There are several problems here. First, even if we agreed that government (as opposed to private) payments for tree-planting made sense, it doesn’t at all follow that the revenue should come from a carbon tax. In general, raising a dollar of revenue from a tax on carbon content hurts the economy more than raising a dollar from taxing labor or consumption. (My article on the “tax interaction effect” gives the economic intuition behind this point.) So if the government needs to raise $x billion in order to pay people to plant enough trees to strike a blow against harmful climate change, then there’s no reason to raise that $x billion through a carbon tax. It would be less harmful to the economy to raise that revenue using taxes that fall on a broader base, so that they caused fewer distortions to economic decisions.
Second, the current estimates of the “social cost of carbon,” upon which the dollar amount of carbon taxes is calibrated, would fly out the window if people began a massive tree-planting campaign (or other type of geo-engineering program). If we took the standard computer models and plugged in a scenario where humans plant enough trees (and/or engage in other geo-engineering techniques) to bring the atmospheric concentration of greenhouse gases back to 1980 levels by the year 2080, say, then in the context of that baseline the additional emission of a ton of CO2 today would have a negligible impact on human welfare.
This is from Robert P. Murphy, “Economist David R. Henderson Changes His Mind on Carbon Tax,” Institute for Energy Research, September 4, 2019.
Think of this piece as a contribution to the very animated discussion on my post and article about this issue two weeks ago.
READER COMMENTS
Rob Rawlings
Sep 8 2019 at 4:53pm
My understanding of the Tax Interaction Effect is that when a Pigovian tax is introduced into a situation with existing distortionary taxes then the optimal Pigovian tax will be lower than in a situation where there were no such existing taxes. In most situations the TIE will not fully eliminate the societal gains from the Pigovian tax especial if the revenue is used to reduce the most distortionary existing taxes.
So the TIE applied to a modern economy will reduce the optimal carbon tax level but will not as far as I can see mean that it will not still be the optimal tax to address carbon emissions. Set to the optimal (TIE-inclusive) level a carbon tax will eliminate the dead-weight loss of over-production of carbon with no other action required. The alternative suggestion that carbon should have no special tax and that carbon-reduction measures (such as tree-planting) be paid for out of general taxation seem to be sub-optimal for 2 reasons. 1) Carbon will continue to be overproduced and 2) the new tax to pay for carbon reduction will add additional dead-weight loss. If tree-planting is a cheap way to reduce carbon in the atmosphere then a Pigovian solution would be for a tree-planting subsidy (possibly paid for our of the revenue for the carbon tax ?).
I do agree however that if it is truly the case that carbon can be eliminated cheaply by tree-planting (and I note that most climate scientists do not seem fully convinced on the scalability of this option) then the social cost of carbon may be much less that it is currently widely accepted a being.
Rob Rawlings
Sep 8 2019 at 5:33pm
Stated more simply:
– A carbon tax set appropriately to take TIE in to account will internalize all relevant costs and lead to optimal carbon production with no need to plant additional trees. A deadweight loss has been eliminated.
– The alternative of a new general tax to fund additional tree planting will fail to internalize all relevant costs and lead to excessive carbon production which will be addressed by unnecessary tree planting. Deadweight losses have been introduced.
David Henderson
Sep 8 2019 at 5:58pm
You’re missing the tree insight. If tree planting is way cheaper than carbon reduction, the optimal carbon tax is zero.
Remember the original point: carbon causes a problem; the solution is not necessarily to reduce carbon if there are cheaper ways of addressing the problem.
Rob Rawlings
Sep 8 2019 at 6:38pm
I am indeed not getting ‘If tree planting is way cheaper than carbon reduction, the optimal carbon tax is zero.’.
I can see that the discovery of a cheaper way of fixing environmental damage caused by carbon production would reduce the cost of carbon production but as long as the non-internalized part of the cost remained above zero (or enough above zero to cover Pigovian tax collection costs) it still feels like a carbon tax to reduce carbon production above the optimal level would be more efficient that a general tax to address the damage caused by this excessive carbon production for the reasons I gave above. (TIE of course makes it more likely that the optimal level for the carbon tax is zero, in which no further action would be needed).
