A brief critical response to the CSIRO study,

S. Hatfield-Dodds, et al., (2015) “Australia is ‘free to choose’ economic growth and falling environmental pressures”, Nature, 5th No.v, 527, pp. 49 - 53.  1doi:10.1038/nature16065 http://www.nature.com/nature/journal/v527/n7576/full/nature16065.html

Ted Trainer


This study (by eighteen authors) concludes that Australia can achieve sustainable levels of resource use and environmental impact by 2050 without interfering with economic growth and without any radical change in values or behavior.  About twenty scenarios are modeled, and reported in many detailed plots in the c. 25 page Nature article plus the Supplementary Information document.  These credentials make it likely that the findings will be widely reported and accepted.

The study contradicts head on the now enormous literature supporting the case that there are limits to growth, that we have gone through them, and that the consequences will be very nasty. It is therefore extremely important that this statement should be examined carefully to see whether it is convincing.  It is in fact easy to show that it is far from convincing and does not provide reasons to doubt the limits case.  It is disturbing that the CSIRO would put out such a flawed document.  Following is a brief outline of the main problems.

The basic fault is to do with the logic of the report. It simply does not even try to do what it claims to have done. Although on first encounter it is difficult to see the logic, what is being said is essentially, if we doubled the annual rate of improvement in materials use efficiency and imposed a carbon price up to sixteen times the present price, then by 2050 there would not be significant reduction of GDP. But the crucial question here is obviously, can we do those things.  Can we improve resource use efficiency at a rate that would cut it to an astounding 4% of the present value by 2050? So one would expect the paper to give a lot of space to evidence and argument showing that this is possible.  But it gives no space at all to this question.  The study is simply an exploration of what would happen if we did those two things.

I wrote to the CSIRO authors asking for their grounds for using such a remarkable decoupling rate assumption and was referred to a single study, by Schandl et al. (2015.) Not only does this paper also contain no evidence on the issue, it concludes (on p. 5):

   “Our results show that while relative decoupling can be achieved in some scenarios, none would lead to an absolute reduction in energy or materials footprint.”  (They do say carbon emissions would go down.)

    “…even strong carbon abatement and large investment into resource efficiency would see global energy use growing from …(416 EJ/y to 1128 EJ/y in 2050, which is enormous, probably the highest prediction you will find.)

“…energy use continues to be strongly coupled with economic activity in all three scenarios.”

There are two measures of decoupling. “Relative decoupling” refers to where growth in one factor can be slowed compared with growth in another.  But given that the planet is suffering resource and ecological insults that are already far too great, what growth advocates need to show us is that  “absolute” decoupling is occurring, or possible, that is that GDP can continue to grow while there is an annual fall in the amounts of energy and materials used and ecological impact. It is clear that at present this is not happening (Burton, 2015.), Indeed all the important measures show alarming growth rates for the worrying items. For instance over the longer term the growth in energy use and carbon emissions have tracked almost exactly in parallel with GWP growth, materials use trebled last century, and just about all indices of ecological damage are rising.

So it is not correct for this paper to say, “We find that substantial economic and physical decoupling is possible.”  (Abstract, and see also p. 49.) That is not what they find; it is what they assume when they take that 4.5% p.a. figure for the improvement in resource use. That improvement rate is in fact a decoupling rate; they are saying they will put into their scenarios an assumed annual 4.5% reduction in resource use per unit of GDP. The study’s optimistic scenario (“Stretch”) just explores the consequences of this decoupling rate plus the carbon price assumption. In other words what the study finds are the consequences of those assumptions. They have clearly not found that decoupling is possible, and it is very important to understand this major fault in the logic of the paper.

