Presidents and the Politics of Agency Design: Political Insulation in the United States Government Bureaucracy, 1946–1997.

By David E. Lewis. Stanford: Stanford University Press, 2003. 224p. $45.00.

In this marvelous book, David Lewis addresses an often neglected but nonetheless important issue: the design of the structure of government agencies. Using the framework of the “new economics of organization,” Lewis argues that many aspects of agency structure—specifically the degree to which bureaucracies are “insulated” from the direct control of the chief executive—can be explained as a function of the interaction and rivalry between the executive and legislative branches. The argument made here is reminiscent of arguments made by Lewis's mentor, Terry Moe, but is a real improvement upon much of Moe's work in that he makes very clear-cut, testable propositions regarding the implications of the conflict between the White House and Congress, and subjects those propositions to empirical examination, using both large data sets of agency creation and termination as well as case studies of small numbers of agencies.

Lewis assumes from the outset that chief executives and legislators have somewhat different interests and preferences regarding the design of public agencies. Presidents are assumed to desire direct, uninsulated control over bureaucracies, while Congress is assumed to be less supportive of presidential management of agencies and more committed to a design that would insulate agency management from central control. Agencies may be insulated by having no layers of bureaucracy over them, by having a multimember governing board or commission, often serving fixed terms, and by requiring specific qualifications for those holding leadership positions within the agencies. Congress is assumed to be particularly likely to oppose presidential control when the legislative branch is controlled by majorities of a party other than that of the president. Lewis hypothesizes that as the size of the majority in opposition to the president's party goes up, the likelihood of insulation in the design of new statutorily created agencies goes up. Yet when there is unified party control, the size of the majority in Congress lowers the probability of insulation. Lewis also argues that as presidential approval for a president goes up under divided government, the likelihood of insulation for newly created agencies also goes up, as Congress tries to weaken a strong president. Agencies created when there is an unpopular president and divided government are less likely to be insulated.

Presidents have responded to congressional refusals to approve agencies in the uninsulated form by creating new agencies by executive orders, departmental orders, or reorganization plans. In turn, Congress may withhold funds from a newly created or reorganized agency, yet this is a dangerous and awkward weapon to use, particularly when the agency or president is relatively popular. In extreme cases, Congress may terminate an agency that displeases them. That was the fate of the National Biological Service created by President Clinton after Congress failed to create an agency design to the president's liking. Lewis's statistical analysis reveals that agencies created by statute have much less chance of termination than do agencies created by executive action during the time period (1947–97) that is the focus of his research.

Lewis does an excellent job of carefully analyzing the interactive effects of divided government, partisan majorities, and presidential approval. Less astute scholars searching for simple and direct relationships would have missed the interactive impact of these variables. Unfortunately, these effects are difficult to explain to undergraduates, so the book may be a bit too advanced for some classes, yet it would be an ideal supplemental text for graduate-level courses dealing with bureaucratic politics, the presidency, and American political institutions.

By Lewis's own admission, he has failed to devote much attention to several issues that could affect the general phenomenon of agency insulation. The specificity of enabling statutes, the details of administrative procedures, particular budgetary techniques, and the role that different kinds of agency functions and interest-group configurations play are not seriously examined in this book. Yet Lewis suggests that later work can and will incorporate these factors into the analysis.

Although Presidents and the Politics of Agency Design is a very good book, a few minor, factual errors pop up in the text. For example, Congressman Bill Emerson of Missouri is identified as representing Mississippi (p. 94). Senator Bob Graham of Florida is inexplicably identified as a Republican (p. 78). These are minor flaws that should have been caught prior to publication. A more important issue is the selection of time periods for the data analysis. Much of the data for the statistical analysis are derived from the U.S. Government Manual, beginning in 1947. Yet many of the most important insulated agencies, the independent regulatory commissions, were created long before that time. Many of these commissions were created during times of united partisan control of the executive and legislative branches. The institutional rivalry argument might need some adjustment to account for the design of those agencies. Lewis also seems to assume that if an agency is originally designed to be insulated, it will almost certainly remain insulated. That may be true, but some skeptical readers would have preferred that such a conclusion be demonstrated rather than simply assumed.

The author clearly has broad interests in American national government, but some readers may recognize that Lewis's argument may fruitfully be applied to the design of state agencies. By examining the effects of divided government and majority power in state legislatures, scholars could subject Lewis's hypotheses to critical tests using far more agencies and more years than in his study of federal agencies. In any case, Lewis's research is likely to motivate other scholars to delve into the origins of agency structure for many years to come.