gailmard [at] stanford [dot] edu
Regulation, Evaluation, and Governance Lab
Stanford Law School
559 Nathan Abbott Way
Stanford, CA 94305
I am a postdoctoral scholar at the Regulation, Evaluation, and Governance Lab (RegLab) at Stanford University.
I study political management of the bureaucracy. My dissertation explores political appointees’ dual roles as agents of the president and managers of the bureaucracy. This view of appointee-careerist relations complicates standard notions of presidential control and bureaucratic power, by recognizing that appointees are reliant on presidential support to maintain their position within an administration. I argue that appointees may undermine presidential control of the bureaucracy to cultivate a good reputation with the president—either by failing to involve expert careerists in policymaking or adopting policies that reflect careerist views.
Prior to joining RegLab, I completed a Ph.D. in Social Science at California Institute of Technology, and received a B.A. in Economics and M.A. in Political Science from UC Berkeley and an M.P.P. from the Harris School of Public Policy Studies at University of Chicago.
Working Papers
The Politics of Presidential Removals
Revise and Resubmit, JLEO
Much scholarship and legal reasoning assumes that the ability to remove and replace political appointees furthers presidential control. However, I argue that the possibility of removal changes the agency problem between the president and her appointees in a subtle but important way that affects the president’s appointment problem. I demonstrate that the president may have incentives to make non-ally appointments in order to encourage reliance on bureaucratic expertise. To show this, I develop a formal model that introduces career concerns for appointees that lead them to distort their use of bureaucratic expertise to appear more expert. The president is uncertain of an appointee’s expertise, but infers it from the appointee’s involvement in policymaking. In equilibrium, non-expert appointees more aligned with the president face greater incentives to determine policy themselves to improve their reputation. By selecting non-ally appointees, the president commits to sometimes dismiss even experts which improves her control over policymaking.
Presidents rely on their political appointees to manage the bureaucracy on their behalf. Appointees often know more about their organizations than the president and, therefore, may be better positioned to generate bureaucratic support for the president’s agenda. Yet, bureaucratic cooperation may be easier for appointees to sustain the more policy reflects the views of careerists tasked with implementation. I consider a model in which an appointee dictates a policy that a bureaucrat exerts effort to implement. The president is uncertain of both her appointee’s management skill and the difficulty of the management problem her appointee faces. Instead, the president must infer the appointee’s skill by observing his policy choice and whether implementation was successful. In equilibrium, both talented and weak appointees may give additional policy concessions to bureaucrats to ensure bureaucratic cooperation and improve their reputation with the president. This incentive exists even when the appointee shares the president’s policy preferences. The results highlight fundamental strategic limitations of administrative tools to preserve presidential control over the bureaucracy.
An Experimental Study of Delegation, with Marina Agranov and Alexander Hirsch
The allocation of formal decision-making authority in organizations has a powerful effect on political and economic outcomes. We examine how individuals delegate decision-making authority to a better informed agent in an experimental setting, testing the key theoretical predictions of the canonical Holmström (1984) delegation model in the lab. While this model has been widely applied to study decision-making within firms and bureaucratic organizations, previous experimental investigations provide only limited insight into its applicability as a model of individual behavior. We develop an experimental interface that more closely approximates the information and choice environment in the model. This innovation allows a more faithful implementation of the model in the decision environment facing the subjects and, therefore, a more complete test of the model.