Removal as a political question.

• Author: Huq, Aziz Z.
• Position: II. Removal Authority as a Means of Bureaucratic Control C. Theorizing Removal's Limits through Conclusion, with footnotes, p. 39-76

3. Theorizing Removal's Limits

   Free Enterprise Fund's equation of removal and control suffers a further deficiency. As a method of exercising control over an agent, removal is simply not all it is cracked up to be. Removal may not only be unnecessary given the extant instruments of agency control wielded by a supervising official; it may also be ineffectual because it is too costly, too clumsy, and too molar a tool for attaining desired policy results. As a result of these limitations, even a supervising official who has no other instruments of agency control will not necessarily find her ability to elicit desirable policy outcomes increased in any meaningful way by a judicial intervention reallocating removal power. Accordingly, for a court to treat removal as a unique Archimedean lever that can move the bureaucratic world would be quixotic. Again, where removal authority is a nugatory addition to the presidential arsenal because of its costs, judicial action in the vein of Free Enterprise Fund may have the perverse effect of creating a semblance of presidential control where little exists, thereby hindering, rather than advancing, democratic accountability. This Subpart explains the theoretical basis for this counterintuitive claim. The following Subpart supplies empirical support for it.

   Developing the limitations on removal power, I draw again on the economic literature on agency costs. Specifically, that literature identifies information asymmetries and transaction costs as reasons to believe that removal will be a systematically less attractive control device in comparison to other generally available tools such as the appointment power. (188)

   1. Information asymmetries

      The principal-agent literature identifies informational asymmetries as an important constraint on principals' ability to police agency slack ex post. Principals typically face two asymmetries that make removal an unreliable crutch: hidden information and hidden action problems. (189) Hidden information problems arise when an agent knows more about the exogenous conditions that affect output independent of effort than the principal. (190) Hidden action problems arise when (1) agents' effort cannot be directly observed and (2) outcomes are imperfectly correlated with effort. (191) Under either set of conditions, a principal's use of removal alone will be suboptimal because the principal will not be able to discern accurately the class of cases in which the agent should be ousted for failing to follow instructions as opposed to cases in which the agent has fallen short for exogenous reasons out of his control.

      Both kinds of epistemic asymmetries occur frequently in public bureaucracies. Agencies often possess more information about external conditions that bear on the optimal selection and performance of policy instruments. Presidents are often unable to ascertain independently whether a given policy failure is caused by agency slack or by an external constraint. Further, agency officials frequently possess subject-specific skills and knowledge that the White House lacks. (192) Given either kind of informational asymmetry, Presidents' exercise of the removal power will be imprecise. Vigorous use of removal risks being overbroad, while its parsimonious employment will be underinclusive. (193) More generally, it will not always be the case that a bureaucrat's task will lend itself to the formulation of "detailed instructions," (194) such that a White House principal can ascertain compliance after the fact.

      Attention to information asymmetries again underscores the virtues of alternatives to removal. In particular, it is plausible that the appointment power will in fact often have lower epistemic costs. As employers can identify desirable candidates by their qualifications and achievements in the private job market context, so the White House can screen potential appointees to mitigate the need for later supervision and discipline. Partisan cues, past employment, and formal qualifications all provide information about preference alignment at the appointments stage. Such information may be easier to obtain and interpret than the noisy signals about agency performance upon which removal decisions rest. Of course, Presidents' appointment power is tempered by the Senate's confirmation role. But empirical evidence suggests that the White House still wields considerable influence. (195)

      In sum, informational asymmetries between a President and an agency impose a cap on the value of removal authority. Whether or not the power to select in a given case is available, there will be a subset of cases in which a President's inability to observe directly an official's actions imposes a constraint on her ability to use (or even credibly threaten) removal in a way that provokes desirable actions. In those cases, it is not clear that removal does much work.

   2. Transaction costs

      The second cluster of reasons to think the appointment power will systematically dominate the removal power as a tool of presidential control turns on transaction costs. These frictional costs come in several forms: political costs, costs to agency performance, epistemic costs, and harmful dynamic effects.

      First and most importantly, the political costs of removal may negate its utility from the White House's perspective. Recent accounts of the presidency have emphasized that it is not so much legal constraints, but the need to maintain favorable "public opinion," that curbs executive discretion. (196) "Without credibility," it is claimed, "the president is a helpless giant." (197) Although some formulations of this claim may be somewhat overstated, (198) it is surely the case that Presidents are highly sensitive to the perception of their actions in the electorate, and to what might be termed the political costs of a given action. Indeed, it is hardly implausible to think that Presidents will frequently be more sensitive to political costs than to policy outcomes.

      Removal often has large political costs, and these may render it an ineffectual supplement to the President's arsenal. (199) Removal is a high-profile means of influencing policy outcomes in comparison to tools such as presidential administration, reorganization, and litigation control. Its use may draw public attention to the fact that a chief executive is attempting to control policy against the wishes of expert agency leadership. Hence, it creates political costs for the President. (200) Consequent transaction costs may be so great in some cases that it is the agent, not the principal, who exercises larger de facto control. By way of illustration, consider the threat in March 2004 by a group of senior Justice Department officials to resign unless changes were made to then-ongoing electronic surveillance programs. (201) President George W. Bush's decision to back down on a program he apparently believed to be central to national security can plausibly be ascribed to the political costs of being seen to have constructively dismissed senior officials with area-specific expertise. The political costs of de facto removal provided subordinates with a lever to prevail against White House influence. This suggests that a subset of cases exists in which transaction costs effectively insulate an official from presidential control by removal. To the extent that appointees are aware that political costs sometimes preclude effective presidential action via removal, a judicial decision awarding the removal power to the White House is ineffectual in terms of the Free Enterprise Fund Court's putative goal of democratic accountability.

      Attention to political costs might suggest that the judicial allocation of removal to the White House might not always improve the President's position, but it also does not show that the addition of removal power makes him worse off. On this view, judicial addition of removal authority to the presidential toolkit is warranted on the ground that, at least in some set of cases, it will be useful in promoting White House control and thereby democratic accountability. At a minimum, the threat of removal (even without its execution) will have a salutary effect on the distribution of policy control within the executive.

      But this moves too quickly. Removal, even if not employed, can have a second set of transaction costs that undermine a President's ability to secure desirable policy actions. The Free Enterprise Fund rule may thus not merely be nugatory, but sometimes positively harmful by fostering a false impression of White House suzerainty that misleads voters.

      Recent empirical work in social psychology finds that "authorities and institutions that exercised authority fairly and that communicated sincere and benevolent intentions encouraged their members to develop supportive dispositions." (202) The use or threat of removal undermines supportive dispositions. It thereby risks rendering agencies across the board less effective in their appointed tasks. Moreover, the availability of an effective removal power may impose costs in terms of diluted agency initiative and a diminished willingness to use expertise. (203) It is, indeed, well documented that eliminating agency discretion decreases the incentives of agencies to acquire information and take policy initiatives. (204) "[B]ureaucratic expertise is endogenous, costly, and relationship specific"; it will be developed only when government induces its agents to invest in relationship-specific skills by granting job security and "some measure of control." (205) Any effectual increase in control, in short, has a price. It reduces the internal stock of epistemic capital within the administration that is often necessary to secure policy goals. There is no a priori reason to believe, moreover, that the loss in terms of agency expertise, initiative, and support will be offset by any gains associated with increased presidential control.

      A third, but...