What is there to say about
cost-benefit analysis that hasn't
already been said about Afghanistan? It looks bombed out and depleted.
I should make a few qualifications. First, I am talking about what I regard as "naïve" CBA. In this view, the policy-maker uses her intuition to come up with plausible policy proposals, and then she quantifies the predicted consequences and uses the results as parameters in some kind of algorithm. This algorithm spits out numbers that purport to assess the costs and benefits of the alternative policies, and the policy-maker implements the policy with the greatest surplus of benefits over costs.
In truth I suspect that CBA functions more like an institutional check on bureaucracy, in that regulations can be challenged and sometimes struck down on the grounds that the regulator didn't properly implement CBA in assessing the policy. This is what seems to be going on in
Business Roundtable vs. SEC (PDF), though in that case the SEC wasn't strictly charged with implementing CBA (rather the SEC was supposed to consider the rule's effects on efficiency, competition, and capital formation). Think of
Chevron + CBA as a sort of business judgment rule for regulators (deference will be given to regulatory decisions so long as the proper procedures are followed—and CBA specifies the proper procedures). In this case, CBA may be a way of disciplining regulators, forcing them to gather and confront the data. The regulatory decision is still arbitrary (and subject to criticism and even statutory repudiation), but it is at least well-informed arbitrariness. This is what I think CBA actually does, and this is why I think it is probably a good thing on balance.
But let's assume that CBA is implemented in the naïve way, and that it works in the sense that the need to specify arbitrary values in some circumstances doesn't render the output unreliable. (Maybe the results are not sensitive to the arbitrary values, or arbitrary values yield small differences in policy and those differences don't systematically skew policy in one direction or another.)
My claim is that CBA is only as good as its results, and that there is no basis to conclude that its results are the best policy in any given case or on average (except perhaps in the special case in which the results are judged from the perspective of a person whose values are fully reflected in CBA). Note that I am
not arguing against the use of quantitative analysis in policy-making, and I am not opposed to delegating controversial questions to unelected officials in some circumstances. I simply think that CBA is valuable only inasmuch as it leads to better results than would otherwise be implemented. And crucially, "better" cannot be judged from the viewpoint of CBA itself (this would be circular). "Better" must be judged from the view of the affected constituency, and constituencies tend to have heterogeneous values and interests. There is no reason to think that you can sterilize an issue of its ideological dimensions by running it through CBA.
The responses to this argument come in two flavors.
First, some CBA advocates essentially say, "DOAN WORRY WE FIXORED IT." The problem is that, just as a cat fixors your internet cable by chewing through it, CBA proponents have fixored intractable ideological issues not by untangling the Gordian Knot, but by cutting it. That is, regulators are empowered to resolve the ideological issues by arbitrary means. This is obviously not a satisfactory solution unless you prefer to substitute the judgment of the regulators for the judgment of the affected constituencies.
Second, some CBA advocates reply that CBA should only be implemented where there is no ideological disagreement. This is problematic both because CBA is actually implemented in cases in which there is ideological disagreement, and because it is difficult to imagine a policy that does not implicate ideological issues. An alternative formulation is that elected (or otherwise politically legitimate) policy-makers should specify what is to be maximized, and the regulators should mechanically implement it. I don't think this is how CBA is practiced anywhere. (Another problem is that if CBA is a panacea, it is unclear why it shouldn't be adopted for all policy-making. Indeed, one of my pro-CBA friends has suggested that CBA could be used to set the minimum wage. I hope we can all agree that there is no level at which the minimum wage would be uncontroversial.)
Here is an example that should clarify my point. Imagine that regulators gather some data, do some cost-benefit analysis, and conclude that the implicit value of a life in airline safety regulation is $100,000,000 and the implicit value of a life in car safety regulation is $1,000,000. The regulators respond by tightening car safety regulation.
Now, your task is to persuade your libertarian friend that the regulators have just made the best possible regulatory decision. Your libertarian friend argues that car safety is not difficult to measure, that car manufacturers compete to make the safest cars and actively market their cars on the basis of safety, and that car safety is one of many variables consumers can consider when choosing the best car for their needs. Therefore, the optimal level of regulation of car safety is zero, and policy gets worse and worse as regulations get tighter and tighter. (Your libertarian friend may also argue that airline safety is difficult for consumers to measure because of small sample size and technical complexity, and that airline safety regulation is therefore justified at some nonzero level. Airlines do not market on the basis of safety, and consumers do not consider safety when buying tickets.)
So if I'm right, you are not going to be able to persuade your libertarian friend that increasing car safety is indisputably the best policy. If you think you could persuade your libertarian friend to abandon his beliefs or principles in these circumstances, then I'm eager to hear the argument you would use.
The final thing you might say in CBA's defense is that your libertarian friend could always show up at a public hearing and demand that the regulators use a higher estimate of the benefits of "freedom" when running CBA. That is, CBA is transparent and invites input about how it is to be conducted. It resolves ideological disputes by being democratically legitimate.
I think this is probably untrue. First, most people don't know how to "speak the language" of CBA. A man whose career is about to be ruined by a burdensome regulation may not know which variable needs to be adjusted to get the result he wants. He just knows that he is about to lose everything. And in any case, public hearings and comment periods hardly bestow democratic legitimacy on regulation. Again, it is more a matter of confronting the regulators with information they may not have already grasped.
Second, people go around saying things like, "CBA results in the best possible policy," which tends to silence people who might disagree. People treat CBA like a science, like an uncontroversial policy tool that only a complete fool would refuse to accept. If someone challenges CBA, he is likely to be told something like, "I'm frustrated that you're opining about CBA without having a good understanding of it. There's an extensive literature in which the various objections to CBA are raised and answered." So CBA is put on a pedestal and its policy prescriptions are shielded from public debate. CBA is
assumed to yield the best possible regulations, and its proponents do not feel the need to produce a shred of evidence in its favor.
This is unjustified. CBA's inputs are inescapably ideological, and its outputs are no less ideological. Only if those outputs are better than the alternative is CBA a worthwhile endeavor.
