Cost-Benefit Analysis and Public Reason
How Applied Economic Normative Analysis Can Contribute to Public Justification
Eric Schliesser has recently published a 3-part series on Joseph Heath’s book The Machinery of Government: Public Administration and the Liberal State (part 1, part 2, part 3).[1] I’ve yet to read Heath’s book but I’ve found Eric’s essays very instructive and stimulating – if the discussion of a book gives the reader the urge to read the book, that’s generally a good sign! I’ve been particularly interested in the third installment of the series, which discusses Heath’s views on cost-benefit analysis (CBA for short). I take it as an opportunity (and an excuse!) to present my readers with my own thoughts on this issue, more specifically how CBA can participate in the process of public justification of policies and institutions by providing intelligible reasons for and against them. As far as I can tell, I end up with a similar position to Heath’s and that Eric criticizes.
It may be worth starting with a relatively obvious but nonetheless important point. While CBA occupies a central role in the liberal state’s administrative practice of conception, justification, and evaluation of public policies, it is fairly unpopular among many academic circles, especially philosophers. From Eric’s discussion, I understand that Heath deploys great efforts to defend the use of CBA within the broader context of the executive function of the liberal state. Closely tied to his interpretation of Pareto-efficiency as a normatively charged social choice criterion, Heath sees CBA as a tool that, in the context of market failures, works by reasoning based on a counterfactual where, in the absence of market failures, parties would have voluntarily contracted for some outcome O. The objective of CBA is precisely to produce O, or more exactly to ensure that the policy under consideration will implement O.[2]
CBA is an applied version of what I like to call the social choice model of normative analysis. In its most generic and general version, the social choice model establishes a link between a set of individual judgments about social states (usually but not necessarily, preferences over bundles of goods) and a social evaluation of those social states. This corresponds to the basic modeling structure that one can find in the fields of welfare economics and social choice theory. The figure below represents graphically the social choice model that is constitutive of mainstream normative economics:
The social choice model is a template for modeling a collective choice rule, that is, a procedure to infer a social evaluation and eventually a social choice from a set of individual judgments. Such a model usually distinguishes between the informational basis (what information is taken into account and how this information is captured) and the aggregation rule (how the information is aggregated to generate an output). Now, it is easy to fit CBA within this generic model, as illustrated by the second figure:
CAB evaluates policies with respect to the changes they trigger in the allocation of goods based on individuals’ subjective welfare judgments, i.e., their preferences. It does so based on a monetary measure consisting of determining the amount of money an agent would be willing to pay for having the policy implemented (and changing her basket of goods from G to G’) or, almost equivalently, how much money the agent would ask for not implementing the policy.[3] These monetary measures (some positive, others negative) are then summed. A policy passes the analysis with success if the sum is positive, it fails if it’s negative.
Several points are worth noting. First, what you do in a CBA is a pairwise comparison between a state of affairs that the implementation of a policy would bring about and the current situation, i.e., what happens if nothing is done. Therefore, the CBA doesn’t compare between possible policies. Second, it follows that CAB doesn’t rank policies and doesn’t tell you if the policy you’re evaluating is the best available. Third, by construction, the evaluation depends on the status quo. Indeed, there is a well-known objection to CBA called the “Scitovsky paradox” according to which, in some cases, a policy P changing the allocation from A to B and a policy P’ moving the allocation from B to A can both pass the CBA.
As I said above, CBA is not necessarily very popular among non-economists and there are many critiques that have been developed against it. Listing them would not be very interesting but it’s worth mentioning the most frequent and obvious. The first obvious critique is that the use of monetary measures introduces a kind of bias. It seems to entail assuming that everything is commensurable and has a monetary value. The risk, however, the critique goes, is to improperly reflect the non-monetary value of some activity or “good.” The second critique is stronger. For reasons that are relatively clear, monetary measures are sensitive to individuals’ initial endowments, more specifically the money that have at their disposal. In general, one’s willingness to pay rises with one’s income and wealth. This introduces a second bias because by construction the willingness to pay of wealthy individuals (be they positive or negative) will be bigger than those of poor individuals. A third standard critique is that CBA is implicitly based on the idea of compensation (a policy is evaluated favorably if the winners could compensate the losers) while in practice it is fairly rare that compensation will indeed occur. This a variant of this critique that Eric develops in his third post when he points out that the unreflective use of CBA may forget about the fact that some people are “structural losers:”
“By structural losers I mean social individuals or groups who lack the political clout or social standing to impose on the state the obligation to pay out actual compensation when the administrative state pursues, in principle Pareto improvements across multiple domains. As I noted, this is likely to occur in conditions of structural inequalities, especially in contexts where the permanent upper civil service is drawn disproportionally from social classes other than those who will be structural losers.”
These three critiques are ordered in increasing degree of seriousness. They can all be somehow answered by situating CBA and, more generally, the social choice model, in a broader framework of public justification. What is basically the point of doing a CBA? It is to a give a public reason for adopting or rejecting a policy. The kinds of reasons that CBA provides are, at least up to some degree, transparent and intelligible. It is transparent because the method for doing such an analysis is publicly known and shared. It is intelligible because it is based on basic notions (monetary evaluated “benefits” and “costs”) that, at least in a market society, we are acquainted with. That doesn’t mean that the reasons CBA provides us with are unobjectionable or the only relevant ones. CBA is a valuable input in the general procedure of balancing reasons that are publicly available to determine if the current state of affairs is justified and if it can be improved upon.
