contagious diseases, despite the fact that they pose serious health threats to others. (Even a child with chicken pox creates an important health risk for any adult who has not previously been exposed or vaccinated.) This suggests the HR2 is not a very strict constraint, but is rather routinely traded off against other interests, such as the liberty interests of those who are sick. But even more prosaically, every time I get into my car I impose any number of “health” risks upon other people. I impose a risk of death and injury upon pedestrians, cyclists, and to a lesser extent other drivers. My car produces tailpipe emissions that, in urban areas, contribute to a large number of health problems (such as asthma and bronchitis), including thousands of preventable deaths – a far more serious impact on human health than those associated with the carbon emissions. Thus whatever right people may have to health clearly does not often play the role of a trump card, but is rather a right of the sort that Joseph Raz had in mind when he described an individual as having rights “if an interest of his is sufficient to hold another to be subject to a duty.”32 Thus the “right to health” does not mean that no one can do anything that risks impairing my health, even in the slightest way. It simply means that my health is important enough an interest that my desire to protect it may generate pro tanto duties for others, e.g. not to recklessly or unreasonably endanger it. But what counts as “reasonable” in this context? It is here that the rights framework requires some sort of supplementation, since as soon as two rights are in tension with one another, and some sort of “balancing” is required, a normative criterion must be introduced in order to determine what sort of a balance should be struck. The alternative is simply “moral gridlock.”33 So, for example, people have a right to a reasonably quiet environment, particularly at night, but at the same time, people also have a right to make a certain amount of noise, particularly when this is a byproduct of some productive activity aimed at satisfying other, essential needs. So how do we decide which “right” should prevail? When one construes the issue as fundamentally a matter of conflicting interests, then it becomes clear that liberal neutrality prevents the state from merely siding with one group. The natural procedure is therefore to try to determine how important the various interests at stake are to the individuals involved, and to weigh them against one 31 Caney, “Climate Change, Human Rights, and Moral Thresholds,” p. 167. 32 Joseph Raz, “Legal Rights,” Oxford Journal of Legal Studies, 88 (1984): 1-21. 33 John Broome, “Trying to Value a Life,” Journal of Public Economics 9 (1978): 91-100. another, subject to the overarching constraint that all the interests of each individual be given equal weight. One way of doing this – not the only way, but a fairly plausible way – is to figure out how much people are willing to give up, in order to protect those interests. This is essentially what the WTP measure in CBA tries to get at. The example I have chosen is trivial, but it is important to recognize that most of the difficult and controversial choices confronting the state have this structure. Both the railroad and the noisy neighbours case are simplified in several dimensions, the most important being that costs and benefits are typically not lump-sum, but rather continuous. Typically there will be a situation where reduced production of the externality is associated with increasing marginal cost, along with decreasing marginal benefit. (For example, cutting the noise level in half may be relatively inexpensive, and significantly improve the lives of residents. Cutting it in half again is a lot more expensive, and generates a lot less benefit.) Figure 5 illustrates the usual structure of such an interaction, with marginal cost increasing with abatement level, while marginal benefit decreases. The point r indicates the outcome that an optimal regulation would achieve. MC r $ MB abatement Figure 5. Increasing marginal cost, declining marginal benefit. Now there are many different normative frameworks that can generate the conclusion that one should regulate. The difficult question however is often how much one should regulate. For example, the days in which particular environmental pollutants could be banned outright are for the most part long gone (primarily because all of the “low-hanging fruit,” from a regulatory perspective, has been picked). The question now is almost always how much of the pollutant should be allowed. Unless costbenefit considerations are introduced, agencies can easily fall victim to what Stephen Breyer calls the “last 10 percent” problem.34 By ignoring the dramatic increase in marginal cost, they expend vast resources trying to solve a problem completely. For example, with respect to environmental cleanup, rather than removing 90 per cent of contaminants from 100 per cent of sites, an agency may insist that 100 per cent of contaminants be removed, but in so doing, restrict themselves to cleaning up only a small percentage of sites. There are many examples to suggest that this was a real problem in American regulatory law, prior to the widespread adoption of CBA. To take a specific example, it is not difficult to arrive at the conclusion that the problem of global warming requires us to price carbon. The question is, how high should that price be? It doesn’t make sense to do nothing, but it also doesn’t make sense to do too much. So we need to be able to say how much we should be doing. Should carbon taxes be set at $30/tC, or $300 t/C? Here the other normative frameworks tend to fall silent, leaving CBA as the “only game in town.” While point r may be difficult to find, empirically, at least it offers a clear and coherent conception of what we are looking for. If you want to solve a collective action problem, and you cannot rely upon the parties to negotiate their way to a solution, then you have to estimate where that the solution would be. CBA is the only known methodology for doing so. The rights framework simply has nothing to say here. CBA not only has something to say, but what it says can be plausibly construed as reflecting our commitment to equality and neutrality. 3.2 The “garbage in garbage out” objection Proponents of CBA sometimes defend the procedure on the grounds that it is more “rational” than the alternatives, or at least less likely to be subject to cognitive bias.35 And yet these arguments tend to cut both ways. The value of CBA as a procedure depends entirely upon the quality of the 34 The tendency to ignore the decreasing effectiveness of additional spending is what Stephen Breyer refers to as the “last ten percent” problem. Breaking the Vicious Circle: Toward Effective Risk Regulation (1994), p. 11. For a certain price, regulation can bring about a 90% reduction in a particular externality. A 95% reduction can be achieved, but at double the price. Even if the cost-benefit tradeoff associated with the 90% reduction would be one preferred by citizens, the political system may generate pressure to achieve the 95% reduction. 35 Cass Sunstein, “Cognition and Cost-Benefit Analysis.” valuations that it uses as inputs, which it must take as given. Thus it is susceptible to a version of the “garbage in garbage out” principle, which states that no matter how good a procedure may be, if the input is worthless then the output will also be worthless. And if cognitive bias is such a serious difficulty that it impairs ordinary deliberative decision-making, as many proponents of CBA claim, then what reason is there to hope that the WTP values that are taken as “input” will not also be hopelessly compromised by many of the same cognitive biases? This is the central contention made by Peter Diamond and Jerry Hausman, in a very influential article entitled “Contingent Valuation: Is Some Number better than No Number?”36 They make a number of extremely important observations. The first is that calling people up on the phone and asking “how much would you pay for x?” is practically worthless as a way of ascertaining the value of an outcome. In particular, it is subject to all of the familiar biases, such as framing effects, anchoring effects, etc.37 Perhaps the most damning example of this comes from a well-known study by Desvouges, which showed that the WTP values for rescuing birds was roughly the same, regardless of whether the number of birds to be saved was 2,000, 20,000 or 200,000.38 Similar problems afflict “revealed preference” methods as well. For example, the idea that worker's valuations of life and injury will be revealed through the wage premium associated with dangerous employments is solidly contradicted by observation.39 There are no doubt many factors that contribute to this, but one of them is cognitive bias. Workers suffer from optimism bias and control illusions, in particular, that result in them underestimating the chances that they will be victim of a workplace accident. It is disingenuous for proponents of CBA to respond to this, as they sometimes do, by pleading modesty, and suggesting that their analysis is only one consideration in a complex decision-making process, which no one should assign greater weight to than is warranted by the quality of the underlying data. There are two things wrong with this response. The first lies in a failure to acknowledge that numbers themselves can have a biasing impact (as studies on the “anchoring effect” have shown), leading to quantitative measures having an outsized role in deliberations, even when this 36 Peter A. Diamond and Jerry Hausman, “Contingent Valuation: Is Some Number better than No Number?,” Journal of Economic Perspectives, 8:4 (1994): 45-64. 37 Amartya Sen makes the narrower, but nevertheless very significant point, that with any sort of public good, the amount that any individual is willing to pay will be, in part, a function of their beliefs about what other people are going to pay. Thus the type of question is one better suited for ascertaining the value of goods that are exclusive in consumption. “The Discipline of Cost-Benefit Analysis,” 38 Desvousges, W. H., et al., "Measuring Natural Resource Damages with Contingent Valuation: Tests of Validity and Reliability,” in Jerry Hausman, ed., Contingent Valuation: A Critical Assessment (Amsterdam: North Holland Press, 1993): 91-164. For discussion see Diamond and Hausmann, “Contingent Valuation: Is Some Number better than No Number?” p. 51. 39 See discussion in David A. Moss, When All Else Fails (Cambridge, MA: Harvard University Press, 2002), pp. 162-169; Cass Sunstein provides a useful summary of studies in Risk and Reason (Cambridge: Cambridge University Press, 2002), p. 174. is unwarranted by the reliability of the underlying data. (For example, there is widespread awareness of the limitations of GDP, and by extension, GDP growth, as a measure of economic progress or wellbeing. And yet growth still exercises something like a magnetic attraction in debates over development, government economic management, etc. Indeed, the effort to develop alternative measures of progress, such as the Human Development Index, was in part a reaction to the fact that just attaching caveats and qualifications to GDP data has not been sufficient to eliminate the biasing effect it has upon deliberations.) The second important consideration is the fact that, in certain jurisdictions, and in certain policy areas, all major regulations will be litigated, and so must be defensible in court. In an adversarial context such as this, quantitative measures are much more defensible than qualitative ones, and so if both are available, courts will gravitate toward assigning the most authority to a CBA, even if the numbers that it is based on are little more than a guess. Thus the “number” provided by CBA is not just one data point among many. What Diamond and Hausman refer to as the “some number is better than no number” fallacy arises in part from the fact that information of equal quality will often be given greater credence, or deliberative weight, when presented in a quantitative form rather than a qualitative.40 Thus if a particular decision requires weighing two considerations, people are more likely to achieve an appropriate balancing if both considerations are expressed quantitatively or both expressed qualitatively. But if one is presented quantitatively, and the other qualitatively, the former is likely to be overweighted. In this case, it is better to have no number than to have some number. Thus the mere insistence upon methodological transparency is not an adequate response to the “garbage in garbage out” argument. A better response lies in the observation that in almost every case in which CBA is used, there is already at least one number available, namely, the number that reflects the cost. This will be true, for example, whenever a regulation is imposed on economic activity, or whenever resources are directed to the provision of a public good. It is usually the benefit that is intangible, because it lacks a market valuation (which is why the state must provide it). Thus the question is not whether we want to have one number or no number, but whether we want one number – reflecting only the cost – or two numbers – reflecting both the cost and the benefit. The logic of Diamond and Hausman’s own position suggests that, if there is always going to be one number available, and furthermore, if this number is always going to be on the same side of the scale, then deliberation will be enhanced by generating a second number, in order to balance out the effects of the first. From this perspective, the one thing that can be said for the WTP value is that, whatever its 40 Diamond and Hausmann, “Contingent Valuation: Is Some Number better than No Number?,” p. 58. shortcomings – and I certainly have no desire to understate these – it represents at least an attempt to measure the right thing. When market prices are used to calculate the cost of a policy, what is essentially being measured is the aggregate willingness to pay for the associated resources in some other employment (keeping in mind that the cost of a particular policy is its opportunity cost, which is simply the foregone benefit of not implementing some other policy.) The danger is that, if one does not construct a similar measure, of willingness to pay for the resources needed to enact that policy (i.e. the “benefits”), then one will not be comparing like with like. Furthermore, misstatement of the benefits will typically err in the direction of underestimation. And this is, in fact, precisely what happens. Consider, for example, an environmental regulation that we now regard as a “no brainer,” such as the ban on leaded gasoline. At the time that it was implemented, the regulation was in fact quite controversial, largely because it was not subjected to proper CBA, which in turn made it difficult to make the case for it. The problem was that the cost of the regulation was easily calculable, because it primarily took the form of “bads” that had market prices. Thus petroleum companies were able to calculate how much their refining costs would go up, in order to achieve the same octane level in their fuels without lead additives (and thus, how much the price of gas would have to go up). Automobile makers were able to calculate quite easily how much it would cost them to add catalytic converters to their vehicles (and how much this would raise the price of a car). By contrast, the benefits of the regulation – primarily health benefits associated with the reduction of airborne lead – had no market value. What CBA recommends in such cases is to take the primary welfare gain and estimate its value, by asking people “how much would you pay to avoid exposure to an atmospheric contaminant that gives you a chance of brain damage and/or kidney failure, according to the following risk/exposure schedule....” But instead of doing this – trying, however, imperfectly, to measure the direct welfare gain – they chose downstream beneficial consequence that did have a market value, such as a savings in health care system from not having to treat those conditions, or increased worker productivity. This is completely arbitrary, because the question of which benefits happen to have a market value is determined by arbitrary features of the world, such as the cost of implementing a system of property rights being lower in one domain than another. Furthermore, the net effect of doing this sort of ad hoc CBA is that it consistently overstates the relative cost of regulation. Thus the choice is not really between doing a CBA and not doing a CBA, but rather between doing a proper, complete CBA and doing an improper, incomplete CBA.41 In the usual case of 41 Another example is that, prior to the introduction of “valuation of life” based on willingness to pay, U.S. Government agencies used a “cost of death” measure consisting of the discounted sum of lost earnings (See Viscusi, “Risk Equity,” pp. 854-855.) The latter sum is morally arbitrary. Future earnings are of no value to a dead person, and as far as society regulation, industry lobby groups can be counted on to do their half of the CBA – the one that adds up the total costs. The question is whether the state will do the other half of the CBA – the one that adds up the diffuse, difficult-to-quantify benefits that flow to the population generally (the population whose interests elected officials are responsible for protecting). Finally, it should be observed that most critics of CBA do not reject it categorically, but rather believe there is some other decision procedure that is in some way superior. The most commonly cited candidate procedure is some sort of participatory democratic deliberation.42 It is worth observing, however, that such a procedure could also be used to increase the quality of the inputs to CBA. There have been some small-scale experiments using community deliberation as a way of improving WTP values, and there is some measure of enthusiasm for the technique among proponents of CBA.43 There is also increased interest in the development and use of standard valuations, constructed by comparing and “cleaning up” many different data sets, in order to develop a model of the impact on the quality of life of the “average” person is from various interventions.44 Thus there are a variety of ways of improving the inputs to the CBA procedure. 