When the term "administrative presidency" was first coined by Richard Nathan (1983), it was focused largely on presidents selecting political appointees who shared their policy aims and wished to alter organizational structures, personnel systems, budgets, and decision rules in their agencies to accord with those aims. Nathan argued that it was a low-visibility strategy offering what recent scholarship calls significant first-mover advantages to presidents (see also Howell 2003; Kagan 2001; Moe 1989, 1993; Moe and Howell 1999). A Congress challenged by collective-action problems could not react to most administrative initiatives because of their sheer number, arcane nature, information asymmetries favoring presidents, and the opaqueness of bureaucratic operations.
Since then, empirical research relevant to the administrative presidency has offered mixed support for this perspective. Employing qualitative, statistically sophisticated, formal modeling, or choice theoretic analyses, some scholars have argued that agency outputs tend to respond to changes in presidential administrations, the comings and goings of political appointees, and other presidential shocks to the American political system (e.g., Wood 1988; Wood and Waterman 1994). In contrast, others have found agencies more responsive, alternatively, to congressional direction (e.g., Chubb 1985; McCubbins 1985; Scholz and Wei 1986; Weingast 1984); to multiple principals rather than just to presidents (e.g., Scholz and Wood 1998; Wood and Waterman 1994); to interest groups, subnational actors, and local contexts (e.g., Epstein and O'Halloran 1994; Scholz, Twombly, and Headrick 1991; Yackee 2006); and to various actors, depending on policy domain (Zegart 1999).
Other researchers have found bureaucratic responsiveness to aspects of the administrative presidency better characterized as a function of contingencies (e.g., Aberbach and Rockman 2000; Durant 1992; Golden 2000; Hammond and Knott 1996; Lupia and McCubbins 1994; Maranto 1993; Michaels 1997; Waterman 1989). These include the extent of change involved for bureaucrats, the extent to which presidential goals are clear, the perceived impact of a delay in complying with presidential missives, the type of agency involved, the culture of the agency, styles of administrative leadership, agency reward systems, and the availability of external employment. Others focusing largely on such unilateral tools of the administrative presidency as executive orders, presidential signing statements, presidential proclamations, and national security directives argue for the potency of these techniques in advancing presidential policy goals (Howell 2003; Mayer 2001; Warber 2006). Many among them worry about inordinate--and even unconstitutional shifts in--power to presidents (Cooper 2002; Fisher 2004; Ginsberg 2007; Pious 2007; Rozell 1994; Rudalevige 2002). Still others see invocation of the unitary executive theory as a relative weakness rather than a strength of presidents (Roberts 2008).
This article argues that despite the significant insights offered by prior research, assessments of the efficacy of the administrative presidency to date are both premature and in need of refocusing to comport with the realities of the American political system. In regard to prematurity, assessments must be withheld until four interrelated limitations of prior research are addressed. First, unlike early qualitative research on the topic, most recent research related to the administrative presidency has used neoclassical economics-based (e.g., principal-agent models) or statistically sophisticated research designs that marginalize the implementation processes accompanying these efforts to advance presidential agendas. Yet, like legislation, administrative initiatives are only the "starting gun" for battle, with interests and stakes for presidents that shift over time (West 1995). Moreover, this reality has not stopped many from praising or decrying the use of, say, executive orders and signing statements, as if issuing them were enough to ensure implementation. Relatedly, scholars have typically examined individual tools of the administrative presidency rather than the interaction of these tools over time as the implementation of presidential goals and priorities shifts to fit changing circumstances. Nor have they adequately built into their designs the reality that policy success requires parts of different national and subnational programs and agencies to work together (sometimes across sectors), making the alignment of various tools essential lest they undercut each other (Durant and Warber 2001; Hjern and Porter 1981).
Second, assessments of the administrative presidency typically are made in ways that suggest four implicit and unrealistic assumptions: (1) presidents and their political appointees think strategically about how best to coordinate the various tools of the administrative presidency across programs and agencies to successfully advance an incumbent's policy agenda; (2) various presidential policy aims pursued administratively are themselves internally coherent and do not work against each other when wielded by appointees; (3) bureaucratic principals and agents do not have policy aims of their own and do not try to turn presidential administrative strategies into tools for advancing their own agendas in targeted or related policy arenas; and (4) bureaucrats affected by administrative initiatives will not quickly alert elected officials and interest groups opposed to presidential goals to these administrative strategies, thus reducing, if not eliminating, the information asymmetry purportedly allowing presidents to frame the policy landscape in their favor.
Third, typical criteria for assessing the administrative presidency fail to incorporate the realities of presidential decision making. Contrary to the rational-choice-driven picture implied and modeled in recent research employing principal-agent theory, presidents' preferences often are not complete, fixed over time, or transitive. Presidents often know less about what they want to do than about what they do not want to do as they take real or symbolic action (e.g., to not run afoul of an important constituency or be embarrassed). Nor do their appointees as they issue unclear policy guidance to bureaucratic subordinates (Aberbach and Rockman 2000). Moreover, even when their preferences are fixed, presidents have multiple preferences that conflict during implementation and must be balanced or traded off. In the process, they often "learn what to prefer" during implementation over time as circumstances shift. In addition, even if adroitly crafted, political, budgetary, and personnel cycles in government mean that resources sufficient to realize presidential goals pursued administratively may be absent. Thus, presidents may initially acquire first-mover advantages, but these may not be compatible, well thought-out, or enduring.
Finally, most recent research fails to employ methodologies that are capable of incorporating these realities or evaluative criteria incorporating them. The common practice of assessing administrative strategies cross-sectionally, one administrative tool and policy goal at a time, and without focusing on implementation processes related to the interaction of tools and multiple presidential goals, violates the realities of presidential leadership in our Madisonian system. Therein, presidential aspirations must be recalibrated to cope with the separation of powers, checks and balances, increasingly overlapping policy domains, and limits on coherent strategic presidential initiatives within and across agencies. Thus, sophisticated statistical analyses may be sufficient to assess bureaucratic responsiveness (but see Meier and O'Toole 2006), but they are insufficient for assessing effectiveness. Nor do they explain the causal mechanisms involved, how much more effective administrative strategies might have been if used more deftly, and how much better any administration could have done under the circumstances.
Most suitable for informing such assessments, this paper argues, are longitudinal case analyses of the subsequent implementation of the tools of the administrative presidency, of their interaction amid shifting and multiple presidential priorities, and of the strategic and tactical savvy used by presidents as they learn what to prefer in shifting contexts. As H. Peyton Young writes, "neoclassical economics describes the way the world looks once the dust has settled; we are interested in how it goes about settling" (1998, 4). So, too, do cross-sectional statistical analyses, no matter how creative, sophisticated, and otherwise informative they might be. As such, research on the administrative presidency needs to return to its early implementation roots at this point to complement the insights made by researchers using more statistically based and economics-grounded principal-agent research designs that have dominated scholarship more recently.
To illustrate why future research needs to retake this path for our understanding of the administrative presidency to improve appreciably in the future, the paper takes as its analytical focus an epic struggle since the late 1940s--and most especially during the Bill Clinton administration--to use the administrative presidency to get the U.S. military to incorporate environmental and natural resources (ENR) values into its day-to-day operations. Chronicled is how and why various tools of the administrative presidency-reorganization, political appointments, decision rules, budgets, personnel structures, unitary tools, and government-wide administrative reforms--interacted over time with national security goals and changing presidential priorities to sorely compromise the president's greening goals for the military.
The paper begins by reviewing, first, how presidents tried to use various tools...