There are strong reasons for thinking that a decoupling effect could not be sufficiently powerful to achieve the outcomes this paper envisages. According to these authors by 2050 Australian GDP can multiply by 2.7 while resource use falls 35%.  That would leave a ratio of resource use to GDP that is around one-fifth of the present level.  No evidence or reason is given to indicate why this is thought to be possible --  in an era when just about all material, biological and ecological resource grades, costs, scarcities problems etc. are deteriorating rapidly. Add the cumulative global resource depletion that will occur in the next 35 years during which they estimate that GWP will multiply by 2.5.

There are numerous well known indices which show how enormous decoupling would have to be if economic growth could continue while resource and ecological impacts become sustainable.  For instance the World Wildlife Fund’s “Footprint” analysis shows that the amount of productive land needed to provide an Australian with energy, food, water and shelter is about 7-8 ha. If 9.7 billion people were to live as we do then we’d need up to 78 billion ha of productive land … but that’s about ten times the amount there is on the planet. And if present loss rates continue we will have only about half the present amount of cropland by 2050.

Similarly, if by 2050 all 9.7 billion people were to have risen to the GDP per capita Australians would have then given 3% p.a. economic growth, world economic output would be about 25 times as great every year as it is now. Is it plausible that “decoupling” could allow GWP, the amount of producing, purchasing and using up going on, to multiply by 20+ while rich world per capita resource use can be cut to one-tenth or one-twentieth of the present total?  What is the case for thinking that anything like this could be done?

Given these kinds of multiples, a 35% reduction in materials demand (i.e., only 25% per capita given that the analysis envisages a 37 million population in 2050) would not get us far towards a global consumption rate that is sustainable and possible for all.

By the way, that energy growth figure is far higher than I have seen anyone predict, even the IEA. Energy demand more than doubles in all three of their scenarios, so to say the least, their best scenario results in anything but an absolute energy decoupling. To quote the Schandl et al. paper again, “…energy use continues to be strongly coupled with economic activity in all three scenarios.” (p.5.) We are left with question, how sustainably could we find 2.7 times present world energy supply.)

Similarly they say that global materials use would increase markedly, from 79 billion tons/y to 183 billion tonnes/y.  Again this is a huge increase, the opposite of an absolute decoupling, and would only be a small decoupling “relative” to GWP. It would be 2.3 times the present burden on the planet due to resource extraction. Yet the paper claims, “Our study suggests that decarbonisation and dematerialization are possible.”  Again this is incorrect; what it shows is if an implausibly high dematerialization rate is fed into the scenarios then by 2050 there will have been a slight relative dematerialization to a global consumption rate of 183 billion tonnes p.a., when the Footprint index suggests what the planet needs is probably a reduction to two-thirds of the present 73 billion tonnes p.a. that to around 49 billion tonnes.

Finally the Nature paper itself says on p. 5, ”Our research shows that while relative decoupling can be achieved in some scenarios, none would lead to an absolute reduction in energy or materials footprint…”

Thus it would seem that a) it is highly implausible that anything like the expected/assumed improvement in materials use efficiency could be achieved, b) no reason is given to believe that it could, c) in fact even when Schandl et al. make implausibly optimistic assumptions they admit decoupling does not result, and d) even if the most optimistic CSIRO rate was achieved it would leave Australian levels of resource and ecological impact far higher than those enabling a sustainable world (explained further below.)


The second of the two big assumptions the paper’s optimism depends on is the assumed potential for bio-sequestration of carbon.  It says that in 2050 large quantities of carbon based energy would still be being used and up to 59 million ha would be planted to take carbon from the atmosphere. (All our cropland is only c. 24 million ha and the authors say all our agricultural land is about 85 million ha.) The yield assumption does not seem to be stated; is it 15 t/ha, or the more like 5 t/ha likely from a very large area of more or less average land?

However the main problem with the use of land to soak up carbon via plant growth is that after about 60 years the trees are more or less fully grown and will not take up any more carbon; what then?  The implications of this do not seem to have been considered. It means that in the second half of the century an amount of new planting would be needed each year that was big enough to take out the amount of carbon emitted that year.  Given that the economy in 2050 is expected to be 2.7 times bigger than it is now, and still growing at a normal rate, the area to be planted each year would be substantial, and increasing…leaving how much for agriculture?

