On Monday this week, Matt Ridley published a blog in the Times claiming that the projected £30bn public subsidy for Hinkley C is too high a price to pay for the reductions in CO2 emissions that it will provide. The Times published my letter pointing out that he should not have used today’s social cost of CO2 in his calculations as Hinkley C will not come into service before 2025, and is planned to continue generating for 60 years. I said that the SCCO2 from 2050 would be more appropriate, which would increase the estimate of Hinkley C’s climate benefits. We then had an interesting but frustrating debate on Twitter, which ended when he quite reasonably pointed out that ‘Twitter [is] no good for such discussions’.
So here’s a first stab at the proper calculation that should be done. According to Matt Ridley, once it is operational, Hinkley C will reduce the UK’s emissions by about 14 million tonnes of CO2 per year. This reduction leads to a small but measurable easing of climate change impacts around the world – crop losses in Africa, floods in Asia, catastrophic melting of ice sheets, and so on. Assuming Hinkley C comes on line in 2025 and operates for 60 years, today’s net present value of the mean climate benefits from its operation are about £56bn. As far as I know no-one has actually done this calculation before. I can do it because I have an integrated assessment model, PAGE09, which is designed for just this kind of thing.
Of course, this estimate is hedged around with caveats. The PAGE09 model has a whole slew of uncertain inputs, from the transient climate response to the pure rate of time preference, which have a profound influence on the results, giving a 90% confidence interval for the benefits of £5bn – 148bn. The lower end of this range is in line with the results that Matt Ridley describes, but he doesn’t talk about the equally likely higher end, probably because he doesn’t personally believe climate change could be that bad.
The calculation also assumes greenhouse gas emissions around the world continue to rise in line with the IPCC’s A1B business as usual scenario, which is close to the path we are on at present. It also takes for granted that Hinkley C will actually prevent 14 million tonnes of CO2 emissions each year. This is far from certain. It’s the saving that comes from operating continuously at base load, displacing gas generation. All of these assumptions would need further consideration if this were an academic paper, or a policy briefing, rather than a quick blog post.
The UK government has said that it will delay the decision until early Autumn to carefully consider the case for Hinkley C. In a rational world, a better, more complete version of this calculation would be a central part of those considerations.
Details: Results are from 10,000 runs of v1.7 of the default PAGE09 model, converted from $US2005 in the year 2008 to £2016 in 2016 using the exchange rate in 2005 and the UK GDP deflator to get to £2016, and the mean consumption discount rate in PAGE09 of 2.8% per year to get to the year 2016.
The argument in your letter to the Times was that I should compare a present cost today of Hinkley subsidies with a mid-century cost of climate change, both per tonne of CO2.
But I don’t follow this because if I take the net present value today of the subsidy (which will be paid by consumers over 35 years from 2025 or 2033) then I should take the net present value today of the cost of climate change. If I take a mid-century cost of climate change, then I should compare it with a mid-century cost of Hinkley subsidy. So I don’t think you were comparing apples with apples.
And by reaching out to the mid-century SCC estimates, you are assuming the IAMs are reliable at forecasting warming many decades in advance. After their performance so far, I don’t share that confidence, and the lower estimates of sensitivity derived by almost all recent empirically based studies gives a very good reason why.
Incidentally, what value do you put on global greening in your model?
I gather one Whitehall civil servant who used to be in charge of this area once said to a friend: “Try as we might, we could never get an SCC higher than the policy costs, so… [fits of giggles] we stopped using SCC.”
Is that why I could find no mention of SCC anywhere on the Committee on Climate Change website?
I realised after our exchange on Twitter that it made much more sense for me to do the actual calculation properly, rather than trying to approximate it with an SCCO2 from any one year. So that’s what I’ve done. The £56bn here is the mean net present value in 2016 of the climate benefits of Hinkley C over its lifetime.
The empirically-based studies giving a lower climate sensitivity are nearly all of short time periods that do not include 2014-2016. It will be interesting to see what they imply once they are updated with the latest years.
Global greening is included in PAGE09 through benefits of climate change for small increases in global mean temperature.
