Last week the Committee on Climate Change brought out new figures for its estimates of the 2020 and 2030 costs of nuclear and solar energy. The last time it produced numbers was in August 2011, just over four years ago. Given the huge differences between today and 2011, we shouldn’t be surprised that the CCC didn’t remind us of its earlier estimates. Or actually even mention the 2011 report. But because it has almost doubled its estimates for nuclear costs and more than halved them for PV, the changes need widespread discussion.

Estimates of costs per kilowatt hour for large scale ground-mounted PV

For 2020, the CCC has reduced its cost forecasts by almost half ( in the case of the 'low' estimates) and almost two thirds (the 'high' estimates). For PV in 2030, the low estimate has been cut by about 30% but the high estimate has gone by two thirds . The reason may be that its 'high' 2011 figure for 2040 is about twice the underlying cost of PV today.

Like almost everybody (including me) the CCC got its 2011 estimates for PV cost declines hopelessly, embarrassingly, laughably, wrong. There’s no shame in this and little point in trying to disguise it. Actually, it would be better to tell everybody and get a discussion going on why the 2011 figures turned out to so inaccurate.

Just in case nobody else wants to do this, I’ll try to start that discussion. The 2011 figures were derived from a really detailed piece of work from Mott MacDonald, the engineering consultancy. This research postulated a learning rate for PV of 22%. This means that every time the accumulated total global installations of PV double, the cost falls by 22%. Mott Macdonald made its forecasts using this number but assumed that there would only be 1.46 doublings by 2020 and 3.79 by 2040.

Those rates of growth probably looked ambitious at the time the work was done. As it turns out, the world has already seen about two and a half times more PV on the ground than Mott MacDonald forecast for 2020. This has driven costs down through learning effects. As importantly, the rates of interest charged by financiers are lower than projected. In capital intensive projects such as PV, this has a striking effect.

Nuclear

In 2011, the CCC projected nuclear costs of between 4.4 and 7.7 pence per kilowatt hour by 2020. These numbers were projected to fall to between 4.0 and 7.6 pence by 2030.

Unlike PV, nuclear estimates have sharply risen. Last week the CCC suggested figures for 2025, half way between 2020 and 2030. The low estimate was 7.8 pence, almost double the midpoint of the previous 2020/2030 figures. The high estimate is 10.2 pence, about 30% higher than in 2011 and far more than PV today.

Obvious conclusions

Expectations of PV costs have fallen precipitously since 2011. Nuclear’s have spiked. Nuclear is now - according to the CCC – more expensive than solar. I am sad that the CCC, a body with intellectual integrity, didn’t point this out.

I’ll also say that if cost trends continue, the CCC’s new figures for solar costs in 2020 and, particularly, in 2030 are far too conservative. There is overwhelming evidence that the learning curve in PV is continuing. A fall from 8.4p to 6.4p suggests little more than one doubling in accumulated total installations between 2020 and 2030. I’d be amazed if there were less than five accumulated doublings in this period, suggesting a resulting cost figure of less than half the CCC’s 2040 number.