This is the abstract of a new paper by Cass Sunstein:
When an agency fails to engage in quantitative cost-benefit analysis, has it acted arbitrarily and hence in violation of the Administrative Procedure Act? At first glance, the question answers itself: Congress sometimes requires that form of analysis, but if it has not done so, then agencies have discretion to proceed as they see fit. But as recent decisions suggest, the underlying issues are far more complicated than they seem. The central reason is that for all its limitations, cost-benefit analysis is the best available method for testing whether regulations increase social welfare. Whenever a statute authorizes an agency to consider costs and benefits, its failure to quantify them, and to weigh them against each other, requires a non-arbitrary justification. Potential justifications include the technical difficulty of quantifying costs and benefits; the relevance of values such as equity, dignity, and fair distribution; and the existence of welfare effects that are not captured by monetized costs and benefits. These justifications will often be sufficient. But in some cases, they are not, and agencies should be found to have acted arbitrarily in failing to quantify costs and benefits and to show that the benefits justify the costs.
Here's a question related to ISDS: If a statute requires a cost-benefit analysis, but the agency does not conduct one and does not provide a justification for not doing so, is that an indicator that the regulation is "arbitrary" and thus the government's behavior is not "fair and equitable"?