Is the nuclear liability limit a subsidy, or not?
© Getty Images

The heat is on. I refer not to the beginning of summer, nor the looming global warming apocalypse for which there is little evidence, nor an election season sure to be characterized by personal attacks aplenty. But instead, the increasing difficulty of defending the absurd argument that the massive explicit and implicit subsidies given wind and solar power and ethanol production yield environmental improvement and reduced costs and improved efficiency for energy markets.


To paraphrase President Reagan: Subsidies are an approximation of eternal life on earth, in that they are created by interest group pressures and they intensify those pressures as well. And so the utter environmental destructiveness of the subsidy programs is irrelevant: The proponents of the wind production tax credit and the solar investment tax credit and the guaranteed market share for ethanol and all the rest — and their representatives in Congress — cannot and will not back away from the subsidy trough. But they have to say something to defend their indefensible pork, and so they now point to all of the supposed subsidies bestowed on other forms of energy, implicitly offering to sacrifice theirs after those others are removed. Which is a prospect that they know will never happen; as a practical matter, the status quo would live forever.

Put aside the reality that most of the tax preferences and other support given conventional energy sources are not "subsidies" properly defined. The most recent example of this misdirection came at a hearing on energy tax policy held by the Senate Committee on Finance, at which Republican Sen. Charles GrassleyCharles (Chuck) Ernest GrassleyHigh stakes as Trump, Dems open drug price talks Senate approves border bill that prevents shutdown Grassley raises voice after McConnell interrupts Senate speech MORE of Iowa (a state benefiting from lavish wind and ethanol subsidies and production incentives) asked repeatedly whether or not the Price-Anderson liability limit for nuclear power is a "subsidy," by virtue of limiting the liability of nuclear plant owners in the event of a nuclear accident causing substantial injury and damage.

A bit of background: Price-Anderson was enacted in 1957 as an amendment to the 1954 Atomic Energy Act, and has been extended several times, the most recent of which is through 2025 as legislated in the Energy Policy Act of 2005. Owners of nuclear generating units must carry $375 million in "first tier" liability coverage per incident; if that amount proves insufficient to pay damages caused by an accident, a second tier creating total coverage of $13.2 billion per accident becomes applicable. The second tier is financed by the owners of all operating nuclear units in the U.S., with the owner of each nuclear unit limited to a liability of about $125 million. Accordingly, the owner of a nuclear reactor causing substantial damage faces liabilities limited to about $500 million.

But suppose that an accident causes damage and losses exceeding $500 million. Is the limit not a subsidy for the owner of the reactor? That is the seemingly unanimous view: Liability should be unlimited for owners of nuclear reactors that cause damage to persons and property. And that may seem utterly reasonable: After all, why should the party causing (massive) damage not pay for it? Should public policies not provide incentives to minimize the likelihood of an accident?

The answer, however counterintuitive, is that the appropriate policy goal is not merely the minimization of that likelihood, that is, maximization of preventative action on the part of the nuclear owners. Instead, we should seek both such preventative steps and avoidance actions on the part of those potentially harmed, that is, minimization of the sum of the damage from accidents and the costs of avoiding that damage, an insight that helped Ronald Coase win the Nobel Prize in economics in 1991.

Consider a highly artificial world in which all actual and potential residents and businesses that would be affected adversely by a nuclear accident could move or locate elsewhere at a cost of zero. In this case, the appropriate liability for the nuclear owners would be zero: Construction of a nuclear unit would induce all those potentially affected to move elsewhere costlessly (or not to move near the nuclear plant in the first place), and any subsequent accident would cause no injuries or economic damage.

Yes, that example, again, is highly artificial. So consider instead the straightforward case of construction in hurricane zones, one analytically identical to the nuclear case just discussed. If the owners of buildings are compensated fully for hurricane damage, with, say, government disaster grants, too much construction will be undertaken in hurricane zones, yielding an increase in damage when a strong storm hits land. Construction in flood zones also is identical analytically. Since it is impossible to prevent hurricanes, we can avoid some substantial portion of the prospective storm damage by limiting development threatened by hurricanes; by forcing the owners of buildings to pay actuarially fair insurance rates, we limit construction to that justified economically given the risks.

The same principle applies to nuclear construction. If people and owners of property are compensated fully for the damage caused by a nuclear accident — if the owners of nuclear facilities bear full liability for the adverse effects of accidents — too many people, businesses and physical capital will be located near the reactors, yielding an increase in damages when a serious accident occurs. This analytic truth holds even for reactors sited in earthquake or tsunami zones; sometimes those are the most appropriate sites for the reactors even given the risks. Accordingly, the Price-Anderson liability limit at least directionally is economically efficient — it conserves resources — by providing incentives for those who can avoid the potential damage from a nuclear accident most cheaply to do so, thus preventing some portion of prospective nuclear damage by limiting location choices near the reactors below the levels that otherwise would be observed if public policy imposed full liability on the nuclear owners.

But, you say, hurricanes are natural phenomena, while nuclear accidents are not. Not so fast. From a cold-blooded economic perspective, the hurricane damage is caused by man: Without the construction in the hurricane zones, there would be no damage. The efficient approach is to limit construction to that justified economically even given the risks.

It is possible to prevent (or limit) the damage from nuclear accidents by limiting approvals of nuclear generating units and/or by making investments in safety systems and the like. Similarly, it is possible to prevent some of the damage by reducing or relocating other economic activity threatened by accidents. The economic question is not whether nuclear owners "ought" to be fully liable. It is instead how to limit damage at the lowest total cost through some combination of limited approvals of nuclear generating units (forgoing some benefits of nuclear electricity), large investments in accident prevention or mitigation (expensive technology), and reductions in other economic activities in nuclear zones (forgoing other valuable productive activity). Are the first two always the cheapest? That is far from obvious. A liability limit in a very rough fashion has the effect of shifting some of the liability onto those who might be able to make adjustments more cheaply, thus improving the overall efficiency of resource use.

Clearly, caveats apply. If the government adopts a liability limit too high or too low, the economic outcome might be worse than that yielded by unlimited liability. If the government subjected to political pressures will make those harmed by an accident whole in any event, a limit on liability would serve no purpose. And it is clear that substantial liability for the nuclear operators is appropriate, in that large investments in redundant cooling systems and containment structures and the like obviously are the cheapest way to avoid much of the potential damage from accidents.

But substantial liability is not unlimited liability. The argument for limited liability is difficult to make and defend in a world of sound bites. But the argument that the Price-Anderson liability limit is a "subsidy" is simply incorrect; it has become an obvious tool with which to divert attention from the indefensible subsidies and cronyism now bestowed upon "renewable" electricity, ethanol mandates and the other such manifestations of wealth redistribution through politics.

Zycher is the John G. Searle scholar at the American Enterprise Institute.