Jon Murphy
Sep 8 2019 at 7:09pm
Remember that the point of the tax is to reduce the quantity supplied of the bad in question (in this case, carbon); that is to say, reduce carbon in the atmosphere. The tax is not costless. If the same goal (reduction of the quantity supplied) can be achieved in a cheaper manner, than the tax, by definition, is not optimal. If it helps, the trees are a different Pigouvian solution.
Rob Rawlings
Sep 8 2019 at 7:23pm
Jon: Remember that the point of the tax is to reduce the quantity supplied of the bad in question (in this case, carbon); that is to say, reduce carbon in the atmosphere.
Yes, Clearly
Jon: The tax is not costless. If the same goal (reduction of the quantity supplied) can be achieved in a cheaper manner, than the tax, by definition, is not optimal
There are collection costs on the tax for sure. And (as Bob Murphy does a good job of pointing out) taxes are a political horror show. But at the level of pure theory what better way is there of reducing the supply of a bad (in a world of frictions and transaction costs) than taxing it so that its producers have to meet the full cost of the bad ? Pushing for non-Pigovian taxes to pay for things that (one hopes) will address the bad are both bad economics and heading in the direction of central planning.
Rob Rawlings
Sep 8 2019 at 7:32pm
And in addition: How does taxing people to pay for tree planting (which absorbs carbon that has been produced) reduce carbon production ? If carbon producers create something with an external costs then they need to either 1) do a Corsean deal with ‘society’ to plant trees (or whatever) to mitigate the external costs or 2) pay a Pigovian tax.
(BTW: Itt quite possible I’m missing something in which case I apologise for my possibly belligerent tone).
Jon Murphy
Sep 8 2019 at 7:51pm
The least costly method, by definition, is the most optimal. Remember that monetary price is merely one of a multitude of margins people can adjust along. If the opportunity cost of the tax is higher than the opportunity cost of other method X, then the tax is not optimal.
Pigou’s model exists in a world where price is the only margin adjusted along.
Rob Rawlings
Sep 8 2019 at 10:06pm
@Jon,
Lets assume there is an optimal level of carbon production and we are currently above it.
You are a central planner and you have 2 possibilities:
You change the world so that the current level of carbon production is optimal. You subsidize carbon-reduction technology to meet your goals.
You change the world so that carbon production adjusts to the optimal levels needed by the current levels of consumer preference. You follow the lead of Coarse and Pigou.
You are a central planner so you can do either, but I think option 2 represents what the free market outcome would be while option 1 represents what people who happen to have ownership-rights in carbon-based fuels might want.
Rob Rawlings
Sep 8 2019 at 10:10pm
(The last comment ignored indentation and made the comment confusing so here it is with ‘-‘ rather than tab)
Lets assume there is an optimal level of carbon production and we are currently above it.
You are a central planner and you have 2 possibilities:
– You change the world so that the current level of carbon production is optimal. You subsidize carbon-reduction technology to meet your goals.
– You change the world so that carbon production adjusts to the optimal levels needed by the current levels of consumer preference. You follow the lead of Coarse and Pigou.
You are a central planner so you can do either, but I think option 2 represents what the free market outcome would be while option 1 represents what people who happen to have ownership-rights in carbon-based fuels might want.
Robert Coffey
Sep 9 2019 at 8:13am
In pure theory, Coasean Bargaining is always superior to Pigovian taxes.
I rank the solutions this way in the order they should be tried:
Coasean Bargaining
Do Nothing
Pigovian Taxes
In most cases where people suggest #3, either #1 or #2 will work out just fine.
#2 is working just fine right now to reduce carbon emissions in the US.
If the tree thing works, Coasean bargaining is easy, instead of paying companies to not pollute, we can just donate money to the Sierra Club (or the Gates Foundation, or whoever) to plant trees.
