Think, Read, and Write: Christy and Emanuel on Climate Change

Re: 4, and assuming this is about the 2.5-9 degree F range posited by Emmanuel:

Emmanuel also thinks that the upper bound of Emmanuel’s range is unlikely – that’s why it’s an upper bound. The chance of hitting the upper bound of a 95% confidence interval is only 2.5%. I think the more interesting question is that Christy thinks the lower bound of Emmanuel’s range is likely, whereas Emmanuel thinks it as unlikely as the upper bound.** I think the hypothesis proposed by Christy, of there being negative feedback loops relating to climate is interesting but assuming that Emmanuel’s models are a valid statistical interpretation of past trends, it would be to Christy to prove it.

**This all assumes that the distribution within the range is normal – it very well may not be.

Russ Roberts spoke towards the end about the vitriolic feedback he gets, and about the tribalism in economics. I support his claim that we should critically examine even (maybe, especially) those views that we agree with. However, I also can understand why people get very worked up over issues in economics, and in regard to climate change. Both of these issues affect people’s lives. When a poor economic policy is followed, people lose their jobs, have less money, and, not to put too fine a point on it, suffer. There are similar worries with respect to our approach to climate change. I don’t intend to excuse insulting or demeaning behavior here, but I do want to point out that the disagreements are not merely academic, but have very serious real-world consequences.

I have really been enjoying the recent climate change series. I got my undergrad degree in Environmental Science, so I do know a bit about climate change, but do not claim to be an expert. I freely admit that my biases lie with Emanuel. While I think that climate scientists like Christy are in the minority, I do enjoy hearing from them. For the pro-climate change side, it is good to hear from people that cannot easily be dismissed that have literal skepticism.

Most people have a world view of government/economics that drives the conversation. The only group of people that I truly cannot have a conversation with are the people starting from the premise that “climate change is a hoax.” That is a similar conversation with young-earthers.

#2. The Moral Question

The moral question strikes me as obscure, yet nicely addressed by the one phrase upon which all participants in this discussion agreed: “we don’t know”.

Armed with this insight, caution regarding dogmatic conclusions to climate challenges we don’t quite understand completely, strikes me as prudent. I believe all agreed that activities which impose damaging higher costs to current inhabitants of this planet so that those not yet here are better off, is a hollow victory. So to this and future generations what is owed is precision in understanding the problem and its remedy, and sober, candid reflection and research in implementing “solutions”. The moral duty is to get this right.

I’ve studied and followed this issue extensively for several years now, so I feel confident in my ability to address these questions:

“1. In what ways is climate science like macroeconomics? How are climate models like economic models? How are they different?”

Emmanuel put it as “we know the equations, you don’t” but I disagree. Economists know plenty of equations. These equations are true by definition, they are called accounting identities. For example we know the equation of exchange. The real difference is more subtle. The physical scientist’s equations are unambiguous in their meaning even without reference to outside assumptions or knowledge. But the economist’s equations are meaningless tautologies without reference to outside assumptions or knowledge.

Of course, Russ seems to have unintentionally caused this conclusion. He was basically saying that an economist will often develop a multiple regression model, and asking whether the development of a climate model is like that, where the relationship between variables is estimated based on the observed data; the answer to that question is “not exactly.”

A climate model is basically an attempt to solve numerically a set of differential equations, the finding of exact analytical solutions to which is essentially intractable as a practical matter. In fact it is not even known that solutions will always exist.

As a result, the results are not unambiguous: specific processes are represented in a simplified manner. This leads to several coefficients whose values need to be estimated, but are not fundamental constants well known to physicists to within several decimal places in accuracy. This is “parameterization.” It bears a little more resemblance to the way an economist might construct a model based on a regression. This introduces to layers of uncertainty and potential for error: the way in which the process is represented, and the value of the parameters. Typically, real world observations will constrain each parameter to within a particular range. So modelers will play around with various combinations of parameters until they can get a set of conditions in the model world that look like the real world in certain respects (but not every respect). This is called “tuning.”

Models are then given certain inputs and output the new conditions that would result. So the choices of parameters ultimately determine the magnitude of the response. The magnitude of response is the primary question of concern: is it large enough-is the system sensitive enough-that it’s state could change so drastically as a result of our actions-that it would cause serious problems, or is it actually not very sensitive at all?

The parameters determine the sensitivity, but not in a simple, direct way. There is a huge multi-dimensional parameter space, and given a particular value of a parameter, setting a second parameter higher could raise or lower the sensitivity, depending upon the values of several other parameters. It is not a parameter itself, but a result of the values chosen for the other parameters.

