W30 - Regulatory Errors under Two-Sided Uncertainty: Or, the Political Economy of Vioxx, Daniel Carpenter (Cohort V - University of Michigan Program) and Michael M. Ting, January 2005.

How do the errors of regulators -- approving bad products, or rejecting good ones -- depend upon the submission strategies and characteristics of submitting private entities or firms? We develop a model of approval regulation in which both firm and regulator are uncertain about the underlying quality of a product, but where the firm is better informed than the regulator. The model predicts that the commission of regulatory error depends heavily upon the induced submission strategies of firms, and in particular on the amount of time that firms take to experiment with their products before submitting them. Specifically, our analysis predicts that when experimentation is short and experimental costs are high, Type I errors (approving bad products) should be disproportionately associated with products submitted by firms with lower experimentation costs (larger firms), and Type II errors (rejecting good products) should be concentrated among products submitted by smaller firms. However, when experimentation is long and experimental costs are low, Type I errors should be concentrated among higher- cost developers (small firms), while Type II errors will be concentrated among products submitted by larger firms. Under all conditions Type I errors should be increasing in the cost of regulatory submission, and the inverse should be true for Type II errors. We test hypotheses from the model in a statistical analysis of regulatory errors committed by the FDA in the approval of new pharmaceutical products. Using two different sets of measures for ``Type I errors," we find consistent support for the predictions of the model. We also find modest evidence that recent predictable reductions in approval times have been associated with a reduced rate of Type I error by the FDA, as our model would predict. The model also sheds new light on more (in)famous cases of regulatory error, including the FDA's delay in approving beta-blockers in the 1970s and the recent Vioxx episode.

Daniel Carpenter is Professor of Government and Director of Graduate Studies, Department of Government at Harvard University(dcarpenter@latte.harvard.edu).

Michael M. Ting is Assistant Professor of Political Science and Public Affairs, Department of Political Science and SIPA at Columbia University.

< Back to home page