I thought I would advertise a post by Steve Keen, that may be of interest to some of my regular readers. It’s about Neoclassical Economics of Climate Change and is extremely criticial of the assumptions used to drive Integrated Assessment Models (IAMs). I’ve written about these a number of times myself, and have also been rather critical.
For example, IAMs don’t self-consistently model the evolution of our economies. They typically assume our economies continue growing and then estimate – using a pre-defined damage function – how much damage is done by climate change. Additionally, even though the damage functions are non-linear, they still produce modest levels of damage even for large changes in global temperature. For example, this paper points out that the damage function in DICE (William Nordhaus’s model) estimates damage due to climate change at 4.7% of world output at 6oC of warming, and that it would only halve world output at 19oC. In a sense, these models would conclude that the impact of climate change is modest, by construction.
On a similar note, in a recent paper, William Nordhaus (who won the Nobel Memorial Prize in Economics) estimated that the economically optimal pathway would lead to 3.5oC of warming (with an standard deviation of 0.75oC). So, a level of warming that many scientists would regard as potentially leading to catastrophic impacts is optimal, according to an economic model.
Steve Keen’s post highlights a few other things that I hadn’t entirely appreciated. It seems that these models don’t take into account how inter-connected our economies are. They seem to assume that there will be some activities that will be largely unaffected by climate change, and that the economic impact of climate change depends largely on the economic value of the activities that are directly impacted. But if climate change severely damages agriculture, it’s not just the agriculture sector that will be adversely affected. If climate change severely impacts regions of the world that aren’t particularly wealthy (as seems likely) it’s not only these regions that will be adversely affected.
Something else I hadn’t appreciated is that the damage function used in IAMs is calibrated by considering how economic activity today varies with temperature. In other words, how economic activity varies with climate now, is used to estimate how global warming will impact economic activity. It may be a reasonable first guess, but global warming doesn’t simply mean that as a region warms it will simply become equivalent to a similarly warm region today. Not only is this clearly too simplistic, there are also many other impacts. You might think that economists clearly don’t hold such simplistic views, but Steve Keen’s post reminded me that we’d engaged in a discussion with Richard Tol in which he appeared to express exactly this view.
Anyway, I’ve already said too much. I encourage people to read Steve’s post. I’ve put another link below, plus links to some other articles that may also be relevant. I should also probably clarify that there are two quite distinct types of IAMs. There are some that are used to actually model the evolution of energy systems, and others that are used to do cost-benefit analyses. The criticisms discussed above refer to the latter, rather than the former. I’m also not an expert at this, so if I do misunderstand how these models work, I’m more than happy to be corrected.
Links:
The Appallingly Bad Neoclassical Economics of Climate Change – Steve Keen’s post about IAMs.
IAMs – other posts I’ve written about IAMs.
Fat tails, exponents, extreme uncertainty: Simulating catastrophe in DICE – paper by Ackerman, Stanton and Bueno about the DICE damage function.
Projections and Uncertainties About Climate Change in an Era of Minimal Climate Policies – William Nordhaus’s paper suggesting that the optimal pathway would lead to about 3.5oC of warming.
Limitations of integrated assessment models of climate change – paper by Ackerman et al.
Climate Change Policy: What Do the Models Tell Us? – paper by Robert Pindyck suggesting that IAMs have fundamental flaws.
...and Then There's Physics 
Having participated with this earlier I will point out – again – that thermodynamics affects economics and in particular defines the limits to how our money can operate. Steve is thinking about this.
In the long run it is going to be obvious to everyone – just as it is obvious to me. I’ll be long dead when that becomes true but it is not possible to ignore the fact that the theories of economics and the “theory” that economists currently have about what money is – violate both the first and second laws. If it isn’t obvious yet it is because it isn’t the laws that break. The thing that breaks can be something else, somewhere else, probably belonging to someone else.
257 Trillion in debt and that was before COVID. Inequality that would gag a Python – and the mainstream economists apologize to the Queen for their lack of foresight and then go back to doing the exact same things that got them into trouble.
There will be no visible recovery from what we’ve done to ourselves for quite a long time. A century may pass before we can retrieve anything like a normal existence, as the COVID crash segues into the climate crunch or as is more likely now – a climate catastrophe. Might be longer when I think about it. We have done ourselves and the planet significant harm.
