Students of bureaucracy applying the logic of rational choice institutionalism (RCI) have offered important and thought-provoking explanations of how and why public agencies are designed and evolve over time in the United States (e.g., Bender & Moe, 1985; Brehm & Gates, 1997; Epstein & O'Halloran, 1999; Gilligan, Marshall, & Weingast, 1989; Huber & McCarty, 2004; Huber & Shipan, 2002; Lewis 2003; McCubbins & Page, 1987; Moe, 1989; Rothenberg, 1994; Weingast & Moran, 1983; Zegart, 1999). The insights and contributions of this scholarship notwithstanding, however, celebrating the explanatory power of RCI as a general theory of agency design (i.e., factors affecting agency origin and design) or evolution (i.e., forces affecting changes in agency design over time) is premature for several reasons.
First, the universe of U.S. agencies studied is quite narrow. With few exceptions, RCI scholarship on bureaucracies focuses on domestic agencies. Second, even when RCI scholarship in the U.S. context has been expanded to include the design and evolution of national security agencies, findings call into question the general application of RCI propositions beyond domestic agencies. Zegart's (1999) pathbreaking application of RCI theory to national security agencies, for example, finds that the key actors in agency design and evolution of the Central Intelligence Agency (CIA), the Joint Chiefs, and the National Security Council are presidents and bureaucrats, not the interest groups and congressmembers frequently identified as key actors in the domestic policy domain.
A third and related issue questioning the explanatory power of RCI is the reluctance of theorists to afford any meaningful strategic role for bureaucrats, a finding disputed not only by Zegart's (1999) research, but also by recent and important research on domestic agencies generally (Krause, 1999; Krause & Meier, 2005). In particular, these studies find that bureaucrats act strategically to influence agency design and evolution; they are not passive bystanders who are merely "acted upon" as most conventional principal-agent models assume. As such, the relationship between elected principals and their bureaucratic agents should be seen as a "two-way" street rather than a "one-way" street favoring principals over bureaucratic agents (Krause; also see Krause & Meier).
Fourth, questions about the general explanatory power of RCI also arise because it typically takes a top-down, Washington-centric focus on U.S. domestic or national security agencies alone. As such, and unlike the use of RCI by comparative politics scholars (see, for example, Crisp & Levine, 1998), it is possible that RCI perspectives on agency design or evolution may marginalize the substantial and growing influence of subnational actors in domestic and national security domains (Agranoff & McGuire, 2003). Again, no logical reason exists to expect that subnational actors in the U.S. also will not try to ensure their access, influence, and power in agency structures, processes, and procedures.
Finally, even were one to accept these recent modifications to RCI theories of agency design or evolution in both domestic and national security agencies, one would have to assume that agencies are either "pure" domestic or "pure" nondomestic agencies. Yet this ignores bureaucratic realities in the contemporary administrative state. A significant and growing subpopulation of agencies work in policy domains where domestic and nondomestic public policies interact (e.g., the Environmental Protection Agency [EPA], the Federal Bureau of Investigation, the Pentagon, and the CIA). Moreover, the existence and import of "intermestic" policies has long been recognized (Shull, 1991), with this overlap becoming more prevalent in the post-9/11 era (Baker, 2005). Because affected agencies are engaged in these kinds of "hybrid" policy domains, both agency design and evolution may be equally hybrid; both domestic and nondomestic actors will be interested in shaping agency structures, processes, and procedures to their advantage.
This study seeks to advance theory building on agency evolution in empirically grounded ways by focusing on the explanatory power of an important theoretical perspective (RCI), in an understudied hybrid policy domain where domestic (environmental protection) and nondomestic (national security) policies intersect, and in an organizational type (the U.S. military) that has drawn scant attention from students of bureaucracy in political science, public administration, or public management. More precisely, the analytical focus of the study is the still-ongoing effort to get the U.S. military to integrate environmental and military readiness values into its day-to-day operations. Central to this organizational change effort has been the alteration of the structures, processes, and procedures that for a half-century were designed by the enacting coalition in Congress to buffer the services from contextual goals (Wilson, 1989), like environmental and natural resources (ENR) protection, that the military saw threatening its ability to complete the services' primary warfighting mission.