4. Road safety Let us suppose now that our municipality receives a large provincial grant for road improvement. There is immediate consensus that that highest priority spending area is road safety, since the rate of motor vehicle accident fatalities in the municipality is above the national average. This category, however, includes not just drivers, but also pedestrians, cyclists, and public transit users. Thus there are a number of different ways that travel in the city could be made safer: one could spend it on new crosswalks, left-turn lanes at intersections, bicycle lanes, improved highway embankments, arborism to improve sight-lines, barriers on level rail crossings – the list goes on and on. The question then is how the money should be spent. What are the high-priority areas? And so a study is commissioned... There are two overarching principles governing the exercise. The first follows immediately is concerned, the individual’s death is a wash, since that individual’s lost contribution to the social product is accompanied by a closely matched decline in consumption. 42 e.g. Anderson. Or Mark Sagoff, “Aggregation and Deliberation in Valuing Environmental Public Goods: A Look Beyond Contingent Valuation,” Ecological Economics, 24 (1998): 213-230. 43 See D. Macmillan, L Philip, N Hanley, B alvarez-Farizo, “Valuing the Nonmarket Benefits of Wild Goose Conservation: A Comparison of Interview and Group-based Approaches,” Ecological Economics, 43 (2002): 49-59; B Alvarez-Farizo, N Hanley, R Barberan, A Lazaro, “Choice Modeling at the 'Market Stall': Individual versus Collective Interest in Environmental Valuation,” Ecological Economics, 60 (2007): 743-751. For discussion see Giles Atkinson and Susana Mourato, “Environmental Cost-Benefit Analysis,” Annual Review of Environmental Resources, 33 (2008): 317-344 at 322. 44 Atkinson and Mourato, “Environmental Cost-Benefit Analysis.” from the Pareto principle, and it is a commitment to maximization. The available funds should be spent in such a way as to produce the greatest attainable reduction of mortality (if this is not done, then a reallocation could improve the condition of at least one person without worsening that of anyone else). The second is an application of the principle of citizen equality, and it suggests that the interests of all users of the road should be assigned equal weight, there should be no privileging of any one category, such as drivers, pedestrians, cyclists, etc. The first requires an evaluation of each potential improvement, in order to see how effective it would be at reducing mortality, and at what cost. In order to compare and rank projects of different scale, one must then calculate the cost of each per life saved. The second requires that one do so comprehensively, examining projects that affect all users, and that one should choose projects with the lowest cost per life saved. (If, for example, one were to recommend a project that reduced driver fatalities at a cost of $7 million per life, over one that reduced pedestrian fatalities at a cost of $5 million per life, then pedestrians could reasonably complain that their interests were not being assigned equal weight to the interests of drivers.) This is implicit in the commitment to maximization, but it is nevertheless worth observing that the cost-per-life calculation is necessary in order to ensure that egalitarian principles are being satisfied.45 One can see here that we are moving into the neighbourhood of one of the most controversial features of cost benefit analysis, which is the valuation of life. Technically at this point we are not yet valuing life, because what is being proposed is a cost-effectiveness analysis (CEA), rather than a fullblown cost-benefit analysis. What makes it a CEA is the fact that monetary valuation is only being used on the “cost” side of the ledger, while some other metric is being used on the “benefit” side. (The fact that one can do this is a reflection of the point, made earlier, that monetary valuation in CBA is simply used as a metric for comparison across individuals. There is nothing essential or important about money. Thus if one can use some other metric to compare different projects, such as “lives saved,” where one can safely assume or stipulate that a life saved is of equal value regardless of whose life it is, then this can also be used as a metric.) Since saving lives is the “benefit” of the policy, while the cost is the actual financial cost, then one is not actually assigning a monetary value to life within this framework, because lives only show up on the benefit side, while money stays on the cost side. CEA is useful if one has a budget that is exogenously determined and a single policy objective (such as “improving road safety”). One can then just look at alternative employments of the fixed budget and pick the most effective projects. So one calculates the cost in monetary terms, but on the benefit side one just uses a certain quantity of “whatever it is that we are trying to accomplish” as the 45 Cass Sunstein suggested, in 1996, that the cost per life-year saved in transportation was $56,000, for occupational regulation, $346,000, and for environmental regulation $4,207,000.”(“The Cost-Benefit State”, p. 10.). unit (in this case, it would be reduction in road fatalities). This eliminates the need to monetize the value on the benefit side of the ledger. There is, it should be noted, an implicit valuation of life here – determined at the point at which the money runs out. So if the least cost-effective project to be funded has a cost-per-life-saved of $5,347,000, then this means that anything beyond this threshold is deemed “too expensive,” and thus life is being implicitly valued at that sum. This decision, however, is not being made in the CEA, it is actually an implicit consequence of the budget decision that was made, when a certain amount of money was allocated to road safety. In this case the CEA just renders explicit what was implicit in this earlier decision. Where the CEA approach does not work so well is if the budget is endogenous to the policy choice (e.g. when the question of how much money should be spent on road safety depends, at least in part, on how expensive it is to improve road safety46), or if the budget is designated for multiple objectives that are rather different in character from one another. Suppose, for instance, that the money is given to the municipality is for “road improvement” in general, a category that includes not just safety improvements, but also maintenance, construction, and various other forms of improvement. One of the questions that must be answered is therefore how much should be spent in each area, and as is typical with these things, spending in each domain is subject to diminishing returns. (The first million dollars spent on pothole repair will produce much more benefit than the second million spent on it, just as the first million dollars spent on safety improvements will save more lives than the second million dollars.) Furthermore, in the case of safety, there is good reason to think that marginal cost tends toward infinity as the level of fatalities approaches zero, which is to say that, absent some stopping rule, one could easily spend the entire budget on safety and still not get the level of traffic fatalities down to zero. There is good reason to think, however, that users of the road are interested in more than just safety. For example, people also want to get to their destination quickly, even when this involves putting themselves at some risk: drivers exceed the speed limit, pedestrians jaywalk, cyclists ride the wrong way up one-way streets. The list could be extended considerably. In each case, the person is trading off the need for safety against the desire for more expeditious passage. Their personal preferences are going to be reflected in their collective preferences as well – people are not going to 46 There is a somewhat esoteric, but philosophically significant point that arises when one considers comparison across policy areas. As long as the overall objective is the same, one could compare by taking the “benefit” side of a CEA and converting it to a monetary value, but one could also take the “cost” side and convert it to a “lives saved” value. How? By looking a how much it costs to save a life in other domains, like health care. Every million dollars spent on road safety is a million dollars not spent in other areas. One can calculate how many lives will not be saved in those others areas – that is the “cost” of spending a million dollars on roads, expressed though in terms of lives saved. One can then convert the monetary cost of a policy into the “lives no saved” equivalent, in order to determine which policy generate the highest net number of lives saved (as opposed to the highest net monetary benefit). want the government to spend all of its budget on safety, if the result is a significant increase in travel times. Thus one must compare the benefits that would come from increased safety with the benefits that would come from reduced travel times, and decide how one should balance the two spending priorities. In order to do this, a CEA will not be adequate, because one must have a metric that will enable comparisons of “lives saved” against travel time improvements. This leaves little choice in practice but to convert to a CBA, by assigning a monetary value to each life saved, as well as a monetary value to a given reduction in average travel time. The state must then determine at what point individuals themselves would want to stop spending on increased safety and start spending on other priorities. The point at which it does so will be equivalent to a monetary valuation of life. Two points should be made here. First, whenever spending decisions are made that affect, in any way, human mortality, or even just longevity, subject to a budget constraint, there is always an implicit valuation of life. States that do not perform cost-benefit analysis still have budgets, which they use to improve road safety, as well as to accomplish myriad other objectives, many of which involve merely enhancing quality of life (such as maintaining parks). To the extent that they do so, they are clearly letting some people die so that others can enjoy what may be, taken singly, relatively minor benefits. Cost-benefit analysis merely articulates, or renders explicit, what is implicit in these decisions; it does not seek to justify them. Second, the only way to tell whether citizens are being treated equally is to compare the valuation of life across policy domains. Where CBA calculations are not performed, the valuation of life is likely to be quite arbitrary. Thus it would not be surprising to find workplace safety rules that impose over $30 million in costs to save a single life, while the transportation safety administration balks at making infrastructure improvements that would cost only $1 million per life saved. To the extent that citizens are exposed equally to all of these risks, the arbitrariness of these valuations does not translate into any significant inequality between citizens. But if exposure to certain risks is correlated with other forms of disadvantage, then the arbitrariness does have the capacity to exacerbate inequality.47 In particular, if budgets are determined through the political process, then “sympathetic victims” are more likely to attract funding for the issues that affect them. So to the extent that low-socioeconomic status (SES) individuals are less sympathetic, they are more likely to receive equal “concern and respect” from the state if all agencies are forced to use a standard valuation for life that is equal across all projects and policy domains. 4.1 The aggregation objection In his influential critique of utilitarianism, Bernard Williams presents a scenario in which an 47 Viscusi, “Risk Equity.” individual (Jim), wanders into a small South American town, where the military is about to execute 20 randomly chosen civilians (the “Indians”), as an act of collective punishment against their village for supporting anti-government rebels. The commander, however, makes Jim an offer: that if he (Jim) is willing to choose one of the villagers at random and personally execute him then the other 19 will be spared. Williams’s intuition (not universally shared) is that it would be permissible, or perhaps even obligatory, for Jim to refuse. This appears to suggest that there is a deontic prohibition on killing the innocent, even when the foreseeable consequences of not violating the prohibition are that others will. Thus aggregationism (e.g. adding up the number of lives saved or lost under different conditions, or the number of rights-violations that occur), ignores morally significant features of the choices that we face. To the extent that CBA is committed to aggregationism, especially with regard to human life, it runs afoul of these intuitions. The standard response to this argument among moral consequentialists is to say “Surely there must be some point at which you would be willing to violate the constraint. What if he was threatening to kill 100 Indians, or 1000?” and then to chip away at the position from that end, with the goal of showing that there is something irrational or indefensible about any strict deontic prohibition. There is no reason, however, for proponents of CBA to get drawn into these sorts of philosophical controversies. After all, in claiming that it is permissible for Jim to refuse to execute the innocent, Williams is granting that it is permissible for Jim to take actions that will foreseeably result in the death of 20 innocents, including the one that Jim himself was unwilling to kill. So what is the difference between these two ways of bringing about someone’s death? Typically, the way that deontologists have tried to articulate the difference is by distinguishing intended outcomes from merely foreseen consequences. If Jim were to shoot one innocent, that individual’s death would be the intended outcome of an action chosen by Jim. If Jim refuses, the same individual will die, but merely as a byproduct of Jim’s action – a foreseen consequence, but not an intended outcome. As the subsequent literature on trolley problems has illustrated, everyday morality imposes deontic constraints in the realm of intended outcomes and actions, but is much less restrictive of aggregative calculation the realm of the foreseen (hence the judgment that it is permissible to divert a runaway trolley onto a side-track where it will kill an innocent maintenance worker, on the grounds that doing so will save a larger number of innocent