Skip to content

Tax Me Baby One More Time!

What is the right carbon tax?

I contemplated a number of doom and gloom topics for my post this week. Under consideration were the $400 proposed rebate for each car owner (bad idea!), Germany’s refusal to stop importing Russian fossil fuels (was ist denn bitte los?) or the finalization of the latest fuel economy rules (yay-ish!). But since we do not talk about or recommend carbon taxes enough on this blog I will talk about a paper that made me think harder about what the level of an optimal carbon tax is. 


Quick refresher. Most things we do emit greenhouse gases and most of us use the atmosphere as a free dumping ground. Basic economics suggests that you should pay for the full cost of your wasteful activities. The optimal carbon tax is supposed to be set at a dollar/yuan/euro/rupee/peso amount that is equivalent to the damage each ton of carbon does at the optimal level of emissions. Or in simpler terms, if your factory dumps gunk into a lake that causes $100 of damages to everything in and everyone around the lake, the tax should be set at $100. Yes, dynamics, uncertainty blah blah blah, we can complicate this any way we want, but the question is, what is the actual dollar amount we should set the carbon tax to?

One candidate has been derived from the so called Social Cost of Carbon (SCC) literature, where we run abstract integrated assessment models and try to calculate the damages one ton emitted does over its lifetime to as many sectors as we can model. To make a long story short, the SCC for a ton of CO2 emitted in 2020 is $52 according to the Biden Administration’s latest estimates. But this is a number that is calculated based on a long list of strong assumptions. 

Another way to get a number is to ask the experts. There is a formal scientific technique, which is called expert elicitation. You basically figure out who knows a lot about a topic and you ask them well phrased questions. Drupp, Nesje and Schmidt did just that moments before the pandemic shut down normalcy. What do they do?

They used algorithms to construct a list of authors that have at least two pertinent and cited publications during this millennium. So nerds like me got surveyed, but it is worth noting that only about half of the respondents were people publishing in economics (which is neither flattery nor criticism). They nagged and nagged and nagged until they got a pretty impressive response rate. Their sample contains responses from all major continents, and the represented countries cover more than 80 percent of global CO2 emissions. Really cool. They then set out to tabulate their results. 

The exact first question asked was: 

“Suppose that a “world government” exists, which seeks to maximize the well-being of all present and future people and plans to implement a uniform global carbon price (measured in real US dollars per ton of CO2). Which carbon price would you recommend to the “world government” for the years 2020 [X], 2030 [X], and 2050 [X]? Which range of carbon prices would you still be comfortable with recommending for the years 2020 [X] – [X], 2030 [X] – [X], and 2050 [X] – [X]?”

The first three panels in the figure below show the distribution of the stated optimal carbon tax by the solicited experts for the year 2020, 2030 and 2050. The fourth panel zooms out (like on your iphone when trying to squeeze the whole family into your holiday picture) on the results for 2050 (look at the much larger scale on the vertical axis). 


There is a lot going on here. The dots indicate individual survey responses. The “X” indicates the average value, the horizontal black line the median, the vertical box the interquartile range and the blob is a kernel density of the survey responses to better visualize the distribution. If you read the paper carefully, the authors dropped a few implausibly high numbers from this figure, but report the full results in the appendix. 

The first thing to note is that the average recommended carbon tax in 2020 is $50, $92 in 2030 and $224 in 2050. The 2020 estimate is almost identical to the SCC that all Federal Benefit Cost Analysis uses. More on that later. The mean advised tax here, however, rises much more quickly than the SCC does for Federal Rules (e.g. the SCC for 2050 in the Federal Rules is $85). When we look at the median numbers, which are more “robust” to extreme expert responses, the numbers for 2020, 2030 and 2050 are $40, $70 and $100 respectively, which is much closer to the SCC numbers currently in use. 

The authors then slice and dice the data summarized in the following four (there are a few more in the paper) findings, which I am just throwing out there. 

  1. There is a strong consensus among experts that a uniform global carbon price should be higher than the existing global average carbon price (today it’s higher in some places, but essentially zero in most).
  2. While there is significant spread in the distributions of stated prices, most experts are in general agreement about short-and medium-term global carbon prices. What this suggests is that experts are not “all over the place”, but that the overall suggested ranges line up reasonably well. 
  3. This one is tricky to condense, but when asked to advise their national governments on what carbon price they should impose unilaterally (assuming that not everyone else does so too), they find that these tax recommendations are higher compared to the same expert’s recommended global carbon tax. This is not consistent with the notion of “free-riding” by one’s own government, where one would expect a lower price. 
  4. When the experts were asked to consider a scenario where border carbon adjustments were introduced (meaning if you impose a carbon tax domestically and other countries want to sell into your market, they have to pay your carbon tax that you collect), this resulted in higher unilateral carbon price recommendations. This is consistent with an expectation that other governments will free ride.

