Over at my day job, I’m in the midst of writing up a rather lengthy paper on cost-benefit analysis. Rereading some of the literature, I was struck by the following claim made by Cass Sunstein, in one of the many interesting things he’s written since retiring from government work and returning to academia (“The Office of Regulatory Affairs: Myths and Realities”):

In the first three years of the Obama Administration, the net benefits of economically significant regulation exceeded $91 billion – more than twenty-five times the corresponding figure for the Bush Administration, and more than six times the corresponding figure for the Clinton Administration.

We are so used to hearing about the costs imposed by regulation that it is easy to lose sight of the fact that, when properly crafted, regulation is a source of enormous benefit to society. And yet it’s not every day that you see government (or in this case, former government officials) standing up and taking credit, unapologetically, for those gains.

It also brought to mind a somewhat more esoteric point, about the value of doing comprehensive cost-benefit analysis on government policy. There is a very influential paper, by Peter Diamond and Jerry Hausman (“Contingent Valuation: Is Some Number Better than No Number”), which criticizes CBA by appealing to a version of the “garbage in garbage out” principle. Supporters like Sunstein spend a lot of time defending CBA on the grounds that it eliminates a variety of cognitive biases that otherwise impair government decision-making. Yet Sunstein seems to ignore the fact that no matter how transparent the procedure may be, if the data that it takes an input – in this case, the “willingness to pay/accept” valuations – is compromised by those same cognitive biases, then the output is going to be just as biased. Diamond and Hausman make a big deal out of this, showing that the valuations on which CBA is based can easily be worse than random.

Fans of CBA often respond to this sort of criticism by pleading modesty, 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, etc. This response is disingenuous, because it fails to acknowledge that numbers themselves can have a biasing impact, leading to quantitative measures having an outsized role in deliberations, even when this is unwarranted by the quality of the analysis. Thus, as Diamond and Hausman suggest, “some number,” if it is not a good number, may be much worse than “no number.”

A better response to the Diamond and Hausman argument lies in the observation that in almost every case in which CBA is used, there will always be at least one number available, namely, the number that reflects the cost. This will be true, for example, whenever a regulation is imposed, or whenever resources are directed to the provision of a public good. Furthermore, industry and taxpayer lobby groups can be expected to make a very big deal out of that number, making sure that it is in all the newspapers. By contrast, it is almost always the benefit that is intangible, because it lacks a market valuation (which is why the state is considering getting involved in the first place). 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, if for no other reason than to counteract the effects of the first.

The fact that I did a double-take when reading Sunstein’s sentence, quoted above, is a reflection of the biasing effect that the constant emphasis on cost can produce. It can even lead a person like me (who is basically sympathetic to CBA, in a faute de mieux sort of way) to forget that regulation actually makes us a wealthier, happier society.

With this in mind, I would like to propose a new slogan, for supporters of CBA: “Cost-benefit Analysis: Two Numbers are Better than One Number.”