I went to a conference held by the Institute for Energy Research yesterday on the carbon tax, where Bob Murphy was presenting. Not surprisingly, I thought the presentations by the economists (Bob and Ross McKitrick of the University of Guleph, in Ontario) were more solid than the non-economists. Kenneth Green was fine although I disagreed with a lot of what he said. David Kreutzer (Heritage Foundation) was who I really had a problem with, which I'll get to in a bit.
Bob's talk, which based on his written testimony is about what he's going to say to the Senate tomorrow, was about the variability of the social cost of carbon estimates depending on (1.) the discount rate and (2.) whether we're looking at domestic or global costs, as well as the variability in the non-discounted estimates themselves. This will probably be familiar to most readers of the blog - if you increase the discount rate (say, to the normal discount rates used by regulatory agencies) and consider only domestic costs (another typical practice for regulatory agencies apparently), the discounted social cost of carbon is a lot lower. Still, this is important to educate the general public about and lay out in the Senate. I'm guessing the Senators are well aware of both of these points too - but if Bob testifies on the issue it will be harder for them to grandstand on larger social cost estimates.
I found a few things particularly interesting about this discussion. First, if you read Bob's testimony (I'm sure he'll post it after today) he really skirts the edge of telling Congress they should only be considering domestic social costs. He doesn't say that. He just makes it clear that the estimates that don't do that aren't following OMB guidelines and he's explaining what following those guidelines would do. But the whole thrust of the testimony is in that direction. I found that surprisingly nationalistic for an ancap! Second, the big discussion in the literature on discount rates is not whether 3 or 7 percent (OMB's ranges) are good discount rates for most regulatory applications - it's whether those rates make sense for looking at the very distant future, or whether they make sense when the precautionary principle should be invoked. I'm not trying to tell Bob how he should testify, but as written it makes it sound like people advocate lower discount rates solely for the purpose of getting a higher social cost of carbon and that they are flaunting standard practice. This of course isn't true. Finally, Bob pushed a hard subjectivist line without really getting into the nitty-gritty details of subjectivist theory, which was nice to see. He pointed out at several points that the social cost of carbon wasn't a single objective value that could be pin-pointed. The moderator asked him if that implied that the discussion of the social cost of carbon was fundamentally political and I think Bob gave the right answer: yes. I'm guessing the moderator took that to be something dismissive about climate politics - I think an economist that appreciates subjectivism would see it slightly differently and just acknowledge that any question of aggregating preferences is ultimately going to be a political question because of the problem with interpersonal comparisons that make any answer justifiable, but also contestable.
Ross McKitrick (who, ironically, kind of looks like Al Gore)
McKitrick's talk was very good. He discussed the tax interaction problem associated with carbon taxes, an issue that Bob took up previously in this featured article at EconLib. To summarize, the problem is that in an economy that has many kinds of non-lump sum taxes, most of which impose some dead-weight loss, the increased factor prices associated with a carbon tax will raise the deadweight loss of the other taxes. This is trickier than it first appears... I had to draw it out to really get it myself. The critical point is that these are ad valorem, not lump sum taxes. If we just had lump sum taxes there would be no tax interaction effect (I think, Bob or maybe Grant can correct me if I'm wrong). These are not from McKitrick's talk, but below I've got factor markets with private and social marginal cost of the factor (the social marginal cost is necessary because things with social costs, like carbon, go into the production of other factors):
[UPDATE: a facebook friend points out that I've got a specific tax here, not a lump sum tax. The logic of the ad valorem tax and the tax interaction there doesn't change, of course. If we're thinking of a lump sum tax on labor supply on the extensive margin it doesn't make much of a difference (although on the intensive margin it would be different). I thought McKitrick referenced a lump sum tax but I could be wrong, and he didn't sketch any of this out in detail. If you want just rely on my right panel.]
In both panels of course the social cost of the taxed factor (remember this is NOT a carbon market - this is the social cost that is passed through because factor costs are a function of other inputs, including carbon) is higher than the private cost. This means that there's deadweight loss associated with the externality. That's highlighted in red in the graph. I think (although I'm not 100% sure) the welfare derived in these factor markets should be included in the market for carbon - so the point in highlighting the welfare loss is not to double-count - it's just to show you where it shows up in this market. Before any factor taxes come into the picture we are at equilibrium point "1".
