Lomborg’s Electric Dreams

Context The University of Western Australia has recently accepted A$4 million to set up an ‘Australian Consensus Centre’ under the auspices of Bjorn Lomborg (Update: Since writing this post, UWA has cancelled the contract, citing widespread staff opposition). Lomborg is a public intellectual known for developing a cost-benefit approach to major global challenges and producing results that often downplay the relative importance of action on climate change. He is, undoubtedly, a bête noire for some climate activists. But in the interest of benefit of the doubt, I decided to try to take an impartial look at his work. I will blog later on the methodology of his ‘Consensus Center’ approach. This blog is the result of a careful reading and ‘fact checking’ of a recent op-ed Lomborg wrote USA Today on electric cars.

Summary

In this article, Lomborg asserts that electric cars (EVs) are more trouble than they are worth economically and that it is ‘time to stop our green worship of the electric car’. His conclusions are based on a rough cost-benefit analysis, drawing on a number of recent academic studies. Much of the article is devoted to calculating the relative costs and benefits of a hypothetical 25 million new EVs in America by the year 2020. Lomborg’s blog asks us to ‘Get the Facts Straight’, so I have taken his advice and fact-checked his argument. The main claims of his article are as follow:

• Electric cars are not cost effective. Each car saves (by Lomborg’s calculation) $27 of CO2 emissions but costs $7,500 in subsidies. Lomborg produces a headline-grabbing figure of $188 billion in subsidies if there were 25 million new electric cars by 2020.
• Electric cars cause twice as many deaths from air pollution as traditional internal combustion (IC) cars.

The detailed fact checking is below. The highlights are:

• Lomborg uses uncorrected worst case estimates for the CO2 emissions of EVs, despite the fact that he demonstrably had access to the revised estimates which showed considerably lower emissions for EVs.
• Lomborg generates wild and utterly unjustifiable cost scenarios for subsidizing EVs.
• Lomborg reduces the benefits of EVs to the current (depressed) market value of the carbon saved but his discussion of the costs of EVs focuses includes wider social costs, including health impacts. The social benefit of the carbon saving is ignored.
• Lomborg appears to wilfully ignore evidence from one of the papers he cites that undermine his claims about the health risks of EVs.

Carbon Dioxide Emissions

First, let’s deal with the CO2 emission estimates. Lomborg produces figures to claim that an electric Nissan Leaf saves only 3 metric tonnes of CO2 emissions over its lifecycle compared with a comparable Mercedes A160 diesel. His source for this is a recent study – and subsequent corrigendum – in the Journal of Industrial Ecology (JIE). The emissions from running electric cars are dependent upon the method used to generate the electricity from which the car is charged. An electric car charged on electricity generated by solar power can approach zero CO2 emissions in its running; a car charged on electricity generated through coal accounts for significant CO2 emissions.

These are the extremes. In comparing the performance of electric versus IC cars, assumptions need to be made about the fuel used to generate the electricity to run the car. Lomborg, reasonably enough, states that his figures are based on the average US fuel mix. The study that Lomborg cites here was originally published in 2012, with a revised estimates published the following year. Lomborg links to the revised estimates. The Nissan electric vehicle (EV) and Mercedes internal combustion models are those used in the study. The study uses a lifecycle of 150,000km, so for a US audience, Lomborg’s figure of 90,000 miles is close enough. The study bases its estimates of CO2 emissions, not on the US fuel mix that Lomborg states but on the European fuel mix. Given that Lomborg was writing for a US audience (in USA Today), it seems reasonable that he should provide estimates based on the US fuel mix. Moreover, from the best data I could find, however, it turns out the carbon intensity of electricity production in the US and the EU are remarkably similar. For 2009 (the latest date I could find data for the US), the US Environmental Protection Agency estimates national CO­2 intensity for electricity production at 1,216 lb/MWh, while the equivalent estimate by the European Environment Agency is 1,230 lb/MWh. We can thus leave aside for the question of US versus European fuel mix. Now to the heart of the matter.

Table 1 provides the original and corrected estimates provided by the source that Lomborg cites. How does this compare with Lomborg’s figures? Lomborg states that the lifecycle CO2 emissions of a Nissan Leaf are 31 tonnes. The only plausible referent here is the high end of the uncorrected study, despite the fact that Lomborg’s own link is to the corrected study, so we know he had access to it. A more reasonable figure to have reported would have been the midpoint of the revised estimates: 27.75 tonnes. This represents a 20% reduction in emissions on the diesel comparison that Lomborg draws, and a 30% reduction on emissions compared to gasoline. Lomborg says EVs perform a ‘tiny bit better’ than diesel on emissions.  20%, by any interpretation, is rather better than ‘a tiny bit’; the 30% figure against gasoline cars might induce even a reticent Brit like me to use a phrase like ‘substantial reduction in emissions’.