passengers waiting on a platform48). Thus the standard defence of CBA lies in the observation that, because of the policy areas in which “embedded” CBA is applied, the deaths that it deals with are entirely in the realm of the foreseen, not the intended. Furthermore, to the extent that these deaths do occur, it is typically the result of an omission on the part 48 See Judith Jarvis Thomson, “The Trolley Problem,” Yale Law Journal, 94 (1985): 1395-1415. of the state (i.e. a failure to spend the extra money through which that death could have been averted). Satisfaction of either of these conditions may be sufficient to establish a context in which strict deontic constraints do not apply. So, when the state limits the amount that it spends on shelters for the homeless, it has the foreseeable result that some people will freeze to death on the street; when it limits the amount that it spends on chemotherapy drugs, it has the foreseeable result that some people will be denied access to life-extending treatment; when it limits the amount that it spends on road safety, it has the foreseeable result that more motorists will die in accidents. But this is not the same thing as killing the homeless, or cancer patients, or motorists, in order to achieve some other policy objective. Thus the actions simply do not fall under the scope of any deontic prohibition. As a result, it is perfectly permissible to ask the question, whether one is really maximizing the number of lives saved by upgrading a highway to provide an additional passing lane, rather than providing expanded outreach services to the homeless. If one discovers that the lives of motorists are being implicitly valued much higher than those of the homeless, and so one decides to reduce spending on road safety while putting new funds into “in from the cold” programs, then one is not killing motorists in order to save homeless people. One is, however, letting motorists die who otherwise might have been saved, so that one can instead save a larger number of homeless people. Thus there need not be any conflict between CBA and the deontic constraints that we feel should govern decision-making in certain policy areas (and there is no need for proponents of CBA to commit themselves to the sort of controversial modifications of everyday morality that are implied by moral consequentialism). Deontic constraints can be accommodated through a set of “input filters” in embedded CBA, that determine the set of problems to which it will be applied. For example, the mining industry is subject to a variety of workplace safety regulations, all of which are eligible for CBA. There are no doubt certain safety procedures, or equipment, or mining techniques, that are not mandated, simply because they would be too expensive, relative to the number of lives that would be saved. As a result, a certain number of mining accidents will occur that could have been prevented, had ‘we’ been willing to spend more money to prevent them. At the same time, once an accident has occurred and there are a group of miners trapped underground, it is not permissible to apply a CBA to the rescue attempt. On the contrary, states are usually willing to spend what seems like arbitrarily large amounts of money rescuing trapped miners (even though that same money, if saved and spent later on preventative safety measures, might actually save the lives of many more miners). Nevertheless, it is impermissible to abandon people to their death merely to save money. That is because the accident triggers a “duty of rescue” that overrides the more general consequential calculus governing budgeting. Of course, reasonable people may differ on whether a particular problem should trigger a “deontic context” that prohibits CBA. For example, after the terrorist attack on the World Trade Centre in 2001, the United States government declared something like a “state of exception” with respect to airline safety, as well as other aspects of anti-terrorism policy. Thus a large number of changes were made to airport security, all of which were exempted from ordinary CBA. This eventually became controversial, simply because the costs imposed by these changes (largely in the form of lost time for travellers) were gigantic, while the number of lives saved seemed to be quite low. This led to widespread speculation that none of the enhanced security measures (with the exception of hardened cockpit doors) could pass a CBA test. (It lead as well to questions about why lawmakers, who in many cases were eager to assign standard valuations to the victims of workplace accidents, should suddenly lose their resolve when it came to valuing the lives of victims of terrorist attacks). The important point, however, is that the state will typically pick and choose which problems to evaluate using CBA, and this gets decided by the political process, an area in which the various deontic constaints governing the treatment of life carry full sway. Thus the commitment to CBA involves much less than a commitment to moral consequentialism, which in turn makes the type of aggregationism it employs less problematic. 4.2 The dignity objection The second major objection to the valuation of life in CBA appeals to the suggestion that human life has an inherent “dignity,” which is incompatible with the assignment of a price. Critics often quote Immanuel Kant in this context: “In the kingdom of ends everything has either a price or a dignity. What has a price can be replaced by something else as its equivalent; what on the other hand is above all price and therefore admits of no equivalent has a dignity.”49 Setting aside concerns about aggregationism (trading off lives for lives), there is a separate concern that arises when one contemplates trading off a human life against something of lesser value, such as a basket of ordinary economic goods. This is thought to diminish the inherent dignity and value of life, and this is thought to be prohibited on strict deontic grounds.50 The preceding discussion has shown that CBA does not justify anything that the state is not already doing whenever it makes decisions affecting human life subject to a budget constraint. There are, however, two further points that are worth making in response to this objection. First, proponents of CBA have tried to explain that when they talk about the value of “life” it is, in most cases, a 49 Grounding to the Metaphysics of Morals. 50 Kelman, “Cost-Benefit Analysis: An Ethical Critique.” shorthand way of talking about something quite different, viz. the disvalue that people assign to very small risks of death. So when one says that the valuation of a “life” is $5 million, it is not actually any one individual’s life that is being valued, but rather it is the aggregate disvalue of very small risks of death distributed over a very large number of persons. This is flagged by the insistence that the dollar figures in question represent the “value of a statistical life” (VSL), not the value of an actual life (VAL). The thought is that if 100,000 people are each willing to pay $100 to eliminate exposure to a pollutant that generates a 1/50,000 risk of death, then the total “budget” of $10 million represents a WTP of $5 million per life saved (since the pollutant can be expected to kill two people out of the 100,000 people affected). When deliberating about whether to ban the pollutant, the question that is being asked is not whether it is worth spending the money in order to save these two people’s lives, the question is whether it is worth spending the money to deliver the benefit of a small reduction in risk to the entire group. Thus one way of responding to the “dignity” objection is simply to state that, while human life may be priceless, risks to human life are not. Even if one could never be justified in killing a person in order to achieve some other good consequence, it is not so obvious that one could not be justified in imposing some risk of death on a person in order to achieve some such consequence, as long as the risk is not excessive. After all, every time we get behind the wheel of a car, we are imposing some risk of death on other users of the road. Thus “embedded” CBA can avoid the general concern about human dignity simply by avoiding any application in areas which would require the use of an VAL rather than a VSL. A second line of argument that defenders of CBA appeal to is the fact that individuals themselves make choices, quite often, in which they trade off risks of their own death against other goods. Nobody wants to spend infinite amounts of money on safety, and very few people take living as long as humanly possible to be central to their vision of the good life. On the contrary, safety is just one good among many that make up a balanced and healthy life. (As Jonathan Wolff puts it, “purchasing – or declining to purchase – devices or services that make small differences to one's safety, or that of one's family, is an ordinary part of life, taking its place alongside other consumer decisions.”51) The appropriate balance to strike between these various goods is an expression of individual autonomy. Yet if individuals, taken singly, have the right to decide what risks they are willing to accept, then there seems to be no reason that individuals, acting collectively, should not have the right to make that same decision. And since the state, in its “public economic” functions (that are the focus of CBA), is just 51 Jonathan Wolff, “What is the Value of Preventing a Fatality?” in Tim Lewens, ed. Risk: Philosophical Perspectives (London: Routledge, 2007), pp. 54-67. acting as the agent of individuals, it seems that it should be able to make such choices that reflect these individual preferences. There are, of course, problems that arise in the transition from individual to collective autonomy, particularly given that the state is often obliged to make “one size fits all” decisions about the levels of risk that individuals will be exposed to. Setting aside these issues, however, it is worth pausing to consider how dramatically handicapped the public sector would be, if it were somehow prevented from performing the ordinary cost-benefit calculations that inform individuals’s private choices, and in particular, their market behavior. It would mean that a very large number of collective action problems simply could not be resolved, because private individuals were unable to contract their way to a solution (i.e. there is a market failure), and yet the public sector is unable to step in and resolve the issue, because doing so would implicitly treat human life as less than infinitely precious. To take just one example, if human life really does have an “inherent dignity” that prevents the state from trading it off risks to individuals against other goods, then the state would be unable to operate public transit systems. Transportation is inherently risky, and so one can spend arbitrarily large sums of money improving safety.52 When choosing private transportation options (e.g. buying a car) individuals trade off the benefits of safety against other considerations. But if the state is unable to make the same calculations, then the cost of state administration of transit would skyrocket. As a result, collective forms of transit like subways or buses would fall through the cracks – too costly to be provided privately, but too unsafe to be provided publicly. This is, of course, not what proponents of the “dignity” argument have in mind. In fact, much of the concern over valuation of life seems not to be a concern over valuation per se, but rather a concern that using standard methods will result in a VSL that is simply too low (and therefore that the benefits of certain kinds of regulations will be understated). In particular, critics focus a great deal on the use of wage premiums in risky occupations as a “revealed preference” mechanism for inferring the valuation that workers put on their own lives. These are perfectly legitimate concerns, but the appropriate response is simply to increase the assigned VSL to a more acceptable level, not to treat it as infinitely large. And in fact, if one looks at the practice of most states that make extensive use of CBA, one can see that the VSL figure is set through what amounts to an exercise in reflective equilibrium. Most importantly, a standard valuation is selected that is applied in all policy domains.53 This value is typically the result of lengthy reflection and negotiation. In Canada, for instance, Environment Canada 52 Jonathan Wolff provides an excellent example of this involving British Rail, which was under pressure to install a computerized signalling system, at a cost of approximately £6 billion, that would have saved perhaps two lives yearly. See Ethics and Public Policy (Oxford: Routledge, 2011), p. 89. 53 Viscusi, “Risk Equity,” p. 855. and Health Canada jointly sponsored a literature review of CBAs, in order to determine the range of VSL values. They found a “mean VSL of $5.2 million with a range from a low of $3.1 million to a high of $10.4 million in 1996 dollars.”54 This mean value ($5.2 million in 1996 dollars), adjusting for inflation, was picked up by the central agencies and became the standard valuation used in all departments of the federal government. 5. Rationing health care In 1992, the pharmaceutical company Bristol-Myers Squibb raised eyebrows when it began charging $4,000 a year for Taxol, a new drug for patients with breast cancer that avoided many of the side effects of traditional chemotherapy. Although there were some complaints, insurers had little choice but to pay the price that was being asked. This had the unfortunate consequence, however, of setting a precedent. In 1998, Genentech introduced another breast cancer drug, Herceptin, at a price of $20,000 per year. In 2002, Bristol and ImClone Systems began charging as much as $100,000 for Erbitux, a drug for patients with advanced colon cancer.55 In 2011, the company Alexion set a new record with the drug Soliris, for which they charged more than $500,000 per year. What is striking about some of these drugs is not just the extraordinary price tag, but how little benefit they offer to patients. A study of Erbitux (Cetuximab) conducted by the National Cancer Institute of Canada Clinical Trials Group found that treatment with the drug led to an average gain of only 0.12 life years, which is to say, 43 days. When quality of life was factored in, the gain was only 0.08 quality-adjusted life years (QALYs). At the worldwide average price – which was at the time approximately half that charged in the United States – this amounted to $199,742 per life year, or $299,613 per QALY gained.56 By way of comparison, colorectal cancer screening costs between $10,000-$25,000 per life-year saved.57 Thus the drug treatment was literally an order to magnitude more expensive than other modalities for combating the same condition (from a population perspective). The introduction of these ultraexpensive drugs has led to a “day of reckoning” for many health insurers, particularly those in the public sector where budget constraints are more sharply felt. There 54 Treasury Board of Canada, Canadian Cost-Benefit Analysis Guide: Regulatory Proposals (Ottawa: Treasury Board, 2007), p. 20. 55 All information on drug prices from Alex Berenson, “Cancer Drugs Offer Hope, but at a Huge Expense,” New York Times, July 12 (2005). 56 Nicole Mittmann, Heather-Jane Au, et. al. “Prospective Cost-effectiveness Analysis of Cetuximab in Metastatic Colorectal Cancer: Economic Evaluation of National Cancer Institute of Canada Clinical Trials Group CO.17 Trial,” Journal of the National Cancer Insitute, 101 (2009): 1182-1192 at 1190. 