Skip to content

Is Taxing Carbon Really the Best Way to Decarbonize the Grid?

Our new research suggests clean electricity standards or subsidies might be more beneficial.

Today’s blog post is co-authored with Ryan Kellogg.

Some economists and policy advocates present a carbon tax (or its close sibling, cap and trade) as the singular answer to climate change, but it’s been clear all along that pricing carbon is not a complete solution. It does little to address innovation, which will be critical to tackling global emissions, particularly from lower-income countries. And it creates large income transfers that in some cases harm the least affluent, and in other cases harm politically powerful market participants who will throw their weight against it. Variations of these critiques can be leveled at alternative policies as well.

But until we wrote our recent paper, Carbon Pricing, Clean Electricity Standards, and Clean Electricity Subsidies on the Path to Zero Emissions, we at least thought that what sets a carbon tax apart is that it is the most economically efficient intervention for decarbonizing a market. Our new paper demonstrates that even this common belief about the benefits of a carbon tax doesn’t always hold, and may not hold in one of the most important industries, electricity.

OK, economics-minded readers, hear us out.  Our finding is not some strange mathematical case that would never occur in the real world or something that depends on quirky beliefs about how people behave.  


In electricity, the alternative climate policies on the table are some form of mandating zero-emissions generation shares (typically called something like a “clean electricity standard” or CES) or subsidies for zero-carbon generation (such as investment and production tax credits).  Among these choices, there are three standard arguments for carbon pricing.

  1. Carbon pricing doesn’t just favor zero-carbon sources over others, it also penalizes extremely-high carbon-emitting sources (e.g., coal) more than other carbon-emitting sources (e.g., natural gas generation). The other two policies promote zero-carbon, but don’t prioritize eliminating coal before gas or eliminating high-emitting gas plants before low-emitting plants.
  2. Carbon pricing adds an explicit cost to electricity, reflecting the true negative externality. A CES, by requiring that every megawatt-hour of dirty power be accompanied by a certain amount of clean power, effectively imposes a cross-subsidy from the dirty to the clean power. The net effect is a smaller overall cost increase than a carbon tax, or in some cases, a cost decrease. Zero-carbon subsidies paid out of the federal treasury just push prices down.  Only a carbon tax increases the cost of fossil generation by an amount that reflects the actual damage.
  3. A carbon tax raises revenue for the federal coffers without the economic distortions of an income or sales tax, and it could finance reduction of those other taxes. It does well for the federal budget while doing good for the planet. A CES, by design, just transfers funds from dirty to clean power without raising any money for the government. Subsidies for zero-carbon sources mean the government has to crank up other distorting taxes.

What we show in the paper is that the first argument probably doesn’t make a lot of difference to total emissions from electricity over the pathway to zero or near zero emissions. The second argument turns out to be a bug rather than a feature due to other distortions in retail electricity. And the third argument, well, that one remains a star in the carbon tax column.

Why doesn’t the carbon tax tendency to kill coal first set it apart? Because a CES or zero-emissions subsidies are also likely to do that. A carbon tax internalizes the GHG externality, so it incentivizes producers to first drop the sources that have the highest social cost, which would be coal. The other two policies tell producers to get rid of fossil plants but without the extra incentive to get rid of the big polluters. In that case, companies dump the most expensive plants first. But, if the private ongoing cost of running a plant is positively correlated with emissions, then the CES or clean subsidy policies also get rid of coal plants first.

We do an empirical analysis of all of the fossil plants in the US in 2019, with the energy prices of 2019, and find this correlation is strongly positive. At those prices, we show that a CES or zero-emissions subsidies would pick off existing fossil plants in almost the same order as a carbon tax. Over the full energy transition to near zero emissions, a carbon tax only saved a few percent of industry emissions more than the other policies.

That result holds, however, only if gas is reasonably inexpensive. You may have noticed that isn’t the case right now. But if we are really phasing out fossil fuels, natural gas is almost surely going to get cheap as its scarcity premium plummets. Maybe that’s why even with gas prices at $7/MMBTU today, the natural gas futures market is pointing to $4 in a few years. (Or maybe that’s just the belief there’s still loads of gas that can be fracked and burned on the road to planet cataclysm.)