Type of Engine	Version	Lifecycle CO2 emissions, g per km	Total lifecycle emissions, metric tonnes CO2
EV – Low estimate	Original	197	29.6
EV – High estimate	Original	206	30.9
IC – Diesel	Original	230	34.5
IC – Gasoline	Original	260	39.0
EV – Low estimate	Revised	180	27.0
EV – High estimate	Revised	190	28.5

The Cost of Carbon Dioxide Emissions

What about the value of that carbon saving, important for Lomborg’s cost-benefit type approach? Lomborg values his 3 tonne saving per EV at $27, based on the current value of carbon emissions on the European emissions trading market. Returning to his 25 million extra EVs scenario, he extrapolates from this a total benefit of around $0.5 billion, which he compares to his subsidy cost estimate of $188 billion.

There are two parts to Lomborg’s calculation. First, the amount of CO2 saved; secondly, the value of this saving. In terms of the amount saved, Lomborg postulates a total saving of 75 million tonnes, derived from the 3 tonne CO2 saving in lifecycle emissions from EV vehicles that Lomborg derives above. We have already seen how unrealistic that estimate is. The equivalent differential in the study Lomborg himself cites is between 10.5 tonnes and 12 tonnes per vehicle, producing, in his 25 million EV scenario, a total avoided CO2 of at least 260 million tonnes. The second part is the value of this saving on the market. Lomborg derives his estimates value from secondary trading of European Emission Allowances in CO2. Secondary trading is a market for European economic ‘installations’ to purchase the right to emit more carbon. Prices have been at around €5 (US$5.60) per tonne for much of the past year but increased to around €7 in 2015, which would give the half billion dollar value Lomborg puts on this carbon saving.

But even this represents a drastic reduction in prior years – largely blamed on the global recession – when prices were often around €20 per tonne. Moreover, a range of post-recession projections suggest the price will be at least €20 per tonne by 2020 (Lomborg’s date), some even as high as €40 per tonne. Using today’s prices, then, with our revised volume (260 million tonnes) and the cost per tonne on 31 April 2015 (€7.39 = US$ 8.30) gives a market value of US$ 2 billion for this saving, four times Lomborg’s estimate. Taking the lower end projections for 2020 of €20 per tonne gives an expected market value of over €5.2 billion, or US$5.8 billion at todays exchange rate. Ten times Lomborg’s estimate.

Even this, however, only takes into account the market value of carbon allowances. There are plenty of estimates around of the total social cost of carbon and given that Lomborg (see below) includes social costs such as air-pollution related deaths in his critique of electric cars, it seems reasonable to provide additional estimates for the total social benefit of the carbon saving from these 25 million EV cars.

There is a huge debate over the social cost of carbon and how best to estimate it. Estimate vary from tens of dollars to hundreds of dollars per tonne. The official estimates that the US government uses put the social cost of carbon in 2020 at $46 per tonne.[1] This figure brings the total value of the carbon saving in Lomborg’s scenario to around $12 billion, twenty times Lomborg’s estimate. A recent study at Stanford University put the value at $220 per tonne. This Stanford estimate would put the value of the carbon emissions saved at $57 billion. One hundred times Lomborg’s estimate.

The Cost of the Subsidy

Against Lomborg’s indefensibly low estimate of the value of the carbon saving of EVs, he postulates an equally indefensibly high estimate of the cost in the form of subsidies for electric cars: $188 billion in his 25 million EVs scenario. Lomborg claims that an extra 25 million EVs by the year 2020 would cost the US taxpayer $188 billion dollars because of a current subsidy in the form of a tax credit for new EVs of up to $7,500. There are two fundamental problems with this claim. Firstly, it is based on a nonsensical counterfactual. Secondly, even on current regulations it is woefully inaccurate. Let us take these two issues in turn.

Firstly, the US tax credit was brought in to encourage demand for electric cars. But if demand for 25 million of these cars existed by 2020, there is no reason why a government would continue that policy. There are currently around 7.5 million new passenger cars sold annually in the US. For there to be Lomborg’s 25 million new EVs on the road by 2020, this means that EVs would account for around two thirds of all new car sales over the next five years. To postulate this radical change in the car market but not to postulate a change in the subsidy regime is ludicrous. Lomborg does indeed caveat his claim with ‘at current subsidy rates’, but this is hardly a defence. It is a nonsensical counterfactual that serves only to generate a huge subsidy sum that can be apportioned to electric cars. Indeed, a much more plausible counterfactual for such a huge increase in demand for electric cars would be a reduction in their cost due to the returns to scale in the production of the vehicles.

