The Congressional Budget Office: The Cost of Liability for Nuclear Accidents

The cost of the Price-Anderson Nuclear Liability act is highly controversial. Here is the Congressional Budget Office's estimate. It should be noted that the accident is calculated for Generation II reactors. The accident rate for Generation III + reactors would be much lower, with casualty-producing accidents occurring in spans of over once in a billion years. The probability of a casualty-producing accident occurring with a LFTR might be something in the order of once every 10 Billion years, or something close to the currently estimated lifespan of the universe. Added LFTR safety features would lower the casualty-producing accident probability even further. At this point concern about nuclear accidents enters into the realm of absurdity.

Among its various provisions, the Energy Policy Act of 2005 extended the Price-Anderson Nuclear Industries Indemnity Act, which limits the industry’s liability for accidents at nuclear power plants. In practice, Price-Anderson subsidizes utilities by reducing their cost of carrying liability insurance. Instead of purchasing full coverage, operators of nuclear power plants are required to obtain coverage only up to the liability limit, which is currently set at about $10 billion per accident.1 The value of the subsidy is the difference between the premium for full coverage and the premium for $10 billion in coverage. On the basis of data obtained from two studies—one conducted by the Nuclear Regulatory Commission (NRC) and the other by the Department of Energy (DOE)—the Congressional Budget Office (CBO) estimates that the subsidy probably amounts to less than 1 percent of the levelized cost for new nuclear capacity.2

To assess the health hazards that existing nuclear power plants could pose, analysts at the NRC estimated the probability of radioactive releases occurring at several nuclear facilities, including the Surry power station in Virginia, and the consequence of such an event.3 Damage to property and possible injury or loss of life caused by a hypothetical accident at that facility could be pertinent to assessing the liability of proposed nuclear plants because several of them would be located in areas of the Southeast with roughly similar population densities. For the Surry power station, the NRC study provides assessments of both internally initiated accidents (which could be caused by malfunctioning equipment or human error) and externally initiated accidents (which could result from a fire or earthquake). According to the study, an internally initiated accident at such a facility that on average caused more than 10 deaths would occur, at most, once every million years. A fire-related accident causing more than 1,000 deaths on average would occur, at most, once every million years. CBO’s analysis adopted those probabilities and results for the sake of determining liability from fatalities. To that, CBO added estimates of injuries and property damage to provide a more complete estimate of liability.

CBO based its assessment of liability from injuries and property damage on the DOE report, which modeled a radioactive release at the Limerick facility near Philadelphia. That scenario includes, in addition to the number of fatalities, estimates of injury and property damage, from which CBO inferred potential liability resulting from an accident at the Surry plant.

On the basis of the probability of fatal accidents estimated in the NRC report and the estimates of damage from such accidents in the DOE report, it appears that catastrophic accidents are possible but likely to be rare; CBO estimates that an accident causing about $500 billion in damages will occur an average of 3 out of every 100 million years.4 Because such potential damages are spread over a long period, the long-run average of damages per year (the expected cost) would be only about $600,000. That figure does not include the cost of nonfatal accidents, which might already be covered by the $10 billion in damages for which the nuclear power industry is held liable under the Price-Anderson Act. If so, the projected annual subsidy is about $600,000 per reactor as well.

Insurance premiums represent a small portion of the levelized cost for a nuclear power plant. Even if the analysis based on the Surry facility understates the expected cost of fatal nuclear accidents by a factor of 10, paying a fair premium would not lead to large changes in the levelized cost. In CBO’s reference scenario, increasing the insurance premium by $6 million per year increases the levelized costs by 1 percent.
______________________

1. That $10 billion in coverage has two layers: The owner of a nuclear plant is required to purchase primary insurance covering liability up to $300 million. In the event of an accident, liability for damages assessed at between $300 million and $10 billion would then be shared among the owners of all U.S. nuclear plants, who would pay a "retroactive premium."

2. See Nuclear Regulatory Commission, Severe Accident Risks: An Assessment for Five U.S. Nuclear Power Plants, NUREG-1150 (December 1990); and Department of Energy, Technical Guidance for Siting Criteria Development, SAND-81-1549 (December 1982). CBO’s estimate was derived to evaluate the sensitivity of levelized costs (or the minimum price of electricity at which a technology generates enough revenue to be economically viable) to limits on liability but should not be interpreted as a precise estimate of the expected cost of liability.

3. A description and evaluation of the NRC’s probabilistic risk assessment models is provided in Nuclear Power Joint Fact-Finding (Keystone Center, June 2007).

4. Each fatality is assumed to lead to $5,000,000 in liability, and each injury is assumed to cause $2,500,000 in liability.

It would not be unreasonable to charge the reactor owners a premium to cover the value of
Price Anderson. On the other other hand it would also not be unreasonable, assuming the extreme unlikelyhood of a reactor event that would require government compensation, that the government fore goe such a premium in the grounds that it would be extremely unlikely that Price-Anderson compensation would ever be paid during the lifetime of these reactors, and that future reactors will be so much more safe, that the payment of future compensation is etremely implausible.