Chemical accident risks in U.S. industry - A preliminary analysis of accident risk data from U.S. hazardous chemical facilities

摘要

In response to the chemical accident that occurred in Bhopal, India, in 1984, and a series of large chemical accidents in the United States in the late 1980s, the U.S. Congress passed a series of laws intended to minimize the likelihood and consequences of catastrophic chemical accidents. The most recently enacted of these laws created a new regulatory program called the Risk Management Program. This program, which took effect in June 1999, requires certain chemical facilities to implement chemical accident prevention and preparedness measures, and to submit summary reports to the government every five years. Approximately 15,000 facility reports have been received to date, and these contain significant information on each facility's accident history, accident prevention program, and the potential consequences of hypothetical accidental chemical releases. These data have been assembled into a searchable computerized database, called RMP~* Info. The full RMP~* Info database was originally intended to be available to the general public via the Internet, so that concerned citizens could use the information to influence local facilities to adopt safer practices, and to allow researchers to identify factors statistically associated with accident-prone or accident-free facility performance. However, the chemical industry and U.S. security agencies raised concerns that some of the data would allow terrorists to easily identify those facilities likely to cause the greatest harm to the public in the event of a release, and target those facilities for attack. These concerns prompted Congress to pass legislation in August 1999, that, along with subsequent federal regulations, currently restricts public access to portions of the RMP~* Info database. Therefore, as of this writing, the complete database is only accessible by a relatively few individuals at EPA headquarters, and, with the exception of one other preliminary study that considered only its non-restricted portions, it has not been subject to the robust statistical analysis that might lead to identifying accident indicators or causal factors. This paper proceeds in that direction by providing some basic descriptive statistics that characterize the database, including its restricted portions, within the limitations set by United States law.