57 K. Robin Yabroff and Deborah Schrag, “Challenges and Opportunities for Use of Cost-Effectiveness Analysis” Journal of the National Cancer Institute, 101: 17 (September 2, 2009):1161-1163 at 1161. was a time when insurers could promise to pay for the “best available” care for their policy-holders in the event of illness. Now it is no longer feasible to pay for all the therapies that patients might be inclined to consume (or doctors to prescribe). Furthermore, very few people would be willing to pay the premiums on a health insurance policy that paid for every life-extending therapy (and few citizens are interested in paying the taxes that would be required in order for public health insurance systems to provide such coverage). The result is that, in public health care systems, patients are sometimes denied therapy, particularly with respect to pharmaceuticals. There may be a drug available, which the patient could be taking, and which would increase that patient's life expectancy, but which the state simply refuses to pay for on the grounds that it is too expensive – and that the money would be better spent elsewhere, providing other, more cost-effective therapies for other patients. One does not need to be a moral philosopher to see that this sort of health care rationing looks as though it rests on a utilitarian calculation. The patient who is being denied treatment is harmed, while the resources that could have been used to help him are being given to some other patient, where they can be expected to “do more good.” And yet as Ronald Dworkin observed, appearances in this case can be misleading. This is because of the role that health insurance plays in mediating the transaction, and the way that it obscures the relationship between the patient's own prior choices and the rejection of the claim. That relationship would be more clear if the market for health insurance were less subject to market failure. Consider, for the moment, what an ideal market for health insurance would look like. It would feature something close to an infinite number of different insurance policies for sale, each providing a specific level of indemnity against a specific range of conditions, and each charging an “actuarially fair” premium, reflecting the precise expected loss that each individual brought to the insurance pool. Thus anyone who wanted to have the option of receiving Erbitux, in the event of developing colon cancer, could buy a policy that offered to pay for that particular treatment. The premium associated with such a policy would, of course, be higher than the premium for a policy that was identical save for an exclusion of Erbitux. In a perfect market, both policies would be available, so it would be possible for a person to see exactly how much of a difference coverage of that specific drug made to the monthly payments. The person could then decide, based on this price information, whether he thought it was worthwhile paying extra for a policy that covered Erbitux. Now imagine that this person opts for the policy that does not cover Erbitux, but ten years later develops colon cancer and finds that he could benefit from a course of it. He requests the drug, but his insurer refuses. He might be tempted to challenge this decision, but in this case it is almost pointless to do so, because the justification for the refusal is too obvious: “You cannot have Erbitux because you chose the policy that doesn't cover Erbitux. Furthermore, you benefited from lower health insurance premiums over the course of the past ten years, precisely because you chose the policy that excluded Erbitux. So you cannot turn around now and request it.” Under these circumstances, it will still be be a fact about the insurance scheme that the money that could have been used to pay for the person's Erbitux will be used to pay for other treatments, for other patients, that are presumably more cost-effective. Thus it will look as though one person is being denied a life-extending therapy in order to provide a greater benefit to someone else. But this is misleading. The greater benefit to someone else does no work in justifying the denial of Erbitux to the individual in question, it is just a byproduct of the way that insurance systems are organized, along with the fact that the policy-holders in these schemes seek maximum value ex ante. The same structure is preserved when the example is changed from an ideal market to the real world. Private insurers are, of course, not able to offer a continuum of policies, and they do not have the information required to charge an actuarially fair premium to each potential policy-holder. Thus they aggregate and average across groups, then offer a limited choices of add-ons and exclusions (and deductibles). Health insurers may offer a choice between “basic,” “comprehensive” and “deluxe” policies, depending upon how much these policies cover. Individuals will then sort themselves roughly into groups, depending upon their willingness to pay for various levels of coverage. In this case, a person may be denied Erbitux on the grounds that she bought the “basic” package, rather than the “no Erbitux” package, but the principle is the same. Nothing changes fundamentally when the state takes over the provision of health insurance, except that now the aggregation and calculation of willingness to pay has to be done over the entire population. Because private health insurance markets are so inefficient, the state is in a position to offer a comparable product at far lower cost (and with less distortion of the market for health care, because of its superior ability to control moral hazard, and thus, to limit cost inflation). In order to do so, the state must calculate approximately what individuals themselves would be willing to pay for health insurance (i.e. what fraction of lifetime income the average person would want to spend). The situation with single payer health care systems is really no different from when the state gets involved in providing public goods like parks. When it comes to choosing the size of their backyards, individuals can find private markets that offer an enormous range of options, and so everyone is likely to find something that is a close match to her particular tastes. Individuals can then further customize, by spending as they like on landscaping, playground equipment, a swimming pool, etc. However, if some individuals want to trade backyard space for access to a communal park, to be provided for the neighbourhood as a whole, the government is forced to make a number of extremely rough-and-ready calculations, in order to determine how to satisfy something like the average preference in the neighbourhood. Thus a variety of decision get made about public park size, landscaping, playground equipment, dog runs, wading pools, etc., which are not likely to satisfy any one individual perfectly, but provide rough satisfaction to a very large number of people. Public health insurance schemes are the same. They provide one-size-fits-all coverage, rather than the continuum of individually tailored policies that an ideal market would provide. This will be based on average health care needs and average willingness to pay. Because we cannot create a market in which individuals are free to choose the exact policy that they want, the state has to choose a level of provision on their behalf that comes as close as possible to satisfying as many people as possible. Such a policy will necessarily include spending limits, simply because people's willingness to pay for health insurance is not unlimited. And when those spending limits are hit, it will look as though the state is engaging in utilitarian redistribution of resources, but in fact it is not doing so. Thus we have seen the emergence of decision-making bodies such as the National Institute for Health and Care Excellence (NICE) in the U.K., which uses an “embedded” QALY framework to decide, inter alia, when the National Health Service (NHS) should pay for pharmaceuticals. This eliminates the problem associated with drugs like Erbitux, since NICE has established a threshold of willingness to pay in the range of £20,000-£30,000 per QALY. Patients are, of course, free to purchase such drugs privately should they so desire.58 5.1 The death panels objection During the debates that preceded the adoption of the Patient Protection and Affordable Care Act (PPACA) in the United States, which sought to extend health insurance to a broader segment of the American population, much of the discussion was sidetracked by the allegation that the proposed legislation sought to implement “death panels,” or committees that would withdraw care from the elderly or disabled. This was all set in motion by an online post made by former governor of Alaska and failed Vice-Presidential candidate Sarah Palin, who wrote that: “The America I know and love is not one in which my parents or my baby with Down Syndrome will have to stand in front of Obama's ‘death panel’ so his bureaucrats can decide, based on a subjective judgment of their ‘level of productivity in society’, whether they are worthy of health care. Such a system is downright evil.” In part to avoid this accusation, and in part due to lobbying by physicians and the pharmaceutical industry, the Patient-Centered Outcomes Research Institute (PCORI) that was created 58 The fact that this option is not available in Canada (under the commitment to “single-tier” medicine) is the source of ongoing controversy in that country. It is available de facto, because any Canadian who can afford a drug like Erbitux can also afford a trip to the United States to purchase it. as part of the PPACA is specifically prohibited to engaging in cost-effectiveness analysis of the sort that is performed by NICE in the U.K. It is limited to engaging in “comparative effectiveness analysis,” which can look at how different therapies compare against each other, but specifically does not take cost into consideration. The legislation specifies that the PCORI “shall not develop or employ a dollarsper-quality adjusted life year (or similar measure that discounts the value of a life because of an individual’s disability) as a threshold to establish what type of health care is cost effective or recommended. The Secretary shall not utilize such an adjusted life year (or such a similar measure) as a threshold to determine coverage, reimbursement, or incentive programs...”59 As we have seen, the mere fact that cost-effectiveness may result in the refusal to fund overly expensive care is not “evil,” but is a legitimate feature of all insurance schemes, whether they are run in the private or public sector.60 Thus the restrictions built into the mandate of the PCORI is probably best understood as primarily a consequence of rent-seeking. There are, however, other issues raised in the famous “death panels” post. One of these is the concern that the cost-benefit calculations underlying the availability of medical care will consider, not just the health benefits to the individual patient, but also the contribution that the patient makes to “society” more generally. In the background, of course, is the spectre of various eugenics movements, including the Nazis in Germany who executed large numbers of disabled people in the grounds that they were a burden upon society. This issue is closely related to a view that has been widely debated among philosophers, about whether health care is “special,” such that its distribution should be largely independent of the distribution of other goods in society.61 Some of this debate is based on a confusion, caused by the fact that health care is paid for through insurance schemes, as a result of which it is distributed in accordance with “need” rather than “ability to pay.” This is not a consequence of some peculiar egalitarian commitment arising from the domain of health, it is just how insurance systems work.62 People whose houses burn down have “need” of a new one, and their home insurance provides them with a new one (whether they are rich or poor, deserving or undeserving, etc.). But this is not because houses are special, it is because the insurance they bought indemnified them against the specific eventuality of losing their home. Similarly, health insurance indemnifies individuals against the specific eventuality of needing health care, and so that is precisely what it provides (whether they are rich or 59 Cited in Alan M. Garber and Harold C. So, “The Role of Costs in Comparative Effectiveness Research,” Health Affairs, 29:10 (2010): 1805-1811. See also Peter J. Neumann and Milton C. Weinstein, “Legislating against Use of CostEffectiveness Information,” New England Journal of Medicine, 363 (Oct. 14, 2010): 1495-1497. 60 The tendency of policy-holders to claim benefits that are not covered also appears to be a universal feature of insurance schemes. See Insurance and Behavioral Economics on flood insurance in Florida. 61 Shlomi Segall, “Is Health Care (Still) Special?” Journal of Political Philosophy, 15:3 (2007): 342-361 at 343. Norman Daniels, ‘Justice, health and healthcare,’ American Journal of Bioethics, 1 (2001), 2–14 at p. 2 62 See Lendell Chad Horne, Health, Risk and Luck, unpublished diss. (2014). poor, deserving or undeserving, etc.) Thus much of the way that health care provision is “insulated” from the patterns of distribution of goods in other domains is a consequence of the way that our purchase of it is mediated through health insurance systems (or quasi-insurance systems, such as one finds in an HMO or a large provider like the NHS in the U.K.). This in turn explains why the allocation of health care resources is done almost entirely as an exercise in “local justice,”63 where there is no compensation across allocative spheres. The rich get the same as the poor, the talented get the same as the untalented, the prudent get the same as the irresponsible, and the disabled get the same as the able-bodied (above and beyond, of course, what they get for their disability). The question is whether CBA threatens this, as Palin suggested, or whether it is possible to preserve health care as a domain of local justice. More narrowly, the question is whether one can preserve health as a domain of CEA, with a health-specific value, such as a QALY, as the metric of comparison, or whether the framework creates an irresistible dynamic toward monetization, in order to facilitate comparisons across policy domains. However, it seems to me that there is nothing inevitable about the use of full-blown CBA, and therefore no reason that health-care resource allocation decisions cannot continue to be made within a CEA framework. Because health care funding is mediated through what amounts to an enormous insurance scheme, and it is possible to ascertain individual willingness to pay for health insurance at a very general level, the health care budget tends to be relatively self-standing. This means that it is not necessary to do CBA in the area of health care, one can instead ask people questions along the lines of “how much of your income do you want to spend on protection from disease and injury?” then do a CEA with the budget that results. Cost-effectiveness analysis is attractive in this domain as well, because there is good reason to think that the health outcomes that are sought provide a plausible basis for interpersonal comparisons. In particular, as mentioned earlier, there are solid normative grounds for assigning a gain in life expectancy the same value, regardless of who experiences it (and regardless of what they are willing to pay for it). This provides the benchmark, against which various states of morbidity can be assessed, generating the well-known QALY framework for assessing gains in health outcome. Using a QALY framework, in turn, allows proponents of CEA to avoid all of the problems that stem from using money as the metric of comparison between individuals. Overall then there is no reason to think that the creation of an institute like NICE represents the first step on a slippery slope that ends with eugenicism. As long as individuals are unwilling to spend their entire income on health care, and they are pooling together their health care spending in response 63 Jon Elster, Local Justice in America (New York: Sage, 1995), p. 133. to the uncertainty of their future needs, then whoever manages that pool is going to have to make decisions about what expenses will and will not be paid by the group. When the pool is managed by the state, then the state has no choice but to make those decisions. The only real question – as is typically the case – is whether those decision will be made in accordance with explicit, fully articulated criteria, as in a CEA, or implicit, poorly articulated criteria. 