The paper goes on to unpack the survey responses in an impressively large number of ways, but the one thing most policy makers will focus on is the $50 number – without a doubt. Some will find it way too low (hej Sverige!) and others will find it unacceptably large as it is not zero (howdy Texas!). That does not worry me at all. It is very nice to see what the consensus figure amongst responding experts across the world is. But assuming for a moment that I am such an expert and was asked that question, I would provide a number pretty close to the official SCC estimate provided by the Federal Government ($52). I would not be surprised if many of my peers would do something similar. We are awaiting, with bated breath, a new estimate of the SCC, which I am willing to bet my spouse’s EV on, is going to be higher. I would love for the authors to conduct a follow up study a little while after the release to see whether experts anchor their beliefs around these all important official figures. 

This is a very nice paper, but any expert elicitation is done conditional on the current state of knowledge. If that knowledge changes, it would be cool to see if that stated optimal carbon tax does as well. And yes, we should tax carbon everywhere.

Keep up with Energy Institute blogs, research, and events on Twitter @energyathaas.

Suggested citation: Auffhammer, Maximilian. “Tax Me Baby One More Time!” Energy Institute Blog, UC Berkeley, April 18, 2022,

Maximilian Auffhammer View All

Maximilian Auffhammer is the George Pardee Professor of International Sustainable Development at the University of California Berkeley. His fields of expertise are environmental and energy economics, with a specific focus on the impacts and regulation of climate change and air pollution.

16 thoughts on “Tax Me Baby One More Time! Leave a comment

  1. I like this study that breaks the marginal impact of each ton of carbon emitted by country and then sums them up to get a Global Social Cost of Carbon – estimated at $417:>

    From what I’ve seen, the $50-100 SCC estimate is not a proxy to measure the damage caused by marginal CO2 emissions, but the price of carbon needed to incentivize markets to accelerate decarbonization efforts at a rate needed to meet Paris Agreement goals:

  2. no need to make t so complicated – just tax the fuel extracted, oil, coal, gas etc. railroad over the coal senators. etc.
    people will yell and scream about competitiveness — so be it. Think of it as a form of ‘affirmative action’ ie paying back for past usage/ emissions. No need to negotiate internationally. If a country like China or India does not ‘behave’ just add x % tariff.

  3. As I have noted in earlier discussions, we should not be considering a tax on carbon until we have defined the amount of energy we will need in 2050 and how we are going to source that energy. I have done the projection and suggested that the energy need will be somewhere near 11.54 Quads. When this estimated was made in 2005, the cost was $44 billion a year for the next 45 years. This would be the energy needs of California with a population of about 60 million in 2050. Source: Energy Planning for Regional and National Needs: A Case Study – The California Forecast (2005 – 2050), Gary C .Matteson, The next Economics: Global Cases in Energy,Environment. and Climate Change. DOI 10.1007/987-1-4614-4972-0_5. Springer Science+Business Media New York 2013.

    Once we have agreed on the amount of energy, then we can chart the paths and the component costs for achieving this goal. To date, various paths have been charted for just electrical energy, but all fall very short of meeting the 2050 need for electricity. Presently, there is little agreement on what or how the non- electricity component is to be met.

    So, we have a lot of work to do before we go to the citizens of California and ask them to pay a carbon tax.

  4. I am increasingly skeptical that we can get to a low carbon/zero carbon future via a carbon tax – especially if one believes (as I now do) that there is little time to waste in moving away from fossil fuels in an attempt to keep global warming below the 2 deg Celsius increase in temperatures, as the IPCC and others strongly recommend. Part of my skepticism is simply that the concept of a carbon tax has been kicked around for many years without much to show for it and even if it were a ‘good idea’ in theory, how long would it take to optimize the tax level to provide the desired results? I have become more convinced that the time scale over which we need to act and that needed to put in place any sort of useful carbon tax are simply incompatible. Moreover, given the political and anti-science climate in this country, the political battles to enact a carbon tax (and messing around with trying to patch/fix the social, economic and environmental equity issues that would surely result) aren’t likely to be any easier that simply taking on the public policy questions straightforwardly through a mainly regulatory/tax incentive approach.