Secondly, even with Lomborg’s ‘at current subsidy rates’ caveat, his claimed figure is deceptive. Firstly, the tax credit is up to $7,500, depending on the amount of income tax paid (as explained in the primer to which Lomborg himself links). [2] But more fundamental is the clear statement – in Lomborg’s own source – that the tax credit is, indeed, limited in scope. Specifically, the credit applies to 200,000 new cars per manufacturer, after which it is phased out. There are currently 18 eligible manufacturers listed on the US Department of Energy website. So even ‘at current subsidy rates’, the absolute maximum that would be subsidized before phase out is $27 billion dollars. And only then in the ludicrous scenario of: EVs accounting for two-thirds of all new car sales; all 18 manufacturers benefitting equally from this sudden growth; the US government not adjusting its subsidy policy in the light of such a growth in demand; and, all purchasers paying sufficient income tax to benefit from the entire subsidy.

Air Pollution

Lomborg’s final claim is that electric cars cause more non-CO2 air pollution than traditional cars; and, that in his 25 million EV scenario, this would cost $35 billion in excess mortality. His source here is a recent a PNAS article. This article does find higher air pollution from EVs charged from the grid or (especially) coal, although not from other charging sources. How Lomborg gets from this to the figure he cites is not clear, but my own attempt to derive an equivalent figure provides very similar estimates, so Lomborg seems on firm ground here. Let us continue to take the grid as the benchmark for charging. The PNAS article estimates for each category of car the impacts of that type account 10% of total projected vehicle miles in the US in 2020 (that is 388 billion miles). They estimate the annual air-pollution related mortality of gasoline cars at around 800 and grid average EVs at around 1300, a difference of around 500. Lomborg, however, is concerned with 25 million cars as items, rather than vehicle miles. We have been using a lifecycle estimate that EVs will drive 90,000 miles. 25 million EVs will hence drive around 2,250 billion miles in their combined lifetimes, costing proportionately about 2,900 extra lives from air pollution. Multiplied by the Value of a Statistical Life used in the paper of $10.1 million gives a monetarized health cost of $29.3 billion, close enough to Lomborg’s $35 billion.

While Lomborg’s extrapolation from this article seems reasonable enough, however, it still raises a number of questions. Firstly, why does he choose to focus only on this social cost of automobile use, rather than the wider social costs? As we have seen, incorporating estimates for the entire social cost of carbon increases the relative benefit of EVs by one or two orders of magnitude. Lomborg ignores this, but focuses on one area where EVs appear to do worse than gasoline and diesel. Secondly, by far and away the best performer in air quality estimates in the study that Lomborg cites, as well as in CO2 emissions from the previous study, is an EV charged off renewables. Yet the conclusion that Lomborg draws from his analysis is to dump electric cars altogether and invest in developing cleaner coal. This is a huge jump of logic, completely unjustified in his article.

Thirdly – and this is the real doozy – while the finding in the study Lomborg cites is important and cautionary, the JIE study that Lomborg relies on for his estimates of CO2 emissions also contains estimates for the relative impact on other forms of pollution.  One is left wondering why Lomborg chose to ignore these estimates from a paper that he knew and, presumably, found to be methodologically rigorous and instead solely reports findings from a different paper.  I am not saying he should not have reported the findings he did: they do seem to raise important questions.  But the question is why he chose to exclude the alternative estimates from the JIE paper. For reference, here are the estimated normalised impact scores for EVs on European fuel mix against diesel and gasoline from the JIE paper: the data that Lomborg chose to exclude.  On these estimates, EVs hold their own against diesel and gasoline cars in all forms of non-CO2 emissions, as well as vastly outperforming them on CO2 emissions (as discussed above).  One is forced to wonder why Lomborg chose not to report these estimates from a paper that he otherwise uses centrally to his calculations.

	EV – European Mix, Li-NCM battery	Diesel	Gasoline
Climate change	0.61	0.77	0.87
Terrestrial acidification	0.62	0.59	0.66
Particulate matter formation	0.67	0.64	0.65
Photochemical oxidant formation	0.58	0.94	0.84

NOTES [1] Based on the 3% discount rate, which the US uses. [2] As a tax credit it represents a reduction in income tax, not a rebate, which means that not everyone would be able to claim the full $7,500. Put simply, you have to pay more than $7,500 in income tax to claim the full amount. To pay $7,500 in federal income tax in 2015, you need a gross income of more than $57,000. The median household income in the US in 2013 was $52,000. Given that at the moment, electric cars are intuitively more likely to be purchased by more well off people, it seems likely that most would be able to claim the full credit. But, again, whether this is still a plausible assumption in a scenario where two thirds of all new car purchases are EVs seems highly doubtful.