Fig. 2 shows that in 2050 a net 200 million tonnes of CO2 would be being taken out of the atmosphere each year.  That is, in addition to taking out the emissions generated by the large amount of fossil fuels still being used in 2050 (which seems to be around 1.825 EJ), another 220 million tonnes would be taken out (the amount from power plus transport), making a total in the region of 450 million tonnes/y. Assuming 10 tonnes/ha/y forest growth (it would be more like 5 t/y for a large area), taking out approximately 36 tonnes of CO2/ha/y, the additional area to be planted each year would be 12.5 million ha. (This is the addition corresponding to the present GDP; by 2050 it would have to be 2.7 times as big.)

How has the carbon embodied in the production and transport of imports been accounted?  It would seem that the 2050 economy would have to be even more dependent on services than the present economy, meaning there would be heavy importation of goods no longer produced in Australia.  The energy, carbon, resource and Third World justice effects of imports is only beginning to be attended to, and the picture is disturbing.  For instance for a rich country the amount of carbon emissions due to imported goods is typically as great as or much greater than the amount released from energy production. (And it shows up on the books of the exporting country, not the rich country consuming the goods.)  Has the amount of bio-sequestration needed to deal with this been included?

In a reply to my draft of this discussion the main author said that “… the carbon sequestered by plantings on currently cleared satiates after a period and does not provide a permanent flow.”  This is difficult to understand because it would seem to admit the inadequacy of their entire bio-sequestration case.  Their defence of the possibility of growth and affluence depends heavily on the capacity of bio-sequestration to take out as much CO2 each year as we are putting in but this reply seems to be admitting that their strategy could only do that until somewhere after 2050.

Randers, one of the original Limits to Growth authors, doesn’t think we will run into limits problems by 2050, but he thinks by about 2070 they will be catastrophic. The time line isn’t crucial; the original book wasn’t concerned with when we will hit the wall; it was concerned that we are going to hit it.  At the best the CSIRO paper provides some reason to think it will be later rather than sooner, but it doesn’t give us any good reason to think we won’t hit it.  Yet the paper is being taken to mean there are no limits to growth to worry about.

            What carbon price will do it?

The study seems to have assumed that power generators will find it economic to shift from carbon fuels to renewables when the price of carbon rises to about $50/tonne (i.e., rises at 4.5% p.a. from $15/t.)  Lenzen’s soon to be published detailed study of Australian renewable potential is likely to indicate that the price needed to drive carbon out of the generating system is $500/tonne.  His colleague working on the German situation says that there the price would be close to $1000/tonne. (The CSIRO paper does not assume close to compete elimination of carbon fuels.)

As noted above the study seems to have made the very common mistake of taking the cost of carbon that would make it more economic for a generator to shift the generation of 1 kWh from carbon fueled power station to a wind turbine. But this is not the right question.  A power supply system with a large fraction of renewable input would have to have a very large amount of redundant generating capacity, most of it sitting idle most of the time, to be able to guarantee supply during periods of low wind or solar energy, or it would have to retain much carbon-fuelled capacity, sitting idle most of the time.  Either way the system has much redundant plant and this means high capital costs. The multiple for a 100% renewable system seems to be in the range of 4 to possibly 10 times the amount of plant that would do the job if renewables worked to peak capacity all the time. So the price of carbon would have to be very high before it became cheaper for power generators to shift to renewable technologies.

Another disturbing point regarding carbon is that the study is saying even an astounding 4.5% p.a. rate of improvement in materials efficiency would only be associated with a halving of emissions by 2050

            No analysis of renewables.

