How cheap is nuclear energy really?

My last post on residual risk was really a longish setup for this one: the “real” cost of nuclear energy. Nuclear incidents, whether Chernobyl or Fukushima, are the classic case of residual risk realized. In the Japan case the possibility of an earthquake and a tsunami at the same time was never really considered and the cost of cleaning up the Fukushima incident is now part of the real cost of nuclear energy in Japan, something that should have been in the economic calculation from the beginning.

How? Residual risk in economic terms is a so-called externality. Externalities are costs that occur when a product or service is produced which are borne by third parties, and thus not part of the price of the product – the cost is “external” to the product.

The textbook example of an externality is pollution, air, water, or ground. Pollution can have all manner of follow-on cost ranging from decreases in property value, rendering a region undesirable as a a vacationing spot, all the way to adverse health effects.

Externalities are market failures in capitalistic systems – the point of free markets is that prices are supposed to reflect cost to achieve optimal resource allocation. When they don’t, the justification for free markets starts breaking down, which is why criticisms of capitalism often point to externalities as examples.

In most cases, though, externalities can be repaired, i.e. internalized through market structural enforcement. That can happen e.g. via direct regulation (“200 tons of CO2 per factory per year allowed”) or taxation (“every kg of CO2 costs US$ x”). These two measures are considered sub-optimal by economic theory. The pollution itself suddenly has a set, fixed price, i.e. the cost to filter out any CO2 more than the 200 tons, or the tax itself. Economists prefer that pollution itself has a market price. That is the basis for “pollution certificates”: the regulatory body sets a region-wide upper limit for emission of pollution and issues certificates that allow a certain quantity of pollution. Producers can then buy and sell these certificates on markets, setting a price that can be compared to the cost of prevention measures.[1]

Let’s get back to nuclear energy. It is often described as “safe” and “clean”. Compared to burning coal, it certainly looks clean, but as mentioned above, history shows that it carries a residual risk.[2] For e.g. Fukushima, the residual risk led to real costs of cleanup that have been borne by society, and should have been internalized in the price of electricity. For residual risk, a way to do that is insurance.

In practice, nuclear plants do not have to be insured; in many countries they are state-owned or -controlled, and even when they are private, governments guarantee a (limited) indemnification in the event of an accident. But that means that as a political decision, the externality is never internalized, and nobody pays the real cost of nuclear energy. For that you would need an actual insurance premium paid.

But what insurance premium? In 2011, a team at the Insurance Institute of the University Leipzig in Germany (actually a research spinoff) presented a study to calculate exactly that.[3] They used several other studies to collect potential risks of operating a nuclear power plant and their probability on one side, and the economic damages cause by a nuclear incident on the other side. Then they laid out how a actuary would calculate the premium, which is tough here for several reasons:

- The base of insured objects is very small. Insurers rely on the the law of large numbers to even out their risk pools. This prevents one event from wiping out the whole portfolio, and premiums from becoming prohibitive.

- The potential damage is very high. The study uses conservative estimates, which do not include health damages or deaths, but purely evacuations, property damage and cleanup. Even using these estimates, a nuclear plant in 2011 would have needed to be insured at a sum of EUR 6 trillion.

- The risk is very low. It’s residual risk, after all. Even the most probable event (a successful physical terrorist attack) was calculated as happening on average once every 1000 years - but when in that millenium? An insurer has to gradually build a reserve fund, and with potential total damages in the trillion Euro range, that does take a bit of building.

To make a long study short: under a fairly conservative scenario, the cost of insuring nuclear energy in Germany would have been about EUR 0.14 Euro per kWh (US$ 0.20), effectively doubling the price of nuclear energy in 2011. I.e. that’s the (minimum) price of the residual risk; under less conservative scenarios, prices would rise all the way up to EUR 67 per kWh. And none of the scenarios include the non-incident-risk related cost like getting rid of the spent fuel.

That’s why I personally oppose nuclear energy: in an honest business case calculation, it’s uneconomical.

------------ Notes:

[1] The market price concern for pollution only applies if the threshold value is above zero. A regulation stating “no pollution with substance x allowed” still works. Also, for the sake of brevity I am omitting fun stuff from economic theory like the Coase Theorem

[2] As a cynic, I personally would say the best gauge of safety and cleanliness for any production facility is to mandate that upper management and their families relocate their home next to said facility and live there for a stretch of years. Their willingness to do that would be a great yardstick.

[3] As chance would have it, the study was published in April 2011 after the Fukushima incident, but started in January, so the authors couldn’t have known the timeliness. The study can be found here (German). It is a quite long, but a good read if you are interested in learning about a lot of basic actuarial concepts, and understand German English translation here – hope this one stays online.