5.2 The discrimination objection One of the striking features of the decision to bar PCORI from assessing the cost-effectiveness of medical spending is that the two major interested parties, viz. the pharmaceutical industry and physicians64, were able to persuade disability rights groups to support their position, by arguing that the CEA framework discriminates against the disabled. This turns up in the language of the statute, as well, where the QALY framework is described as a system that “discounts the value of a life because of an individual’s disability.” Thus another objection to the use of cost-benefit considerations in the allocation of health care resources is that it is discriminatory. This is a spurious charge. So long as individuals are concerned not just to avoid death, but also to avoid pain, suffering and indignity, then they are going to want to allocate resources in a way that reflects not just mortality risk, but also the disvalue of different states of morbidity. Hip-replacement surgery, for instance, does not extend anyone’s life, and yet it vastly improves a patient’s quality of life. Chemotherapy, by contrast, extends life, but often at great cost, and with significantly reduced quality. So how much should be spent on hip-replacement surgery versus chemotherapy drugs? In order to make this decision, there must be some basis of comparison between the two, so that changes in the quality of life can be weighed against changes in quantity (i.e. expectancy). The core idea of the QALY framework, again, is to defer to patients’ own autonomous choice in this domain. The fact is that individuals living in various states of morbidity (or disability) typically would prefer to be restored to full health (or normal ability). In the standard run of cases this preference is strong enough that they would be willing to accept some reduction in life expectancy, in order to be able to live out their remaining years without the condition. (Or conversely, and more commonly, people would like to live longer and are willing to accept some reduction in their quality of life in order to extend their life.) Through a standard titration procedure, it is possible to determine how much lifeexpectancy people are willing to trade off against various state of morbidity. So if a person is willing to 64 In the United States, it remains common practice for physicians to be paid a commission for prescribing drugs, often a percentage of the sale price. This one of the reasons that drug prices are so high in the United States, and that physicians often prescribe brand-name pharmaceuticals instead of generics. Politically, it means that there is a strong alignment of interest between physicians and the pharmaceutical industry. accept a reduction in life expectancy from 10 years to 9 years, in order to alleviate a particular condition, it suggests that she values a year spent living with that condition as worth .9 or less the value of a year spent living without it. By finding the precise point at which that person becomes indifferent between a reduction in life expectancy and an improvement in quality of life, it is possible to assign a QALY value for life with the condition. This can then serve as a metric, for comparing gains or losses in quality of life with gains or losses in life expectancy. So, for example, the “badness” of being confined to a wheelchair might be reflected in a year lived with that condition being assigned a QALY value of .8. As a result, offering hip-replacement surgery to a population with an average life expectancy of 10 years, although it would not increase anyone’s actual quality of life, would be equivalent – for resource-allocation purposes – to increasing each person’s life expectancy by 2 years (because the QALY value of living with the condition makes the 10 years confined to a wheelchair equivalent to only 8 years with full mobility). Now this all sounds a bit like saying that the lives of people who are confined to wheelchairs are somehow worth less than the lives of those who aren’t, but this is not in fact what is being said. “Life” is being used as a metric of comparison, not as a unit of evaluation. The simple fact is that people care about more than just survival, they care about their quality of life. The phenomenon of “living wills” shows that many people would rather die than be kept alive under certain circumstances. (In particular, most people value being alive yet unconscious, with no hope of regaining consciousness, as being without value, or even as having disvalue.) Thus every health care system is going to make tradeoffs between investing in interventions that extend life and those that improve quality of life. The QALY framework represents an attempt to make explicit, and hence to systematize, the criteria for doing so. Far from being discriminatory, the use of QALY measures in health care resource allocation actually reflects a commitment to a certain sort of egalitarianism. It is based on the view that the opportunity to live life for a given duration, in a particular state of health, is of equal value, regardless of whose life it happens to be (or conversely, that death and disease are of equal disvalue, regardless of whom they may strike). But that having been said, it should be acknowledged that there are a number of other egalitarian intuitions at play in the health care sector, which are clearly in tension with the egalitarian intuition that is at the heart of the QALY framework. For example, the American Association of Retired Persons has at various times been mobilized in opposition to CEA, on the grounds that the use of QALYs discriminates against the elderly – as does any framework that assesses benefit by considering the duration of life-expectancy gains. This is because young people have greater life-expectancy, so an appendectomy performed on a 30-year old will produce much more “benefit” than an appendectomy performed on a 90-year old, even if it has the immediate effect of restoring each to perfect health. Now it is important to issue a small caution, in order to clarify that CEA is not used to make individual patient care decisions. It is used rather to determine which therapies will be made globally available through the insurance system. Thus it would not be the case that two patients might be lying in hospital beds side-by-side, one of them younger the other one older, and a particular therapy would be offered only to the younger one, not the older one (on the grounds that offering it to the older one does not produce enough benefit to justify the cost). Similarly, a therapy would never be denied to a disabled person, on the grounds that the extra years added to that person's life counted for less, because of the pre-existing disability. On the contrary, the value of a therapy is calculated using the average life expectancy gain of the average patient receiving it, and is then made available to everyone or to no one, on the basis of that assessment. Thus there is no danger of differential treatment based on, say, age or disability.65 What is true, however, is that the use of a QALY framework at the global level will tend to favour funding for therapies that disproportionately benefit the young. On a per patient basis, for example, treatment of childhood leukemia is going to be assigned higher priority than treatment of Alzheimer’s disease, because “saving the life” of a child winds up saving more life than “saving the life” of an elderly person. In this sense, the CEA does discriminate against the elderly, in the sense that it does not treat all persons as having equal claim on a given quantum of resources. This gets confusing, however, because many people find this sort of “discrimination” to be both intuitively correct and fully justified. Many people’s thinking about these questions is informed by what is called the “equal innings” intuition – that everyone is entitled to a certain amount of life, and that there is something more unfair (or at least tragic) about a child contracting a deadly disease than an older person contracting it, since the child hasn’t had his fair share of life yet. From this perspective, there is something untoward about the elderly “hogging” vast quantities of health care resources, despite having already led full lives. Complicating things, however, is the fact that some people also have a luck-egalitarian intuition, which says that in cases when people are not responsible for suffering from various conditions, the fact that these conditions are more or less expensive to treat should not be allowed to determine their prospects. From this perspective, one person should not be entitled to less health care than some other, just because the disease he happens to have caught is more expensive to treat than the other person’s. 65 Again, this is why it is important to emphasize that CBA is a procedure used to guide correction of the large-scale market failure besetting the insurance system, as opposed to moral consequentialism, which is a theory that could be used to guide the treatment of individual patients. People also have a strong tendency to direct resources to the person who is in the worst condition, not necessarily to the place where it will do the most good.66 At the limit, this takes the form of the “duty of rescue,” which takes the form of a strict deontic prohibition on “letting people die,” and therefore entails a willingness to spend arbitrarily large quantities of resources to keep someone alive. Both of these allocation formulas are of course in tension with the principle of triage, which requires pretty much the opposite – that “lost causes” be identified and denied resources altogether, so that attention can be focused on cases where there is greater opportunity for improvement. The point of all these examples is to show that people have a number of very strong deontic intuitions in the domain of health care resource allocation, and that although many of these intuitions are in tension with the basic principles that inform QALY-based CEA, they are also in tension with one another. As Larry Temkin and others have shown, it is extremely difficult to get any consistent set of principles for the distribution of resources in this domain.67 Thus it is typically not the case that the aggregative framework recommended by CEA runs headlong into a single, unified, egalitarian or deontic principle. It tends rather to run into a morass of confused and contradictory principles. The way that the tension is therefore diffused is largely institutional. The QALY framework is used only as a basis for very high-level budget decision-making by the state. Medical decision-making on the ward, by contrast, is made on a far more deontological basis, with a set of principles that constitute the traditional professional ethic of doctors and nurses. This can generate a variety of tensions, but not ones that tend to undermine the validity or usefulness of CEA as the basis of high-level budget decisionmaking. 6. Saving an ecosystem The municipality is considering expanding its highway network, by building a new expressway in a ravine that cuts through the city. Doing so, however, will disrupt the natural ecosystem in and around the river that formed the ravine, producing significant habitat destruction for local wildlife. The municipality decides to do a CBA, in order to determine whether the expressway project is worth undertaking. The advantages and disadvantages of the roadway – including pollution, noise, and impacts on downtown congestion – are considered. Questions arise, however, over how to handle the environmental impacts of the project. Recreational use of the ravine is absolutely minimal. It is too inaccessible to be used as a park, hunting is not permitted, and no one fishes in the river. Thus the impact on human welfare of the wildlife habitat and ecosystem destruction are quite small. 66 Eriki Nord, Cost-Value Analysis in Health Care (Cambridge: Cambridge University Press, 1999), p. 64-69. 67 Larry Temkin, “Equality or Priority in Health Care Distribution,” (unpublished). At the same time, many citizens are profoundly concerned about the environmental consequences of the project. Thus the planners decide to incorporate an “existence value” into the CBA.68 Residents of the municipality are asked how much they would be willing to pay to preserve the ravine ecosystem, regardless of whether they make any direct use of it (e.g. such as through recreational use). This provides an estimate of an additional cost associated with construction of the roadway. Once this cost is taken into consideration, it turns out that construction of the expressway fails the CBA test. This is of course controversial, and soon a group of citizens who had supported the expressway project step forward to challenge the decision in court. They argue that the use of “existence values” in the CBA was illegitimate, and unfairly tipped the scales against the project. Such values are, according to their brief, a type of “intrusive preference,” which should not be counted when it comes to determining the level of “social welfare” produced by a particular change. The inclusion of “existence values” represents a form of gerrymandering of the CBA by environmentalists, based on the introduction of a type of preference that no one would be willing to accept in other areas of decisionmaking. For example, a person might have very strong views about what colour his neighbour’s house should be painted, and might become quite distraught when the neighbour decides to change it. Or a person might have very strong views about what books other people should read, or what sort of food they should eat. But unless it is possible to show some tangible “harm” – as John Stuart Mill argued long ago – then whatever loss of welfare this person suffers should not count as an argument in favor of social regulation. In other words, for people to argue that the destruction of a natural ecosystem lowers their welfare, even though they have never in any way interacted with that ecosystem, nor do they have any concrete intention to do so, is an abusive concept of welfare. Allowing it to “count” as a consideration that speaks against the proposal, on par with the concerns of drivers wanting to reduce their commute time to work in the morning, opens up a Pandora’s box, making the entire CBA framework entirely unworkable. For example, when people eat pork, even in the privacy of their own homes, should the disgust that is evoked among observant Jews and Muslims, at the very thought that someone is doing so, count as an “externality,” whose cost must be taken into consideration by the state (justifying, perhaps, a special tax on pork?) Presumably not. However, trying to specify why this disgust reaction, along with the attendant loss of welfare, should not count is surprisingly difficult.69 There are several 68 B. Weisbrod, “Collective Consumption Services of Individual-consumption Goods,” Quarterly Journal of Economics, 78 (1964): 471-477. For critical discussion, see John Quiggin, “Existence Value and the Contingent Valuation Method,” Australian Economic Papers, 37 (1998): 312-329. 69 For my reflections on a similar set of issues, see Joseph Heath, “Envy and Efficiency,” Revue de philosophie economique, 13 (2006). candidate theories, but what they all have in common is some reference to the lack of close connection or involvement between the event that occurs (the consumption of pork) and the loss of welfare suffered by the individual who is upset by it. And yet, this would seem to apply to the environmental case as well. The mere fact that some people don't want a piece of land to be developed, and will get upset if it is developed, doesn't seem like the right sort of preference to count as an argument against development. In fact, it seems more like a moral judgement, to the effect that the piece of land ought not be developed. But this is what the CBA is supposed to be determining. In order to make that determination, the CBA needs to start with some characterization of the interests that are at stake in the decision. To allow people's moral attitudes towards those interests to count as interests creates a variety of problems, including opening the door to an objectionable form of double counting. Supporters of the “existence value” practice respond, however, by emphasizing that the desire to set aside land for conservation purposes is an extremely common preference, which markets already cater to quite extensively at both an individual and a club level. Many wealthy individuals buy ecologically sensitive land, in order to protect it from destruction. It is well-known, for instance, that CNN founder Ted Turner purchased 128,000 acres of Patagonian wilderness with this in mind. More significantly, wealthy conservationist Douglas Tompkins owns more than 2 million acres in Chile and Argentina, with the goal of protecting biodiversity.70 Many celebrities and wealthy environmentalists have been buying large tracts of Amazonian Rainforest. There are also a large number of popular nonprofit organizations, such as the World Land Trust, the Nature Conservancy, the Rainforest Conservation Fund, or Nature Trust in Canada, which allow less-wealthy individuals to pool their funds to purchase land, which is in turn set aside in perpetuity as a private conservation trust. Most of these people, it should be noted, will never have occasion to visit these properties, or to interact with them in any significant way. Thus unlike the case of the books that other people read, the food that they eat, or the colour that they paint their houses, the interest that people have in preserving land for conservation is one that private markets already cater to. This suggests that it cannot be overly intrusive. Of course, some of the conservationist impulse seems directed toward “existence value,” but at the same time, a certain element of it can also be construed as as the preservation of an “option value,” particularly when it comes to protection of biodiversity (an often expressed aim). Ecosystems tend to be rather easy to destroy, but extremely expensive, difficult and time-consuming to rebuild. Thus even a very small possibility that one might want to make use of it in the future can make it rational to preserve it. (This issue arises quite often with rainforest biodiversity, where many of the plant species have not even been 70 http://www.theguardian.com/environment/2008/feb/13/conservation catalogued, and there is some suspicion that they may have useful medicinal properties). Furthermore, in the same way that having a hospital, or a fire station, in the neighbourhood can be of value to residents, even if they never have to go to it, the presence of a nature preserve can also generate value for those who never directly enjoy its charms or benefits. The knowledge that it is there can have positive value. Thus the burden of proof, in the case of ecosystem preservation, is not to show that the relevant existence values are legitimate, it is merely to show that there is some sort of a market failure, which explains why the state should be undertaking the conservation effort, rather than a non-profit organization or an individual conservationist. 