In 1989, Cass Sunstein1 published an article entitled On the Costs and Benefits of Aggressive Judicial Review of Agency Action. 2 Sunstein apparently meant the words" costs" and" benefits" in an informal sense, as the article considered the advantages and disadvantages of aggressive judicial review without pretense of explicit quantification. That article was several generations ago in Sunstein scholarship, al-most 100 articles and over a dozen books. The central concerns of that article, however, are relevant to an assessment of Sunstein's latest book, whose title, The Cost-Benefit State, uses the words" costs" and" benefits" as labels for quantitative assessments of the effects of governmental actions. Sunstein, though a democratic theorist rather than an economist, enthusiastically urges agencies to make decisions based on numerical assessments of regulatory consequences, factoring in variables ranging from effects on consumer prices to lives saved. The book is not so much a primer on cost-benefit analysis as a manifesto, concluding without apology that cost-benefit analysis" is for everyone"(pp. 19-20).

It may seem ironic, then, that The Cost-Benefit State is written in much the same easygoing, engaging style as the 1989 article. Though illustrating with examples how cost-benefit analysis might prevent regulations whose costs exceed benefits, 3 and perhaps more startlingly how it might spur regulations whose benefits exceed costs, 4 Sunstein makes no attempt to quantify the costs and benefits of cost-benefit analysis itself. At one point, Sunstein muses about whether a version of cost-benefit analysis survives cost-benefit analysis (p. 121) but his analysis has less quantification than even some of the agency analyses that he criticizes in the book. 5" The answer is that we cannot be sure," he concedes." But the current situation is not nearly as good as it could be, and if the analysis is done well, there is every reason to expect that it will lead to improvements"(p. 121). My intent, however, is not to criticize Sunstein for writing the book in a conventional form, without mapping his arguments into a series of calculations. After all, even a study assessing cost-benefit analysis quantitatively might not survive cost-benefit analysis (and so on to in-