accident liability

Cover to Japanese edition of "Learning from Fukushima"
Cover to Japanese edition of "Learning from Fukushima"

With the eighth anniversary of the Fukushima accident having recently passed, I wanted to mention a few Fukushima-related threads.

1.  A Japanese translation of Learning from Fukushima was released in February, making the material accessible to more people within Japan.  Namatame Norifumi was the main translator, with assistance from Suzuki Tatsujiro.  The original volume was edited by Peter van Ness and Mel Gurtov, and covers a wide variety of issues related to the accident and to nuclear power in Asia. 

My chapter focuses on the scale of subsidies and some of the reasons that the most important subsidies to nuclear often get missed. 

The Japanese version can be purchased here.  The English version can be accessed (free PDF downloads) here.

2.  Cost estimates for cleaning up after the Fukushima accident continue to grow; highlight inadequacy of Price-Anderson coverage levels in the US.  The Tokyo-based Japan Center for Economic Research (JCER) has continued to revise its estimates of Fukushima accident costs upwards, with their latest figure from March 2019 of between between 35 trillion yen and 81 trillion yen ($315 billion and $728 billion).  The government estimate by the Ministry of Economy, Trade and Industry, was much lower, though still a sobering 22 trillion yen ($198 billion).

The low-end of the JCER range involves entombing the most damaged plant (Fukushima 1) in concrete and releasing radioactive water to the sea.  The higher cost part of these estimate range includes treating contaminated water and soil.  The implications for nuclear liability coverage globally is instructive:

  • The US liability system under the Price-Anderson Act, which provides the largest pool of insurance for nuclear accident damage in the world, will generate a gross insurance pool of less than $12.5 billion.1  This assumes all reactors will be able to pay in their retrospective premium in full, and ignores the facts that payments are made over roughly six years so have a significantly lower value in present value terms, and that the pool shrinks as reactors close.
  • Even assuming the official Japanese government estimate (i.e., the lowest one) for Fukushima cleanup costs is accurate, liabilities in Japan exceed the maximum US insurance pool by more than 15x.

The implication is that any moderately-sized nuclear accident in the US will quickly exhaust the available insurance coverage and taxpayers -- rather than reactor owners or their insurers -- will shoulder most of the burden.  Because they are too low, the Price-Anderson caps on US reactor accident liability provide a recurring and significant subsidy to reactor operations.

What about Japan?  As of 2016, the government of Japan had lent more than $120 billion to TEPCO, all of it interest free.  The TEPCO long-term borrowing costs listed in financial statements around that time are less than 1 percent.  The values are so low for a firm in distress, strongly suggesting that they are skewed by the subsidized government loans themselves. 

A better proxy is the weighted average cost of capital (WACC) for Japanese firms -- though even those values are likely also to be too low given that TEPCO would be deemed extremely high risk or insolvent absent the government bailouts.

Data from PWC for 2015 provides some benchmarks, with an average WACC of 5.6% for the JPX-400 index of stocks;  6.7% for the electric sector; and 3.0% for the power sector.  The capital subsidy from this loan alone is between $3.6 and $8.0 billion per year.  Normally, interest would compound, resulting in even larger subsidies over time.  Even without compounding, the capital subsidy alone provided by Japan to TEPCO exceeds the total value of the nuclear accident insurance pool in the US every 1.5 to 4 years. 

3.  Fukushima health effects, widely conflicting claims.  Staunch nuclear advocates such as Michael Schellenberger frequently discuss how safe they view nuclear power as being.  Often, a comparison is made to coal.  It is true that the coal fuel cycle does trigger massive numbers of deaths worldwide both through mining accidents and air pollution.  Future-looking comparisons need to look at natural gas and renewables more than coal, however, as coal investments are on a strong downward trajectory and existing plants continue to close. 

Further, the concerns with nuclear are primarily linked to worries about large incidents caused either by accidents at civilian reactor or fuel cycle facilities; or from military activities resulting from proliferation leaking from civilian activities.  This is in contrast to other fuel cycles where deaths associated with extraction and emissions dominate.

