Moral Hazard and the Role of Users in Learning from Accidents1

Abstract

Most technological accidents studied in the literature consist of a single event occurring in a single location (Challenger, Chernobyl, Ariane 5, Bhopal, etc.). However, a significant number of accidents comprise a series of incidents taking place in multiple and unconnected locations. This may happen whenever a given technology is used by multiple decentralized users – as is the case for medical devices or diving equipment, for instance. In this paper, we argue that such a structural characteristic raises specific problems and issues in terms of learning from accidents. We focus on a specific issue and problem: information asymmetries between the users and the manufacturer in favor of the manufacturer and the risk of “moral hazard” that they can induce. Using second hand information on a well known case (Therac-25), we show how the manufacturer behaved opportunistically when it came to learn from a series of decentralized incidents and we argue that this opportunism was made possible by information asymmetries. Following predictions from the principal-agent theory, we then argue that for users to monitor manufacturers' behavior in a sense favorable to the taking place of learning from accident, it is necessary to ensure communication among users of risky technologies. Besides reducing information asymmetries conducive to moral hazard this would moreover improve risk prevention. Yet, as communication among decentralized users is unlikely to arise spontaneously in most cases, we argue that this makes the case for public intervention in the form of publicly sponsored user groups.