However, given this situation, sometimes scientists will go a little more “low tech” and treat the system in terms of a smaller set of equations thought to simplify the problem. In this case, sensitivity itself becomes a parameter. And here we can get to the bare bones, what exactly is the sensitivity?. The answer is that it is merely the rate at which heat loss by the planet increases as it’s temperature increases. If the rate is lower, sensitivity is higher (because the temperature must increase more in response to an increase in energy input or decrease in the efficiency of heat loss). You’d think it would be easy to find this. Well, actually, for a given degree to which a body reflects incoming energy, and a given efficiency with which it radiates energy, this is a simple matter of physics. From physics we know that if everything else were held constant, but you decreased the rate of energy loss by the amount a doubled amount of CO2 is thought to, the temperature would need to increase about 2 Fahrenheit degrees to restore energy balance. However, everything else is not held constant-and here I can see where economists can sympathize “Ah! a ceteris parabus problem, eh?” The things that are constant, but expected to vary as a result of changes in temperature-and not independently of temperature, that can change the degree to which sunlight is reflected off to space, or the rate at which heat is lost to space, are considered “feedbacks”

I would compare feedbacks to the infamous “multiplier.” In several respects actually, not only because it’s magnitude or even sign are a point of contention and controversy (although you aren’t likely to find many who think a meaningful value doesn’t exist at all) but because it also amounts in effect to a sum of a geometric series, so it’s mathematical form is very similar as well!

Of course, just because most of the climate models are not constructed using regression, doesn’t mean that no one in the field does that. In point of fact there have been a number of studies using simplified models and observations to attempt to estimate the value of sensitivity based on regression. And modelers can also claim their models fit past data fairly well, but although they did not decide the value of parameters based on that data, the inputs are sufficiently uncertain that they can match past history even though models vary by plus or minus 50%-if you have a more sensitive model, just assume that the cooling effect of particulate pollution is on the higher side.

“2. Christy claims that the debate over climate change really entails a moral question, not a scientific one. What does he have in mind? To what extent do you agree?”

I believe Christy has in mind the question whether we can ask current and future generations in developing economies to continue to live in poverty for the sake of averting the possibility of some climate change.

And I agree this is a moral question, and that it is ultimately the question we have to ask ourselves. This is because, even is one decides that people in already relatively wealthy societies can afford to sacrifice wellbeing today, to avert some degree of damages in the future…it’s not enough. Even if the US stopped emitting all CO2 tomorrow, the difference in future climate trajectory would not even be measurable. That’s a point which is not in dispute, by the way. Most of projected future warming is caused by emissions from developing economies. The moral question is, do we have the right to tell those people they can’t try to attain standards of living like we have? I think we do not. And if we do not, and they don’t chose to, there is nothing can be done to actually change the trajectory of future climate.

Note that innovation and technology is not “something we can do.” It’s a determinant of the trajectory that is itself something that will or will not happen.

“3. In discussing temperature trends over the past ten to fifteen years, Roberts says, “one person’s eon is another person’s blink.” What does he mean by this, and how does it help illustrate the differences in the perspectives of the two guests?”

If I had to guess, he’s getting at the fuzzy question of what constitutes the long term versus the short term. Is the recent period sufficiently long to call into question the claimed effects of added CO2 or is it too short, where it may be admitted that short term effects dominate but maintained that over the long term the system is ruled by the physics equivalent of a tendency toward Walrasian equilibrium. I think he is in fact getting at an important point, but I’m not sure that the guest’s perspectives are necessarily that far apart on this matter. Regardless of the timescale you put on what are presumably natural fluctuations up or down, you have to agree, if you accept the observations, that is evident we have had one such. And if you accept that there is an underlying warming tendency due to increasing CO2 etc. then you implicitly accept the ability of nature to generate fluctuations at least as large as the short term effects of those increases, on at least a timescale between ten and twenty years. Emmanuel’s positions seems to be that this is likely a such a fluctuation. On the other hand, based on what Christy said, I think he applies this observation more broadly. After all, if over a ten years, the system can cool independently of CO2 even as CO2 increases should be warming it, it should also be able to warming above and beyond what CO2 would actually cause for a period. After all, he thinks some portion of the warming that was attributed to CO2, could have been a natural fluctuation. One needs to assume a strange asymmetry in order come to the view that nature is allowed to “mask” warming with random fluctuations, but it is not allowed to “exaggerate” warming. I suspect that if pressed, Emmanuel would concede the point. But if less of past warming is due to CO2 and more of it due to natural fluctuations, we have less to be worried about because the sensitivity must be lower.

It does not prove that the effect does not exist at all, and I don’t think Christy would assert that the effect doesn’t exist, or even might not exist.

“4. Christy thinks the upper bound of the Emanuel’s temperature change forecast is unlikely to occur. How do you think he might justify that claim?”

Many recent studies indicate that the mean estimate is higher than the evidence suggests. This makes the upper tail of the distribution less likely. But one should be aware, the issue with the sensitivity and the upper end of the distribution, is that the distribution of estimates tends to be skewed and have a “fat tail” at the high end. This will happen when value in question is the multiplicative inverse of something about which the uncertainty is normally distributed. But with a lower mean, the tail will be less fat and the distribution less skewed. So the probability of the high end is already low, but it’s probability can drop dramatically if the estimate of the mean value goes down.