Schellenberger argues that even with catastrophic reactor events such as Fukushima, the accidents ain't no biggie.  In an article he wrote for Forbes last month, he says that no radiation-related deaths have been linked to the Fukushima accident.  He argues that health impacts from radiation have been overstated, with cancer incidence even from the Hiroshima and Nagasaki bombs much lower than predicted, the implication being that the impacts of reactor incidents will be smaller still.  And he notes that a financial settlement on litigation involving a worker who died from lung cancer was a political settlement, not one based on good science, and his illness was unrelated to the reactor. 

Tilman Ruff of International Physicians for the Prevention of Nuclear War (and who also wrote a chapter in the Learning from Fukushima book) has a very different take:

By 2017, a total of 40,000 workers had been involved in the extensive decommissioning work which will be required for many decades.  About 8000 work at any one time. Over 90% of these are subcontractors, who have poorer training and conditions and receive on average more than twice the radiation exposure compared with TEPCO employees. Maximum exposures for subcontractors in Jan 2018 were documented at over 10 mSv/month. Thus far 5 cases of cancer among clean-up workers have been officially recognised as occupationally-related – including 3 cases of leukemia, one thyroid cancer, and 1 case of lung cancer...

By Sep 2018, the Japan Reconstruction Agency identified 2202 deaths as related to the nuclear disaster – principally through suicide and interrupted or diminished medical care. However comprehensive long-term prospective mechanisms linked to radiation exposure have not been established to monitor population health impacts of the nuclear disaster. If you don’t look, you won’t find. Given the fragmented and incomplete nature of cancer registries in Japan, it is quite possible that health effects would not be detected.

Tracking of thyroid cancers should have been an area with strong and consistent data, Ruff notes, but it is not being properly tracked by the Japanese government.  Monitoring of animals and plants in the accident area is indicating effects from the radiation as well.

So either the Fukushima accident has caused no cancer deaths, but many deaths from ill-advised evacuation orders by the government in 2011 triggering dislocation, stress, and loneliness; or cancer effects are not so small and yet government actions are pushing people back into radioactive areas that are still not safe and failing, whether on purpose or not, to track critical health data.  Given a lack of transparency on past nuclear sector problems in Japan, I put more faith in Ruff's assessment of the situation.

  • 1. Coverage requirements are periodically adjusted for inflation, and decline as reactors are retired. This estimate reflects mandated coverages and reactor counts as of November 2018.

In one of their recent blog posts, the Nuclear Information Resource Service (NIRS) called attention to the apparently amazing efficiency of Chinese new build nuclear reactors.  Two new units are being built at a stated cost of roughly $2.5 billion each, significantly less expensive than what it would cost for similar projects in other countries (access the NIRS posting here).  Plant Vogtle in the US State of Georgia, for example, will cost well more than $8 billion for each of two planned new reactors.  Olkiluoto-3 in Finland is now estimated to cost nearly $12 billion (8.5 billion euros).

The Western and Chinese cost figures are literally worlds apart.  Yet the core question with most heavy industry in China is what drives this difference.  What proportion of the cost differential is due to real competitive advantages, and what proportion is the result of simple, old fashioned industrial subsdization policies?

The US nuclear industry likes to point to Asian costs as the "real" economics of nuclear builds.  They frequently blame nuclear opponents for their economic troubles, arguing that irrational opposition drives up costs by elevating safety standards and plant delays unnecessarily.  If only they could just get on with their work, the argument goes, new reactors would be very inexpensive even in the United States.

Such arguments are actually unhelpful, even to the industry itself.  They deflect attention from real structural problems in the way they operate and manage risk, and the sooner they fix those problems, the sooner they have a chance of a competitive market offering without the coddling hand of government. 

