Formal and Empirical Analyses of Bureaucratic Delay: The Case of FDA Drug Review

Harvard University, Cambridge MA

Investigators

Abstract

Why do bureaucrats delay? Why do regulatory choices made under identical administrative procedures exhibit highly varying decision times? The investigator conducts a formal and empirical analysis of the duration of bureaucratic decisions, with applications to drug approval times by the U.S. Food and Drug Administration (FDA). The speed with which drugs are reviewed is now one of the most controversial issues in federal regulation, and is arguably the most salient case where the duration, more than the content, of bureaucratic decision making is at issue. While the political control of agencies has been well studied, relatively little is known about why agencies wait, and whether the duration of agency decisions is subject to political control and from what sources. The project tackles this problem through the elaboration of a generalizable formal model of the approval decision. The investigator models product approval as an optimal stopping decision by an uncertain regulator subject to political demand for the drug. The regulator seeks to preserve a reputation for consumer protection and sees drug approvals as irreversible (or reversible only at a significant reputation cost). The payoff of approval is also a function of the political "demand" for drug approvals from consumers (patients) and producers (firms). The model predicts that even under risk-neutrality, quick approvals are highly unlikely, not all drugs will be approved, and that larger firms receive "protection" (quicker approvals) even when firms are not politically organized. The investigator tests this model using maximum likelihood duration analyses of FDA drug review times. The formal model suggests that the proper hazard function for estimation is the inverse Gaussian distribution, which has a non-monotonic hazard function. Drug approval times will be modeled as a function of political preferences, budgetary shifts, media coverage and congressional hearings on the diseases treated by the drug, and characteristics of the submitting firms. Preliminary results on a partial sample of 67 drugs approved from 1985 to 1995 support the model's predictions, and also suggest that political "control" in drug approval occurs less through "principals" (Congress and the presidents) than through organized interests (patients and firms) and the media.