The second argument goes up in smoke, so to speak, based on a paper Severin wrote with Jim Bushnell a couple years ago. It found that in much of the country – the places that are most effectively decarbonizing – retail electricity prices are already way above their full social cost, which includes the cost of pollution emissions. Those utilities still have to pay for grid infrastructure and other non-fuel costs. California is the poster child, as previous blog posts here have discussed at length, but any grid that decarbonizes is very likely to end up with retail electricity prices that are well above social marginal cost.

In those locations, driving up the retail price with a carbon tax would further over-price electricity. This is a particular problem if we think that “electrify everything” is the way that we take the biggest bite out of climate change. A CES would have less effect on retail electricity prices during the transition, though at full decarbonization, prices would be just as high under a CES as under carbon pricing since at that point both policies are zero-carbon mandates. Zero-emission subsidies, however, would lower wholesale electricity prices and thus would actually help get retail prices in line with social cost.

So, does “carbon taxes ain’t so efficient after all” carry over to gasoline and decarbonizing transportation? NO! The two key arguments that make a carbon tax problematic in electricity are points in its favor when it comes to cars. What sort of autos does a “clean car mandate” or EV subsidies crowd out? So far, at least, they look to be substituting for other small and medium sedans, rather than substituting for gas-slurping SUVs, muscle cars, and trucks. Is gasoline already overpriced? Another paper by Severin and Jim last year showed just the opposite: almost everywhere in the US, gasoline is priced below its social cost.  And gas being too cheap not only means EVs are less attractive, it also means that people drive more than they would if gas prices reflected the full social cost. The same logic applies to large industrial users of coal and natural gas, for which they pay well below social cost. 

For years, many in the environmental community have said that we need multiple arrows in the quiver to tackle climate change. Taxing carbon, mandating clean energy shares, and subsidizing clean energy each has pros and cons, from both economic and political perspectives. Our paper shows that the economic case for carbon pricing in the US electricity sector isn’t as strong as we and many others once thought. But it still has an important place in the quiver—especially for decarbonizing transportation and industrial energy—and implementing any of these three policies would be far better than sitting on our hands and doing nothing.

If you would like to hear more about this research or discuss it with the authors, please join us for a webinar this Thursday, July 28, at 9:30 AM PDT.

On Twitter, follow Severin @BorensteinS and Ryan @RyanMKellogg

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

Suggested citation: Borenstein, Severin and Kellogg, Ryan, “Is Taxing Carbon Really the Best Way to Decarbonize the Grid?”,  Energy Institute Blog, UC Berkeley, July 25,  2022,

Severin Borenstein View All

Severin Borenstein is Professor of the Graduate School in the Economic Analysis and Policy Group at the Haas School of Business and Faculty Director of the Energy Institute at Haas. He received his A.B. from U.C. Berkeley and Ph.D. in Economics from M.I.T. His research focuses on the economics of renewable energy, economic policies for reducing greenhouse gases, and alternative models of retail electricity pricing. Borenstein is also a research associate of the National Bureau of Economic Research in Cambridge, MA. He served on the Board of Governors of the California Power Exchange from 1997 to 2003. During 1999-2000, he was a member of the California Attorney General's Gasoline Price Task Force. In 2012-13, he served on the Emissions Market Assessment Committee, which advised the California Air Resources Board on the operation of California’s Cap and Trade market for greenhouse gases. In 2014, he was appointed to the California Energy Commission’s Petroleum Market Advisory Committee, which he chaired from 2015 until the Committee was dissolved in 2017. From 2015-2020, he served on the Advisory Council of the Bay Area Air Quality Management District. Since 2019, he has been a member of the Governing Board of the California Independent System Operator.

