Congress has a dual role in the American system of government. In relation to executive agencies, it serves as principal. Congress enacts laws which create, structure, authorize, and fund executive agencies. Congressional committees and subcommittees are charged with monitoring and disciplining agencies within their respective purview. On the other hand, in relation to the American electorate, Congress serves as agent. Each member of Congress is electorally accountable to a well-defined geographic constituency and serves at the pleasure of that constituency. Members of Congress must employ the bureaucracy as an effective instrument of constituent service to achieve re-election. The complexities of the policy environment inform Congress’ methods of employing the bureaucracy to provide the public goods on which each member’s electoral fate depends. Importantly, limited time and cognition among members of Congress combined with the selective attention of the public leads Congress to pursue a decentralized, largely automatic oversight approach which shifts much of the costs of monitoring and enforcement to third parties. This method is largely effective but susceptible to the introduction of bias in policy making.
Early scholarship on congressional control of the bureaucracy assumed agency independence under perfunctory and ineffectual congressional oversight. Principal-agent theory provided support for this proposition. Information-asymmetry between a preoccupied, non-expert Congress and the career bureaucrats under its charge makes monitoring agency compliance difficult and costly. An agency’s direct, routine contact with both the objects and putative beneficiaries of regulatory output contributes to this asymmetry. Even when noncompliance is detected, congressional sanctions require the costly coordination of members of Congress to pass new legislation. Taken together, these principal-agent problems and the apparent cursory nature of observed congressional oversight activities engendered a widely held view of a bureaucracy largely unanswerable to Congress and insulated from the democratic process. For scholars optimistic about the possibility of objective scientific policy making predicated on emerging welfare economics, this insularity was a great boon. For many others, insulation from the democratic process rendered the work of a popularly unaccountable bureaucracy inherently suspect.
By the 1980’s, several scholars began questioning the assumption of agency independence. Weingast and Moran (1983) postulated that the traditional model of bureaucratic discretion and their then-heterodox model of congressional control yielded facially equivalent observations. Using the paradigmatic FTC controversy of the late 1970’s as an empirical test-case, Weingast and Moran show that—far from being a runaway agency—the FTC’s consumerist policies from the mid-1960’s through the mid-1970’s were in harmony with the preferences of its direct overseers in Congress. The discordance in the late 1970’s between the commission and its congressional principals resulted from an abrupt change in the preferences of the principals arising from a nearly complete turnover in the Senate oversight committee after 1976. “The 1979 and 1980 hearings were simply the most visible culmination of this process” (777). In a logit analysis, Weingast and Moran used ADA scores as a measure of subcommittee panel preferences and test for the influence of changes in subcommittee preferences on the degree of consumerist activism in FTC case selection. Rather than a runaway agency, they found that “the FTC is remarkably sensitive to changes in the composition of its oversight subcommittee and its budget” (792). This examination of the FTC yielded evidence of latent congressional control.
How could such effective oversight be overlooked? McCubbins and Schwartz (1984) posit that “what has appeared to scholars to be a neglect of oversight…really is a preference for one form of oversight over another, less effective form” (165). They distinguish two oversight techniques that Congress might employ. The first, police-patrol oversight, is a costly command and control technique requiring such activities as “reading documents, commissioning scientific studies, conducting field observations, and holding hearings to question officials and affected citizens” (166). The second, fire-alarm oversight, is a decentralized incentive-based technique which co-opts affected parties and the courts as monitors and enforcers of agency policy. This indirect and less visible oversight strategy has its advantages: “although [it] can be as costly as police-patrol oversight, much of the cost is borne by the citizens and interest groups who sound alarms and by administrative agencies and courts rather than by congressmen themselves” (168).
How is this indirect, cost-shifting form of oversight implemented? McCubbins, Noll, and Weingast (1987) provide the more elaborate answer. Congress, they argue, carefully crafts—and, when necessary, tweaks—administrative procedures to enfranchise favored groups in an agency’s decision making process. This selective enfranchisement may take the form of shifting the burden of proof in agency determinations, granting standing to certain parties to challenge agency action in court, calibrating the level of judicial review, and requiring the collection and dissemination of relevant information. This strategy ‘locks in’ a stable yet adaptable arrangement capable of self-perpetuation: “Whereas political officials may not know what specific policy outcome they will want in the future, they will know what interests ought to influence a decision and what distributive outcomes will be consistent with the original coalitional arrangement” (255). This indirect oversight method also gives members of Congress the ability to distance themselves from unpopular policy determinations as well as the option of performing constituent service on an ad hoc basis by applying pressure to an agency where and as needed. Finally, “the courts are the key, for without them political actors could not rely on decentralized enforcement” (id.).
But some scholars warn that fire-alarm oversight by election-conscious political leaders can have unintended consequences for the content of policy. If Congress is not monitoring agencies directly, it must receive relevant information from others. The sources of information would include affected parties, the media, and, through procedural reporting requirements, the agency itself. These information flows certainly help congressional principals cope with information costs. However, as Hopenhayn and Lohmann (1996) observe, “the amount of information received by the principal need not be independent of the agency’s decision” (197). Hopenhayn and Lohmann ask us to consider an example from Kazman (1991):
There are two types of errors the FDA can make in reviewing a new drug application: it can approve a drug that turns out to have unexpectedly adverse side effects, or it can delay or deny a beneficial drug. From a public health standpoint, both of these errors can be equally deadly, but from a political standpoint, they are worlds apart. Incorrectly approving a drug can produce highly visible victims, highly emotional news stories, and heated congressional hearings … Incorrectly delaying a drug, on the other hand, will produce invisible victims and little more … Not surprisingly, the FDA’s fundamental approach to drug approval is designed to reduce the likelihood of the first type of error while paying little attention to the second. The well-documented result of this excessive caution is drug lag-the frequent unavailability of major new drugs in this country long after they have been approved elsewhere (Kazman 1991 as quoted in id.).
Kazman’s observation can be taken further, I think. Due to the availability bias observed above, it is not only the incorrect approval of a drug that can be politically toxic, but approval of any drug likely to result in a significant number of adverse outcomes, even if those outcomes are greatly eclipsed by the beneficial effects of granting access to the treatment. Adverse consequences traceable to agency action is likely to bring heat on the agency in the form of media scrutiny and public outrage. Even if the agency’s actions are defensible, election-conscious members of Congress might prove more likely to grandstand against an agency under fire than take the heat on themselves.
Agency officials throughout the bureaucracy are certainly aware of the vacillating tendencies of their congressional principals and are, therefore, likely to exhibit extreme risk aversion in their decision making. This bias is likely to escape the notice of a congressional principal concerned mainly with a cost-effective method of pleasing its own principal. Thus, while Congress may have found a cost-effective means of employing the bureaucracy in support of its own electoral goals, the true cost is perhaps imponderable: the countless “invisible victims” described by Kazman.
How does the re-election incentive affect Congress’ ability to control the bureaucracy? I argue here that Congress pursues an indirect, incentive-based approach to controlling the bureaucracy which uses administrative procedure to shift the costs of monitoring and enforcement to other parties. This is a cost-effective method for Congress to employ the bureaucracy as an instrument of constituent service, but also may introduce welfare reducing risk-aversion bias into some areas of public policy.