To these ends, the study assesses the explanatory power of RCI in three major initiatives to advance or protect past progress in "greening" the military in the post-Cold War era. These initiatives include: (i) holding the military accountable for natural resource management (NRM) planning during the Clinton years; (ii) moving away from the military's preferred incineration strategy when demilitarizing chemical weapons (CW) during the Clinton and Bush administrations; and (iii) preventing the military from obtaining broad exemptions from ENR laws after the War on Terror began during the Bush administration. Analysis of these cases reveals that the explanatory power of key elements of RCI in pure domestic or nondomestic policy domains is less robust in explaining agency evolution in agencies involved in hybrid policy domains.
In each case, the study treats the military services as the career bureaucracy. As Barzelay and Campbell (2004) demonstrate in their work, the Pentagon is "an organization dominated by career officials" (p. 129). Moreover, this was especially true during the Clinton years. Former Pentagon officials, press observers, academics, and interviewees note that the Clinton administration's rough beginnings with the services over Somalia, Haiti, and gays in the military led the White House to craft a modus vivendi with the military: don't push us too far and we won't push you too far (Priest, 2003). As such, Pentagon positions were, minimally, informed by those held by military leaders in the services and, maximally, driven by them.
The data informing the analysis are culled from extensive archival research of documents afforded by the agencies and interest groups involved, as well as a computerized and systematic review of articles appearing between 1993 and 2005 in the leading industry newsletter on military affairs and the environment, Defense Environment Alert (DEA). Published biweekly by Inside Washington Publishers as a newsletter for individual, group, and institutional subscribers in government, business, industry, and academia, the DEA's reputation is unparalleled for its neutrality, accuracy, and commentary. These data are supplemented by off-the-record background interviews conducted between 1993 and the present with over one hundred individuals either working in, or authorities on, the greening of the U.S. military.
The study begins by elaborating in greater depth the logic, tenets, and potential constraints on RCI explanatory power in domestic, nondomestic, and hybrid policy domains. From that discussion, the article derives and tests propositions related to the explanatory power of RCI in the NRM planning, CW demilitarization, and regulatory relief cases. The article concludes by discussing the implications of the findings for the explanatory and predictive power of RCI when it comes to agency design and evolution in hybrid policy domains.
RCI offers parsimonious and provocative deductive theories of politics generally, and of agency design and evolution in particular. As a result, it offers students of the bureaucracy testable "ideas about which actors most influence agency design and development, how this influence works, and what this means for public administration" (Zegart, 1999, p. 19). As Wood and Bohte (2004) state, the central premises of RCI regarding agency design are that enacting coalitions use structures, procedures, and processes to increase the political and administrative transaction costs for policy opponents seeking to change an agency's direction in the future (Wood & Bohte).
These past "democratic coalitions" are uncertain about what the future may bring, and fear that "coalitional drift," "bureaucratic drift," or unsympathetic presidents will try to change their original intent (Wood & Waterman, 1993). Consequently, they try to impose administrative structures, processes, and procedures when creating agencies that they hope will make significant changes in organizational behavior quite difficult (Wood & Bohte, 2004). To do this, proponents of creating a given agency have a variety of tools at hand. They can set the agency on autopilot (i.e., by granting little discretion to agencies and by reducing their own monitoring costs); increase transactions costs for those seeking to intervene judicially (e.g., by limiting standing to sue and when suits can be brought); and stack the deck (e.g., by creating advisory councils to ensure that citizen groups in the enacting coalition have access to agency decision making).
But as Moe (2005) argues, "the fact is that the enacting coalition consists of only one faction of legislators and interest groups. Other factions are losers and may well be worse off because of the coalition's choices (p. 220)." Importantly, these losing factions must be contended with by the enacting coalition because of the need to bargain and compromise throughout the legislative process. Indeed, before any enacting coalition "wins," it is likely to have to make agency design compromises and concessions that allow opponents room to enhance their future power, access, and influence in program operations and decision making. Institutionalized in these designs as a consequence is …