The Optimal Architecture of Bright-Line Rules

Abstract

I model bright-line rules which obey the decomposability principle (Simon, 1962). The principle implies that rule followers are not asked to actively balance the rule’s criteria, but instead check off whether each of the rule’s criteria has been met. I show that adding a new criterion to a rule’s definition may weaken the correspondence between the rule and its underlying standard (under- and overinclusions), thereby serving to limit the optimal rule’s list of criteria. In the extreme, optimal rules totally ignore local information (e.g., a command), even when that information has no administrative cost and would otherwise be meaningful to experts who interpret the rule’s underlying standard. Extending the model shows that a criterion has incremental value for a class of nearly decomposable rules if and only if it has value for a decomposable rule. Collectively, these rule-driven results help to explain an organization’s limited demand for information and the need to keep things ‘simple.’