When we add factor taxes (but before we add carbon taxes) we shift to equilibrium point "2". In the left panel we have a
lump sum specific tax. Tax collectors collect the same tax amount for each unit of the factor. The deadweight loss from that tax is in blue at the "2" equilibrium point. On the right panel we have an ad valorem tax - so tax collectors collect a constant share of the price, not a lump sum specific. At equilibrium point "2" I've made sure that the deadweight loss associated with the factor tax is equal to the deadweight loss of the lump sum specific tax.
Now let's say we have carbon taxes and factor taxes. That shifts us to equilibrium point "3" because now we are incorporating all the social costs of carbon and that trickles down to all these factor markets. The
lump sum specific tax case at equilibrium "3" is straightforward - we are collecting the same tax per unit of a good so the deadweight loss is the same. Adding the carbon tax does no additional distortionary damage, and since it eliminates the deadweight loss of the externality it does some good.
However, on the right panel with the ad valorem tax the blue triangle has to be bigger (this isn't necessarily to scale but it does have to be bigger) because we're maintaining the tax rate, not the size of the tax. So you've dealt with the externality and that's great but you've increased the deadweight loss from the rest of the tax system, so the benefit of the carbon tax is reduced.
McKitrick cited literature in the 1990s on this, but then noted that Agnar Sandmo came up with the result in 1975. Interestingly enough, in Sandmo's history of thought textbook (which I'm assigning, on Grant's advice, for my class), Sandmo points out that Pigou made a similar point many decades earlier - that if revenue is collected through distortionary taxes then the optimal level of externality correction has to be adjusted by the marginal cost of financing government (i.e., by the distortion). The version Sandmo discusses with respect to Pigou pertains to spending, not taxes, but the tax interaction is exactly the same. So really we've had this result from the very beginning of the externality discussion, it's just a little trickier (it has a few extra steps) so it takes a little more work to communicate.
Kenneth Green and David Kreutzer
I don't have a lot to say on Green's presentation, although it was interesting. He was talking about his own history with carbon taxes and the politics of second best policy. When he was at AEI he apparently promoted the benefits of a carbon tax (without actually advocating it) because at that time everybody wanted cap-and-trade (which has always been less popular among economists, for obvious reasons). The carbon tax was not as bad as cap-and-trade so he figured it was better to advocate that if some kind of climate change legislation was going to get through. His thinking has "evolved" on that, and he's more convinced now that he gives and inch people will take a mile. I'm not so sure I agree on all points or that "give an inch, take a mile" is an excuse not to advocate good policy, but it was a nice talk.
Kreutzer, I thought, was much worse. He basically gave a litany of the costs of a carbon tax that he thought people were ignoring - jobs lost, output lost, etc. etc.
Well there's a reason why economists at least don't focus on this - if we think carbon imposes a negative externality then that production represents overproduction that imposes a deadweight loss on society. This is like saying we shouldn't quit using cocaine because we wouldn't be able to do all that extra work we're able to get done when we're using. Sure, that extra production is a benefit, but when you make the assertion that carbon imposes externalities you're not ignoring it - you're saying that the costs associated that extra production exceed the benefits.
This is bothersome because this sort of analysis has a lot of legs in Washington. It's very easy for a politician to grand-stand on a negative economic impact without considering the broader welfare issues at stake. You see this a lot in high skill immigration discussions too - lots of people talk about the costs to the tech industry without even thinking about whether there is deadweight loss associated with a visa policy only available to a select class of workers.
My asked, but unanswered question
So I actually submitted two questions, both for McKitrick, but the second one wasn't quite answered. I asked what the implications of tax interaction were for subsidies for green energy (might they even work in the opposite direction since we're dealing with a positive externality and reducing factor prices?). I couldn't figure it out myself sitting there in the audience because while we're reducing factor prices on the one hand we also have to finance the subsidy. So I was wondering if he knew the answer.
The moderator asked it but then added some discussion of the comparative efficiency of carbon taxes vs. regulation, which wasn't my question at all. Of course regulation is even less efficient than carbon taxes, which is precisely what McKitrick said in response.
So if anyone has any thoughts on my question (which I'm still scratching my head over) let me know!