The reality is that nuclear has a number of characteristics that make it difficult to build cheaply, regardless of where the plants are built.  The facilities are capital intensive and complex, a combination that often triggers significant construction delays and overruns well beyond the nuclear sector.  Think of big tunnel, highway, and building projects. Financing costs are high not only because the project is big, but also because investors worry a great deal about overruns, and therefore require more compensation for risk.  Interest compounds mightily because large investments take years to begin producing revenues even in the best of times.  This makes the levelized costs of power sensitive to financing costs and delays, as interest on a large investment base continues to compound during delays.   It also means that taxpayers and investors alike are right to favor smaller, quicker-to-build sources of energy or energy efficiency to very large nuclear plants.

Subsidies in the west: cheap money, shifting risks to ratepayers and taxpayers

In Western countries, the industry has focused on trying to get subsidized debt, both through using tax-exempt bonds and through government guarantees on borrowing.  The government loan guarantees subsidize nuclear projects in two ways:  by allowing higher risk projects to borrow at a lower interest rate (the "risk-free" Treasury rate rather than a rate reflective of the project itself); and by enabling them to use much more debt (which is less expensive than equity) than would be possible in a market transaction.  Purchase guarantees have also been popular in the UK.

The states currently developing new reactors have also employed favorable Construction Work in Progess surcharges that "avoid" interest charges altogether by forcing ratepayers to fund the interest costs upfront.  This works, at least for the utilities.  They even get to keep the money if the plant is never completed.  It's not such a great deal for ratepayers though, as can be seen in a $1.5 billion bill faced by customers in the cancellation of Duke Power's planned Levy reactors in Florida.

Chinese approach:  direct government ownership, very little transparency

Tapping into cheap sources of capital and CWIP has been effective in the US projects that continue to move forward.  But nuclear projects do still face market risk and scrutiny from bond underwriters.  Government subsidy programs are at least moderately visible, and subject to public challenge.  These factors help explain why the vast majority of US nuclear projects announced 7-10 years ago have been cancelled. 

In China, the state is directly involved with key parts of the supply chain and the scale of their interventions is murky at best.  While the World Nuclear Association is often a mere cheerleading squad for the expansion of nuclear power, they do have a good summary of the structure of the Chinese industry.  Many, many of the entities have substantial government involvement, sometimes with ownership shares by both national and provincial interests.  It is likely that somewhere in China, somebody knows how much money has been fronted by the government to each part of their nuclear infrastructure.  One should not expect that analysis ever to see the light of day, however.

Suffice it to say that if the Chinese government views nuclear power as a strategic industry (as it does), the costs they report on any part of the nuclear market need to be taken with a grain of salt.  How much capital is the government putting into not only the reactor projects directly, but into key components in the nuclear fuel chain and related infrastructure?  Are those funds entirely free or at terms more favorable than would be offered in an arms-length transaction?  When there are plant delays or changes in market conditions affecting the value of future power sales, are plant returns and costs adjusted appropriately, or are these costs merely socialized? 

Costs are kept low through other subsidy mechanisms as well:  state funded R&D; less expensive labor, part of which is likely due to fewer options for workers to unionize or organize; and woefully inadequate requirements for accident liability (roughly $45 million --see page 809) -- the losses from Fukushima are $500 billion and counting.  Additional support likely comes through how Chinese nuclear waste is managed and government involvement with uranium mining and enrichment.  Whether Chinese nuclear is really less expensive than reactors in the West, or simply more heavily subsidized, remains an open question.

This challenge is not unique to nuclear investment:  it is very difficult to get data on how China subsidizes any of its energy industries.  This was a clear conclusion of our review of data sources on Chinese fossil fuel subsidies a few years ago.  The trade cases on solar probably identify many of the instruments the government is also using in the nuclear sector to subsidize its investments, though doing a detailed exposition to the nuclear sector would be a fair bit of work.  One interesting aspect that I'd expect to emerge from such a comparison is that the same exact subsidy instrument would turn out to be more valuable to nuclear new build than when it is applied to renewable energy.  This is because the market risks of nuclear are higher (so the difference between subsidized and market interest rates would be bigger on a nuclear project) and the scale of construction is larger as well (so any subsidy will apply to a larger base of money, and often remain in place over a longer period of time).