Nuclear Costs Reconsidered

‘Negative Learning’ Not Inherent to Nuclear Power

Last month in Paris, the cognitive dissonance between environmental demands for immediate and rapid decarbonization of the global economy and the long standing rejection of nuclear energy by environmental NGO’s and advocates reached the breaking point. Four climate scientists, led by Dr. James Hansen, flew to Paris to reiterate their call for environmental leaders to reverse their opposition to nuclear energy. “The future of our planet and our descendants depends,” the four scientists wrote, “on letting go of long-held biases when it comes to nuclear power.”

In response, Harvard historian and Merchants of Doubt author Naomi Oreskes penned an op-ed in The Guardian accusing them, bizarrely, of being climate deniers. Advocating for nuclear energy, Oreskes claimed, simply distracts the world from getting on with the business of building an energy system run wholly with wind, water, and solar energy. Irrespective of its environmental benefits, nuclear energy, Oreskes argued, has proven too costly to build.

In this, Oreskes and others who have rejected Hansen’s call for a major reconsideration of nuclear energy simply repeat a long accepted environmental trope. Nuclear plants, a generation of environmentalists has argued, are characterized by “intrinsic negative learning” - the more of them we build, the more expensive they become. It is an extraordinary claim. There is little experience or evidence in any other domain for the proposition that the more people do something, the worse they get at doing it, particularly when it comes to highly technical and complex tasks.

But then, nuclear energy has always been an exceptional technology, at least insofar as public perceptions of it are concerned, and so the claims were taken, even among many scholars and analysts, at face value. There were, of course, plenty of anecdotal examples of nuclear plants costing a lot to build. And over the last several decades, a growing body of empirical evidence also seemed to support the claim. Analyses of nuclear costs in the United States and France appeared to show evidence of negative learning.

But the basis for the sweeping claim that has been made from these studies, that negative learning is inherent to nuclear energy turns out to have been extraordinarily narrow, based upon nuclear construction costs in two countries constituting less than a third of the total global nuclear fleet over the course of only two decades, the 1970’s and 1980’s.

In fact, the evidence for dramatic cost escalation associated with the scale up of nuclear energy was entirely associated with only one of those countries, the United States. After the Three Mile Island accident, construction delays increased dramatically and costs of plants that were under construction at the time, about a third of the entire US nuclear fleet, went through the roof. As a result, the United States stopped building nuclear plants.

But the rest of the world did not. In a new paper, published this week in Energy Policy, I, along with Breakthrough's Jessica Lovering and Carnegie Mellon researcher Arthur Yip, collect and review a much larger set of data from a larger number of countries and time periods. All told, we account for the cost of almost 60% of all plants built globally between 1950 and 2015, more than twice the number covered by earlier studies.

We collect full histories of nuclear construction costs from Japan, Germany, India, and South Korea, as well as plant construction costs from earlier periods in the United States and France, to give a full accounting of the history of nuclear construction in those countries.

The results tell a strikingly different story than that told in the “negative learning” literature. Cost trends vary significantly by country and time period. Overall, nuclear construction costs globally have been relatively stable since the mid-1980’s. Countries that have continued to build nuclear power plants such as Japan, India, and South Korea have seen stable or declining costs. South Korea, in particular, which was a relatively late entrant into the nuclear game, has seen consistent cost declines through virtually its entire nuclear construction experience. Even in the United States, there was over a decade of falling nuclear construction costs before nuclear costs began to rise in the 1970’s while in France nuclear construction saw very mild cost escalation, consistent with other large construction projects.

What a broader, longer, and global view of nuclear costs suggests is that many of the key drivers of nuclear cost trends lay beyond the technology itself. Nuclear cost trends appear to be more a function of the policies, institutions, and perhaps even the engineering cultures of national economies than technological learning, positive or negative. The cost of nuclear power, and whether it becomes more or less costly over time, is not intrinsic to the technology itself but rather is something that we have some control over.

That doesn't mean that new plants won't be expensive, particularly first-of-kind plants such as the EPR's and AP1000's currently under construction in the US and Europe. But what the evidence pretty consistently shows is that if you build multiples from the same design and within a sensible regulatory and institutional context, costs from those initial plants do come down and stabilize.

It’s not hard, of course, to see why anti-nuclear environmentalists seized upon the notion of inherent negative learning with such enthusiasm. If cost escalation is inherent to nuclear technology, then there isn’t anything we can do at the level of policies and institutions to address it. This is a distinction that must be made if one is to advocate for the need to remake the traditional utility business and regulatory model, build an entirely new electricity grid, and revolutionize the ways in which we consume energy, all to accommodate intermittent renewable energy technologies on the one hand, while dismissing all efforts to create policies and institutions to accommodate nuclear energy with the other. If nuclear energy has inherently negative learning than all efforts to promote it are, unavoidably, doomed.

Thankfully, many sensible environmentalists and renewable energy advocates are coming around to the idea that there is not only ample opportunity but an urgent need for an all-of-the-above, low-carbon energy strategy. What our new paper suggests is that with the right policies and institutions, nuclear plants can be built quickly, safely, and cheaply. Given the scale of the climate challenge, the many environmental benefits that clean, cheap, and power-dense nuclear energy brings, and the challenges associated with scaling renewable energy to meet our energy needs, we would be well served to design policies that can promote and accommodate as many plausible pathways to deep decarbonization as possible.