Our friends at the Washington Business Alliance are giving a bit of push-back against carbon taxes:
Only a high carbon price, in excess of $50/tonne, will materially alter electricity generation given the dispatch order of plentiful, cheap coal and natural gas. Transportation fuels are relatively inelastic, similarly requiring a high price and long-term commitment to meaningfully impact emissions.
But is their analysis correct? The short answer is “Almost Certainly Not”. In a previous post I looked at transportation, but in this post I focus on electricity.
Because this is a long post, let me summarize by quoting the Brookings Institution’s Charles Frank:
In the United States, where the price of natural gas is low compared to most other countries, the price for CO2 emissions had to be about $5 or more in 2013 in order to tip the short-term balance in favor of shutting down coal. [Emphasis added.]
The summary of this post is that there’s a huge gulf between what WaBA believes (that you need $50 or more to “materially alter electricity generation”) and what just about everybody else on the planet believes (that a much lower number is correct, perhaps as low as the $5 cited by Charles Frank).
The most well-known energy model is the NEMS model used by the Energy Information Administration. Their latest projections (from the 2014 Annual Energy Outlook) show that a carbon price starting at only $25 per ton of CO2 and growing at 5 percent per year will drop electricity emissions by about one-third immediately and by about two-thirds by 2035.
Source: Energy Information Administration.
The EIA doesn’t model the effects of a carbon price great than $25—although surely the effects of a $50 carbon price would be significantly greater than those of a $25 carbon price—but another well-known national model does: the Haiku electricity sector model used by the Resources for the Future, a high-powered environmental economics think-tank. Their recent study (see Figure 1 on PDF page 15) concludes that a carbon price of about $60 would cut electricity emissions in half almost immediately.
In short, neither of these models provide any support for WaBA’s claim that “Only a high carbon price, in excess of $50/tonne, will materially alter electricity generation…”
“Okay,” you might say, “but those are models. What about empirical evidence?”
Thanks for asking. (And BTW, yes that is a sly reference to the media-shy Marshawn Lynch of the Seattle Seahawks!) It turns out that events over the last decade have provided some terrific evidence. Fracking and the “shale gas revolution” have dramatically reduced the cost difference between coal and natural gas, as shows in the green line below, which charts how much more expensive—per million BTUs—natural gas is than coal. (Note that natural gas plants are cheaper to build and can also command premium prices during “peak periods” because they’re easier to turn on and off; these advantages help offset the cost difference favoring coal throughout the period shown.)
The black line below shows that the amount of US electricity produced from coal during this time period has fallen by an impressive 20 percent. (Note that the relevant axis is on the right-hand-side and does not start at zero; the units are billions of mega-watt hours.) Other data—not shown, but also from the US Energy Information Administration—show that the drop in coal production is matched by an increase in natural gas production (plus smaller changes in renewables and the overall level of production).
Source: US Energy Information Administration, “Net Generation for Electric Utility” and “Receipts, Average Cost, and Quality of Fossil Fuels for the Electric Power Industry”
The bottom line is that a reduction in the natural gas price premium (natural gas minus coal, measured per million BTU) from over $5 to under $2.50 caused a huge shift in electricity production from coal to natural gas.
And here’s the kicker: adding a $50 carbon tax would have effectively eliminated the natural gas price premium. That’s because coal produces about 0.1 tons of CO2 per million BTU, and natural gas produces almost exactly half as much, so at $50 per ton the price of coal power would go up by $5 per million BTU and the price of natural gas power would go up by only half as much.
Now, I personally can’t say for sure what the effects would be of effectively eliminating (or perhaps even reversing!) the price premium for natural gas over coal, but the data shown above suggests they would be substantial. A strong case could also be made that a $50 carbon tax would allow renewables to make big gains over both coal and natural gas.
What I can say for sure is that the evidence presented above—both from models and from real life—casts tremendous doubt on WaBA’s claim that “Only a high carbon price, in excess of $50/tonne, will materially alter electricity generation given the dispatch order of plentiful, cheap coal and natural gas.”
My compatriots at Carbon Washington asked WaBA on Twitter for their sources (memo to WaBA: it would be helpful if you included citations in your posts), and here’s what they said:
@carbonwa Carbon price dispatch order chart; Convos w/ WA utility pros; Resource levelized cost; Specific plant $/kWh pic.twitter.com/a6ZMmKYUx8
— WA Business Alliance (@WABizAlliance) December 8, 2014
Let’s focus on the primary citation, the “carbon price dispatch order chart.” It comes from an article published in the utility journal Fortnightly. It’s not even clear that the chart was intended to be accurate—it’s a hypothetical “simplified three-plant example” used to demonstrate how carbon pricing affects different generators— but even more important is that the article was published in 2007. So the chart ignores whatever technological progress has been made in the intervening years, and—as shown by the second figure above—it ignores the huge drop in natural gas prices (and the natural gas price premium) produced by the shale gas revolution. As the article itself states in a discussion of the electricity market in Texas (see also Figure 4 in the article): “When gas is selling for around $8/MMBtu [million BTU], even a CO2 value of $40/ton produces little emissions reduction. On the other hand, with gas at $4/MMBtu, a $20/ton CO2 price could generate a 20-percent reduction in emission.” The price of gas when the article was written was about $8; the price now is about $4.