Jon Murphy
Sep 9 2019 at 8:13am
Well, given this is a “central planner” situation, neither one is a free-market outcome 😉
But that aside, we do not have enough information from the choices on which is least costly; thus, despite your intention, we are not following Coase. Remember: part of Coase’s insight in The Problem of Social Cost is that, even with the presence of an externality, a tax is not necessarily optimal, which is to say not necessarily the least costly. Indeed, as Buchanan and Stubblebine show in their 1962 paper “Externality” and Carl Dahlman elaborates on in his 1979 paper “The Problem of Externality”, sometimes the optimal thing to do is nothing even with the presence of an externality.
Rob Rawlings
Sep 9 2019 at 10:16am
Thanks Jon. Let me see if i can find those 2 papers.
Jon Murphy
Sep 9 2019 at 10:40am
You can find the Dahlman paper here: https://www.jstor.org/stable/725216
The Buchanan paper is behind a paywall, unfortunately. Forgive me for not posting the links sooner: I was on my phone
Rob Rawlings
Sep 9 2019 at 1:21pm
Thanks, I took a look at the the Dahlman paper and based on that I think the argument being presented here is along the line of:
– If there were no transaction costs then individuals could strike deals with carbon producers that would lead to all production costs being internalized and the optimal level of carbon being produced
– However as there are transaction costs these must be taken into account. If they are sufficiently high then the cost of the process of enacting the internalization of the costs of carbon may be higher than the benefits of doing so.
– If there are other ways of addressing carbon emissions (such as tree planting paid for by private donations or a general tax) that are cheaper than the likely ‘transactions costs’ of implementing a carbon tax then this become the economically best option.
Is that the case being put forward here ?
If so, then (while it does seem to let the polluters off the hook)’ I guess I can’t argue with the economic logic.
Jon Murphy
Sep 9 2019 at 1:45pm
Essentially yes. A Pigouvian tax doesn’t make sense in a world without transaction costs (that’s Coase). However, Coase says that when transaction costs are too high to allow effective negotiation, then a Pigouvian tax can come into play. But Dahlman argues even that is incorrect as we need to ask why the negotiations haven’t already occurred. If the costs are too high, then any effort to reduce the externality may be too costly and thus we are already at an optimum. Further, under Pigouvian rationale, the tax is only welfare-enhancing if: 1) both parties failed to negotiate 2) One or both parties failed to properly estimate the costs/benefits of negotiation, and 3) the party(ies) that incorrectly estimated should have known better.
So, as it applies to your comment: In a world of pure theory and no transaction costs, a Pigouvian tax never makes sense. In a world of costs, a Pigouvian tax may make sense, but only under very specific circumstances.
Rob Rawlings
Sep 9 2019 at 8:05pm
My main takeaway from the Dahlman paper was the importance of transaction costs. It is the existence of these that results in the free market apparently failing to address situations of non-internalized costs and that led Pigou to develop his tax theorem. The paper does a good job of showing that the view that the state will do a better job of identifying areas of non-internalized costs and then address them with lower transaction costs than the free market was able to do is a matter of faith rather than good economics.
I wouldn’t say that this paper (plus Jon’s notes) have made me give up on the idea that a carbon tax is the optimal way to address climate change but I have now joined David in the ‘questioning’ camp.
robc
Sep 10 2019 at 8:42am
This is one of the hard thing’s to understand about the Coasean perspective, it was/is tough for me.
Externalities should be thought of like Newtonian forces, equal and opposite. So there is no one to be let “off the hook” as no one is ever on the hook. There is just the direction with low transaction costs and the direction with high transaction costs. And if the transaction costs are zero in both directions, it doesn’t matter at all who is “at fault”.
That isn’t entirely true, of course, while it doesn’t matter so society as a whole, it still matters to internal players. And in the zero transaction cost case especially, you can use justice to determine which way the costs flow. Or they can negotiate and split the difference.
Robert Murphy
Sep 8 2019 at 7:36pm
Rob,
It sounds like you are not aware of the possibility that the Pigovian approach can lead to an inefficient outcome, under certain circumstances. This was one of the things Ronald Coase demonstrated in his famous paper, though it’s admittedly hard to distill that conclusion from it (because his paper is a tough read). If you read the first half of this post, I explain the idea in an example about river pollution.
