Another Lesson from the BP Disaster: The Need for Better Risk Assessment
Apparently, the lease grant to BP was exempted from environmental review, according to the Washington Post’s Juliet Eilperin:
The decision by the department’s Minerals Management Service (MMS) to give BP’s lease at Deepwater Horizon a “categorical exclusion” from the National Environmental Policy Act (NEPA) on April 6, 2009 — and BP’s lobbying efforts just 11 days before the explosion to expand those exemptions — show that neither federal regulators nor the company anticipated an accident of the scale of the one unfolding in the gulf.
The reason was the allegedly minimal risk of harm from an oil spill:
In one assessment, the agency estimated that “a large oil spill” from a platform would not exceed a total of 1,500 barrels and that a “deepwater spill,” occurring “offshore of the inner Continental shelf,” would not reach the coast. In another assessment, it defined the most likely large spill as totaling 4,600 barrels and forecast that it would largely dissipate within 10 days and would be unlikely to make landfall.
This is actually all-too-typical of the lousy job that agencies do at risk analysis in conducting environmental reviews, as I have discussed elsewhere. Here are some prescriptions for fixing this problem.
- Where possible, confidence intervals should be provided for critical data.
- When the agency relies on formal modeling, validation issues should be directly addressed.
- Whether or not a formal model is used, the agency should discuss the limitations of current understanding of system dynamics and conflicting models found in the scientific literature.
- Rather than relying solely on model output as a basis for evaluating risk, the agency should give explicit attention to model uncertainty.
- The agency’s reasoning should be transparent and model assumptions should be clearly stated.
- Where the agency has proposed a major project or regulatory initiative, and a possible catastrophic risk could attend that action, the agency should at least obtain a peer review of its analysis and ideally should procure a risk assessment from an independent body.
- Courts should not second-guess an agency’s scientific judgments, but neither should they allow expertise to function as a smokescreen for any agency’s failure to probe the relevant science in depth, explore opposing viewpoints, and candidly disclose analytic uncertainties.