Jim Manzi replies to Ryan Avent

The following is a guest essay from Jim Manzi, CEO of Applied Predictive Technologies (APT), an applied artificial intelligence software company. He writes occasionally for National Review and blogs at The American Scene.

—–

Last week on this site Ryan Avent presented a thoughtful response to my recent article at The American Scene arguing against a carbon tax. Grist has graciously invited me to reply.

As I understand it, Ryan had three basic criticisms of my logic:

1. the impacts of global warming will be more messy, unpredictable, and heartbreaking than I let on,
2. I don’t understand the economic trade-offs that make a carbon tax an elegant solution to the problem, and
3. the technology-focused approach to the problem I propose is insufficiently conservative.

I’ll try to address each of these in turn, all with a spirit of open-minded inquiry.

The first objection highlights the fact that productive global warming debates almost always hinge crucially upon predictions of the future. Consider three generic types of predictions: deterministic (“If I let go of this pencil, it will fall”), probabilistic (“If I flip this coin, it has a 50% chance of coming up heads and a 50% chance of coming up tails”), and uncertain predictions, for which we can not specify a reliable distribution of probabilities (“There will be a military coup in Pakistan in 2008”). Economists will immediately recognize the distinction between probabilistic and uncertain forecasts as, in essence, Knight’s classic distinction between risk and uncertainty.

Strictly speaking, all predictions are uncertain, but as a practical matter we treat different predictions differently based on the observed reliability of the relevant predictive rules used to generate them.

No serious person believes that even the physical science projections for climate sensitivity (i.e., how many degrees hotter the world will get if we increase atmospheric carbon dioxide concentration according to some emissions scenario), never mind our predictions for how fast the world economy will grow or the economic impact of various degrees of climate change, are deterministic. This is why climate modelers and integrated climate-economics modelers spend so much time developing probability distributions for various outcomes and combining possible outcomes via odds-weighting to develop expected outcomes. When we hear a modeling group say “the expected outcome is X," it doesn’t mean they’ve assumed only the most likely scenario will occur; it does mean, however, they assume their distribution of probabilities is correct (not being idiots, of course they constantly work hard to try to test and improve this distribution of probabilities).

For the moment, let’s assume that predictions of global warming outcomes are probabilistic. I go into all this in my posts and articles in much greater detail, but if we take Nordhaus’s DICE modeling group at Yale as a benchmark, we can make the following observations about global warming:

• The total costs of global warming are expected to be equal to about 1% of present value of future human consumption.
• Even if we could implement a “perfect” global carbon tax that operates optimally to reduce emissions at minimum cost, we should still just let about 75% of this warming damage happen, because it would be more expensive to prevent it than simply suffer the damages.
• The net benefit of such a “perfect” carbon tax would be about 0.17% of present value of future consumption.

Think about that: the total benefit available to the world economy from addressing global warming is projected, at best, to improve present value of human consumption by less than 0.2%. Now, not everything that matters in life can be measured by money, but this is nothing like the rhetoric of “New York destroyed in a disaster” would suggest. It makes it very hard to just wave away all of the efficiency costs that would likely be imposed by an effort to force rapid conversion away from fossil fuels.

Even this net benefit is unlikely to be achieved in the real world. Think of what it would mean to enact a harmonized global carbon tax (or some alternative regulatory scheme, even less theoretically efficient than a carbon tax). You’d have to get, among others, the French National Assembly, the Parliament of India, the Brazilian National Congress, the Chinese Politburo, and Vladimir Putin to go along. What kind of side-deals do you think would be required to get them to do this? Further, even if we got to an agreement de jure, we would then have to enforce a set of global laws for many decades that would run directly contrary to the narrow self-interest of most people currently alive on the planet. How likely do you think a rural Chinese official would be to enforce the rules on a local coal-fired power plant?

And this is just on the international front, assuming the U.S. is on the side of the angels. It doesn’t require complete cynicism to observe that political coalitions are not entirely composed of philosopher-kings debating the good of humanity. Trillions of dollars of assets and millions of jobs will be threatened by pushing a huge portion of a currently widely-distributed tax burden onto a subset of the economy. What do you think the reactions of the good people of Wyoming, North Dakota, and Alaska will be to the idea of paying for huge tax cuts for their beloved countrymen in New York, California, and Florida? The senators for any 20 states can block most legislation. In my original NR article, I described how, in a presidential election, this dynamic would likely play out to punish harshly any candidate foolish enough to propose such a tax.

What do you think are the odds that the real, not theoretical, regime created by such a process would produce global economic drag (i.e., politically-directed, inefficient allocation of resources) that is greater than 0.17% of present value of consumption?

Of course we can’t be certain that our probability distribution of outcomes is correct. That is, we may be facing uncertain, rather than merely probabilistic, predictions. I’ve written extensively about why, in fact, we should take such concerns seriously in the case of global warming. There is some chance — by definition non-quantified, since if we could quantify it, it would be incorporated into our probability distribution — of catastrophic climate impacts well beyond the distribution of possibilities contemplated by consensus models. It is insuring against such an outside chance that is the most valid justification for potential large-scale global warming interventions.

