Bureaucracy’s Masters and Minions. Eleanor L. Schiff

of individuals and their environments that are most worthy of investigation and it integrates them into a logically coherent whole” (p. 757). The theory enables scholars to observe a series of interactions between a single principal and a single agent, and to understand the different incentives, agency loss, and moral hazard inherent in particular situations. To the extent that these observations are generalizable to other, similar situations within the bureaucracy is part of the power of the model. Yet, Moe also argued that the model’s parsimony was also one of its weaknesses with the empirical reality of multiple, competing principals, different behavioral expectations for workers, and different organizational contexts. He cautioned that application of the simple principal-agent framework “may severely distort rather than clarify” (p. 773). In short, the principal-agent model is valuable, but must be used cautiously. It is critical to recognize the inherent complexity with multiple principals and also the diversity of tasks across organizations. Competing principals’ political control is an important factor in understanding which one is more influential over bureaucratic policy.

      Accepting Moe’s proposition that bureaucratic control rests with multiple principals, the current theoretical framework focuses entirely on the principals and completely ignores the bureaucracy. He (Moe 1987) comments that the congressional dominance thesis is

      almost entirely Congress centered. It offers strident arguments that Congress controls the bureaucracy, but it develops nothing that can be called a logic of control and pays precious little attention to the bureaucracy. It builds on theories of legislative and electoral institutions without putting them to effective use in understanding bureaucratic behavior. (476)

      I argue, as an extension of Moe, that previous explanations within the principal–agent framework, including the congressional dominance and the presidential sovereignty theories, are limited in extending scholarly understanding to the bureaucracy; rather, they help clarify how different institutional principals exert control. Moreover, the extant literature offers competing and often contradictory theories about which principal exerts more, or less, control over the bureaucracy—but offers no overall framework to disentangle when one principal might have more or less influence over bureaucratic policy-making, or when the bureaucracy might assume control over itself. In effect, current theories are all principal oriented in the principal-agent framework, and do not provide evidence on how the agent, the bureaucracy, might respond given competing principals’ agendas, or when political control is largely absent. Scholars using this framework may conclude, based on empirical observation, that political control exists but fail to consider the alternative explanation that the outcomes observed may be a result of the characteristics internal to the bureaucracy itself (Meier and O’Toole 2006; Kennedy 2014; Selin 2015). To this end, the principal-agent framework is underspecified and therefore can lead to erroneous inferences.

      In this study, I will take a different approach and look at bureaucratic politics from the “agent” perspective. I will work to peek under the “black box” of bureaucratic policy-making. In effect, rather than a principal-agent framework, I will utilize an agent-principal model to understand when the preferences of different principals are more or less important to the bureaucracy. In this framework, using the agent as the starting point for analysis requires bottom-up investigation within the bureaucracy to understand how internal characteristics can explain variation in top-down control from different principals. This is the overall goal of this study in taking a bottom-up view of the bureaucracy: recognizing the diversity across the bureaucracy and examining variation in the internal characteristics of the bureaus themselves.

      This study will attempt to answer: Why and when do political actors exert different levels of political control over the bureaucracy? Under what conditions do different principals have their preferences reflected in bureaucratic policy? The answers to these questions have implications for the degree of democracy and democratic control the public can expect to exert on the bureaucracy. In analyzing this relationship, the extant literature has often relied on a case-study approach (Moe 1987; Weingast and Moran 1983; Ringquist et al. 2003; Balla 1998; Carpenter 2001) in order to make broad generalization about bureaucratic behavior without recognizing the enormous heterogeneity across the federal bureaucracy in terms of workforce structure, composition, and responsibility. Without attentiveness to the inherent variation across the bureaucracy, generalizing from a handful of case studies cannot lead to a broader understanding of how the bureaucracy works, or a positive theory of bureaucratic control (Lakatos 1968).

      Following a discussion in the next chapter about the shared power of budgeting between Congress and the president and the nature of budgeting, reworking the principal-agent theory to accommodate a broader relationship between principals and agents is the motivation behind the agent-principal model presented in the chapter 4.

Chapter 3

       The Business of Agency Budgeting

       Incrementalism

      Budgeting as a Shared Power