6.1 The intrinsic value objection For many theorists who approach these issues from a background in environmental ethics, the cost-benefit approach to environmental valuation is an absolute non-starter. In many ways, the cornerstone of modern environmental ethics is the critique of “anthropocentric theories of value,” of which all major schools of thinking in normative ethics are taken to be variants. As J. Baird Caldicott described it, “An anthropocentric value theory (or axiology), by common consensus, confers intrinsic value on human beings and regards all other things, including all other forms of life, as being only instrumentally valuable, i.e., value only to the extent that they are means or instruments which may serve human beings.”71 He criticizes this, claiming that it represents an arbitrary limitation on the scope of moral concern. It seems obvious that standard CBA embodies an anthropocentic approach of this sort, because in order for any sort of consideration to count, in the analysis, it must first be translated into an object of human concern. Thus when a CBA considers the “cost” involved in the destruction of a natural ecosystem, it does not consider the cost to the ecosystem, but rather the cost to humans of the destruction of that ecosystem. The problem with this is often expressed by saying that the CBA treats the ecosystem as having only instrumental value, failing to recognize that it possesses intrinsic value. (Kant’s distinction between that which has price and that which has dignity is, again, often quoted in this context.) As a result, while CBA may be an appropriate guide to decision-making when the primary impacts of a policy involve human welfare, it is inappropriate for thinking about environmental questions, where the primary impacts of a policy involve the natural world. Yet while this argument may appear straightforward enough, it does involve a few 71 J. Baird Callicott, “Non-Anthropocentric Value Theory and Environmental Ethics,” American Philosophical Quarterly, 21 (1984): 299-309 at 299. complications that are worth flagging. First is the assumption, often made, that the categories of “intrinsic value” and “instrumental value” map onto the natural and the artificial, so that anything that has been made by humans will have been built for some purpose, and so will have instrumental value. This is not correct, however, since under any reasonable definition of intrinsic value, many human artifacts will also have intrinsic value. Great works of art are the example most often given.72 Consider, for instance, the way that Richard Routley illustrated the concept of intrinsic value, using a “last man” scenario: if you were the last person alive on earth, about to die, would it be permissible for you to kill and destroy as much of the natural world as you could, before the lights go out?73 He regards the fact that many would regard this as impermissible as a sign that the natural world has intrinsic value, which in turn calls out for new environmental ethics that transcends our species chauvinism. And yet many people would be equally alarmed at the thought of the “last man” going through the art galleries and library, slashing paintings and burning books. It seems much less obvious that these artifacts of human creation could have a value that is completely independent of human valuation. These reflections suggest that intrinsic and instrumental values do not correspond to “natural” and “human” values, they represent merely two different modes of human valuation. Further argument would have to be given, in order to show that it is possible for values to be ontologically independent of human valuers – merely pointing to the existence of things with intrinsic value is not enough. Now it may be possible to provide such an argument. What is important for present purposes, however, is the observation that the basic approach taken by advocates of CBA is completely independent of the outcome of these arguments. Because even if it were possible to show that there were values “out there,” completely independent of human valuation, it is not clear what sort of a role they would play in human deliberation over questions of environmental policy. In order for these values to play any sort of effective role, people would have to know what they were, and would have to somehow advocate or represent them. And whatever did ultimately get put forth would undoubtedly be controversial. For example, even among environmental ethicists there is deep disagreement over whether individual animals have intrinsic value, or whether value lies in animal populations, or species, or rather entire ecosystems. The difference becomes an important one when it comes to deciding, for example, whether there is a moral obligation to intervene to prevent predators from killing other animals.74 Thus what John Rawls described as the “fact of pluralism” obtains just as much in this domain of value as in any 72 See Elizabeth Willott and David Schmidtz, “Introduction,” in Elizabeth Willott and David Schmidtz, eds. Environmental Ethics (New York: Oxford University Press,) p. xvii. 73 Richard Routley, “Is There a Need for a New, an Environmental, Ethic?” Proceedings of the XVth World Congress of Philosophy, 1 (1973): 205-210 74 Mark Sagoff, “Animal Liberation and Environmental Ethics: Bad Marriage, Quick Divorce,” in Elizabeth Willott and David Schmidtz, eds. Environmental Ethics (New York: Oxford University Press,) pp. 38-44. other (and perhaps even more so, since the emphasis on the objective independence of these values makes it much more difficult to know what they are, and makes it practically impossible to agree on any procedures for determining their significance). As a result, what is typically taken as a commitment to an anthropocentric theory of value in CBA is better understood as just a consequence of liberal neutrality (or that the decision-making procedure is “political,” in Rawls’s sense of the term). When it comes to ecosystems, for instance, there is disagreement about what sort of value they possess – with some people valuing them instrumentally, others valuing them intrinsically, some valuing them quite highly, others assigning them only negligible value, etc. The liberal state is committed to remaining neutral on these questions, and so is not willing simply to “take sides” on the basic question of how natural ecosystems should be valued. Instead, it looks to how individuals in the society value them, to try to gauge the intensity of their commitments. It does so by trying to determine how much the individual would be willing to give up for the sake of those values (which is to say, that the individual’s WTP is). However, in recognition of the fact that certain environmental goods are value both instrumentally and intrinsically, the CBA takes into consideration not just the use value of the land to drivers, but also the existence value of the ecosystem to conservationists. Thus the inclusion of “existence value” in a CBA is essentially a response to the criticism that CBA focuses only on instrumental value to the neglect of the intrinsic values. Since money has only instrumental value, the use of a money as the metric of comparison is misleading, suggesting that everything is being “reduced” to instrumental value. In fact, it it only being used to make instrumental values commensurable with intrinsic values – something that must necessarily be done, not matter what the decision procedure, since no one thinks that intrinsic value categorically trumps instrumental value. This has, of course, been the subject of considerable misunderstanding, and it is entirely possible that the “optics” of using money and WTP values in environmental CBA are so irremediably bad that some other metric should be chosen. When one scratches the surface, however, what becomes apparent is that much of the opposition, among environmentalists, to the use of CBA as a guide to decision, is not actually based on anything specific to CBA, it is really an opposition to liberal neutrality as such. In other words, it is really a demand that certain first-order values be used as a direct basis for legislation. Formulated as such, the demand obviously has troubling implications – which take us, however, beyond the scope of these reflections on CBA, and relate back to the more fundamental arguments that have been developed, over the years, for the wisdom of liberal political arrangements. 6.2 The precautionary principle objection A second objection that is often made to the use of CBA in environmental decision-making, particularly with respect to ecosystem protection, is based on an appeal to the “precautionary principle,” and the suggestion that this principle is more appropriate than the aggregative principle employed by CBA. Unfortunately this is a difficult criticism to assess, because there is no single, widely accepted formulation of the precautionary principle.75 The general idea, stated at a high level of abstraction, is that if there is some possibility of harm from an action, and yet there is some uncertainty as to whether this harm will materialize, the burden of proof should fall upon the proponents of that action to show that the harm will not materialize, and until this burden has been discharged the action may not proceed. This is an all-purpose principle for dealing with situations of epistemic uncertainty, but it is often thought to be particularly relevant to actions that involve making changes in complex ecosystems, because it is so difficult to anticipate what the effects of an intervention will be. The precautionary principle is also often appealed to as an objection to the introduction of new synthetic chemicals to the environment, or genetically modified organisms. Opponents insist that scientists should be obliged to “prove that it is safe,” before they are allowed to act. Not knowing what the exact consequences of, say, a new pesticide will be should count as an argument against introducing it, not as an argument in favour. When formulated in this way, however, the principle is far too restrictive – since it is seldom possible to rule out entirely the possibility of some “unknown unknown.” Furthermore, if formulated as a response to Knightian uncertainty, it does not offer much guidance when dealing with the known probability of harms. Most importantly, it does not permit any balancing of possible harms with possible benefits. As a result, depending upon the exact formulation, it might wind up blocking an action with enormous, known benefits, because of a miniscule probability of harm. Finally, there is the fact that, in many cases, both action and inaction may have unknown effects, and so precaution winds up rather arbitrarily privileging the status quo. Thus there have been many attempts made to formulate a somewhat more moderate version of the principle. Stephen Gardiner, for instance, has argued that the precautionary principle can be understood as the application of maximin reasoning to particular cases.76 The idea is that, when making a decision, one must focus on the “worst case scenario,” and select the action that is best in that case. (A similar idea informs the Final Declaration of the First European “Seas at Risk” Conference: “If the ‘worst case scenario’ for a certain activity is serious enough then even a small amount of doubt as to the 75 See discussion in Cass Sunstein, Laws of Fear: Beyond the Precautionary Principle (Cambridge: Cambridge University Press, 2005): 18-20. 76 Stephen M. Gardiner, “A Core Precautionary Principle,” The Journal of Political Philosophy, 14 (2006): 33-60 at 48. safety of that activity is sufficient to stop it taking place.”77) And yet the idea that one should assign lexical priority to the worst case scenario seems far too extreme. It is one thing to be risk-averse in one’s decision-making – this can easily be incorporated into standard CBA. To assign lexical priority one’s concern over a potential loss requires ignoring all other scenarios, and thus, ignoring foregone gains of arbitrarily large magnitude. It is difficult to avoid the suspicion that, in the standard sort of environmental case, where there is a small risk of great harm to the environment that is being weighed against a reasonable certainty of producing a moderate economic benefit, the reason for assigning an outsized importance to the small risk is not that anyone considers this an appropriate way of managing small risks, but merely that some people assign much greater value to the environment than they do the economic benefits. Thus the decision principle is essentially being gerrymandered, in order to ensure that the right set of values wins – while nevertheless maintaining the appearance of neutrality among the competing values. The easiest way to determine whether this is the actual motive underlying one's own endorsement of the precautionary principle is to take the proposed decision rule and apply it in a policy domain in which one is not sympathetic to the values that win out when an outsized importance is given to small risks. For instance, many environmentalists might hesitate to endorse the use of the precautionary principle for the management of national security risks. Under the Bush Administration, Vice-President Dick Cheney articulated a version of the precautionary principle for dealing with terrorism. “If there was even a 1 percent chance of terrorists getting a weapon of mass destruction — and there has been a small probability of such an occurrence for some time — the United States must now act as if it were a certainty."78 Thus the invasion of Iraq (along with its $700 billion price tag) could be justified as a “precautionary” measure, given that no one was able to decisively rule out the possibility that Saddam Hussein had weapons of mass destruction, and was willing to give them to terrorists. Many people would not hesitate to describe Cheney’s principle, not as prudent, but rather as paranoid. The real problem with it, however, is that it does not allow any sort of balancing, in this case between the probability of terrorists acquiring weapons of mass destruction, on the one hand, and the enormous cost that Americans were asked to shoulder in order to rule out that possibility. So while this version of the precautionary principle is clearly an alternative to CBA, it is not clear that many people would regard it as very compelling as a general principle of decision-making under uncertainty. The more likely explanation is that people find it compelling in the environmental case because, as a matter 77 Final Declaration of the First European “Seas at Risk” Conference, Annex 1, Copenhagen, 1994. Cited in Sunstein, Laws of Fear, p. 20. 