It’s Friday so can I interest you in awith a highly incomplete and self-serving history of the “optimal” taxation of carbon dioxide emissions?

*proceeds without waiting for answer*

Let's do this! (1/x)

100 years ago (literally!), Arthur Pigou published the Economics of Welfare, teaching our great grandparents that when private transactions impose costs on 3rd parties, we'll produce “too much” of the product. By taxing the product we can reduce production to the optimal amount.

As long as we know everything (ASSUMPTION), that "optimal" production is the level at which the costs and benefits of one more unit of production are equal. So, in theory, we can just set a tax at the level that produces exactly the optimal production amount. Voila!

Cool and useful theory from Pigou. The literature evolved, as you might expect, in a more practical direction. Here's a paper published 50 years ago (not literally but close!) by William Baumel in AER on optimal pollution taxation (h/t to @bobkopp) https://pdfs.semanticscholar.org/97da/0c7676fbab1c4f6a27384e20a1219333fbff.pdf

Baumel says policymakers obviously can't implement Pigou’s approach, so he proposes a humbler method. Use taxation to control pollution efficiently, but (I'm simplifying) set the standards as best as we can and use an iterative approach to identify the acceptable level to society

And this is basically what we do for most policy problems (especially involving risk). If we're talking about car safety or nuclear deterrence or whatever, nobody says do nothing, and nobody says “optimize” these policies perfectly based on marginal benefits and costs.

So this is naturally how we might have treated CO2 emissions also, were it not for a new, and in my opinion fascinating and useful, strand of research started by William Nordhaus 40 year ago (literally!)
http://cowles.yale.edu/sites/default/files/files/pub/d05/d0565.pdf

Nordhaus took Pigou's approch, not Baumel's, to the climate problem, and begin to estimate the optimal CO2 tax (or social cost of carbon) with precision.

But if you read Nordhaus' early papers, it's pretty clear that he understands that this is just a way to *think* about the climate problem, and real-world policy might look somewhat different.

And most policymakers didn't take the results too literally. Until, that is, we Americans, ever confident in our abilities, began to use our social cost of carbon estimates in policy. First, in regulatory RIAs, which is harmless enough...

But more and more as specific carbon prices. SCC estimates determine how nuclear plants are subsidized in NY, carbon tax proposals in Congress were pegged to them, etc.

Not to mention roughly a kazillion academic papers and dissertations devoted to tweaking SCC models (including a portion of mine!) https://link.springer.com/article/10.1007/s10584-011-0140-7

Most in the climate world, including economists, understand all this. IPCC reports don't take the SCC models all that seriously. Many economists focus on CO2 pricing in an emissions target context (like many EMF studies or this Stern/Stiglitz study) https://www.carbonpricingleadership.org/report-of-the-highlevel-commission-on-carbon-prices

But, among economists, these are still considered two separate strands of literature. Emissions-target CO2 prices and "optimal" CO2 prices from benefit-cost analysis.

This has led to some confusion. People conflate these strands of literature and criticize carbon pricing for problems with the SCC. Like this Bloomberg piece, for example (I don't mean to pick on @Noahpinion, who is excellent, b/c this is common)

So one goal of our @NatureClimate paper (you knew I'd make this about me, right?) is to suggest that the last 40 years of Pigou-type "optimal" CO2 taxation as been a long detour, and we consider refocusing efforts on the Baumel-type approach. https://rdcu.be/b6jIH 

Again, so many great economists are already doing this. We didn't "invent" anything here.

But, in my opinion, re-focusing a lot of the effort that currently goes into SCC models to more practical applications, like the NT2NZ approach described in a paper, could go a long way. There is still so much we don't know! /end