One reason to think true value is more likely to be lower than the central tendency of mainstream distributions like those cited by Emmanuel, is exactly the fact that the trend has been so close to zero for many years now. And in fact that models have generally overestimated warming on average even over the long term-the trend in the last few decades is lower, also, than the average of models.

Somewhat bizarrely, Emmanuel suggested models underestimated warming in the decades prior either the most recent decades-that is, “sure they’re overshooting for the most recent 30 years, but they undershot in the previous 30 years”-or in the decades prior to the most recent decade-which would be more like asserting that models are overshooting in the last ten years, but undershot the previous twenty. Whichever he means to assert, he’s wrong on both counts. Models typically have the rate in the 80’s and 90’s about right, they are too high in the 40’s through the 70’s, because temperatures were going down then and most models have them flat to slightly warming, and they did indeed undershoot from about 1910 to 1950-ish.

I’m not sure if he mispoke or hasn’t carefully examined the historical runs.

“5. Emanuel concedes that current climate models are imperfect. How does he justify his predictions in the face of that imprecision? How does this influence the policy applications Emanuel thinks should follow?”

If I had to guess, in his view, it’s as likely models are flawed to underestimate the problem as overestimate.

Which, if all you know is that models have flaws, seems like a logical a priori position. But it’s really not. It’s the thinking of what I like to call the Walter Wagner school or probability. After Walter L Wagner who, in an interview with the Daily Show, declared the probability that the Large Hadron Collider had a 50% chance of destroying the Earth because it either would, or wouldn’t. Two possibilities, so it must be 50-50.

This is fallacious reasoning. Although both over and under estimates are possibilities, they are not a priori equally probable. The relationship between the two probabilities is not something that can be known a priori. So instead of saying it could go either way, you should instead ask, which way does it go?

I believe in Christy’s research, he actually tries to find answers to the question, and based on that research, he thinks that models overestimating warming is more likely than them underestimating it.

On the other hand, Emmanuel’s area of research has nothing to do with the magnitude of warming so he hasn’t tried to answer the question whether models are under or over. His research area is Hurricanes. Most people in this field are in that position, and as a result it is natural for many of them to gravitate toward the view of assuming the models are good enough in spite of their flaws. It’s just easier.

Replying to Andrew_FL

1. Re the equation of exchange: are you seriously suggesting that MV = PQ is the equivalent of e = mc2? Your equation of exchange is never true, anytime, anywhere. But e = mc2? Always true. Big difference, my friend.

2. No responsible person advocating action on climate change who I know of suggests poor people should continue to live in poverty. Quite the opposite, in fact. This is just a red herring popular amongst your clan.

3. Russ, Christy and you show your true colours with this kind of nonsense. Every decade is warmer than the last decade. ACG dates back more 150 years. Yes, natural variability can, short term, overwhelm the AGC signal. But look at the long term trend. Think hockey stick. And remember that the planet’s orbital setting should be sending us into a cooling phase.

4. Models consistently underestimate reality. Your take is just wishful thinking.

5. Models do not define global warming. Go to James Hansen’s website and enjoy his recent model-free paper.

I thought Emmanuel didn’t do climate science any favours, which is as I expected from him and possibly why Russ approached him. Emmanuel either ignored or is ignorant of a substantial body of published work that would have rebutted all Christy’s points. I would have done a better job.

I have developed the view that core problem with the ‘climate change’ debate is that it is mired in the impossible question of causation within an immensely complex system.

So I look for areas in which uncertainty is at least reduced. There is, I suggest, some level of ‘certainty’ that high levels of concentration of carbon based gasses in the atmosphere is bad. How bad, is again a matter of opinion. But the reason we call these gases ‘greenhouse’ is the archaeological, geological and other evidence that high concentrations of these gasses in the atmosphere have caused stark temperature rises in the earth’s past.

Therefore, my view is that it is imperative that the pollution cased by humans must be reduced. Clearly, there are contrary views based on the uncertainty about causation; i.e. – why stop polluting if we cannot be sure we are destroying the atmosphere?

So much of the ‘do nothing’ argument is based on ‘certainties’ about the damage that reducing atmospheric pollution will do to our economies. Therefore, the ‘do nothing’ argument effectively says that the economy must be preserved even if that destroys the environment.

But we do not live in an ‘economy’. We live in an environment, in societies and in communities. That we have become servants of the economy is the tragedy here. It seems that many have lost the perspective that any economic system must serve society and that is no better demonstrated than in this debate.

The irony is that, if the more dire predictions about atmospheric warming are correct, the economic system will also fail catastrophically. Then we will have both social and environmental chaos.

The alternative is to radically change economic incentives in favour of a cleaner future. Then we have both bets covered. This is not a matter for the heroic gambling which the backers of economy over environment effectively advocate.

There was a time when many thought, and advocated, for example, that slavery was good and female suffrage bad. But those who were wrong, and harmed society by their influence, died before they could be held accountable. I wonder whether the same impunity will apply in relation to the environment.

Regards,
Jim Feehely