Not saying you have to agree with my views about a carbon tax, but I’m saying your comments make it sound as if you don’t see how–even in principle–taxing emissions could be the wrong thing to do, if we stipulate that there’s a negative externality involved. And yet, as Coase showed, that attitude is simply not correct.
Rob Rawlings
Sep 9 2019 at 10:13am
With regards to ‘It sounds like you are not aware of the possibility that the Pigovian approach can lead to an inefficient outcome, under certain circumstances’.
I am (I hope) aware of this. My takeaway is that where ownership rights are sufficiently well defined and transaction costs are sufficiently low then market forces alone will lead to all costs being efficiently allocated. In an ideal world the market (and not a central planner) would find the best way to minimize and allocate all costs.
When (as it nearly always the case in the real world) those condition are not met then whoever (the government or the courts) acts to address unallocated external costs then they need to take Coarse’s insight onboard when looking for a solution.
But I think I am missing how Coarse gets the producers of carbon off the hook. Maybe (and maybe not) massive tree planting is the best way of dealing with carbon emissions, but the costs of this tree planting must count as a cost in the production of carbon and (Coarse or Pigou) be reflected in the ‘price’ of carbon.
Rob Rawlings
Sep 9 2019 at 8:17pm
@Bob,
You may be pleased to know that today I have:
– 1. Learned to spell Coase’s name correctly
and
– 2. Come round to the view (after my exchange with Jon , above) that there are indeed some serious issues with the Pigovian approach that I had previously failed to spot.
Bob Murphy
Sep 8 2019 at 7:32pm
Rob,
I think you took my argument to be, “Because of the Tax Interaction Effect, planting trees is a better way to fight climate change than taxing carbon.” And you’re right that that is a non sequitur.
However, that wasn’t my argument. Rather, I was saying that the Pigovian tax framework assumes that there is too much CO2 emission, and that we need to induce people to emit less. But, it’s possible that it’s cheaper to subsidize the planting of enough trees to absorb a ton of CO2, rather than taxing businesses to emit a ton less.
So if it is actually correct to say that subsidizing tree-planting is the least-cost way to combat climate change, then there’s no reason to raise the required revenue via a carbon tax. Putting aside issues of climate change, raising a $1 billion via a carbon tax has more deadweight loss than using an income tax.
Now to understand *why* my last sentence is true, refer to the Tax Interaction Effect literature.
Rob Rawlings
Sep 8 2019 at 9:39pm
#Bob,
This is the sentence I have problems with:
“But, it’s possible that it’s cheaper to subsidize the planting of enough trees to absorb a ton of CO2, rather than taxing businesses to emit a ton less.”.
If there were no external costs then (in equilibrium) all goods would sell at cost and economic activity would be optimized.
If a single good (for whatever reason) started generating an external cost then some goods would sell at less than cost and economic activity would no longer be optimized as such good would be over produced.
In that situation then either a Coarseian agreement between producer and losers (from the external cost) or an optimal Pigovian tax would be needed to bring costs in line with price and reoptimize production levels.
I cannot see how an alternative approach using general taxation to pay the external cost of the production of the good in question would optimize the outcome. This would leave the price of carbon below its true production cost and lead to its over production.
Technological advances that minimized the external costs of production of this good would reduce the Coarsean transfer levels (or Pigouvian tax) required but the basic intuition would apply. TIE likewise would not change anything significant.
I am happy to proven wrong but I feel (TIE notwithstanding) that I am just applying basic economics here.
robc
Sep 9 2019 at 8:21am
I think I agree with you that the general tax is not optimal.
Coasean bargaining would be optimal, but I think you have it backwards. The lower transaction cost way of the bargain is that people who are worried about the extra carbon should plant trees and/or donate to organizations to plant trees, skipping the government step altogether — neither carbon tax nor general tax.
Christophe Biocca
Sep 8 2019 at 8:03pm
True, but that just means the tax should be capped at whatever the lowest-cost mitigation/fix is, not become zero.