I can’t really comment on your anecdote except to say that once you have decided what your CO2 reduction target is, the focus naturally shifts away from SCCO2 to the costs of meeting the target. But I would argue it still makes sense to keep calculating the SCCO2 so that you can check your targets continue to be justified as your knowledge changes. At the moment, the mean SCCO2 from PAGE09 is higher than the costs for many climate policies.
I have updated the Lewis and Curry (2015; online 2014) observationally-based climate sensitivity study to use data through to 2015. See https://judithcurry.com/2016/04/25/updated-climate-sensitivity-estimates/ . Including recent data does not change the estimate of transient climate response (TCR), which is the sensitivity measure most relevant to the SCCO2. The 0.01 C increase in the TCR estimate to 1.34 C is due to using a later version of the HadCRUT4 temperature dataset. The period used (1859-82 to 1995-2015) spans a long enough period to estimate TCR reasonably well. The TCR estimate using an alternative base period of 1930-50 is 1.33 C.
It makes no economic sense to set an carbon reduction target that does not properly (if at all) reflect the SCCO2 and is not varied to reflect unanticipated changes in the SCCO2. Surely you can see this.
You say “£56bn here is the mean net present value in 2016 of the climate benefits of Hinkley C over its lifetime.” May I ask how exactly you arrived at the £56bn figure, which looks high to me? Assuming it is derived using PAGE09, is a copy of all the relevant model input data and settings, and output, available somewhere?
‘It makes no economic sense to set an carbon reduction target that does not properly (if at all) reflect the SCCO2 and is not varied to reflect unanticipated changes in the SCCO2. Surely you can see this.’
I agree entirely, and made just this point in my reply to Matt Ridley’s comment.
The input data to the default PAGE09 model can be found in the supplementary materials to this paper http://link.springer.com/article/10.1007/s10584-012-0633-z
The modal TCR is 1.3, but with a long tail.
Thanks, Chris. I will study the material that you refer to.
Not in a million years. You’re dating yourself. Go back to school and take an Earth Science class. And a bit of ecology.
Does that mean that you agree that the letter in the Times was wrong then, because of failing to compare like with like? If so, I think the Times should be told so that a correction can be printed.
The key low-sensitivity studies use long periods, not short, as I understand them. That is certainly true of Lewis and Curry 2014, which used 1859-1892 and 1995-2011. See https://judithcurry.com/2014/09/24/lewis-and-curry-climate-sensitivity-uncertainty/
Re using SCC, I presume you agree with my point that the cost per tonne of CO2 mitigated is far too high for renewables, if not for Hinkley. Constable and Moroney calculated it as at least $380 per tonne for small solar, $274 for offshore wind, and $137 for offshore win, rising to $470/tonne offshore and $350/tonne onshore if system costs are taken into account.
I have to say I remain amazed by the indifference to cost so often on display in the climate change and renewable energy lobby.
No, the letter to the Times is correct as far as it goes. If you are going to use one year’s SCCO2 it should be one from the middle of the operating period, 2050 (or actually 2055, but the US EPA don’t give values for that year). But the calculation reported in this blog post is a much superior way to find the climate benefits, as it accounts for the full time profile of the emission reductions.
Simon de Jambon
I can see how bien pensants ended up supporting Mao and Stalin now
Thank you Chris for this fascinating and thought provoking post. Although I work here in the UK (at Leeds Uni), I was particularly interested because this analysis may also apply to a debate happening where I am from in Canada over a large hydro plant: https://sitecstatement.files.wordpress.com/2016/07/1-site-c-comparative-ghg-analysis-report-final.pdf
What I am interested in understanding better, is why you have used an IAM to do this analysis. (Forgive me if I am ignorant).
An IAM can give you cost/benefit estimates – but these are based on a damage function that’s one of the assumptions of the model. And the discount rates used are also endogenous (no?). Why use the IAM at all in this case when you could just pull out the damage function and discount rate assumptions?
In my mind it would make more sense to pose the question this way: there is a big cost to Hinkley but a carbon savings. At what combinations of carbon prices and discount rates are the value of carbon savings greater than the upfront costs.