Henry E. Smith

Harvard University

Laws can be written along a spectrum of specificity, ranging from vague standards to more detailed rules with particular examples. Behavioral and legal scholarship each present conflicting views about the optimal degree of specificity with which laws should be designed. From a behavioral standpoint, specificity is important to help people understand their goals and use their cognitive resources in a focused manner. At the same time, ambiguity in the law can even encourage good people to engage in creative interpretations of legal requirements, allowing them to justify unethical behavior, with limited awareness of the meaning of that behavior. By contrast, theories of crowding out, trust, and cooperation suggest that specificity can create resentment and lead to under-compliance and under-performance. These conflicting views about the effects of specificity serve as the background for this experimental project. This paper studies the effects of specificity on behavior in response to a directive that shares important features with the law. First, we examine the effect of specificity on compliance (following a directive) versus performance (beyond a minimum threshold). Second, we compare the controlling, limiting effects of specificity with its instructive, informative effects by comparing the interaction between specificity and monitoring with the interaction between specificity and good faith. We hypothesized that the combination of specificity and monitoring enhances the effect of specificity on compliance but harms performance and trust, whereas the combination of specificity and good faith enhances both the informative goal-setting aspects of specificity and people’s sense of commitment. The study employs an experimental design in which subjects edit a document after being exposed to detailed (vague) instructions, with (without) a reference to good faith, and with (without) monitoring (through sanctioning). The assignments were designed in such a way that people could engage in various levels of editing (both required and not required, reasonable and more than reasonable), allowing us to measure distinctly both compliance and performance. Our results suggest that when participants require information and guidance, as in the case of editing a document, specificity increases performance even beyond what is required relative to a vague standard condition.