What does this imply for the three critiques mentioned above? First, to see CBA as participating in the process of public justification implies that the reasons it provides have weight in so far as we agree that efficiency matters. In this perspective, as David Schmidtz points out in his book Living Together, CBA offers stronger reasons “for blocking inefficient takings rather than for licensing efficient ones.”[4] We can in general agree that efficiency is an important, maybe necessary condition to be satisfied by public policies, but we may not agree that this is the only one that matters. Hence, CBA is very good at providing decisive public reasons against policies, not necessarily for policies. The first critique above, that CBA reduces everything to commodities and pretends to compare values that are incommensurable loses its strength once we realize that one way or another, we cannot make (collective) decisions without comparing alternatives and values. For that, it’s better to provide a justification for the way the comparison is done than no justification at all. The alternative to CBA is to use other metrics to compare choice options. Whatever we think about incommensurability, we cannot make decisions without assuming that things are commensurable, if only implicitly. As I noted above, CBA has at least the merit of being transparent in the way it makes things commensurable.
Regarding the second critique, since there is no doubt that other considerations than efficiency matters, we may want to take into consideration that monetary measures are biased toward the rich. There are technical ways to circumvent the problem, for instance by attaching distributive weights to monetary evaluations,[5] but note a more fundamental point. In a market economy, individuals’ decentralized economic decisions will tend to converge toward the outcome of an unweighted CBA. So, for instance, if you use CBA to determine whether to build waste treatment facilities in some area, the policy has more chance to pass the test if the facilities are planned to be built in a poor region than in a rich one. That’s not surprising: wherever the facilities are built, rich people will be more willing to pay for an expensive house far from the nuisance than poor people. Hence, the market processes tend to make poor people live closer to waste treatment facilities than rich people. We may find this disturbing and ethically problematic, but this has nothing to do with CBA per se. This is more generally a feature of market economies. So, while this “bias” may provide a reason not to take the reasons provided by CBA as decisive, this reason is not stronger than the one objecting to market mechanisms on the ground that they reflect unequal economic situations.
This leads us to the third and strongest critique. CBA, in the same way as market processes, is unlikely to remedy structural inequalities. Because of that, some people risk always being on the wrong side of the winners/losers divide. That’s a strong argument against the systematic and unreflective use of CBA to evaluate public policies. But here, again, the problem is not so much with the tool than with the underlying practice. As Schmidtz puts it:[6]
“CBA is an important response to a real problem. However, it is not magic. There is a limit to what it can do. CBA organizes information. It can be a forum for eliciting further information. It can be a forum for correcting bias. It can be a forum for giving affected parties a voice in community decision-making, thereby leading to better mutual understanding and acceptance of tradeoffs involved in running a community. CBA can be all these things, but it is not necessarily so. CBA can limit a system’s tendency to invite abuse, but CBA is prone to the same abuse that infects the system as a whole. It is an antidote to abuse that is itself subject to abuse. Biased inputs will tend to crank out biased outputs. However, although the method does not inherently correct for biased inputs, if the process is conducted publicly so that people can publicly challenge suppliers of biased inputs, there will be some tendency for the process to correct for biased inputs as well. We can hope there will be opportunities for those who see a bigger picture to challenge those who have tunnel vision and can’t see anything beyond their narrow concerns, but we cannot guarantee it.”
More than “being a forum,” CBA is more generally part of a public forum where we decide what we should collectively do. As a contributor to public deliberation, what CBA can at best establish is that a policy is justified in so far that it brings more benefits than costs. Other considerations are relevant too, and they should probably have stronger weights in some context. The point however is that there is no strong case for doing without CBA at all. There is no clear substitute for CBA (or similar tools that ultimately rely on market evaluations) in a market economy where it is widely agreed that resources should be allocated to a large extent based on agents’ willingness to pay. If you challenge CBA, you also challenge the very principle of market allocation. It’s of course possible, but then comes the big question: is there any other general system for allocating resources that can be publicly justified?
[1] Joseph Heath, The Machinery of Government: Public Administration and the Liberal State (New York, NY: Oxford University Press, 2020).
[2] Maybe a bit more precisely, if market processes can produce a range of Pareto-efficient outcomes O (graphically, all the points on the contract curve that are located on or above the indifference curves of the market participants passing through the initial endowments point), then CBA must make sure that the policy P under consideration implements an outcome O belonging to O. CBA is indeed largely silent regarding the distributive aspect, a point on which I return below.
[3] The two measures correspond respectively to the “CV” (compensating variations) and EV (equivalent variations) that appear in the figure. They are not equivalent because of income effects.
[4] David Schmidtz, Living Together: Inventing Moral Science (New York, NY, United States of America: Oxford University Press Inc, 2023), p. 165.
[5] See Matthew Adler, Well-Being and Fair Distribution: Beyond Cost-Benefit Analysis (New York: Oxford University Press, 2011), pp. 88-114.
[6] Schmidtz, Living Together, pp. 174-5.
Interesting to hear that any diciplines but economists had ANY opinions on CBA (or even heard of it), much less that they were generally negative. The "part of the discourse" point is just the standard Neo-Social Democratic POV separating efficiency analysis from distributional analysis.
Thank you for your kind shout out. (I almost tagged you in today's post on the strategic use of overlapping consensus.)
Your defense of CBA is not far from Heath's (and he engages with some of the criticisms, too--his responses to the first two criticisms are very nicely developed in him). I think Schmidtz is actually more willing to acknowledge the limitations of CBA in practice than Heath is.
Now, let me get to the nub, that there is no alternative generates, in fact, an obligation to be quite transparent about the limitations in use, and to emphasize these to the audience of practitioners. And also to sponsor work on how to remedy the structural defects.