Working Paper No. 03-36:
Information Markets, Administrative Decisionmaking, and Predictive Cost-Benefit Analysis

Author(s):

Date Posted: 2003

Availability:
Abstract (below) | Full text (most recent) on SSRN

Abstract:

FutureMAP, a project of the Defense Advanced Research Projects Agency, was to involve experiments to determine whether information markets could improve Defense Department decisionmaking. Information markets are securities markets used to derive information from the prices of securities whose liquidation values are contingent on future events. The government intended to use such a market to assess the probabilities of potential political assassinations, and the indelicacy of this potential application contributed to a controversy leading to the cancellation of the program. In this Article, Professor Abramowicz assesses whether information markets in theory could be useful to administrative agencies, and it concludes that information markets could help discipline administrative agency predictions, but only if a number of technical hurdles such as the danger of manipulation can be overcome. Because the predictions of well-functioning information markets are objective, they function as a tool that exhibits many of the same virtues in predictive tasks that cost-benefit analysis offers for normative policy evaluation. Both approaches can help to overcome cognitive errors, thwart interest group manipulation, and discipline administrative agency decisionmaking. The Article suggests that the two forms of analysis might be combined to produce a "predictive cost-benefit analysis." In such an analysis, an information market would predict the outcome of a retrospective cost-benefit analysis, to be conducted some years after the decision whether to enact a particular policy. As long as the identity of the eventual decisionmaker cannot be anticipated, predictive cost-benefit analysis estimates how an average decisionmaker would be expected to evaluate the policy. Because the predictive cost-benefit analysis assessment is not dependent on the identity of current agency officials, they cannot shade the numbers to justify policies that the officials prefer for idiosyncratic or ideological reasons.