Controlling for Oversight in Studies of Oversight

ABSTRACT In honor of the creation of the Congressional Oversight Research Database (CORD), we explore some of the selection effects that observable measures of oversight may be subject to. We present a model in which Congress endogenously chooses whether to hold a hearing or to preempt the need for one by overturning an agency's policy in advance. These endogenous oversight choices imply that observable outcomes arise through two pathways: Congress may overturn a policy immediately or only after acquiring additional information through a hearing. Failing to account for this information‐gathering stage can obscure the underlying mechanisms that drive these outcomes. Using data from ProQuest and CORD, we test two implications of the model, finding evidence that Congress selects into conducting oversight hearings and implementing subsequent policy reversals more often when in opposition to the President, with no impact of the information gathered during a hearing on subsequent policy reversals. This implies that Congress uses committee hearings primarily as a mechanism for sharing information about the executive branch when it is politically convenient, rather than as a mechanism for gathering information on policy alternatives. We then conclude with broader implications for empirical research on oversight, including the importance of distinguishing between information gathering and information sharing in oversight.