Renewable energy is claimed to provide a significant proportion of the power and transport energy but there is no reference to the many, difficult and unsettled associated problems of intermittency, redundant capacity, and storage, and the resulting total system capital costs.  Schandl et al. say (p. 11) “Our modeling suggests that a global price for carbon emissions (…rising to $236/t) would be sufficient to drive decarbonisation of the energy system…”  But it is not possible at this point in time to make pronouncements like this. We are not close to knowing what amounts of what renewable technologies would be needed to replace fossil fuels, or the cost of such a system. These factors are far from being clearly understood or settled.  At the very lest we would need to see detailed derivations and justifications, or reference to studies of these. It is very likely that this statement from Schandl et al. is based on many assumptions that cannot be supported. (It is likely that they have made the common mistake noted above, i.e., taking the capital cost of producing 1 kWh of power, when the important cost is for a system containing a great deal of redundant plant to deal with intermittency. Even for Australia the power system capacity factor might be no more than 10-20%.)

In addition generalisations are not warranted because renewable potential varies greatly with the region under discussion.  The very scarce evidence available at present indicates that Australia probably could provide all power from renewables, although at a high production cost (Elliston, Diesendorf, Macgill, 2012, Lenzen, 2015.), but it is very likely that Europe would fall far short of being able to meet all energy demand. (Trainer, 2013.)

            Conservation potential?

The plots show that it is being assumed that demand and impacts can be greatly reduced by conservation and efficiency effort.  This is commonly assumed but few if any optimistic pronouncements take into account the significant energy, resource and environmental cost of saving energy, resources and environment. In other words claims are often only about gross reductions achievable and not net reductions.

Powerful examples of this are given by figures on housing and vehicles.  Much attention is given to the German Passivhaus which it is said can reduce energy consumption by 75% or more. However this kind of claim usually refers only to energy consumed within the house, and does not take into account the energy used to install the typically elaborate insulation and heat transfer equipment.  The issue seems to be unsettled but a recent study by Crawford and Stephen (2013) found that the total life-cycle energy cost for the Passivhaus is actually greater than for a normal German house. 

Even more common is the claim that electric vehicles (assumed to make up 25% of transport energy use in this study) can reduce energy use by 75 – 80%, but this does not take into account the considerable energy costs in producing EVs. The State Government of Victoria's trial of EVs found that they reduce emissions only if powered by renewable energy.  (Carey, 2012.)  Otherwise life-cycle emissions taking into account all factors in addition to fuel are actually 29% greater than those of petrol driven cars. Mateja (2003) finds that electric cars involve much higher embodied energy costs than normal cars. Bryce (2010) says 60% of the life cycle energy and environmental cost of these cars is to do with their production and disposal, not their on-road performance.

Again it would be important to see what assumptions are being made by these authors in arriving at the optimistic conservation and efficiency claims being made.

            What would the cost be?

It is one thing to show that something could be done but it is another to show that it could be afforded.  The paper claims that no significant cost to GDP would be involved. Even if the decoupling and sequestration assumptions were valid we would want to know the cost of doing those things, e.g., of maintaining and harvesting 59 million ha, and of producing half the power by renewables. My understanding of Lenzen’s current study is that it seems to be indicating that a fully renewable power supply system would result in a production cost around four or five times the present cost of fossil fuelled power. (The CSIRO paper does say the cost of power production could double.) This would seem to be affordable, but it would have major disruptive effects, especially on GDP as energy costs feed into everything and have multiplier effects. 

The post GFC stagnation, and wild fluctuation in oil prices seem to have shown how surprisingly fragile and sensitive the global economy is to resource input factors.  Tverberg (2015) argues persuasively that resource limits to do with the increasing difficulty of providing oil and its deteriorating EROI led to the recent spectacular rise in its price, which in turn depressed the economy, which led to the present low oil demand and prices.  This suggests how disruptive a significant rise in electricity price might be.  Th CSIRO paper adds questions to do with the probable costs for all that bio-sequestration, and especially regarding the EROI assumed for biofuels which are assumed to provide 25% of transport energy. (Various studies find that it is around 1.4 or less for corn based ethanol, which suggests that option is not worth bothering with.)

Would it scale to 9.7 billion people?

The amount of land planted for bio-sequestration would not. The area assumed for the optimistic scenario, up to 59 million ha of forest plantation for sequestration plus 35 million ha for “biodiversity planting” would total 2.2 ha per person (assuming population will reach 37 million by 2050.)  But Australia has much more potential forest area than most countries and the amount of forest on the planet now averages about only 0.45 ha per person, and is heading for .25 ha by 2050 (and much less if the rate of forest loss is taken into account.)

The paper’s expected 2050 consumption of petroleum and gas is considerable, adding to about. 35 GJ per person.  The authors give no attention to the strong probability that there will not be much of either left by then. Thus for 9.7 billion people demand would be 340 EJ which is about 1.7 times present world oil consumption … and therefore far from a plausible amount all could be consuming in 2050.

The figures for materials use also show that the expected 2050 Australian use rate could not possibly be had by all people.  Thus even if the optimistic scenario could be achieved it would fall far short of one that could save the planet.  It would still leave Australians living at per capita levels of resource use that were many times higher than all could share.


As noted above, it would be difficult to suggest an issue that is more important than whether or not the limits to growth thesis is valid.  The case for it has been accumulating weight for at least fifty years and in my opinion has long been beyond serious challenge.  All resource stocks are being depleted at significantly unsustainable rates, summarized by the WWF conclusion that 1.5 planet Earth’s would be needed to provide them sustainably.  And only about 2 billion are using them; what happens when 11 billion (the UN’s 2100 expectation) rise to our levels of consumption … let alone the levels we will have then given 3% growth … that is, levels that might be ten times as high as they are now. 

This is the kind of arithmetic that is now leading considerable and increasing numbers of people to see the dominant obsession with affluence and growth and tech-fixes as absurd and suicidal, and to join the De-growth and associated movements such as Voluntary Simplicity, eco-villages and transition towns.  We who are working in this area believe we know how to save the planet and we know the only way it can be saved.  It is to shift to ways that do not create the problems now destroying the planet, depleting resources, condemning billions to deprivation, causing resource wars and damaging the quality of life in even the richest countries. Our “Simpler Way” vision (http://thesimplerway.info) would be easily and quickly achieved, if that was what people wanted to do. It isn’t and it will not be considered until the conditions presently devastating the lives of billions begin to impact supermarket shelves in the countries now living well on their grossly unfair proportion of world wealth. By which time it will probably be too late.  The CSIRO paper is saying what just about everyone wants to hear, i.e., that there is no need to worry about any need to take The Simpler Way seriously.

Ayres, R. and V. Vouroudis, (2013), “The economic growth enigma; Capital, labour and useful energy?”, Energy Policy, 64 (2014) 16–28.

Bryce, R., (2010), Power Hungry, Public Affairs, New York.

Burton, M., (2015), “The Decoupling Debate: Can Economic Growth Really Continue Without Emission Increases?”, The Leap, October 23.

Carey, A., 2012. Electric cars make more emissions unless green powered. The Age, 4th Dec.

Crawford, R., A. Stephan, (2013), “The significance of embodied energy in certified passive houses”, World Academy of Science, Engineering and Technology, 78,  589 –595.

Mateja, D., (2000), 'Hybrids aren’t so green after all'www.usnews.com/usnews/biztech/articles/060331/31hybrids.htm

Schandl, H., et al., (2015), “Decoupling global environmental pressure and economic growth: Scenarios for energy use, materials use and carbon emissions.” J. of Cleaner Production, (In press.)

Trainer, T., (2013), "Can Europe run on renewable energy? A negative case", Energy Policy, (63) 845 – 850.

Tverberg, G., (2015) “Oops! Low oil prices are related to a debt bubble”, Our Finite World,