Our theoretical argument examines how four factors combine to influence the legislative design of bureaucratic discretion. We explored the role of two of these factors – policy conflict between the legislature and agency and conflict across legislative chambers – in the previous chapter. The issue of policy conflict has been central in studies of Congress and delegation, and our case study of Michigan confirms the importance of divided government. The second issue, bicameral bargaining, has received much less systematic attention, as have the other elements of our theory that could not be explored in the Michigan case study, such as the legislature's capacity to write detailed, constraining laws and the role of nonstatutory policymaking mechanisms.

In this chapter we conduct a systematic empirical test of the main arguments of our theory using a novel data set that looks at all bills affecting Medicaid recipients that were enacted by U.S. states over a two-year period. The data permit a test of the comparative elements of our arguments, as the U.S. states have considerable variation in the theoretical variables we have identified. In addition to using these data to test our theoretical argument, we test alternative hypotheses by operationalizing a number of other factors, including the degree of electoral competition, changes in the partisan control of government, and partisan preferences over the level of governmental activism, that might influence the level of discretion that legislatures choose to allow.