78 Ron Suskind, The One Percent Doctrine (New York: Simon and Schuster, 2006), p. 62. of private comprehensive doctrine, they assign very high value to environmental preservation. Thus the real question is whether there is anything specific to the environment that would justify precaution in that area, but not others. If the answer is no, then there will a strong suspicion that the common appeal to it in this area is really an attempt to put a “thumb on the scale” in favour of environmental values. Thanks to objections such as these, there has been some impetus to weaken the principle further. Many of these modifications, however, have the effect of dissolving the tension between the precautionary principle and CBA. For example, some theorists focus on the epistemic issues that arise, emphasizing the way that the precautionary principle assigns the burden of proof when it comes to determining the impacts of a policy. Most obviously, the precautionary principle rules out using Laplace’s principle of insufficient reason to assign probabilities to outcomes (in the style of subjective Bayesianism). There is an obligation to go out and actually figure out a reasonable basis for assignment of probabilities – when someone challenges a particular action, saying “how do you know it's safe?” one cannot just reply “how do you know it's not safe?” CBA, however, is only intended to address the question of which policy should be adopted, given the information that we have. The question of how much information we should collect before making a decision is, as Adrian Vermeule puts it, “a distinct and logically prior question.”79 Thus, to the extent that the precautionary principle addresses this prior question, it is not really in tension with CBA. To see a clear-cut example of this, consider the basic operating procedures of the Food and Drug Administration (FDA) in the United States, which has adopted something very much like an epistemic precautionary principle when it comes to assessing new drug applications. For a very long time now, the FDA has placed the burden squarely upon pharmaceutical companies to demonstrate the safety and efficacy of new drugs before it will even consider approving them for sale.80 Thus it is the pharmaceutical company that is obliged to fund the research, run the clinical trials, collect the information and submit it to the government. Only once this information has been provided will the FDA proceed to consider the merits of the drug – by considering its costs and benefits. Since almost nothing is entirely safe, and almost nothing is entirely effective, assessing the “safety and efficacy” of a drug involves, on the one hand, considering the probability of side effects versus the probability of therapeutic effects, and on the other hand, weighing the seriousness of the side effects against the significance of the therapeutic effects. As a result, the FDA will be willing to approve a drug with very serious side effects if it is, for example, a chemotherapy agent, whereas it would not be willing to do so if it merely offered headache relief. In other words, at the level of the drug approval it is doing what 79 Adrian Vermeule, “Rationally Arbitrary Decision,” Harvard Public Law Working Paper No. 13-24, p. 5. 80 See Daniel Carpenter, Reputation and Power (Princeton: Princeton University Press, 2010). amounts to a CEA, even though this analysis is preceded by an epistemic procedure that is essentially precautionary. Finally, it should be observed that the version of the precautionary principle most popular with governments – the formulation accepted by the United Nations Conference on Environment and Development, held in Rio de Janiero in 1992 – is popular precisely because it builds in reference to cost considerations. The principle is as follows: “Where there are threats of serious or irreversible damage, lack of full scientific certainty shall not be used as a reason for postponing cost-effective measures to prevent environmental degradation.”81 This is the formulation that has been adopted in Canada to guide environmental decision-making.82 One can see why – the explicit reference to costeffectiveness means that the principle does not require disregarding the cost that precautionary measures may impose. At the same time, the principle is clearly vacuous, because what counts as a “cost-effective” measure depends upon how one handles uncertainty and risk. Thus the force of the principle is really just an epistemic one – it prevents opponents of a particular environmental protection policy from reversing the burden of proof, and demanding that proponents of that policy establish “full certainty” that it is needed before any action can be taken. 7. The three-step procedure I suspect that no one who has followed the argument so far can fail to be impressed by the amount of flexibility and variation there is in the way that CBA is applied to real-world decision making. Many of the objections presented above are perfectly legitimate as objections to the use of bald CBA as a decision procedure, or to the use of a plain-vanilla WTP/WTA calculus. What has happened in practice is that the basic calculus and procedure have been adapted to the specific needs of different policy domains. One can see the divergence of methods this creates most clearly in the growing use of QALY-based CEA in the area of health care resource allocation, compared to the use of VSL-based CBA, with standardized valuations of life, in areas of health and safety regulation. The decision to focus on life years saved as the central outcome, as opposed to lives saved, has very significant implications (not to mention that fact that in neither of these two domains is there any use of expressed WTP values). From a utilitarian perspective, this might seem troublingly ad hoc. In this final section, I would like to suggest that these variations are not the results of path-dependency, gerrymandering, or any other sort of arbitrariness, but rather that they flow from the interpretation and application of the liberal principles articulated at the beginning of the chapter: to 1. Pareto 81 See Principle 15: http://www.un.org/documents/ga/conf151/aconf15126-1annex1.htm (accessed Sept. 11, 2014). 82 Environment Canada, Planning for a Sustainable Future: A Federal Sustainable Development Strategy for Canada (Ottawa: Ministry of the Environment, 2010). improvement, 2. liberal neutrality, and 3. citizen equality. Governments want to put the resources that they have to best use (1), and yet in many different policy areas, citizens disagree in their evaluation of the outcomes that might be achieved. Instead of just picking a winner based on first-order values (2), governments strive instead to show equal “concern and respect” for the values of all citizens, by giving their evaluations equal weight (3) in determining the overall ranking of outcomes. What accounts for the variation in the way that embedded CBA is carried out is largely a consequence of variation in the way that the principle of citizen equality is interpreted and applied in different domains. For instance, the fact that we use a VSL calculus in certain areas reflects the fact that we tend to think of accidents in terms of unnecessary injury and death. Thus we articulate our policy objectives in this area by saying things like “we want to reduce the number of deaths that occur.” But of course, no one is in a position to prevent death, the most that we can do is delay it. So when we say that we are “saving lives,” technically all we are doing is extending them. In the case of workplace safety, however, it doesn't seem relevant to ask how much we are extending a particular life by. If we did, then we would tend to favor initiatives that reduced the accident rate among employees who were younger, or female, on the grounds that doing so resulted in greater benefit. Of course, the thought that we might take into consideration factors like “how much longer a person expects to live” when it comes to workplace safety is not entirely beyond the pale. In the wake of the Fukushima nuclear disaster in 2011, a 72-year old engineer assembled a volunteer corps of over 160 retired civil and nuclear engineers to work in dangerous radiation zones. The rationale was that, being already quite old, most of them would not to live long enough to develop the various forms of cancer likely to be caused by the radiation exposure. Of course, it is crucial to the public acceptability of this arrangement that the work group consisted entirely of volunteers. What the example shows, however, is that these sorts of considerations about life expectancy are not irrelevant to the question of who can be exposed to which workplace hazards. And if one were a utilitarian, it would be difficult to explain why it would ever be permissible to avoid taking them into consideration. In a standard safety CBA, however, we choose not to take them into consideration, in way that reflects a particular, relatively formal conception of equality. We are, in most of these areas, talking about the risk of a highly unexpected, extremely negative event, which no one has done anything to deserve. The template that such a problem evokes is something like the “choosing straws” procedure, where everyone gets an equal chance of choosing the short straw, regardless of any differences there may be among individuals. It is therefore this conception of equality that is reflected in the way CBA is carried out, with the use of a standard VSL. The situation with respect to health care is quite different. Here we are dealing with a situation in which almost everyone has suffered from a particular bit of bad luck, creating a need for medical resources, but where there are not enough resources to fully satisfy everyone. There is, however, significant gradation in the extent to which needs can be met. For example, patients who are denied very expensive chemotherapy drugs will still receive “standard of care” cancer treatment. Since there is largely a fixed envelope of health care spending, sliding one patient up on the scale of treatment will almost necessarily involve sliding some other patient, or patients, down on the scale of treatment. Thus there is reason to think that the relevant conception of equality will be one that focuses much more on ensuring that the system is impartial with respect to individuals, when determining what sort of a claim each individual can make on the pool of resources that are being set aside to meet health care needs. Both of these cases are quite different from the ones in which resources are being claimed by individuals who disagree more profoundly about what sort of use these resources should be put to. In the case of regulatory intervention, for instance, or environmental preservation, there is often deep disagreement underlying the conflict. Unlike health care, where people are seeking roughly the same outcome, in these cases there is no shared outcome to serve as a metric for interpersonal comparison. One could just hold a vote, of course, and count heads to see who should win in the contest over resources. The problem with this procedure is that it fails to take into consideration how strongly people feel about the issue. Thus a large group with very little at stake in an issue might easily outvote a small group with much at stake. A CBA tries to ascertain how much people care, or how much benefit they would derive from a particular use of resources, by asking them how much they would be willing to give up in the way of other resources to secure it (which is to say, what their WTP/WTA value is). The concern for efficiency is obviously in the foreground here, but one can see as well a concern for equality in the insistence that no one group's interests should be privileged. This shows that CBA should not be thought of as a generic procedure, that would ideally be applied in the same way in all domains. My suggestion is that it reflects a combined commitment to three distinct principles, each of which admits of different interpretations, depending upon the situation to which they are being applied. Thus the application of the approach involves, either implicitly or explicitly, answering a fairly lengthy series of questions. This is why, as I suggested earlier, I think it is useful to treat the core procedure as being framed by two “filters,” one on the input side and another on the “output” side.83 (More informally, these can be thought of as “questions to be asked before you 83 There are several good discussions in the literature, providing lists of “questions” that should be asked when contemplating a particular CBA exercise. See, for instance, Michael F. Drummond, “Allocating Resources,” International Journal of Technology Assessment in Health Care, 6 (1990): 77-92 at 81, or Dan Brock, “Ethical Issues in the Use of CostEffectiveness Analysis for the Prioritization of Health Resources,” in G. Khushf, ed. Handbook of Bioethics (Dordrecht: Kluwer, 2004), pp. 352-380. My purpose here is just to provide a bit of structure, by organizing the questions into a sequence. open up your spreadsheet,” then “questions about how to do the calculations in your spreadsheet,” and finally, “what to do with the results produced by your spreadsheet.”) 7.1. Input filters This is not an exhaustive list, but the following represents a series of questions that it is useful (and in come cases essential) to pose before embarking upon a CBA. There are also, of course, a set of completely pragmatic questions that must be asked as well, involving for instance the threshold of significance (a proper CBA is costly to carry out, and so may not be worth doing, simply because the cost involved in identifying the best policy exceeds the largest possible benefit of adopting that policy). What I will be presenting, by contrast, are principled considerations, dealing with the question of whether a CBA represents the right conception of justice to be applied to a particular problem, and if it is, precisely how it should be applied: Is this an appropriate problem for CBA? The most valuable heuristic for determining the appropriateness of CBA is to ask, “Where’s the market failure?” As we have seen, policies that involve primarily considerations of distributive justice, rather than correcting a market failure, are not appropriate candidates for CBA. For example, in Canada, social transfers to the elderly are delivered by two programs. The first, the Canada Pension Plan, is structured like a pension plan, based on lifetime contributions through workforce participation. It exists primarily because of inefficiency in the market for private life annuities. It is therefore quite reasonable to look to cost-benefit considerations in determining a benefit level. The second program, Old Age Security, is a means-tested transfer program, designed to alleviate the problem of poverty among the elderly. It clearly is not motivated by any sort of market failure, but is rather intended merely to improve the distribution of income from the standpoint of equality (or sufficiency). It would therefore be inappropriate to determine the benefit level of a program such as this through appeal to cost-benefit considerations. Are there individual rights that trump the policy? During the highwater mark of the “law and economics” movement, certain legal theorists were fond of claiming that the criminal law could be justified through reference to cost-benefit considerations. Richard Posner made the unfortunate suggestion that rape was illegal because it was inefficient.84 If men and women were to negotiate, in the Coasian style, women would be willing to pay more for the right not to be raped than men would be willing to pay for the right to rape them, and so the efficient outcome requires the assignment of the 84 Richard Posner, Sex and Reason, pp. 357-358. right to control their own bodies to women. Many people found this line of reasoning troubling, for a variety of reasons, but perhaps the most obvious is that control over one’s own body is not usually regarded as something that is “up for grabs” – so that if someone else is able to make better use of your body than you are, he might be granted control over it. On the contrary, the rights that we exercise over our own bodies are normally regarded as “trumps” with respect to projects that others might have that make use of them. Thus there is no point doing a CBA on a project that involves violation of