Query: Do those who rely heavily on error cost analysis also account for accuracy benefits? Should they?
Error costs are the costs of making a mistake. We generally would like to minimize them. But how do we minimize them? We might be more careful and invest more time or other resources in making a decision. Or we might make fewer decisions. The tradeoffs between these two solutions depends upon the benefits realized when a decision is made accurately. That is, positive positives and positive negatives occur as well and need to be accounted for. (Right?)
Let me try out an example from antitrust (off the top of my head, so bear with me). So shifting from per se rules to a rule of reason approach may be understood preliminarily as adopting option 1 (being more careful and investing more in gathering information before making a decision about liability). Moreover, the shift from per se rules to a rule of reason approach also entails a shift from simple rules to a more complex rule and a tradeoff between different types of error–potentially, a reduction in Type I errors but an increase in Type II errors. In antitrust law and scholarship, the belief is that Type I errors are more costly than Type II because the latter are more likely self-correcting while the former are not. So we might applaud the move from per se rules to rule of reason because the move reduces error costs. But the shift itself entails costs–the direct costs of investing more time and resources, and if those costs are too high and lead to a reduction in decisions (e.g., because of court congestion or perhaps because the increased burden is on the plainitff), opportunity costs from foregone accuracy benefits. These costs ought to be weighed against the marginal improvement in reduced error costs.
It seems to me that ignoring accuracy benefits while adopting a strategy of minimizing error costs may drive toward underenforcement. Presumably, society has adopted a rule and delegated its application to a decision maker (say, a judge) because the accuracy benefits (substantially?) exceed the error costs. Maybe this presumption is a stretch, but I don’t think so, although perhaps it depends upon the process by which the rule itself was created.
In any event, I raise this issue because it is something I’d like to learn and think more about.