CONGRESS IN THE ADMINISTRATIVE STATE.

• Author: Feinstein, Brian D.

In an era of increased concern over presidential power, congressional oversight of the executive branch constitutes a substantial--but underappreciated--means of influencing agency decision-making. Scholars too often have overlooked it, and Congress is sub-optimally designed for its provision, but oversight has a significant impact on agency behavior.

This Article provides a corrective. It presents the legal mechanisms that give oversight hearings their force and situates these hearings in their historical and legal context. In light of this framework and historical practice, the Article posits that ex post oversight hearings facilitate political control over the administrative state. Because oversight gets its bite from an implicit threat of legislative sanctions should an agency not change its behavior following hearings, however, committees' decisions whether to pursue oversight hinge on the credibility of this threat.

To test this theory, the Article introduces an original dataset of over 14,000 agency "infractions," i.e., agency actions that are potential subjects of hearings. Analysis of these data reveals, first, that oversight is most likely to occur when the particular preference alignment of Congress, the relevant committee, and the agency make the threat of new legislation credible. A second empirical analysis finds that, when oversight hearings do occur, they can get results; infractions that are subject to hearings are 18.5% less likely to recur compared to otherwise similar infractions that are not subject to hearings.

These findings call into question the received wisdom regarding Congress's role in governance. Whereas scholars focused on the political branches 'formal powers see Congress as a branch in decline, a more nuanced picture emerges when one also considers "soft powers," like oversight. These findings offer a blueprint for greater congressional involvement in administration: to increase Congress's role in governance, committee membership rosters should be representative of the larger legislature and committees with overlapping jurisdictions should be established. By redesigning its internal structure, Congress can promote more frequent oversight and, because oversight can be consequential, thereby strengthens Congress as a check on presidential administration.

Table of Contents Introduction I. Fundamentals A. Hearings and Alternatives B. Historical Practice C. Legal Authority II. Occurrence A. Theory B. Research Design C. Results III. Impact A. Theory B. Research Design C. Results IV. Implications A. Committee-Chamber Relations B. Major itarianism C. Jurisdictional Redundancy D. Checking the President E. Administrative Democracy Conclusion INTRODUCTION

In retrospect, it was only a blip on the media's radar screen. But in the summer of 2000, tire safety held the public's attention. (1) That summer, the nation learned that failed Firestone tires were responsible for over one hundred deaths during the previous several years. (2) Concerned about the perceived inability of the National Highway Traffic Safety Administration (NHTSA) to identify and adequately address the defect, (3) Congress enacted legislation requiring the agency to establish a data-reporting and analysis system by mid-2002 under which manufacturers must submit to NHTSA information on accident-related claims. (4)

Yet NHTSA, with more industry-friendly officials at the helm following the 2000 election, (5) dragged its feet. (6) In 2002, a House subcommittee convened a hearing where several legislators sharply criticized NHTSA's administrator for the agency's inaction concerning the defect information system. (7)

Following the hearing, NHTSA made swift progress, completing the first phase of the system just nine months later. (8) Two years after that, the agency issued the first recall based on analysis using the new system--which, incredibly, had become the government's largest non-military computer database. (9)

This sequence of events--Congress passes a law, the agency delays implementation, Congress critiques the agency's inaction, and the agency improves--suggests that congressional pressure caused an otherwise recalcitrant agency to act. (10) Yet the episode stands outside of the accepted view of congressional power. When scholars typically discuss Congress's role, they tend to focus on the branch's well-known, direct powers: primarily its lawmaking function, along with appropriations and appointments. (11) Recent work on Congress's other powers--most notably Josh Chafetz's study of Congress's "soft powers" concerning the freedom of speech or debate and each chamber's powers to establish cameral rules and discipline its members--has begun to challenge this conventional focus on the institution's legislative powers. (12) Yet mechanisms, like oversight, that lie beyond those delineated in the Constitution remain underappreciated--despite the significant resources that Congress expends performing these functions. (13) Given this incomplete picture, it is not surprising that the received wisdom holds that Congress's role in policymaking, relative to that of the President, is diminished. (14)

This Article provides a corrective. It contends that, as NHTSA's response to congressional oversight hearings exemplifies, hearings provide Congress with a powerful tool to influence administration. This Article tests this theory with an original dataset of 14,431 agency "infractions," which, as explained infra, comprise the set of issues from which Congress tends to select its subjects for oversight hearings. These infractions include critiques regarding a wide variety of regulatory implementation, enforcement, and personnel issues across all executive departments and major independent agencies, as raised in inspector-general reports, Government Accountability Office "top challenges" lists, and newspaper editorials. For each infraction, I identify, first, whether Congress held a hearing on the subject within one year after its mention and, second, whether the infraction reappeared in the dataset in the next year.

The use of this large-scale dataset allows for the comparison of agency actions that are subject to oversight hearings with otherwise similar agency actions for which Congress does not hold hearings. After all, one cannot know the independent effect of the Transportation Recall Enhancement, Accountability and Documentation (TREAD) Act implementation hearing on NHTSA's later actions without comparing that episode to a (hypothetical) other NHTSA implementation issue on which Congress did not hold hearings. This effort, the first large-scale, quantitative study of congressional oversight, answers two questions: under what conditions will oversight occur, and is this activity consequential? Taken together, answers to these questions will shed light on the broader question of whether oversight enables Congress to exert a degree of ex post control over the administrative state following legislative enactments.

Empirical analysis concerning the first question shows that the particular preference alignment of Congress, the relevant committee, and the relevant agency affect whether oversight occurs concerning a given infraction. This finding is attributable to Congress's bifurcated structure: committees are empowered to convene hearings, but only the full legislature may sanction agencies for continued non-compliance following hearings. This structure encourages committees to ignore some infractions that Congress might prefer to probe, based on the committees' fears that convening hearings could motivate Congress to enact legislative changes that the committees oppose. Essentially, committees--mindful that their parent chamber's preferences may differ from their own--make strategic decisions concerning which agencies they take to task and which they ignore. (15)

A second analysis finds that, when it occurs, oversight often is consequential, changing agency behavior for a statistically significant 18.5% of infractions, relative to otherwise similar infractions for which oversight does not occur. To put that figure in perspective: agencies commit an average of 656 infractions per year, of which 239 infractions continue (or reoccur) the next year; by holding oversight hearings, Congress prevents an additional forty-seven infractions per year from reappearing in the dataset in the next year on average. Oversight alters agency behavior--moving it towards congressional preferences on issues ranging from the level of regulatory enforcement to the creation of programs that stretch agencies' statutory authority, as well as concerning more run-of-the-mill issues such as waste, fraud, and abuse--an average of 89 times per year.

These findings have implications for our understanding of the roles that all three branches play in the administrative state. First, the finding that committees strategically decline to hold hearings based on the preference alignment of Congress, the committee, and the relevant agency shows a subtle majoritarian dynamic at work in Congress's internal organization. Although committee-based oversight can be remarkably impactful, outlier committees are less likely to engage in oversight. Thus, the existence of a bifurcated congressional principal provides a majoritarian check on unrepresentative committees--and cuts against arguments favoring strong presidential administration based on the premise that congressional control supposedly involves control by outlier committees.

Prescriptively, that finding suggests that those interested in enhancing Congress's capacity ought to do away with two of the branch's institutional features: legislators' self-selection onto committees and the granting of exclusive jurisdictions to committees. The current practice of allowing legislators to select their committee assignments yields committees that are unrepresentative of floor preferences. As explained infra, outlier...