such rights, as it is ruled out ab initio. It is worth noting as well that even when rights do not render a CBA entirely otiose, they may significantly constrain the policy space within which the CBA is conducted, by ruling out certain options. Are there other deontic constraints that preclude the policy? Even if a policy is not ruled out by individual rights, there may be other deontic constraints that preclude the sort of balancing, or consideration of consequences, that is involved in CBA. For example, when it comes to dealing with the loss of life, it is important that CBA involves “statistical lives,” not known individuals, and that the deaths are not the intended outcome of the policy. When these conditions do not obtain, policy may be dictated by the set of rules or deontic constraints. Alpine search and rescue, for instance, is an area that is probably not cost-benefit justified, but that continues to be carried out because the situation triggers a deontic constraint, viz. the “duty of rescue.” (The case, however, is not reducible to one involving individual rights, because individuals are not normally taken to have a “right” to expensive rescue operations.) Some argue that the CBA should be done anyway, so that everyone knows how much the commitment to that particular deontic constraint costs us, as a society. The perception, however, is often that such an exercise is a thinly-veiled attempt to erode public commitment to the constraint. Thus anyone committed to the constraint might argue against the use of CBA in the policy domain. Is there a compelling metric to establish interpersonal comparability? As we have seen, since a CBA fundamentally involves comparing the proposed use of resources by one set of actors against the proposed use by some other(s), the value of the analysis depends heavily upon the quality and fidelity of the metric used as a basis for interpersonal comparison. Every measure has serious deficiencies, but the seriousness of those deficiencies depends in part on the domain in which they are being used. For example, money (and hence WTP/WTA) has major defects, but these are far more severe in the area of medical decision-making than they are in the area of consumer goods. For example, one study found that, after being asked to rank different medical interventions based on their perception of the benefits, participants were asked how much they would be willing to pay for each one. For a significant fraction of respondents, the WTP values failed to respect the earlier ranking, i.e. subjects were in some cases willing to pay more for interventions that they considered less beneficial. Thus if there were not something like the QALY measure available, and one was forced to use WTP/WTA values, one might easily conclude that it was not worth doing a CBA for medical resource allocation, simply because the decisions are so far removed from the everyday context of consumer choice, and there are so many other complicating factors, that money simply does not serve as a plausible basis of interpersonal comparison (i.e. it generates so much “noise” that the “signal” is drowned out). What is the scope of the analysis? Finally, there is a preliminary question that must be asked about what the scope of the CBA will be, and thus, what factors are to be taken into consideration. Unlike moral consequentialism, which offers essentially a “global” standard of justice – used to evaluate the total benefit that individuals receive over the lifetimes – much of the way that we apply norms of equality is responsive only to “local” justice considerations. This in turn affects what counts as a cost or a benefit. Preventing the death of a worker with a family to support may produce more benefit than preventing the death of a worker who is single, but the benefits to the family do not “count” in a typical CBA. Thus it is important, before initiating a CBA, to identify clearly the range of effects that one is planning to take into consideration.85 One might also decide, rather than strictly limiting the scope of an analysis, to give greater weight to certain effects. 7.2. Performing the analysis Once the policy area has been defined, and a CBA has been selected as both appropriate and feasible for that domain, a series of further questions arise that are, as it were, “internal” to how the CBA is to be conducted: Which costs and benefits should count? Selecting the policy domain already involves excluding certain costs and benefits from consideration, particularly if the CBA is undertaken as an exercise in local justice. And yet even once that is complete, serious questions may arise about which cost and benefits should be taken into consideration within that domain. One of the most significant dangers – commonly warned against in all textbook treatments – is that of “double counting” either costs or benefits (e.g. counting the increase in property values due to construction of a new school), or simple confusion 85 One of the most obvious places where this arises involves the treatment of citizens and non-citizens. Historically, there has been a tendency for national governments to take into consideration only the effects on their own citizens. This, however, generates obvious absurdities, effectively rendering certain cross-border collective action problems unsolvable. It is also a non-starter when it comes to dealing with issues like global climate change. about what counts as a benefit (e.g. counting the creation of jobs as a benefit). In other cases, certain preferences may be excluded, on the grounds that they are either anti-social or inauthentic. For example, many CBAs on cigarette smoking do not count as a benefit the enjoyment that smokers get (presumably on the grounds that they are addicted). Also, as Wolff has observed, people who are exposed to certain hazards because they have broken the law (e.g. by trespassing), might have their interests assigned less significance than those who have respecting it.86 What metric to use? It may only be possible to make a final decisions about which metric to use after the CBA has been initiated and some data collected. For example, if there is an enormous tension between expressed and revealed WTP (e.g. residents claim that a transit project would confer an enormous benefit, and yet there is no reflection of this elsewhere in property values), then one might want to look for some other metric. For example, some transit planning is done as a CEA using “commute time” as the metric, rather than as a CBA with a monetary measure. How to measure preferences, given the metric? No matter which metric is chosen, there are different ways of getting at the information one is looking for. Furthermore, the different methods for getting at it can yield dramatically different results. With WTP/WTA, for instance, one can choose between stated, revealed, standardized or negotiated values. Similar indeterminacy affects the QALY measures, where one can get dramatically different results on the “quality of life” scale depending on whether one asks patients before they have the condition or after. Indeed, both before and after valuations are so compromised by cognitive bias that some QALY scales are determined by asking caregivers to do the quality evaluation. What corrections should be made to establish comparability? As we have seen, some metrics require adjustment in order to establish a normatively compelling basis for interpersonal comparison – most obviously, any metric that is itself subject to diminishing returns. This is a major issue with the use of money as a basis for comparison, and it generates a set of problems that are, in my view, not taken seriously enough by most proponents of CBA. On the other hand, the simple attachment of distributional weights, as the U.K. Green Book recommends, is not an adequate response in all cases – especially if one adjusts only the stated WTP values but not the market prices. What social discount rate should be applied? In some cases, the question of whether to apply a social 86 Wolff, “What is the Value of Preventing a Fatality?,” p. 63-64. discount rate, and if so, what that rate should be, is one of the most difficult to arise “internal” to the CBA procedure. In medical decision-making, it is common to apply no discount rate (which is, to say, a discount rate of zero). In other cases, such as infrastructure investment, a simple “opportunity cost” framework seems unproblematic. Then there are a variety of mixed cases, including most environmental ones, where neither approach seems appropriate (zero is too low, but the rate of return on capital is too high). The literature on this is too vast to be usefully summarized here, what is important is simply to flag the question as one that must be answered, in order to determine how the CBA should be carried out. 7.3. Output side After the CBA has been performed and the best option identified, there are still a number of questions that must be asked before that option can be recommended. If potential-Pareto is satisfied, is it possible to achieve actual Pareto? As we have seen, despite using the Kaldor-Hicks criterion, the underlying commitment in CBA is actually to resolving collective action problems, which is to say, promoting Pareto-efficiency. Satisfying Kaldor-Hicks shows that it is possible for the “winners” under the policy to compensate the “losers.” In many cases, a policy can only go forward if it is possible to actually compensate them as well. For example, landowners who suffer expropriation so that a public infrastructure project can go through are given actual compensation. If it is for some reason not possible to compensate the losers – e.g. if it is impossible to know who they are, or to sort them out from those who merely claim a loss – then that might count as an argument against pursuing the project. Or, in cases where the losers are not to be compensated (as with many regulations), some distributive justice argument would have to be adduced to show why there is no obligation to do so. Would implementation conflict with rights, or other deontic constraints? Even if the policy is not precluded ab initio by individual rights or other deontic constraints, conflict may turn up once the CBA is conducted. In particular, with rights that do not function as “trumps,” but rather have a threshold of application, it may not be possible to know if a threshold would be surpassed without conducting a CBA. A policy might also have a negative effect on gender, racial, linguistic or religious equality or raise issues pertaining to Aboriginal treaty rights. Would the policy have adverse distributive impacts? A CBA may find a particular policy to be worth implementing based on people’s expressed willingness to pay, but the question of how it will actually be paid for is a quite different one. Whether it is financed through income taxes, consumption taxes, payroll taxes, or user fees will have important implications for the distributive effects of the policy. If the impacts under any feasible modality were considered unacceptable, then that would provide reason not to proceed with the policy. Is it in tension with other qualitatively expressed considerations? There are a variety of other, purely qualitative considerations, which might speak in favour of or against a policy. For example, it is entirely possible that state subsidization of amateur athletics is not cost-benefit justified, but that it plays an important role in promoting national unity. Rather than trying to quantify these benefits and insert them into the calculus, it might make sense just to introduce them as qualitative considerations on the “output” side. Sunstein suggests, along these lines, that there are qualitative features of certain risks, such as “the voluntariness of the risk, its potentially catastrophic character, whether it is especially dreaded,”87 that one might want to take into consideration after the CBA is complete. That having been said, it is important not to engage in double-counting. For example, it is sometimes suggested that CBA articulates our commitment to just one value, viz. efficiency, and that once its results are known these must be balanced against all other values. This is incorrect, insofar as the CBA already takes into consideration of the value-commitments of all citizens (to the extent that these inform their WTP). The efficiency and equality principles at the heart of CBA are not just “values” like all others, they are principles that we employ in order to provide a basis for public decision-making in the context of first-order value pluralism. Thus whenever one applies “qualitative considerations” as an output filter on a CBA, it is important that one be able to specify quite clearly why those considerations could not be represented within the CBA. 87 Sunstein “The Cost-benefit State,” p. 39. This overall deliberative procedure is illustrated in Figure 6. Methodological issues Input filters Is this an appropriate problem for CBA? (e.g. is the structure primarily winwin, not win-lose) Are there individual rights that “trump” the policy (e.g. is it obviously unconstitutional) Is there a compelling metric to establish interpersonal comparisons of utility? (e.g. if WTP/WTA is the best option, are expressed preferences meaningful?) Which preferences should count? (e.g. are “intrusive” preferences to be discarded? Should“existence values” be considered) Use expressed, implied, standardized, or negotiated values? What corrections must be made to achieve interpersonal comparability? (If using WTP/WTA, what weighting should be used to correct for diminishing marginal utility of money?) Should distributive weights be imposed? What social discount factor should be applied? Output filters If potential-Pareto is satisfied, is it possible to achieve actual Pareto? (i.e. can the “losers” from the policy actually be compensated by the “winners”) Is the policy in conflict with other constraints, particularly deontic ones? (i.e. does a health policy violate a “duty of rescue,” does it conflict with treaty rights?) Does it have adverse distributive impacts? Is it in tension with other values? Does it have impact on specific lives, or statistical lives? Figure 6. Input and output “filters” on CBA 8. Conclusion One of the most striking features of the slow, steady advance of CBA across a range of jurisdictions is that it typically arises through the initiative of the executive branch of government (e.g. via an Executive Order in the United States, or as Treasury guidance in the U.K.) Legislatures, by contrast, seldom require it, and if they mention it at all in a statute it will often be with the goal of prohibiting it. Similarly, while courts have increasingly come to regard it as the “gold standard,” when it comes to evaluating administrative decisions in areas of delegated or secondary legislation, they also stop well short of imposing it – and are often quite aware of the limitations of their own abilities, when it comes to conducting such analyses. Thus I think it is not unreasonable to suggest that CBA arises out a set of normative commitments that are endogenous to the executive branch of government. This would be troubling, if it were the case the CBA reflected a commitment to full-blown utilitarianism, simply because utilitarianism is a controversial doctrine, elements of which more properly belong in the domain of political ideology, and hence of democratic contestation. Thus it would be a violation of civil service neutrality for the executive branch push for the use of CBA. What I have tried to show instead is that the commitment to CBA stems from a set of much more minimal principles, all of which can plausibly be construed as structural features of a liberal political order. So while the use of CBA has obvious normative implications, these norms are ones that transcend, or stand outside of, democratic contestation, precisely because they provide the structure within which democratic contestation can occur. And yet CBA is obviously controversial, which would seem to be in tension with this claim. Thus my objective in this chapter has been to show that these controversies arise mainly from the “optics” of the procedure, and that most objections are based on rather diffuse intuitions that are difficult to sustain under analysis, or that apply with equal force to any other feasible decision procedure.

Log In

Cost-Benefit Analysis as an Expression of Liberal Neutrality

Cost-Benefit Analysis as an Expression of Liberal Neutrality

Free Related PDFs

FREE RELATED PAPERS

RELATED TOPICS