,
27 tweets,
7 min read
Read on Twitter
Did you see the claim that $200/t carbon tax would only cut a tiny impact on climate change, specifically 4%?
Well, there's a bit of "science news cycle" telephone going on... 1/n
Well, there's a bit of "science news cycle" telephone going on... 1/n
Let's start with the abstract.
Pretty clear, right? Actually, the details really matter...
Heal and Schlenker 2019 model the impact of a Pigouvian carbon tax on cumulative oil-associated CO2. Let's step through what this means term by term.
Pretty clear, right? Actually, the details really matter...
Heal and Schlenker 2019 model the impact of a Pigouvian carbon tax on cumulative oil-associated CO2. Let's step through what this means term by term.
First, let's focus on "oil".
We already know changing price of oil/gasoline has little impact on its use. The paper uses/cites elasticities of 0.2-0.8, meaning +1% price => -0.2-0.8% demand.
But even a tiny c-tax would kill a whole lot of coal. Huge impact on climate change.
We already know changing price of oil/gasoline has little impact on its use. The paper uses/cites elasticities of 0.2-0.8, meaning +1% price => -0.2-0.8% demand.
But even a tiny c-tax would kill a whole lot of coal. Huge impact on climate change.
Here's one estimate: $50/t cutting 2030 US emissions by 40% compared to 2005 (see how 97% of that is from non-transport reductions?)
Next, "cumulative".
Annual and cumulative use/emissions are very very different. Cumulative calculations require a lot more assumptions.
See Fig 2 in paper itself. It projects $50/t tax reduces 2020 oil use by 10%, $200/t by 30%. Now why does cumulative differ so much?
Annual and cumulative use/emissions are very very different. Cumulative calculations require a lot more assumptions.
See Fig 2 in paper itself. It projects $50/t tax reduces 2020 oil use by 10%, $200/t by 30%. Now why does cumulative differ so much?
Cumulative means counting all future oil use and its emissions. This paper builds on the green paradox literature, which claims that oil that is saved now (e.g. due to c-tax) will just get used later. In this case, in 2097-2110.
See those solid lines getting pushed out? Those are the scenarios where we haven't "run out of oil" by 2097 b/c we saved so much of it due to the high c-tax. So we keep burning, and very little is saved cumulatively.
Fig 4 shows the cumulative reduction. The c-tax stops working as effectively as the price adjusts to the final $250/bbl price, and then it has no effect at the end. In these projections, the 2097-2110 oil use erase most of the reductions from 2020-2097.
This is intertemporal leakage, also known as "green paradox". It assumes we will use up almost all the oil that costs less than $250/bbl, and a carbon tax only delays its use. But where the graph ends (when we use up all the oil&stop using oil) really depends on the backstop..
Short detour on elasticity before I talk backstop. The authors say it doesn't affect cumulative use/GHG, only timing, bc of green paradox. But see what elasticity of 0.9 (right) instead of 0.6 (left) does to projected trends? Much larger reductns. Eventual use pushed to 2120-2160
OK, "backstop"! Probably the most important assumption. Which makes it weird why they didn't do sensitivity analysis on it. The backstop is the price of the "most costly reserves" before substitution. Maximum price of oil before we stop using it, based on the Hotelling model.
@vsiv noticed R=250 was assumed to stay constant, but I think the problem is bigger than that. Even though the paper mentions renewables/storage in several places, there doesn't seem to be any calibration to RE/battery prices.
What is a perfect substitute to oil? Some argue we don't have any, only imperfect ones & high cost (electrification, or demand cuts e.g. efficiency, transit). But these seem critical to R=250 assumption. It determines "cumulative" oil use, particularly if how much we use in 2100.
The green paradox is controversial & there r strong assumptions/dependencies on backstop tech & substitutes. Authors need to test range of R, & also authors should clarify how R represents cost of oil substitute/alternative, e.g. electrification (or if it does / why it doesn't).
By making simple numeric conclusions about cumulative emissions, authors leave unsaid how that depends on what might happen in 2097-2160 and at $250/bbl for oil. The green paradox effect that drives this should be clearer in the abstract.
They decided to put numbers on $200/t and 4%, but didn't put numbers on "some of the initial reduction in consumption [2020-2097] will be offset through higher consumption later on [2097-2107]". This deserves part of the blame for the "science news cycle"/telephone effect.
Takeaways:
Qualitatively, true that c-tax won't hurt oil producers too much in short &perhaps longterm. For foreseeable future, oil will still be in high demand 4 transport,& there's lots of supply too. But c-tax will still kill a lot of GHG from coal,&if high enough, also natgas
Qualitatively, true that c-tax won't hurt oil producers too much in short &perhaps longterm. For foreseeable future, oil will still be in high demand 4 transport,& there's lots of supply too. But c-tax will still kill a lot of GHG from coal,&if high enough, also natgas
Quantitatively, please be more careful citing "$200/t => -4%!!!" as evidence of how weak a carbon tax will be / is. It seems there's a lot of intuition abt ineffectiveness of carbon tax that this article plays to.
There are many reasons why a carbon price isn't politically feasible or may be "ineffective" in the real world (while being the most economically efficient policy). But from what I gather, this article and its projections may not be showing what is being popularly believed.
See here for some of my brief thoughts on carbon pricing, cost-effectiveness, policy alternatives, and my current research: Now time to get off Twitter and back to that research!
also: "Some of the initial reduction" would be more accurately described as large amounts, high proportions, 50-90%+ ...
oops this was not quite a Pigouvian tax - it's not modeled/set at social optimal level (social cost of negative externality).
higher price-elasticity of oil demand would come from changing technologies, preferences, and behavior. past range of elasticity may not be indicative of future.
P.S. from policy point of view, we need all of it: carbon pricing, flexible regulations, R&DDD. They all play important roles in our fight to drive down GHG to zero ASAP.
oops, forgot to explain how cumulative quantity = area under annual quantity on the time graph.
I agree this was more of a modelling methods paper, so they should have been much more careful in drawing out specific numeric results and putting them prominently in the abstract
oops