13 thoughts on “Is Taxing Carbon Really the Best Way to Decarbonize the Grid? Leave a comment

  1. Multiple good comments.
    I am an engineer and 40+ year power industry senior manager. The complexity of our decarbonization transition renders it perhaps the most difficult societal transformation we have ever attempted. Our current politically-driven approach is an utter failure to date – global GHG emissions are increasing, not decreasing. The notion that politicians can optimally and accurately choose appropriate technokogies in the appropriate timeframes is beyond ludicrous.
    A carbon tax is technology neutral, and most importantly can be designed to reach across the world to China, India, and other countries that are rapidly increasing emissions. For every ton of CO2 emissions the U.S. and Europe have reduced in the past 5 years, China and India appear to have added 4 to 5 tons. Until we address global emissions, we are just whistling in the dark.

  2. You do not seem to consider a carbon fee and dividend on carbon emissions. This solves many of the social problems as low-income people may well be financially better off with this system, see:
    As to a carbon tax increasing government revenue, this is only useful if the money is spent on efficient cost-effective projects, which is not guaranteed. Those of us who live through the pre-proposition 13 period in California remember a huge increase in government revenue with no obvious benefit. As it is California is one of the highest taxed states but still doesn’t manage to spend as much on, for example, schools as other states. Carbon tax revenue might well be wasted on ridiculous projects like high-speed rail that are not cost effective at reducing carbon. Hence at least a fee and dividend would not penalize low-income people, while providing a clear market message on reducing carbon emissions. It is also more politically feasible than a carbon tax that would just go into government revenue.

    • Paul
      The paper on the blog references on the inability to fully compensate economic losers with a carbon tax is based on the Pareto optimality efficiency criterion, not distributional considerations that support the fee and dividend analysis. These are different considerations. However, you raise a point that economists too often focus solely on efficiency, even when that criteria is applied inappropriately. For example, the critique in the referenced paper also implicitly highlights that the Kaldor-Hicks criterion that is an underlying premise in support of standard benefit-cost analysis violates standard efficiency criteria. K-H should be reevaluated as an ironclad metric used by economists in the way that it is. In fact the BCA approach is a distributional analysis in which the distributional changes are rarely reported and never evaluated against some standard. Politicians and other stakeholders are really much more interested in the distributional impacts than whether a choice is the most efficient. Witness the debate over the rooftop solar NEM tariff–it only got traction when NEM opponents put forward claims of large wealth transfers from lower income ratepayers to higher income ones.

  3. A good look at the complexity the electricity industry that creates so many market distortions that the standard economic principles and solutions don’t fit well. Another is that the various reliability requirements blunt any scarcity pricing premium and suppress the apparent marginal cost that should be showing up in the organized wholesale markets under standard theory. More studies that delve into this divergence between standard economic paradigms and what’s actually happening in the utility industry that has accreted many institutional and engineering structures over the decades.

    “You know it’s said that an economist is a man who, when he finds something works in practice, wonders if it works in theory.” – Walter Heller, Chair of the Council of Economic Advisers, 1961–64.

    (BTW, you should encourage Chris Knittel to finish his analysis of how gasoline prices drove fuel efficiency rather than CAFE standards until the price collapse in 1984. He had a side analysis on the issue related to his paper on the trade off in fuel economy vs. horsepower. That’s the sort evidence that might get more attention for a carbon gas tax.)

  4. There was a time back in the 80s and 90s where coal and nuclear power had about the same bus bar energy cost. Bumping up the cost of coal a bit would allow nuclear to be a bit lower cost thus driving the decision to build nuclear instead of coal. Coal dropped below nuclear and TVA built more coal plants and stopped their nuclear building program. Meanwhile the rest of the nuclear industry ran in to a lot of problems and the NRC tightened up and opposition to nuclear increased. The coal industry took off big time with the demise of nuclear. In hindsight this was a climate change disaster. Today old coal plants are being phased out and only a few new ones are being built. The darling energy source is natural gas. Putting a tax on natural gas isn’t going to change the dispatch of gas plants one bit. A tax on CO2 will just hurt gasoline prices; something we have witnessed is not a practical way to lower gas powered cars emissions in the short term. Its going to take a lot of EVs on the road to displace gasoline powered cars. Incentives for EVs is better than taxing ICE cars. We need a way to replace gas powered generation on the grid. I have some scenarios posted here for how to do that, especially case 22C.