I think that intellectually serious advocates of rapid, aggressive abatement of carbon emissions (via whatever regulatory, tax, or other means) need to confront this reality. If we are taking action to prevent an outcome that is not predicted either in a deterministic or probabilistic way, but is an unquantifiable risk of something that might go horribly wrong, how do we compare it to costs and other risks in order to price it? Simply pounding the table and saying that humanity is imperiled isn’t very convincing outside of your own choir loft.

This is not an academic concern. If we introduced a tax high enough to keep atmospheric carbon concentration to no more than 1.5X its current level (assuming we could get the whole world to go along and implement a perfect tax), we would expect (PDF) to spend about $17 trillion more than the benefits we would achieve. That’s a heck of an insurance premium — something like one-quarter of the world’s total annual output — for an event that it is literally outside of a probability distribution. Of course, I can find scientists who say that level of atmospheric carbon dioxide is too dangerous. Al Gore has a more aggressive proposal that if implemented through a perfect carbon tax would be expected to cost more like $23 trillion in excess of benefits. Of course, this wouldn’t eliminate all uncertainty, and I can find scientists who say we need to reduce emissions even faster. Once we leave the world of odds and handicapping and enter the world of the Precautionary Principle, there is really no principled stopping point. We would be chasing an endlessly receding horizon of zero risk. Is your answer that you just “know” the right practical amount of risk to take?

The same basic point can be made by noting that there is nothing magic about global warming that should lead us to privilege its costs over all the other bad things that might happen to us. We face lots of other unquantifiable threats of at least comparable realism and severity. A regional nuclear war in Central Asia, a global pandemic triggered by a modified version of the HIV virus, or a rogue state weaponizing genetic engineering technology all come immediately to mind. Any of these could kill hundreds of millions of people. Is your answer that you just “know” how seriously to take each of these risks, and you just “know” that global warming risks, beyond those specified in consensus modeling distributions, are more serious than these other risks?

Now, a reasonable response to this is, “Addressing global warming doesn’t mean that I can’t also work out smart security arrangements in Asia, invest in biological science, and contain rogue regimes and so on.” This is true, but each of these things cost money (or economically equivalently, imposes efficiency costs). I don’t have to make this list very long before I start to eat up the bulk of GDP. In a world of finite resources, a responsible advocate must have some integrated framework for weighing relative risks and opportunities.

I’ll be clear about my conceptual framework: in the face of massive uncertainty on multiple fronts, the best strategy is almost always to hedge your bets and keep your options open. Wealth and technology are raw materials for options. Because it is easy to predict the future, but extremely hard to predict the future correctly, we should avoid irrevocable commitment to any course of action until as long as it seems possible (not just for global warming, but in general). Adding a large tax burden to the economy is very difficult to reverse, and holds large regrets (just as under-estimating the severity of global warming would). The loss of economic and technological development that would be required to eliminate literally all theorized climate change risk would cripple our ability to deal with virtually every other foreseeable and unforeseeable risk, not to mention our ability to lead productive and interesting lives in the meantime. It’s possible to buy so much flood insurance that you can’t afford fire insurance.

Rapid, aggressive emissions abatement is not economically justified by probability-weighted consensus forecasts of the distribution of modeled global warming impacts. Advocates for this policy need to do a better job than they have to date explaining how we justify an enormous loss of consumption in return for chasing this one uncertainty in a world of troubles.

Ryan raises a very smart objection to my analysis of costs vs. benefits of a carbon tax, which I think I could paraphrase as, “Since we have to tax something, why not tax something we want less of. If we offset the carbon tax with other tax reductions, why don’t I get a free lunch?” I’ve written another article at The American Scene (“Coase Club“) about why I think this is a seductive fallacy, so I won’t repeat all the arguments here. I’ll just make two points that are described at greater length in that article.

First, as I’ve already touched on this response, the actual carbon tax that a real government passes is very unlikely to be nearly as efficient as the tax swap envisioned by an academic economist.

Second, there is an unending set of externalities created by all energy sources, and there is no logical reason to privilege AGW-related costs over any others. If I die in an AGW-caused flood or I die from cancer caused by inhaling the fumes from somebody else’s car, I am in both cases equally dead. I reference research in Coase Club showing that we can’t measure the external costs (including AGW plus everything else) of carbon and non-carbon-intensive fuels within an order of magnitude, and the range of total external cost estimates overlap between every energy source. How can we set an initial price rationally and then engage in trial-and-error learning? How will we know whether we have decreased or increased total external costs once we have shifted X units of production, say, from LNG to biomass? In practice, prices would be set entirely politically, via interest group bargaining.

To the third objection — that my proposals show that I am not a true conservative — I can only say: get in line. I doubt anybody reading this post also listens to Rush Limbaugh (I don’t either), but I can tell you from experience that it was no fun when the guy basically called me an idiot in front of his audience of 14 million dedicated fans. Our good friends at FoxNews, naturally enough, piled on, along with many bloggers who adorn their homepages with pictures of assault rifles.

I take seriously the possibility that global climate change could become severe. In that eventuality, we would want to have done the long lead-time work to enable us to deploy technology to blunt or eliminate the problems it would create. That is why I favor a technology-investment program. It seems to me that, here at Grist, “insufficiently conservative” would be more in the spirit of a compliment than a criticism. If the charge is trying to think for myself, I plead guilty.