More to the point, the economics professors of the past 50 years have done significant harm. The number of times I have HEARD “not another engineer” from one or another of them is – for someone who does not often interact with them – a significant bit of data. Their lamentable lack of attention when they hear the disagreement – almost invariably based on the violations of thermodynamics inherent in their assumptions – is a real thing. Eddington was right about what happens to a theory that tries to violate those laws.
Mainstream economics is a poison affecting the body politic. Scientists and Engineers have a strong responsibility to confront their pronouncements and shred their pseudo-scientific fantasies.
Now – back to the book.
Just to be clear, because when it is framed as you do, the casual reader may not realize the implied math behind these output delines: Those % economic output declines due to damages are relative to what they otherwise are assumed to have been.
So, for illustration, say it takes us 200 years to warm the planet 19°C. The economic damage of 50% of GDP is relative to what it is assumed to be otherwise in an unchanged climate. Assuming normal economic growth of 2%/year, economic out would be 52x current output. So, the a 50% damage would suggest that economic output at 19°C would be 26x current GDP. And, working backwards, annualized growth over the period would have been ~1.65%.
Which just makes the conclusions even more starkly absurd.
Now, this is somewhat an artefact of the Nordhaus model being applied outside of scenarios for which it was ever intended, and, as I understand it, has been modified for since. But it has always been an alarm bell that “something is wrong here”.
I saw Stern talking about this in Copenhagen in 2009, during an early-year conference in the run up to COP15. (Nordhaus had spoken the previous day, for what it’s worth, and it.)
I mention this just because IAM deficiencies have been highlighted for quite a while – including ones far more problematic than the example above. It gets tedious pointing some of it out. Case in point, ATTP, you mention that you “didn’t fully appreciate” several points Steve Keen is (rightly) highlighting. You’ve seen the same critiques for close to a decade in the Skeptical Science authors’ forum. Where a good deal of the response was little better than “We don’t like Richard Tol, but Chris Hope is the man!”.
For instance, on the “agriculture as %of GDP”, here’s me in 2013:
Not a complaint. Just saying.
rust,
Yes, that’s a good point. I have indeed heard some argue that even if it is 10s of percent, it’s 10s of percent of a much bigger global economy and so people are going to be wealthier in the future anyway. This ignores that economic growth is assumed, rather than self-consistently modelled. There’s a 2015 paper by Frances Moore and Delavane Diaz that does try to estimate the impact on economic growth itself. They conclude that
Should this be linked to the previous post? I guess the question is – are modelers promoting results of IAMs directly to inform an idea of optimum policy? Or do they recognise the limitations of these models, but suggest they provide some insight?
For example, the Nordhaus paper finds 3.5C to be optimal but also that there isn’t a huge difference in economic outcome compared with scenarios reaching net zero by 2040. Even ignoring damages the cost to pursuing 2040 net zero looks to be only about 3% relative to the baseline. So the paper can support an interpretation that strong mitigation is very justifiable on a cost-benefit basis once you take into account the obvious long-tail risk potential not included in the IAM.
I’ll also point out that there are some oddities with regards the optimal scenario looking at Table A-7 in the paper. The CO2 emissions profile is very much in the realm of 4.5W/m2 scenarios, which produce 2.5-2.8C warming in the SSP database. The highest 2100 CO2 concentration in 4.5 scenarios in the SSP database (I think all using MAGICC median configuration) is 575ppm, compared with 633ppm in Nordhaus’ model, so looks like the carbon feedbacks are on the strong side, possibly accounting for permafrost etc. But also the total forcing calculated as of 2015 is 2.96W/m2, which suggests a very weak or non-existent aerosol forcing.
A more standard and complete implementation of Nordhaus’ optimal scenario would produce 2.5-2.8C warming at 2100. Presumably that would affect the damage calculations, but I think it would still shift the optimal down to a lower temperature.
One important facet of this that some may not be aware of (I wasn’t until very recently) is that IAMs produce a strongly logarithmic relationship between carbon price and temperature, and between economic growth and mitigation scenario level. Looking at SSP2 scenarios in the database, it actually seems to be pretty easy to get down from the 7.0 baseline to a 6.0 mitigation path with negligible cost, which is a larger drop than it seems given that 6.0 scenarios really tend to reach 5.3-5.5W/m2 at 2100 (The 6.0 refers to forcing at stabilisation in the 22nd Century). Then it’s a little more difficult to get from 6.0 to 4.5. But then going from 4.5 to 3.4 the cost can be 10 times as large as going from 7.0 to 5.3-5.5, despite being almost half the forcing distance. If you don’t take into account non-linear damage potential on the other side of the equation you’re always going to find an optimum pathway stuck around 4.5W/m2.
Paul,
I think a related issue is that IAMs assume a baseline. I think this is often assumed to be somewhere close to RCP8.5. Even though the RCP8.5 discussion was remarkably frustrating, the critics are probably right that this isn’t a reasonable baseline. One consequence of using it as a baseline is that it then makes it seem much more costly to mitigate than is likely to actually be the case. If RCP6 is a more reasonable baseline, then getting to RCP4.5, or lower, will be estimated to be less costly than if RCP8.5 is used as a baseline (at least, I think that’s right).
ATTP,
In general a higher baseline is probably more difficult to mitigate but SSP3 seems to cause the most mitigation problems in the SSP database, and that’s RCP7 level in most models. That’s relevant for the old Riahi et al. 2011 RCP8.5 because that was basically SSP3 with slightly greater economic growth. None of the SSP IAMs consider it feasible to go below 3.4W/m2 under SSP3 and even at that level there’s a 10-13% economic hit relative to baseline by 2100, mostly in poorer countries. Perhaps the primary focus right now should be avoiding the SSP3 world since we seem to be drifting in that direction.
I’m not sure these kind of studies generally do use a RCP8.5 level baseline. Nordaus’ baseline is RCP7 level, I think that’s been widely recognised as a median baseline level for many years.
“Mainstream economics is a poison affecting the body politic. Scientists and Engineers have a strong responsibility to confront their pronouncements and shred their pseudo-scientific fantasies.”
Now, now– how is Piketty going to make a living if scienists and engineers notice he’s part of the climate economics discourse?
https://vvattsupwiththat.blogspot.com/2020/06/death-and-carbon-taxes.html
What is needed:
Kate Raworth, Doughnut Economics – https://www.kateraworth.com/doughnut/
It gets a little textbook-ey but a bullseye imnsho. That and Ostrom’s circular economy,* Annie Leonard’s (now with Greenpeace) Story of Stuff, Monbiot’s trenchant criticisms.
I realize I’m usually OT on the technical and intellectual focus of this comment community. I know talk is cheap. I am aware of the value of aTTP’s immensely valuable service in using the language of academia to demonstrate the shortcomings of continuous expansion models of economics and other common assumptions that lead to a dead end for humanity.
But the above is where I think we must go if we want anyone with a life expectancy of more than 20 years to have a chance.
I came here to make a blunt comment strongly recommending Raworth, but as I search around I am finding this field a little richer (which should be good) than I thought it would be. So I’ll stop for now and see where it goes next. Listening to one Johan Rockström’s Ted Talk on the way.
*This is an odd result of a search, note the inclusion of Alan Turing, and the Santa Fe Institute, a project of my father and Ken Arrow, and others. I found it thought-provoking.
Click to access What-is-complexity_Ed-version.pdf
Ostrom’s connection with the circular economy appears to have been largely erased from the internet. I find this curious. Am I making a mistake here?
—
Further wanderings, just for fun, quoting Bilbo Baggins:
Roads go ever ever on
Under cloud and under star,
Yet feet that wandering have gone
Turn at last to home afar.
Eyes that fire and sword have seen
And horror in the halls of stone
Look at last on meadows green
And trees and hills they long have known.
Would that it were that simple. But the forces of evil are all too obvious.
Not the ordinary political scientist:
https://en.m.wikipedia.org/wiki/Elinor_Ostrom
Despite the claims that working together solves resource limitation problems, there’s massive evidence of failures. Here is a multi-list of links to make the point:
https://bravenewclimate.proboards.com/thread/159/climate-change-emergency
Since the carbon dioxide concentration now obtained was last experienced in the mid-Pliocene, when sea stands were about 25 meters higher than now, we expect sea level rise.
From the last link on
https://bravenewclimate.proboards.com/thread/159/climate-change-emergency?page=7
over 10% of the world population lives within but 10 meters of the current sea stand. So expect continuous disruption.
Optimism is only for the Polyannas.
I thank Keen for his concerns and suggest he not give homework to others.
Paul,
Thanks, I wasn’t sure. That does somewhat weaken some of the complaints about RCP8.5.
I find these economist’s views of the future lacking in imagination and in historical perspective; this narrowness is codified and enforced by the tools and language of economics including IAMs. They see the last 70 years of history as a straight line on a log plot of GDP, then extrapolate that line. It is perhaps an error to enter that frame at all, but is that extrapolation likely? The next 70 years could be like the last 70, but what if they aren’t? In that scenario, creating a massive upheaval in the natural world seems like a spectacularly bad strategy.
Only in some techno-optimist view of the future where we are no longer dependent on the external world does ‘nobody really needs to go outside anyway’ and ‘farms can be moved to a different bit of the world’ look OK.
If you picked a random 80 year period of the last couple of millennia, a lot of them were pretty bad. Plague, war, and misrule feature prominently. Can we really escape nature and societal failure? Disease is messing with us pretty hard right now. What if world GDP stagnated? How do you model conflict caused by flooded people having to move, or crop failures? Instruments like IAMs as made and played by economists aren’t what you need; maybe a historian or anthropologist would give a more faithful account of this future. Or for that matter, people who are paid to imagine the future in words, images and games.
Also, taking world per-capita GDP as a proxy for human well-being is pretty limited; can you really paint human experience with that one hue? What this does is ensure that only things rich people buy and sell are valued much. Then you discount the future beyond 20 years as largely irrelevant (at least in Nordhaus’ but not Stern’s). The conclusion is baked into the methodology and assumptions.
“What this does is ensure that only things rich people buy and sell are valued much.”
B-I-N-G-O
“The next 70 years could be like the last 70, but what if they aren’t? ”
Here: https://surplusenergyeconomics.wordpress.com/
is someone who argues that :
‘The first principle is that all forms of economic output – literally all of the goods and services which comprise the ‘real’ economy – are products of energy. . . .
The energy that is consumed in the supply of energy therefore comprises both a capital (investment) and an operating component.
This principle is central to the established concept of the Energy Return on Energy Invested (EROI or EROEI), in which the consumed, cost or invested component is stated as a ratio. In Surplus Energy Economics (SEE), the cost element is known as the Energy Cost of Energy or ECoE, and is stated as a percentage. . . .
it is that ECoE (the Energy Cost of Energy) is rising, relentlessly and exponentially. The exponential rate of increase in ECoE means that this cannot be cancelled out by linear increases in the aggregate amount of total or gross (pre-ECoE) energy that we can access. The resultant squeeze on surplus energy has been compounded by increasing numbers of people seeking to share the prosperity that this surplus provides.
As a result, prior growth in prosperity per person has gone into reverse. People have been getting poorer in most Western advanced economies (AEs) since the early 2000s. With the same fate now starting to overtake emerging market (EM) countries too, global prosperity has turned down. One way of describing this process is “de-growth”.’
The conventional economists’ answer?
“This reflects a question I posed here a while back – if electricity is (say) 4% of GDP, how much GDP would we have left if power supply collapsed? The conventional economists’ answer is 96%. The real answer is somewhere near 0%.”
The gist that I get from this long ramble is that this guy doesn’t like neoclassical economics. Neoclassic economics is likely not good at predicting the economic effects of climate change, but this is not because neoclassic economics is bad. It’s because the world’s economy is an even more chaotic system than the climate and nothing will be able to predict it. He chides NordHaus for conducting only two surveys of “expert opinions”. Has this guy ever heard of Philip Tetlock?
> Has this guy ever heard of Philip Tetlock?
The first conclusion one should take from Philip’s work on expert opinion is to disregard it.
@mike dombroski
Actually Mike – Neoclassical IS bad. It is based on the use of supply and demand driving production, consumption and pricing of goods. Then it leaves the prices entirely up to the market. Two out of Three ain’t bad but using supply to determine price can be analyzed as a violation of the first law.
Ownership of a thing cannot make money, no more than energy can make more energy. Yet using a shortage of supply as a reason to drive up the price does just that. Money represents work done, energy and you cannot make more of it you have to do work. Having something is not work.
That is a fundamental error that the neoclassicals make, but now that they have been subverted by the neoliberals – the market-uber-alles crowd – the results are even worse. Stiglitz pointed out – correctly in my opinion – that the externalities that the market uses to dismiss all the market failures are in fact almost ubiquitous. Things like CO2 happening to cause warming is a market failure the size of our civilization and the market wants to ignore it and is funding the people trying so hard to deny it.
Neoclassical economists and others worked pretty hard to ignore the truth that Henry George explained to them. Keynes denied the monetary innovation provided by Silvio Gesell on the basis, basically, of money being a store-of-value which is a violation of the SECOND law.
Neoclassical economic theory founders on the rocks of thermodynamics. Eddington had a point.
Should also point out that Keen accurately predicted the GFC.
@postkey
Thank you for mentioning Surplus Energy Economics (which I had not seen before today). I’m trying Dr Morgan now. It is clear that he understands what is happening and why it is not going to be the same in the future. The neoclassical mob can’t even guess at the scale of failure they are attempting to get our civilization to embrace.
Feeling warm yet?
https://phys.org/news/2020-06-uk-increasingly-40c-temperatures.html
The level of skill that IAMs show might be revealed by the projections they have made, and the policy implications that derive from them, when considering a global pandemic.
Before COVID19 there were attempts to model the economic impact of such and event,
Slightly depressing reading, as its recommendations match well with what some Nations have NOT done.
Click to access cesifo1_wp8016.pdf
–
“Several suggestions that can minimise the economic risk of epidemics, as is measured here,
include: Investment in prediction of disease emergence and appropriately designed Early
Warning Systems that can shorten the period in which there are declines in economic activity;
active minimization of transmission pathways thereby reducing exposure to the disease in
areas that were not exposed to the initial hazard (for example, by timely global reporting);
reducing vulnerability to disease outbreaks by improving public health systems or decreasing
other ‘root causes’ of vulnerability (such as poverty);”
there are now ongoing efforts to model the actual impacts,
Click to access SP257.pdf
and economic modelling is even being applied to estimate the advantages of specific actions, such as mandatory mask wearing.
https://markets.businessinsider.com/news/stocks/goldman-sachs-national-mask-mandate-save-gdp-hit-economic-impact-2020-6-1029355543
I have tried to post some links to past and current efforts at modelling the economic impact of a global pandemic.
For some reason the posts disappear, too many pdf links ?
But the takeaway is that they are not skilful in projecting the impacts, but do make recommendations about establishing global monitoring and information sharing, along with improving healthcare systems and reducing wealth disparities as key factors in reducing the impact.
They also emphasise the importance of National governments managing the behavioral changes that would be needed to reduce transmission, even at the expense of economic damage from such measures.
Perhaps this will make it through the filter as an example…
https://markets.businessinsider.com/news/stocks/goldman-sachs-national-mask-mandate-save-gdp-hit-economic-impact-2020-6-1029355543
A national mask mandate could potentially substitute for lockdowns to curb the spread of coronavirus, Goldman Sachs Chief Economist Jan Hatzius wrote in a Monday note.
Goldman’s baseline estimate is that a national mandate could raise the percentage of people who wear masks by 15 percentage points and cut the daily growth rate of confirmed cases by 1 percentage point to 0.6%.
That could save the US from a 5% hit to gross domestic product that could result from renewed lockdowns, according to Goldman Sachs.
post with links are vanishing…
But this story is spreading in the business media –
A national mask mandate could potentially substitute for lockdowns to curb the spread of coronavirus, Goldman Sachs Chief Economist Jan Hatzius wrote in a Monday note.
Goldman’s baseline estimate is that a national mandate could raise the percentage of people who wear masks by 15 percentage points and cut the daily growth rate of confirmed cases by 1 percentage point to 0.6%.
That could save the US from a 5% hit to gross domestic product that could result from renewed lockdowns, according to Goldman Sachs.
> post with links are vanishing…
Strange. I fished out from the spam comments from izen, Dave, and jack. Haven’t gone through the other pages. There’s a lot of spam.
“The gist that I get from this long ramble is that this guy doesn’t like neoclassical economics. ”
he is welcome not to like it. Until he builds an IAM that does economics differently he should be thanked for his concerns.
IAM is a tool, the same way a GCM is a tool and covid models are tools.
Skeptics will always have a field day with models because models are always wrong.
Their job is to make a better tool. Until they do, you can simply thank them for their concerns
and carry on.
> Until he builds an IAM that does economics differently he should be thanked for his concerns.
Not sure if auditing only delivers concerns, Mosh. Specific criticisms ain’t manifesto slogans.
Steven,
The point I would make is that modellers should be willing to highlight their model strengths and weaknesses, and also where their model is more likely to be valid and where it probably fails. It seems obvious that the damage function is not appropriate above some temperature changes, but I’ve yet to see a demonstration of what range of temperature changes it is valid for. Similarly, are they really suggesting that they can estimate damages in 100 years time, or are these models best used to calculate an social cost of carbon today?
I’ve noticed something else about Nordhaus’ model setup. As pointed out above, his Optimal scenario is basically a 4.5W/m2 scenario, which would place it at about 2.5C in 2100 in standard assessments, rather than 3.5C. I pointed to a high-end carbon cycle and apparent lack of aerosol forcing as contributors to this discrepancy, but another major factor is how Nordhaus incorporated other forcings.
It seems his model is only able to predict CO2 emissions so all other forcings are added from external sources. The paper states that he uses AR5 projections to do this. Based on the numbers it looks like he used RCP8.5. As we know this is a high end scenario which is not really in-sync with his baseline CO2 emissons, but more importantly he appears to implement exactly the same numbers in his optimal mitigation scenario. This means that the mitigation policies which reduce fossil fuel CO2 emissions to 20% of the baseline by 2100 have zero effect on methane emissions, which is impossible. Fossil fuels are the dominant source of methane emissions in RCP8.5.
And that doesn’t just affect the reported temperature for the Optimal scenario, it would affect the optimisation procedure itself. Mitigation is less valuable if it isn’t allowed to affect a large proportion of the forcing. Perhaps better than looking at the eventual temperature is to look at the carbon price they find optimal, and it isn’t as weak as 3.5C would suggest. It features a sharp increase to $35 at 2020, far in excess of what we actually have in 2020, then continuing strong increases through the century to $260 by 2100. Comparing with the SSP database this is very consistent with 4.5W/m2 scenarios and a small number of 3.4W/m2 scenarios. The question is would the optimisation have supported a stronger carbon price if this had actually been allowed to affect non-CO2 forcings?
The summary is that the problem of the 3.5C finding actually has not much to do with economics, but rather the suboptimal way economics were translated to final temperature change. It can be reasonably argued that the damages used/produced by IAMs are too weak but an important point to make is that, even with these weak damages, IAMs still support following a mitigation policy trajectory far in excess of where we’re currently heading.
Thanks Willard.
Well, that’s the Internet for you 😦
-I think Keen’s argument mostly boils down to the damage function being wrong (or poorly justified): an important input parameter, rather than that the IAM approach per se, is wrong.
-The social cost of carbon today is mostly dependent on the expected damages in the future so you can’t really do one and not the other.
But obviously the Nordhaus ones don’t care at all about what happens in 100 years time because the ~4.5% effective discount rate makes anything past 2040 irrelevant.
PaulS – interesting. To add to your point, models struggle to compare the future costs of technologies on learning curves (solar, wind+batteries) to the cost of natural resources (fossil fuels). Already fossil fuels are struggling to compete, and they will lag even further with time. Our current path is highly sub-optimal.
Paul,
Yes, I think this is an important point. IAMs suggest that we are still essentially paying too little for the use of fossil fuels. Even those who aren’t in favour of strong mitigation should at least recognise that this implies that we are developing a less effecient economy than if we were try to properly price the use of carbon. It doesn’t seem, though, that this is a common conclusion.
Ben,
True, and is one reason why these types of analyses often suggest that the long-term cost of cimate change will be relatively small, almost by default.
I am not sure if I have seen it explicitly stated or derived from IAMs, but there is often the implication that the cost of abandoning hydrocarbons as our primary source for energy and transport systems would be far higher than the cost of climate change.
At least in the short (5 year plan) term that business is concerned with.
Pingback: 20.07.02 Neoklassiset mallit ilmastonmuutoksesta ovat älyllinen itsemurha – Kirjoituksia
Pingback: 2020: A year in review | …and Then There's Physics
Pingback: Deferential? | …and Then There's Physics