  5. I would draw several different conclusions from the same data.

    First, however, I will reiterate my criticism of the Borenstein / Bushnell paper concluding that electricity is priced above its social marginal cost in most places. That paper effectively treats marginal transmission and distribution costs as near-zero — less than 1 cent/kWh. I think that is utterly flawed, due to a focus on short-run marginal costs in a capital-intensive industry.

    If one substitutes long-run marginal costs for transmission (about 4 cents/kWh) and for distribution (about 3 cents/kWh), the “Electricity” map above shows that nearly every part of the country has retail rates that are at or below societal long-run marginal costs. Only New England, California, and one utility in Colorado (those shown in dark blue on the Borenstein/Bushnell map) remain “above” societal marginal costs. New England and California are expensive places to do business (they also have high gasoline prices, high housing costs, high labor costs, and high state taxes). The Colorado utility is a very rural cooperative with very high distribution costs.

    The first conclusion I reach is that, yes, a bunch of subsidies for “good” things can work about the same as a tax on “bad” things to promote utility investment in clean energy. But you cannot know in advance all of the good things. In particular, energy efficiency investments BENEFIT from a carbon tax, but setting up subsidies to encourage them efficiently is fiendishly difficult, because the technology evolves so quickly. Subsidizing good stuff with money from elsewhere suppresses the price of electricity (something that some equity advocates like), but suppressing the price of electricity with hidden subsidies means that energy efficiency is competing with a subsidized product.

    This can be solved by requiring all load-serving entities to budget a minimum of 5% of revenues for energy efficiency investments ( a level that RAP and ACEEE found adequate to fund all cost-effective efficiency about a decade ago). That internalizes the subsidies within utility rates, but compounds the problem that Borenstein/Bushnell assert in their map full of blue (the map I think if fundamentally flawed.)

    Which brings us to an interesting quandary: if Borenstein/Bushnell stand by the “rates are too high” mantra of their previous paper, and we need to subsidize all the good stuff, that subsidy money (a uniform system benefit charge) drives rates up further. If it is collected in a non-volumetric fashion, it then falls heavily on those who have invested in energy efficiency and on-site generation, certainly not an equitable result.

    My conclusion from this is that carbon pricing is necessary, but not sufficient. We need carbon pricing to level the playing field between energy efficiency, on-site renewable generation, central station renewable generation, and fossil fuel resources. But that does not really speed innovation in alternatives. We can solve that by investing the carbon tax revenues in energy research, development, demonstration, and market transformation.

    We call that “Cap and Invest.” You set a cap on carbon pollution. You impose a tax on emissions. You invest that tax in energy technology and energy efficiency. The combination of the carbon tax (pushing dirty resources OUT) and the investment (pulling clean resources IN) can be tuned to meet the cap.

    Which is pretty close to what has been done by the electric utilities in the Pacific Northwest for two decades, with a uniform system benefit charge funding energy efficiency and utility revenue funding for the Northwest Energy Efficiency Alliance to develop and commercialize new efficiency technology. Combined with a Renewable Portfolio Standard to pull in clean generating resources.

    And it is almost exactly what the Washington State Clean Energy Transformation Act does, with a ban on coal after 2025, a GHG-neutral mandate by 2030, a 100% clean mandate by 2045, a carbon fee on excess emissions, a low-income energy assistance fund, a statutory mandate for utilities to “acquire all cost-effective energy efficiency” and other elements.

    But, without a fee on carbon, there is no money for the energy efficiency, for the RD&D, or the subsidies for clean energy technology.

    A carbon fee is necessary, but not sufficient.

  6. Here in New England people are already worried about their heating bills this winter. And I know people who cannot afford to gas up their cars. Yes I know get an EV but EVs run in the $60,000-$80,000 range. I read just yesterday GM is building a $300,000 Cadillac Escalade EV; I thought it was a joke, bit it’s no joke — and nobody is laughing. In all seriousness I don’t know how people will afford to heat their houses in Central New England this winter, especially if it’s a bad winter. Any Suggestions?