  5. Tax Carbon? Utilities want to “tax the Sun” with fees to homeowners that own solar panels that have no carbon footprint. This is how utilities will get the money to pay their carbon tax.😕

  6. An interesting analysis but incomplete without determining where the tax will be spent. The most cost-efficient way of reducing carbon is to invest the money in the cheapest way of reducing carbon emission. This means using the money to fund the lowest cost reduction technology using cost per tonne of carbon reduction. Unfortunately, so far some of the money has been wasted on unproductive pork barrel projects. Consider that California’s high speed rail project. This has so far used some of the cap-and-trade carbon money to pour huge amounts of concrete producing huge amounts of carbon. This for a project that has no realistic chance of every being finished or cost effective. So, the California carbon tax is being used to emit even more carbon! This is why the carbon fee and dividend is more realistic. This returns any carbon fee equally to every taxpayer. This is not only more politically viable and socially equitable, but realistically the most cost-effective way of reducing carbon emissions.

  7. In regard to point 3 in the article: “tax recommendations are higher compared to the same expert’s recommended global carbon tax.” I would suggest that is because of the nature of the experts questioned. I would expect most are academics whose focus is on global warming and press for change in their country. I do not think that position would be supported by the majority in many / most locals.

    In the US, I would posit that one side of the aisle would be very much in favor of the higher price and the other would be looking for a lower price. Would you like to guess which side of the aisle most of the academics would sit on?

  8. To what extent are social cost submissions influenced by other authors? The authors each have two pertinent papers – curious if their papers reference one another (could show relationships through some sort of network analysis – i.e. “low” social cost submission authors have higher network “edge” connectivity amongst other “low” social cost submission authors)? Seems like a good way to filter interdependence of this information –

  9. The figures cited in this blog — $50 – $224/ton are remarkably similar to:

    a) the estimates published by the Carbon Tax Center a decade ago (; and
    b) the estimates prepared by the Congressional Budget Office in their analysis of where the cap-and-trade mechanism in the Waxman-Markey legislation would lead a decade and a half ago.
    c) the carbon tax actually proposed by Congressman Bob Inglis (R, SC) in his carbon tax bill, introduced in 2009.

    It seems that if we ask the same question of careful analysts, we pretty consistently get a similar answer.

    With California Carbon Allowances now selling for $31/tonne, we’ve actually gotten into the right order of magnitude in that particular market.

    EU Carbon allowances are currently selling for 79 Euro, about $90/tonne. A level that this blog post seems to suggest is “about right.” But even at that price, the EU remains heavily dependent on oil and natural gas — from Russia (without love).

    Any carbon price above $30/ton will quickly remove coal from the electric generating mix — pretty much as soon as wind and solar can be built, and, for central-station resources, the associated very expensive transmission lines sited and built, and storage systems and demand response mechanisms put in place.

    But even $100/tonne will do relatively little to address automotive carbon emissions. That’s a buck a gallon. The price of gasoline went up a buck a gallon two months ago, and gasoline consumption dropped only a few percent.

    Yes, as people replace their vehicles, they will choose more efficient models, or electric cars (which, outside of California’s inflated electricity pricing, have MUCH lower fuel costs per mile). In the short run, a buck a gallon is about a 30% increase in the cost of fuel per mile. But the cost of fuel is only one-third of the cost of driving (two-thirds is interest, depreciation, insurance, tires, oil changes, repairs and the like). So even a buck a gallon is only a 10% increase in the cost of DRIVING. People will not easily give up their steel cocoon with satellite radio providing weather-protected point-to-point transportation to take the bus or ride their bike in an elastic response to a 10% change in the cost of driving.

    Now, if we could convert substantially ALL costs of driving to a variable cost (as smart rate design does for electricity), then people would be facing a variable charge of about $1/mile, and we might see some movement. See Todd Litman’s work from the Victoria Transport Policy Institute, for measurement of the total cost of driving. Litman includes 23 cost categories in his Transportation Benefit and Cost Analysis, totaling about a buck a mile — more in urban peak periods. Of these about one-third are internalized variable costs (on which many people decide how much to drive), one-third are internalized fixed costs, and one-third are external.

    Perhaps once we internalize the external costs (of which air pollution is a tiny fraction — subsidized parking, congestion, and land use impacts are much larger), and move to electric robotaxis providing transportation-as-a-service (TAAS), perhaps people will give up their ownership, and actually see all costs as variable.

%d bloggers like this: