Emission scenarios and economic impacts of climate change

[waldo]

I didn't even read the article. Someone else linked to the image on another blog.

ok... that's fair! Of course, all I've got to go by is the link associated to the image you posted... which was from the site and author as I described. You could have prevented any confusion by either linking to the graphic from the study itself... and/or identifying the actual source in the first place... which you didn't do. As you read, I posted particulars from that study... a health/health care related study; findings that present an assortment of options to explain the presentation within that graphic.

now, again, you have yet to explain why you've posted the graphic, particularly in relation to the thread's, your thread's, focus on emission scenarios and the economic impacts of climate change... still waiting:

[-1=e^ipi]

I skimmed this paper today when I was looking for empirical estimates of how CO2 emissions enter the production function.

http://ageconsearch.umn.edu/bitstream/103412/2/Zhao-1-the%20impact%20of%20The%20cost%20of%20CO2%20emission%20cuts%20on%20income.pdf

"To reduce emissions 50% below 1990 levels by 2050, the economic cost per year for developed countries is about 0.3% reduction in GDP per capita which represents a 15% slowdown in economic growth."

Over a period of 60 years, 0.3% reduction in GDP per capita results in society being 16.5% poorer.

Edited September 30, 2015 by -1=e^ipi

[Bonam]

I skimmed this paper today when I was looking for empirical estimates of how CO2 emissions enter the production function.

http://ageconsearch.umn.edu/bitstream/103412/2/Zhao-1-the%20impact%20of%20The%20cost%20of%20CO2%20emission%20cuts%20on%20income.pdf

"To reduce emissions 50% below 1990 levels by 2050, the economic cost per year for developed countries is about 0.3% reduction in GDP per capita which represents a 15% slowdown in economic growth."

Over a period of 60 years, 0.3% reduction in GDP per capita results in society being 16.5% poorer.

Not sure I agree that this paper has any real validity. For one, what is the justification behind their production function (eq 1)? Secondly, what is the reasoning for assuming that alpha + beta + gamma = 1 (page 6)?

Regardless of the math there, I think it's fundamentally not possible to predict the economic impact to anywhere close to this level of precision because we simply do not know how technology will change over the next 35 years. It could be that in 5 years we'll discover the secret to cheap, plentiful, easily deployable energy storage and build solar panels to provide 100% of all our power. Or it could be that in 5 years we'll have another nuclear disaster, all nuclear power plants on Earth will be shut down as a result, energy storage research will stall, and the only viable source of energy will be fossil fuels. The costs of cutting emissions under these two scenarios are orders of magnitude different.

[-1=e^ipi]

For one, what is the justification behind their production function (eq 1)? Secondly, what is the reasoning for assuming that alpha + beta + gamma = 1 (page 6)?

That's a Cobb-Douglas production function. It's pretty standard/common. The assumption of alpha + beta + gamma = 1 is one of constant returns to scale (so if you have twice as much labour, physical capital and energy then you will have twice as much economic output). Constant returns to scale is not an unreasonable assumption. Obviously there are some limited factors of production such as land, which means that alpha + beta + gamma < 1, although there are also some benefits to having more population density.

I could go into detail of why people tend to use Cobb-Douglas production functions, but basically it is easy to use and has desirable properties.

Obviously, the functional form is likely incorrect (and I've seen a fair amount of evidence that it is incorrect, particularly how energy enters the production function), but it isn't that bad. You have to remember that there is a tradeoff when making a model more complicated since you would make it harder to estimate, use and share results with other people, but you might be able to better explain observations with a more complicated model. That's why people like Karl Popper argue that various theories have predictive power and it makes sense to give preference to the null hypothesis / simpler beliefs until proven otherwise and not to accept unfalsifiable beliefs. That's why things like Occam's razor and Akaike's information criterion exist. It's why in physics, people will often use Newtonian physics or neglect air resistance when making calculations.

Regardless of the math there, I think it's fundamentally not possible to predict the economic impact to anywhere close to this level of precision

I don't think the paper claims super high amounts of precision. Anyway, you can still make predictions using empirical data which have high levels of uncertainty. And you can use those predictions to determine a unique social welfare maximizing CO2 emission tax assuming you define the social welfare function to have a reasonable level of risk aversion (say a coefficient of relative risk aversion somewhere between 0.5 and 2; I would suggest a value of 1 for reasons I have given in other threads and also because logarithmic consumption utility combined with a Cobb-Douglas production function makes maximization of the social welfare function relatively easy to perform).

Edited September 30, 2015 by -1=e^ipi

[waldo]

I skimmed this paper today

your referenced paper is not peer-reviewed... the author certainly has credentials and has very good 'rate my professor' ratings - it just doesn't appear she has much of a publishing record. Out of the, quite literally, thousands of papers related to the economic impact of CO2 mitigation, why would you settle in on such a paper?

.

[-1=e^ipi]

Out of the, quite literally, thousands of papers related to the economic impact of CO2 mitigation, why would you settle in on such a paper?

Top of google search engine results. And the methodology was roughly what I was looking for, since having CO2 directly in the production function makes it easier to predict economic impacts of CO2 mitigation policies and determine an optimal level of CO2 emission taxation.

The fact that the production function is Cobb-Douglas makes the empirical analysis easy to follow / reproduce and, if combined with a social welfare function where consumption utility is logarithmic, it can make it relatively easy to determine an optimal level of CO2 emission taxation. So that was an added bonus.

Edited September 30, 2015 by -1=e^ipi

[waldo]

Top of google search engine results. And the methodology was roughly what I was looking for, since having CO2 directly in the production function makes it easier to predict economic impacts of CO2 mitigation policies and determine an optimal level of CO2 emission taxation.

The fact that the production function is Cobbs-Douglas makes the empirical analysis easy to follow / reproduce and, if combined with a social welfare function where consumption utility is logarithmic, it can make it relatively easy to determine an optimal level of CO2 emission taxation. So that was an added bonus.

not peer-reviewed... no peer responses... and no chance to be formally cited (even if worthy). Just your kind of paper, hey!

[-1=e^ipi]

not peer-reviewed... no peer responses... and no chance to be formally cited (even if worthy). Just your kind of paper, hey!

I care about the quality of the research, not so much if it is peer-reviewed. As shown by Christopher Monckton, nonsense can get peer-reviewed.

[waldo]

I care about the quality of the research, not so much if it is peer-reviewed. As shown by Christopher Monckton, nonsense can get peer-reviewed.

sure, sure... particularly when you (and your presumed expertise) are the arbiter of "quality", hey!

[Bonam]

That's a Cobbs-Douglas production function. It's pretty standard/common. The assumption of alpha + beta + gamma = 1 is one of constant returns to scale (so if you have twice as much labour, physical capital and energy then you will have twice as much economic output). Constant returns to scale is not an unreasonable assumption. Obviously there are some limited factors of production such as land, which means that alpha + beta + gamma < 1, although there are also some benefits to having more population density.

There are also major benefits to economies of scale. In fact, my admittedly cursory reading on the subject has suggested that as you concentrate more and more capital, labor, knowledge, etc into one area, you get exponential rather than linear returns. I don't think there's any reason at all to assume a+b+g = 1, it could easily be more than 1 or less than 1 depending on the situation. Nor is it even at all convincing that all the relevant factors have been identified.

I could go into detail of why people tend to use Cobbs-Douglas production functions, but basically it is easy to use and has desirable properties.

Sure but that doesn't say anything about whether it is correct or not. Do we have existing data sets that are well matched by the model's predictions? Has the model been used to make a set of predictions that has then proven out to be true? Can the constants in the equation be derived in some fundamental way rather than simply being taken as fitting parameters to match existing data?

Obviously, the functional form is likely incorrect (and I've seen a fair amount of evidence that it is incorrect, particularly how energy enters the production function), but it isn't that bad. You have to remember that there is a tradeoff when making a model more complicated since you would make it harder to estimate, use and share results with other people, but you might be able to better explain observations with a more complicated model. That's why people like Karl Popper argue that various theories have predictive power and it makes sense to give preference to the null hypothesis / simpler beliefs until proven otherwise and not to accept unfalsifiable beliefs. That's why things like Occam's razor and Akaike's information criterion exist. It's why in physics, people will often use Newtonian physics or neglect air resistance when making calculations.

People use Newtonian physics because in the overwhelming majority of calculations for anything happening on Earth, they give results that are sufficiently close for all practical purposes to what you would get from any more complex model. As for making calculations neglecting air resistance... this actually introduces major errors in many cases and makes predicted results differ from reality quite substantially in many applications. Even simple every day things like throwing a ball or something are significantly affected by air resistance and neglecting it will give you an observably wrong result. It's mostly neglected just so people can run through simpler calculations as a teaching tool.

That said, in both of these examples, the person is consciously deciding to use a simplified model of a more complex overall set of known physical laws to get a result of a known level of applicability, with a known set of errors being introduced. On the other hand, purely numerical analyses in economics like the model we are discussing don't seem to be based on any fundamental principles or laws or simplifications from a more complete theory. They are simply conjured up and have enough fitting parameters (fudge factors) to make them fit some data.

[-1=e^ipi]

There are also major benefits to economies of scale. In fact, my admittedly cursory reading on the subject has suggested that as you concentrate more and more capital, labor, knowledge, etc into one area, you get exponential rather than linear returns. I don't think there's any reason at all to assume a+b+g = 1, it could easily be more than 1 or less than 1 depending on the situation. Nor is it even at all convincing that all the relevant factors have been identified.

You are right, it might be different from 1. I don't think the assumption is that bad given the application though. Similar analysis could be performed without the restriction. Disagreements about economies of scale can be resolved using empirical data.

Do we have existing data sets that are well matched by the model's predictions? Has the model been used to make a set of predictions that has then proven out to be true?

Yes and no. For some applications a Cobb-Douglas production function yields okay results and in other cases it doesn't. For example, generally such functions predict that physical capital's share of national income will be relatively constant over economies of varying levels of GDP per capita; this is observed to be roughly true where physical capital's share of national income is roughly 1/3. On the other hand, I have seen a paper recently that suggests that entering energy in the production function as Cobb-Douglas yield predictions inconsistent with empirical data (I'll see if I can find this paper). However, a Cobb-Douglas does have falsifiable predictions so can be tested against empirical data. It is a decent place to start; think of it as a first order approximation.

If a Cobb-Douglas production function is found to be inconsistent with empirical data, then the next step would be to try to use a generalization of a Cobb-Douglas function, such as a CES function or a translog function. And if those are inconsistent with observations, then one can try a more complicated model.

Anyway, I think you will agree that trying to determine an optimal level of CO2 emission taxation is difficult, although I'm getting the impression that you think it's not possible. So it makes sense to try to keep things as simple as possible while being consistent with empirical observations.

Can the constants in the equation be derived in some fundamental way rather than simply being taken as fitting parameters to match existing data?

Can the gravitational constant be derived in some fundamental way?

I don't know how one can derive the value of the constants in a fundamental way.

Or is this a typo and are you asking if the functional form can be derived in a fundamental way?

https://en.wikipedia.org/wiki/Cobb%E2%80%93Douglas_production_function

"Crucially, there are no microfoundations for it. In the modern era, economists try to build models up from individual agents acting, rather than imposing a functional form on an entire economy. However, many modern authors have developed models which give Cobb–Douglas production function from the micro level; many New Keynesian models, for example."

On the other hand, purely numerical analyses in economics like the model we are discussing don't seem to be based on any fundamental principles or laws or simplifications from a more complete theory.

Well it would be super nice if measurement error was super low like it is for physics in most applications and if all humans were identical like how all electrons are identical. But realistically, that isn't the case, so models have to be made given what is available otherwise decisions can't be made.

They are simply conjured up and have enough fitting parameters (fudge factors) to make them fit some data.

You mean like they do in climate modelling?

Joking aside, what is wrong with selecting the best model using things like Akaike's information criterion?

[waldo]

makes one wonder if there any actual peer-reviewed/peer-responded papers that use recognized/accepted economic macro models, rely upon the Cobbs-Douglas production function and speak to carbon taxation levels What's that you say waldo... there are... a brazillion of them you say!

[-1=e^ipi]

makes one wonder if there any actual peer-reviewed/peer-responded papers that use recognized/accepted economic macro models, rely upon the Cobbs-Douglas production function and speak to carbon taxation levels What's that you say waldo... there are... a brazillion of them you say!

DICE 2013 model uses a Cobb-Douglas production function of physical capital, labour and energy, where energy can either be CO2 based or non-CO2 based. The price of CO2 emitting sources of energy is modeled as a function of total consumption of fossil fuels (resources get more scarce) plus a possible tax on CO2 emissions. So that's a bit more complicated than putting CO2 emissions directly in a Cobb-Douglas production function.

http://www.econ.yale.edu/~nordhaus/homepage/documents/DICE_Manual_103113r2.pdf

Edit: Well more accurately, it has abatement costs in the production function and determines abatement costs depend on substitution costs to non-CO2 forms of energy.

Edited September 30, 2015 by -1=e^ipi

[-1=e^ipi]

Okay, upon more thought, putting CO2 emissions directly in a Cobb-Douglas production function is probably not the best idea, since mitigation costs approach infinity as CO2 emissions approach zero.

But putting energy in a Cobb-Douglas production function, and then treating mitigation costs as a power function of percentage of energy from non-CO2 intensive sources, as Nordhaus does, is far more reasonable.

Heck Nordhaus was doing this all the way back in 1992.

http://www.econ.yale.edu/~nordhaus/homepage/Optimal.science.1192.pdf

Anyway, this suggests that Xiaobing Zhao's estimate of mitigation costs is likely an overestimate.

So one could get an estimate of mitigation costs as follows:

- First estimate the Cobb-Douglas production function, where energy is one of its parameters, using empirical data such as differences in output by country; this gives energy's share of national income. One could even treat the parameters in the model as functions over time and over level of economic development to take into account change in energy use due to economic development and technological progress.

- Next, use energy's share of national income, GDP PPP by country and energy consumed by country to get an implicit price of energy by country.

- Use implicit price of energy by country and CO2 emissions per unit energy by country to get an estimate of energy costs as a function of CO2 emissions per unit energy. The functional form can be a power form as suggested by Nordhaus, and one could even have it vary as a function of time and/or by level of economic development.

- The last step would be to look at mitigation policies by country in recent years and use only the no-mitigation data to get an idea of CO2 emissions per unit energy under no-mitigation. One can then use this to convert implicit price of energy by country as a function CO2 emissions per unit energy to a function of percentage of mitigation performed (between 0% and 100%).

[socialist]

Climate change has been happening for thousands of years.

[-1=e^ipi]

Thought that I would perform a basic calculation of the Cost of the NDP's plan to reduce emissions by 2025 by 34% from 1990 levels.

2025 CO2 emissions under BAU is about 622 gigatons of CO2 (I am linearly interpolating from 2020 and 2050). (https://books.google.ca/books?id=5cwE1AY1vNcC&pg=PA27&lpg=PA27&dq=abatement+cost+function+coefficient&source=bl&ots=aONrVIbYzE&sig=lnhtHAR7PCnBmM3AQ2WOuL3iWHU&hl=en&sa=X&ved=0CCcQ6AEwAmoVChMIrOLexNmnyAIVRBkeCh1V6gcr#v=onepage&q=abatement%20cost%20function%20coefficient&f=false)

1990 levels were 613 gigatons. So a 34% drop is 404.6 gigatons. So NDP target is 35% below BAU emissions.

If I use Nordhaus' RICE model of carbon abatement costs (which has 0.0435 as roughly the abatement cost parameter and an exponent of 2.8), then this suggests a cost to GDP of 0.0435*(1-0.35)^2.8 = 1.3% of GDP.

Of course, the NDP isn't choosing to go with the least costly method to reduce CO2 emissions. They are going with cap-and-trade instead of a pigouvian tax, and will likely target specific industries rather than treat polluters equally. So I'd say maybe multiply that by a factor of 2 at least.

So under the NDP plan, we would be 2.6% poorer.

Perhaps that 2.6% would be better spent on international aid.

[Michael Hardner]

Climate change has been happening for thousands of years.

This post could only come from somebody who hasn't looked into this matter at all. Yes, things happen... and they have causes sometimes too. You didn't even ask.

[-1=e^ipi]

Here is an interesting paper on trying to evaluate sea level rise costs:

http://www.pnas.org/content/111/9/3292.full.pdf

"Under constant protection damage, costs are 0.3–5.0% of global GDP in 2100 under RCP2.6 and 1.2–9.3% under RCP8.5.Under enhanced protection, impacts are about 2–3 orders of magnitude lower."

The enhanced protection scenario is the post reasonable though (obviously you would modify dikes in response to sea level rise).

"While the median global mean sea-level rise is projected as 35 cm for RCP2.6 and 74 cm for RCP8.5."

If I assume the upper bound of the confidence intervals for the costs, and assume the enhanced protection is 10^2.5 times less than the constant protection, then by 2100, RCP 2.6 costs 0.016% of global GDP and RCP 8.5 costs 0.029% of global GDP. Since sea level rise costs are a roughly linear function of sea level rise, we are looking at ~0.0004% of global GDP per cm of sea level rise at best. If I use the best estimates (assuming lognormal distributions) instead of the upper bounds then this reduces to about 0.0001% of global GDP per cm of sea level rise.

[socialist]

This post could only come from somebody who hasn't looked into this matter at all. Yes, things happen... and they have causes sometimes too. You didn't even ask.

Please let me know when you feel like making an actual point.

[-1=e^ipi]

I'll try to do a rough calculation to convert this into external cost due to sea level rise. Suppose an equilibrium sea level rise of 3m/C (justified by what we know about the Eemian) and that climate sensitivity is ~2C. 1 ppm of CO2 corresponds to 7.81 gigatons of CO2. Current atmospheric CO2 is roughly 400 ppm.

This means that equilibrium change in sea level is roughly
1/(7.81*10^9)/400/ln(2)*2C*3m/C
= 2.77 * 10^-12 m / metric ton

For a cost of 0.04% of global GDP per m, this corresponds to a cost of 1.11 * 10^-13 % / metric ton

If I assume that sea levels will approach equilibrium with a decay time of 500 years, use a social rate of time preference of 2%, assume a social welfare function with logarithmic consumption utility and use UN medium variant population projections till 2100 (and assume that population plateaus after 2100) then I can determine the net present external cost as a percentage of GDP due to future sea level rise damages.

This corresponds to -8.87 * 10^-13 % / metric ton.
Global GDP PPP in 2014 was $107 trillion. https://en.wikipedia.org/wiki/World_economy

Using this suggests that the marginal external cost due to sea level rise of emitting CO2 today is $0.95 / metric ton of CO2.

This is consistent with the results of post #67, where costs of sea level rise are less than a dollar per metric ton.

Edited October 5, 2015 by -1=e^ipi

[Michael Hardner]

Please let me know when you feel like making an actual point.

My point was that your post had nothing to do with current climate change, and you didn't even think to explain why.

[socialist]

My point was that your post had nothing to do with current climate change, and you didn't even think to explain why.

It's not my fault you don't understand the significance of the graphic I posted.

[Michael Hardner]

It's not my fault you don't understand the significance of the graphic I posted.

It's not significant. It's common knowledge that temperatures have changed on earth since it was formed. It has nothing to do with the current situation, though, and you don't seem to have any understanding or you would post your rationale for putting it here.

[socialist]

It's not significant. It's common knowledge that temperatures have changed on earth since it was formed. It has nothing to do with the current situation, though, and you don't seem to have any understanding or you would post your rationale for putting it here.

Climate change is natural. I don't buy your "sky is falling" mantra. Sorry.

[Michael Hardner]

Climate change is natural. I don't buy your "sky is falling" mantra. Sorry.

You could only post these things if you were uninformed about the causes of climate change, which you don't seem to be curious about. The greenhouse effect causes warming, and that's what we're dealing with today. It caused it in the past too.