For an update on what this chart might look like now, let’s consider the latest EIA data on “Variable O&M (including fuel)” from Table 1 of this study. It shows coal at about $30 per megawatt-hour (MWh) and advanced combined cycle natural gas at about $45 per MWh. The difference is only $15 per MWh, which is 40 percent less than the $25 per MWh difference in the Fortnightly chart. Add in the impact of a $50 carbon tax (which the Fortnightly chart correctly implies would raise the price of coal by about $50 per MWh and the price of natural gas by only $25 per MWh) and you get natural gas coming in at $10 per MWh less than coal.
In short: a simplified hypothetical from an article written in 2007 does not provide convincing evidence that “Only a high carbon price, in excess of $50/tonne, will materially alter electricity generation given the dispatch order of plentiful, cheap coal and natural gas.”
The next reference in WaBA’s tweet was Twitter-speak for “conversations with Washington utility professionals”. We asked one such professional and here’s what he said:
I can’t dispute their claim because… of CTAM, the model they used to compute it. Their bottom-line numbers are consistent with my own results using CTAM.
Now, CTAM (for Carbon Tax Analysis Model) is the standard model for analyzing carbon pricing policies in Washington State; see the reference to CTAM in the section on “Taxing Carbon” in this WaBA post. CTAM is an Excel spreadsheet developed by Keibun Mori as his masters thesis at the UW Evans School of Public Affairs. (Keibun also worked on CTAM for the state Department of Commerce, and at some point Sightline ended up hosting the model on its website. I think that’s the version I used, but note that there are other versions and that CTAM is under revision.)
CTAM has one brilliant advantage that explains why it is the standard model in Washington State: it is incredibly tractable, i.e., it’s remarkably easy to understand the model and play around with it. This tractability explains why CTAM has been adapted in Oregon and Massachusetts and perhaps elsewhere, and it stands in marked contrast to the NEMS model used by the Energy Information Administration.
But tractability comes at a cost. Consider for example the treatment of electricity in the NEMS model:
The electricity market module (EMM) represents the generation, transmission, and pricing of electricity, subject to: delivered prices for coal, petroleum products, and natural gas; the cost of centralized generation from renewable fuels; macroeconomic variables for costs of capital and domestic investment; and electricity load shapes and demand. The submodules consist of capacity planning, fuel dispatching, finance and pricing, and load and demand (Figure 9). In addition, nonutility supply and electricity trade are represented in the fuel dispatching and capacity planning submodules. Nonutility generation from CHP and other facilities whose primary business is not electricity generation is represented in the demand and fuel supply modules. All other nonutility generation is represented in the EMM. The generation of electricity is accounted for in 15 supply regions (Figure 10), and fuel consumption is allocated to the 9 Census divisions.
For the record, here’s Figure 9:
Now consider the treatment of electricity in CTAM: there’s basically just one number.
That number (-0.46, in case you’re curious) is an estimate of what economists call a price elasticity, and it means that CTAM assumes that a 1 percent increase in the price of electricity produces a 0.46 percent reduction in electricity consumption in the long run. (The numbers can be scaled as desired, e.g., a 10 percent increase in price produces a 4.6 percent reduction in electricity consumption in the long run.) Reasonably enough, CTAM gets that number from a weighted average of long-run price elasticity estimates from the economics literature; and, since these are long-run estimates, CTAM phases in the elasticity over twenty years: -0.026 in year 1, -0.052 in year 2, etc. (Technical note: CTAM actually has three numbers, one each for residential, commercial, and industrial users; but the numbers are so close [-0.43, -0.47, and -0.49, respectively] that I’m taking the liberty of combining them to get -0.46.)
Point #1 is that this is pretty crazy. Notably, there’s nothing to account for fuel-switching from coal to natural gas. (CTAM does have a “Fuel Mix Change” assumption, but that is a jerry-rigged effort to account for fuel-switching from fossil fuels to renewables, not from coal to natural gas.)
Point #2 is that CTAM doesn’t even include the major source of electricity-related emissions in Washington State, namely the “carbon-by-wire” from coal-fired power plants in Montana and elsewhere.
Point #3 is that CTAM (for what it’s worth, which in my opinion is not much when it comes to electricity) actually does suggest that there will be a significant reduction in carbon emissions as a result of a $25 carbon tax. In the two figures below I’ve modified CTAM to model US electricity emissions with a carbon tax like the one considered by the EIA: $25 per ton CO2 and rising at 5 percent per year. The first figure below shows the CTAM results (in red) compared to a no-carbon-tax baseline (in blue); the second figure is an overlay of this figure with the EIA model results shows at the top of this post.
Source: Author’s calculations based on CTAM; Energy Information Administration.
In short: who knows exactly what kinds of modifications WaBA made to the CTAM model to get their results, but it’s unlikely to provide convincing evidence that “Only a high carbon price, in excess of $50/tonne, will materially alter electricity generation given the dispatch order of plentiful, cheap coal and natural gas.”
The final references in WaBA’s tweet are to the levelized cost of various types of generation and to the cost per kilowatt-hour of “specific plants”. These references are incredibly vague, but see for example what the EIA has to say about levelized cost; these are discussed above and there’s no evidence to support WaBA’s claim.
In conclusion: WaBA’s citations are very poor and WaBA’s conclusion is completely at odds with national modeling and with the empirical evidence of the past decade.
Talk to your friends and neighbors about I-732