Toy model:
– An unmitigated ton of CO2 causes $10 in damage.
– Planting enough trees to reduce CO2 by 1 ton costs $2.
– Electricity can be generated by options:
— A: $50 and 5 tons per MWh
— B: $60 and 0.1 ton per MWh
— C: $55 and 1 ton per MWh
Then:
– The do-nothing policy ends up with electricity generated by option A with $50 seen cost and $50 externalities.
– The tax-at-unmitigated-cost policy ends up with electricity generated by option B with $61 in seen costs.
– The plant-trees-paid-from-general-revenue policy ends up with electricity generated by option A with $50 seen cost, and $10 in externalities (where the externality is now borne by the general taxpayer who’s paying for tree-planting, rather than whoever gets hit with climate-related-costs).
– The tax-at-tree-planting-costs-then-plant-trees ends up with electricity generated by option C with $57 in seen costs.
The reason this happens is that a carbon tax’s incentive-changing-effect is greatest for the CO2 reductions that are cheapest to make (those that cost much less than the tax). To the extent that there are opportunities to reduce CO2 emissions that are cheaper than tree-planting, we’d still like to use them.
Note that what matters is the specific CO2 cost slope experienced by people, not the specific mechanics in play. You can get the same outcome in theory being willing to also pay for emissions reduction, at $2 per ton of CO2, out of your tree planting budget. Or, in the other direction, by taxing at the unmitigated cost but allowing people to deduct tree-planting they paid for from their emissions for carbon-tax purposes.
I don’t know enough about the tax-interaction effect to know how it would change the above result, but if someone does I’d be interested in an example.
Jon Murphy
Sep 9 2019 at 8:14am
Unless the alternative measure accomplishes the exact same outcome cheaper.
Christophe Biocca
Sep 9 2019 at 6:29pm
That sentence is ambiguous.
Are you claiming that tree-planting is, ton-for-ton, cheaper than any possible reduction of emission, by any amount, anywhere in the US economy? If so, then you are right that a non-zero carbon tax is less efficient. But I suspect that premise to be false; simply due to the enormous number of ways to reduce CO2 emissions across the entire economy, we’d expect at least a few to be very cheap.
If instead you are claiming that it is merely cheaper than most of the reductions that would happen under a full-price carbon tax, then for that subset of the reductions costing (for example), less than half as much, ton-for-ton, as tree-planting, how could it be more efficient to plant trees than to reduce (those particular) emissions?
The toy model I wrote above demonstrates exactly this effect. You end up with the cheapest overall cost by setting the carbon tax to the cost of the cheapest available mitigation option. That way, reductions that cost less than the mitigation get done, those that are more expensive don’t.
Thaomas
Sep 9 2019 at 6:00am
It seems very unlikely that tree planting alone would remove enough CO2 from the atmosphere to slow and revers the increase, especially if there were no other incentive (e.g. a carbon tax) to reduce the incentive to continue emitting CO2 from other activities. Looked at the other way, if there were a revenue-neutral tax on net CO2 emissions equal to the marginal harm done by the emissions (this would be a trajectory of rates that presumably change over time as technologies changed and the concentrations eventually started to decrease) would the economy react to that by planting trees and nothing else?
On the other hand, if there is not reason to want to reduce and eventually reverse the accumulation of CO2, then plating a lot of trees is just a waste of resources however the revenue were raised and if it were to be undertaken ex hypothesi, it should be financed with the least distorting tax possible, presumably a progressive consumption tax.
Mark Barbieri
Sep 9 2019 at 1:56pm
I see the biggest risk from climate change being an overreaction. Today, there seems to be tremendous pressure to “do something”. The biggest benefit I see to a carbon tax, provided that the tax costs are made transparent to consumers is that it will help people see the costs to decreasing carbon output.
You have me convinced it isn’t necessarily to optimal solution from an economic standpoint, but I’m still concerned about the politics. If not having a carbon tax leads to worse policies, it might still be the best solution.
Comments are closed.