  7. This is an important argument and perhaps a basis for a paradigm shift for the most cost-effective way to reduce GHG emissions in the electricity sector. I look forward to the EI@H webinar this Thursday in which Severin and Ryan Kellogg will expound on this argument.

    In an EI@H blog post nine months ago by Bushnell, et al., (The CEPP is not a Clean Energy Standard), the authors highlighted the design challenges in establishing an effective CES through an analysis of the Biden Administration’s ill-fated CEPP proposal: “Economists usually complain that clean energy standards pay people to buy more clean energy, when what we really want is for them to buy less dirty energy. Most of the time, this distortion is small enough that it doesn’t create big inefficiencies. But by cranking up the “buy clean” incentive for supposedly new clean energy purchases, the CEPP takes an arguably modest incentive problem and transforms it into a really big problem.”

    Surely a lot of very smart people worked on the design of the CEPP. As the old saying goes, the devil is in the details, and that can — and should — send designers back to the drawing board. So we need to get this right — as subjective as that determination may be in the politically-charged public policy realm. But it seems to be worth the effort. Afterall, it has often been noted that the primary reason that CATP has not proved more effective in CA is because the carbon price is too low. Raising that price would no doubt improve the efficacy of the CATP but also further add to the cost of already over-priced electricity. A well-designed CES could be just as effective, and maybe more so — AND less costly. But I’m not sure what political gauntlet awaits a serious CES proposal. Hopefully we’ll find out.

  8. The argument against a carbon tax is backwards and “turns out to be a bug rather than a feature due to other distortions in retail electricity” The authors state that a carbon tax will not do the job because the cost of electricity is already too high. Leaving out the quibble about calculating the societal cost, the reason for the high cost is because we are NOT taxing for externalities but rather to pay for government programs that do not have much impact on incentives.

    That is, subsidizing low-income families or subsidizing the purchace of solar panels or insulation may well be good public policy, but it should not be part of setting rates. If we value these activities for their social benefits we should pay for them out of general taxes and not as a transfer from electric ratepayers-. Doing that creates the perverse incentives that lead to these conclusions that defy normal econcomics. The reason the situation is different for tranportation is that we don’t require that funding for mass transit be exactly paid for out of use taxes on cars and trucks.

    Carbon taxes ARE an important arrow in the quiver as they will set up the correct incentives, but only if we move the perverse incentives out first. The fact that they don’t make a lot of sense in the electricity market at the moment, speaks more to the inappropriate regulation in that market rather than to any fundamental issue.

  9. Thanks Severin – glad to see respected energy and climate economist say that perhaps carbon taxes are not a single cure for decarbonizing the electricity sector. You mentioned equity issues but did not get into them in the posting – how are the poor and renters to reduce their electricity use in response to higher electricity prices? And your point that governments can use the carbon tax revenue for programs to reduce carbon emissions – not sure if the record for that is particularly good – see uses of tobacco taxes and the priorities of Republican (or West Virginian) politicians. Another item not discussed is the uncertainty of how much of a reduction a carbon tax gets us versus the greater certainty of a CES. And I look forward to your further discussion on the elasticity of energy prices and the impact on consumption and changing investment decisions, whether electricity or gasoline. The energy burden for most households and businesses is very small, not a big driver to change behaviors unless considering low-income households, and again, for them, not much they can do to battle rising energy costs other than suffering. Having said all this, I’d take a carbon tax or a straightforward CES over the Clean Power Plan (which I worked on for years) – a rational Congress would dismiss the need for overly complex decarbonization mechanisms.

  10. Why not let the ‘free market’ do its job?

    Remove all policies and practices that support any technology over others. Separate social policy [clean] from taxation. Consider only the environmental impact and mitigation in approval of mining/ drilling. Factor in the risk of leaks. Set federal leases at fair market rates. Do not get involved in wars to assure oil supplies.

    Naive? Maybe. But probably more effective in the long run that tax and social policy contortions.

%d bloggers like this: