Saturday
Feb072015
by 
Bishop Hill
Bishop Hill Climate rhetoric
Feb 7, 2015 Economics The ongoing war of words over Richard Tol's summary of the various estimates of the costs of climate change seems to have come to an end, with the Journal of Economic Perspectives publishing another revision to the paper. The key change is in Fig 1, which previously looked like this:
The new version essentially reproduces the IPCC version of the figure, which excludes the best fit line and the confidence intervals.
;)
To my mind the rhetorical effect is to make the alarming estimates look like non-credible outliers. But perhaps that's just me. Unintended consequences eh?
Reader Comments (47)
"Meaningful" is a useful word; in this context I suspect (no more than that) that it in fact means "helpful".
1. Will it affect the price of ice cream?
2. Will their be more lolly for climate science economists?
It does seem that those two studies published after AR4 indicate a divergence of certainty with respect to the economic impact.
So it can be concluded that the earlier 'settled science' was incorrect and certainly unsettled.
Aren't we lucky that no-one took the claims seriously and that Copenhagen was a bust.
This should be forwarded to Roger Harrabin for his next report on the BBC about Paris.
Milder winters and nights, longer growing seasons, climate change to date has been a benefit.
More atmospheric CO2, surely a benefit to plant growth and food production.
Where do I go to get a grant to publish this in a journal?
Maybe I am a bit dim, but this study strikes me as one which looks at how the health of grouse on the moors affects the price of fish. As temperatures have risen since the Little Ice Age, so has the economic health of the world. However, correlation is not causation; economic growth started during the LIA (perhaps even before), as manufacturing processes developed. Increasing temperatures might have benefitted growth, but can we be sure it caused it? How do the economies of two countries with completely different climates compare, with all else being equal? Let us look at, say, Canada (cold) and Australia (hot).
To my mind, the rhetorical effect, is to make the non-credible alarming estimates, look like outright lies.
There are two things going on.
First, data were incorrect. This is embarrassing. However, correcting the errors did not lead to results that are qualitatively or indeed significantly different.
Second, new observations were added. These hardly affect the estimates of modest warming, but they sharply shift the estimated effect of more pronounced warming away from alarmism.
As people have said before, Bob Ward is paid for by oil money.
Richard Tol:
Wrong! Bob Ward is oiled by big money!But we are going to get at least 30 years of Global Cooling: http://notrickszone.com/2015/02/06/the-sun-and-ocean-cycles-drive-global-temperature-natural-factors-bringing-cooling-for-next-30-years/
Ah, paleontology, where warming is good and cooling is bad. Heck, let's invent 'optimal' and 'minimal' and apply them to temperature regimes. Revolutionary!
Naw, kim, doomed, useless, rhetoric.
==========================
Warming is a net benefit, since a warmer world sustains more total life and more diversity of life. Paleontology has not yet tested the upper limit of warming, and demonstrates the harm of cooling at each such event.
You can draw your lines really wherever you like. You are coloring outside the lines if you're not within the above cartoon.
========================
@Tol Respect to anyone that admits their errors. It is OK to make errors it is part of the process of learning & moving towards the truth. What is wrong is when people try to hide their errors for fear of letting a side down ..and others try to help by circling the wagons, that is a hindrance to the truth
I forgave Richard Tol years ago, once I figured out that he had thought that the science was settled. But, please, this is just bargaining.
Excellent, effective, bargaining, mind you.
=============
OK. First, there is just too much input into these numbers which is claimed to be relevant but (solely in my opinion as a Texas housewife) amounts to nothing more than guesswork, dressed up as data.
Second, what it does prove if taken as true is that it doesn't really matter. Take the number of years it must require even at the most alarming warming rates to get to five or six degrees of warming. How many would you like? Fifty. Even that's way beyond the extreme end. How much lost growth does a few percent matter over that period? Nothing, that's how much. Given the guessed input, it's just immaterial.
Has anybody calculated the improvements to the health and wellbeing of the worlds population (and their grandchildren) if we stopped wasting money on unproven theories about what might happen, if world temperatures do what "experts" say they should be doing, even though their predictive skills have been less than guessing "heads or tails" on which end of a seal gets eaten first by a polar bear.
Ronald Reagan was subjected to ridicule when it emerged in later years, he had made decisions based on Nancy's astrolger's advice.
Barack Obama is currently relying on John Holdren.
Who will be better remembered?
With Holdren, Erlich, and that sort, the stars are fixed, about half a century ago. Was there an alien invasion about then? Childhood's Fixation, not Childhood's End.
==============
The label on the y-axis (vertical) has been changed.
First chart: Welfare impact (%GDP)
Next chart: Impact on welfare (equivalent income change, percent)
Why has this change been made?
~~~~~~~~~~~~~~~~~~~~~~~~
@12:15 Radical Rodent - - re: countries by type and all else NOT being equal
Here is my pair: Democratic Republic of the Congo versus Norway
“Maybe I am a bit dim, but” I don't imagine these places would change much if the global temperature changes by ½ degree up or down.
~~~~~~~~~~~~~~~~~~~~~~~~
The use of “centigrade” in the first chart suggests the person doing it is more-or-less my age. That's getting on in years. Further, these things ought to be expressed as “C degrees” and not as “degrees C” – one is a temperature change, the other is a temperature.
Kim, in the original TV series "Star Trek", Scotty, the Engineer, was always insistent that the shields/di-lithium crystals etc could not take further punishment. Jim Kirk always ignored him, and was proven right.
As someone with a bit of engineering knowledge, I resent this cavalier attitude towards engineers, however few climate scaremongers seem to have engineering backgrounds, apart from one notable railway engineer, though the punctuality of Indian railways is not something you would wish to set your calender by.
As science fiction writing has developed, with the assumption of travel faster than the speed of light, so climate science fiction writing, has developed with assumptions that are not sound, and should not be given the light of day.
"First chart: Welfare impact (%GDP)
Next chart: Impact on welfare (equivalent income change, percent)
Why has this change been made?"
The first is a short-hand for the latter, but readily misinterpreted.
Question is, who are the geniuses publishing the maroon diamond papers?
The Maroon Diamonds.
John F. Hulquist: ?
Little of what you say makes any sense. You obviously have little idea of the point I was trying to make: both countries in my example were of similar economic standing, with radically different climates – i.e. even with such different climates, their economies are not very different. Your choices have nothing whatsoever in common.
These attempts to globalise economic impacts are about as pointless as the attempts to globalise temperature variations. The only possible reason for it is to give some kind of absurd legitimacy to the UN's attempt to globalise world government. The impact of any weather, climate or other natural disaster will be felt most by the poorest countries. Hence few people die in a Miami hurricane when many die in the Manila version. The solution is simple in theory but impossible in practice and that is to raise the living standards and hence the infrastructure in poor countries to match their counterparts in the West. This will not happen in my lifetime as long as corruption rules the world as it surely does from the despotic rulers in Africa right through to the despotic bankers of the world bank. There is no solution to corruption just as there is no solution to so called global warming, just an awful lot of people making an awful lot of money at the expense of the poor by pretending to offer the pot of gold at the end of the rainbow. I am off now to stoke the fire with lovely life giving coal in an attempt not to become one of the 40,000 extra people predicted to die in fuel poverty this year in the UK.
I’d like to see a diagram of each of these economic models backtested.
The implication of the worst-case trajectory is that human welfare reached its optimum sometime in the 18th century.
If climate modellers genuinely required weatherproof IT to do their job properly their job would be done properly.
If tax payer funded economic climate modellers worked on a no-win, no-fee basis, tax payers would be happier.
If no one had made up man made climate change, no one would have noticed.
This post is incredibly bad. First, it somehow fails to highlight the single most significant change between these two graphs: The complete lack of regression in the newest version. The first graph was widely used to claim moderate global warming will be beneficial. The new version implicitly acknowledges that claim was baseless. That means years of talking points were predicated on junk science, and this post completely ignores it.
The post then makes the absurd claim:
Which is only true because the new version of the figure is still terrible. This post is trying to make a "Gotcha!" moment out of Richard Tol screwing up the data for the umpteenth time. The biggest outlier in this graph is the data point at -12.4. This point is given for the Maddison and Rehdanz (2011) paper. It was originally given as -11.5, but Richard Tol refused to provide calculations justifying that value. When he finally provided calculations, they gave a different value so they just changed it.
More importantly, however, is Maddison and Rehdanz (2011) don't estimate the same thing the other papers estimate. There are two types of GDP: nominal and purchasing power parity (PPP). Every paper but Maddison and Rehdanz (2011) estimated damages in nominal GDP values. Maddison and Rehdanz (2011) estimated damages in PPP GDP. That is the reason it is such an outlier. It is such an outlier entirely because Tol conflated two different things in that figure. If one insists on including the paper, one should at least try to convert the results to nominal GDP. When I did so, I got a value of -4.7%, far lower than the -11.5/-12.4% given by Tol. Tol responded to this by telling me:
Geray-Khamis dollars are used to measure PPP GDP, meaning Tol told me I should not mix PPP and nominal GDP, which is exactly what he has done.
There is much more to be said about this figure, including a number of obvious and stupid errors introduced in the newest version (some discussed here), but I give up. All of what I'd say ignores the fact this "correction" relies upon the IPCC report which in turn relies upon the work being corrected. If Tol didn't give answers you guys liked, everyone here would be condemning this. Just compare what was said with the Jesus Paper to what's been said about this work. And hen realize this figure has been changed time and time again to correct errors in a list of only ~20 data points, still has obvious errors and now no longer supports the primary talking point it was used for.
But nobody cares. I could talk until I am blue in the face demonstrating how pathetically incompetent this work is, how dishonest the promotion of the work has been and how hypocritical "skeptics" are being about it, but what's the point? People like Tol's conclusions, so whatever it takes to defend them is okay. It doesn't matter if they have to defend the very behavior they condemned just a few years ago.
Skeptics should stand for high-quality work, openness and transparency. This work fails on all three counts. That's really all anyone needs to know.
Don't we already have a 0.8 degree of warming? Where is that measurement - not an estimate? Oh, I forget that the dismal science never measures anything.
Curious George, if you want a really confusing problem, try figuring out what that 0.8C of warming means in relation to Richard Tol's figure. In one case, he presented the figure as being baselined to pre-industrial temperatures. In another case, is presented it as being baselined to current temperatures. In another case, he said he thought it is impossible to give a baseline for the graph because the different papers all use different baselines.
The IPCC WGII SPM uses the figure to make a claim about economic effects at certain amounts of warming, but it picked a strange baseline value for it. The baseline temperature it used wasn't used in any of Richard Tol's publications. What the IPCC did was pick one of Tol's baseline values, claiming his baseline was for current times, then subtract out .5C of warming and claimed the result was the baseline since pre-industrial times. That doesn't make sense.
There is no way to calculate a baseline value for Tol's figure since every data point on it uses a different baseline, and he didn't even attempt to account for that. Ignoring that, if there has been .8C of warming, subtracting .5C from current times wouldn't give us the baseline value for pre-industrial times. That means even if we ignore the problems with Tol's work, the values the IPCC came up via it are nonsensical.
On a different matter, it would definitely be interesting for people discussing the economics of global warming to tell us just what effects global warming has had on the economies of our planet. You won't see that happen though. There's really no good information on the effects global warming has had on economies, much less what effects it will have in the future.
I recommend reading Brandon's writings on this plot to anyone who hasn't done so.
But even if you take the plot at face value, do the underlying analyses behind each data point have any credibility? Personally I'm one of those who regard them às not meaningful - others have alluded to some of the reasons why.
Over and over again, paleontology shows the benefits of warming and the harms of cooling. Now, this demonstration is in terms of the biome, so we would probably need vegetable economists to put it into the terms needed for humans to understand.
===========
@Brandon
The IPCC left out the regression results for a simple reason: The added data require a different model specification. That was deemed "research" rather than "assessment". The paper with the new research was accepted a few months after the IPCC deadline.
I like plots like this that try to show the big picture. I can't comment on whether the data points are in the right place in Y but I prefer the version with error bars and a regression line. The value at 0,0 is inferred and is a "hard" data point - maybe the only hard data point there...(?). It also seems reasonable to assume a quadratic form - or at least a curve with a single maximum and no minima. Unless though one subscribes to the view that today's temperature is the optimum, then it is also implied that the maximum Y is at X>0.
I would like to see the temperature scale as an absolute, with values below the present average temperature shown (I appreciate data to the left of X=0 probably don't exist). Obviously the effect on GDP of temperatures prior to the end of the last glaciation would be significant.
Fascinating, but for policy it's obvious that we wish to predict Y from a given scenario in X, so a curve of some sort is required. It would also be informative to be shown just how likely, or how far away in time, are the outliers in X. Is +5.5 reasonable on any timescale?
Lord, he's aStern, and the vessel much enlightened with his loss.
The BRICs bucked up and tossed him off the hoss.
The bell Tols softly, muffled at two decibels.
Salt water taffy, tongue enclosed in crucibles.
=========================
I’d like to see a diagram of each of these economic models backtested.
The implication of the worst-case trajectory is that human welfare reached its optimum sometime in the 18th century.
Feb 7, 2015 at 7:59 PM | Chris Hanley
Love it!
Actually the so-called revision serves to give an impression that the latest IPCC means there will be a worse economic impact. The grouping of charlie foxtrot reports the alarmist economists have put together are on balance worse. And are allegedly based on the latest from the gold standard of climate science, the IPCC.
Why is the only data point that justifiably has zero uncertainty, i.e. by definition, (0,0) been excluded from the second graph? All models should have to go through that point and therefore its omission should be questioned. Of course, the reason is that if it is included the perceived trend/s is not as alarmist.
@ Brandon Shollenberger
As I write, I have yet to determine whether your arguments prove what you say they do. Colloquially, whether you're right or Richard is. But that's not the reason I'm writing right now.
Instead, I need to tell you "You're wrong and you're right."
You're wrong to say "nobody cares". We do care, very deeply.
You're right to say Richard has been telling us what I want to hear. I'm saddened to learn that papers that support my view might be open to more question than I had thought. But obviously I needed to hear that, so I care. I'm sure many others do too.
But so what? Don't write for me, or for them. Don't care about my caring. Write for the sake of truth. For integrity in the debate. For your own self respect. Write because you care, not because you think I do or do not.
As an aside, you'd written that nobody cares (about your arguments ) before I'd even read them. Possibly your contention was a bit premature?
When nobody dissents, I can zone out on the fine detail of work outside my area of expertise, confident that the overview is all I'll remember anyway, and that it must be right because no-one challenged it. Now I must examine the arguments more carefully, to determine which of you I judge to be right, or to admit to myself that I cannot tell. Blast. And thank you.
If anyone is interested; I've reworked the second graph adding point 0,0 and moved outlier as per Brandon Shollenberger, Bishop Hill, Feb 7, 2015 at 10:07 PM. A model fit has been applied with 95% confidence band.
https://pbs.twimg.com/media/B9VjaqGCYAAAZCz.png:large
I understood Radical Rodent's point and believed my comment would also be clear. My mistake. My intent was to make the point that the society (call it by another name if you like) is much more important than a 1 or 2 K change in average global temperature. Suggesting variations in climate as the “cause” of development has been done; see Environmental Determinism.
@JFH
Good comparisons are between Singapore and Malaysia, North and South Korea, Haiti and the Dominican Republic, Israel and Jordan, Botswana and Zimbabwe.
Thanks Jonathan.
DaveS:
Thanks! I haven't discussed the "underlying analyses" all that much, but I think the idea of basing any important conclusions on them is laughable. One of them was nothing more than asking people what they thought would happen, and that was ~20 years ago. In what world would we base our conclusions of the effects of global warming on some people's opinions of it from ~20 years ago?
Jit:
Sadly, the value at 0,0 isn't a "hard" data point at all. It's arbitrarily chosen and has no basis in anything. Each data point in that figure has its own temperature baseline (as well as several other baselines). Richard Tol didn't even attempt to account for that. He just ignored it. That means the figure's temperature baseline isn't the same as the temperature baseline used in the papers, meaning the point at 0,0 isn't "hard." It's purely arbitrary.
Johnathon Drake:
I don't agree the fit you do is remotely appropriate. I've explained multiple reasons for why doing a regression on these data points is inappropriate, but leaving that aside, there are several glaring problems. First, as I explained to Jit, the point at 0,0 is not a real data point. It's purely arbitrary, and any calculation which relies upon it is incorrect.
Second, you offer absolutely no justification for the model you chose. Richard Tol used a quadratic fit (though he called it a parabolic fit). That means it had three parameters. You used a cubic fit, meaning you have four parameters. Neither of these models has any inherent validity, and neither of you have done anything to show they are appropriate. Even worse, you've simply changed the model being used without bothering to say so, much less explain way. Ultimately, you two haven't shown the regression or uncertainty margins you display have any bearing on reality, and the models you two used are known to be inappropriate for such limited data sets with wide distributions.
Third, the entire basis for your figure is misleading. I never said the Maddison and Rehdanz (2011) data point should be used. I indicated I believe it shouldn't be used because it represents something different than the rest of the data points. I suggested converting the data point is a minimum of what should be done, but if we do it, we need to be clear about what we've done. It is highly misleading to claim the point is an outlier without explaining the point doesn't represent the same thing as the rest of the points.
Fourth and finally, I've repeatedly pointed out there are multiple problems with the figure. Correcting one of them and leaving the rest while pretending you've somehow addressed what I said is incredibly disingenuous.
Put simply, you made no attempt to justify the inappropriate curve-fitting you performed, made no effort to correct for a number of problems in the data which have been pointed out, used a data point which wasn't real and after all that, merely concluded a data point which doesn't represent the same thing as the other 19 data points is different from those other 19 data points.
Leo Morgan:
Comments like mine are a bit unfair in that there are a lot of people who read material but don't comment on it. This silent majority could well care about things the people I see commenting don't (seem to) care about. It's like the old saying, "The absence of evidence is not evidence of an absence."
So while I can only comment on what I can see, maybe I ought to think more about the possibility what I see isn't all that representative. I don't know. Going that route just seems fraught with difficulties. Limiting myself to what I see people talk about may not be as right, but it is certainly more practical.
That's largely the plan. I write about what interests me. However, the thing which interests me the most is the world seems, quite literally, insane. As such, when I see things that seem crazy, such as the reactions my criticisms of Tol have received, I care. I care because they show something of what bothers me about this world: People often don't care about being right, consistent or even coherent.
Eh. I've written about this topic for something like a year now. It is easy to verify most of the things I've said on this topic. It is easy to show many parallels between what I've said on this topic and what people have said many times in the past. Despite that, there has been almost no discussion of any of this. Practically nobody has (as far as I can tell) spent any time looking into the matter. If I didn't keep bringing it up, there'd probably be nobody on the "skeptic" side who knew anything about it.
All told, there has been one blog post written because of my efforts to examine this issue. It was on this site, which I think is to the credit of this site. Still, I could pick any of a hundred less important subjects I know about in the global warming debate, put far less effort into examining them yet get far more discussion of them.
There is history which shows this is a pattern. I think I'm already saying more than I ought to so I'll keep it brief. I've criticized Tol on three different issues. In one case, I defended an unimportant "skeptic" paper against absurd criticisms he raised. It received attention and support. In another case, I defended Cook et al's consensus paper against absurd criticisms Tol raised. It received almost no attention or support. In this final case, I criticized Tol for absurd behavior and work he did. I received almost no attention or support.
In other words, when Tol did stupid stuff for the "warmist" side, my responses were liked and appreciated. When Tol did stupid stuff for the "skeptic" side, my responses were disliked and ignored.
You're welcome, though if you truly do examine your beliefs because of what I've said, I owe you some thanks. People do do that from time to time, and it is quite gratifying.
What an entertaining little pillowfight between Tol and Schollenberger, arguing over which of their junk economic models should be used in conjunction with the junk climate models. It's like watching a pair of astrologers debate over whether the position of Betelgeuse is auspicious for Capricorns or Geminis.
Bloke in Central Illinois, your comment is baffling:
I don't have an economic model of my own. I've never promoted any economic model. Pretty much the only things I've ever said about economic models were negative.
I don't know what you think you've seen me say, but it isn't anything I've actually said.
@ Brandon Shollenberger
“Correcting one of them and leaving the rest while pretending you've somehow addressed what I said is incredibly disingenuous.” I have never claimed anything of the sort. I commented that I moved the apparent outlier to the value you stated and referenced it back to you. If a different interpretation has been made then it is unfortunate and unintentional.
After that and adding the 0,0 point, I applied the specified local polynomial kernel regression (kernel smoother/filter) to hopefully find a reasonable representation of the possible underlying trend. The kernel was Gaussian. “You used a cubic fit, meaning you have four parameters.” – no, I didn’t.
In principle, the change in impact on welfare has to be zero for zero change in temperature. It is the point of reference. I do not know whether that has been done correctly; I assumed it was the one of the few things that would be. If as you say, this is not the case (re your comment to Jit) and, “It's purely arbitrary, and any calculation which relies upon it is incorrect”, is accurate then the whole exercise probably is worthless, but then so are IPCC claims derived from it. Indeed, arguing over curve fitting is therefore just as meaningless.
Jonathan Drake:
You're right. I didn't see your caption (I didn't think to scroll down to see it) so I went by the form of the regression line. The problems are still the same though. You've done nothing to establish your choice of model is appropriate. You haven't even explained why you used it instead of a different one (such as that Richard Tol used). Even worse, that model is known to be inappropriate for this sort of data set.
I've made this point over and over and over. I have repeatedly said Richard Tol's figure is meaningless as the data points cannot be compared the way he has. I've said it should not have been added to the IPCC report because of this, and I've said it certainly shouldn't have been added after the last round of external review so nobody could point out the problems with it. I've also said the IPCC WGII SPM statement which relies upon it is unsupported because of this.
In other words, you're saying the same thing I've been saying for months. I agree with what you're saying. I'm just one of the few people you don't need to say it to :P
Though I'm not sure why you say, "but then so are IPCC claims derived from it." The entire reason this discussions has been happening is I've criticized the IPCC report for using this work. There is no "but then" to this. The IPCC changed its report to add bad work and unjustifiable conclusions after the last round of external reviews, and I've been criticizing them for it. It's the sort of thing we should all be able to get behind.
A most powerful questioner of the reports of harm, which has been greatly exaggerated.
========================