TABLE 6-4 Average Number and Volume of Oil Spills on U.S. Soil, 1990-1998a



Annual Number per Year

Volume per Year (gallons)





Spills by size

<100,000 gal




>100,000 gal



Spills by source

Tanks and barges




Other vessels




Facilities and unknownb



aData from the U.S. Census 2006.

bData exclude spills from pipelines—information on pipeline spills is contained in the following section of the report.

pleted (for example, Cohen 1995; Garza-Gil et al. 2006). Talley (1999) estimated the property damage costs from tanker accidents but did not attempt to consider environmental damages. Overall, little research is available to estimate the expected damages from spills that could be appropriately attributed on a per gallon transported basis. One exception is the study by Cohen (1986) in which estimates of the per gallon benefits of avoiding oil spills are given. On the basis of the compensation payments from 11 spills and in-depth environmental damage studies from four large spills, Cohen estimated a $6.08/gallon benefit of avoided spillage for reduced environmental damages, $1.72/gallon benefit for avoided loss of oil, and $6.93/gallon for avoided cleanup costs (values converted to 2007 dollars). The first of these categories clearly represents an externality of interest to this committee, and the second is a private cost and therefore not relevant for this study. The third component is also not an externality—it represents the costs of cleaning up an externality once it has occurred. However, if the optimal level of cleanup is chosen, then the marginal damages averted through cleanup would equal the marginal costs of cleanup and we could use this third component as a proxy for the damages averted through cleanup, giving aggregate marginal damages of $13.01 per gallon spilled.8

We need to convert from damages per gallon spilled to damages per gallon produced or consumed. According to data reported in Huijer (2005, Table 2),9 tanker spills worldwide averaged 115,810 barrels between 2000


To be clear, Cohen (1986) estimated average damages but assumed they were a proxy for marginal damages. A convex damage function using average damages underestimates marginal damages.


Paper cited at

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement