The labor theory of value is ill-suited to incorporating inputs from the natural environment into either a theory of capitalist prices or a theory of social opportunity costs. But to blame Marx for this would be gratuitous. Marx inherited the labor theory of value from “classical economists” such as Adam Smith and David Ricardo, and like them, Marx lived in a world where there was only one person for every five people alive today, nature’s “services” still appeared to be bounteous, and the costs to society of using inputs from nature consisted mostly of the time needed to find and extract them. Besides a few passages where he condemns capitalism for exploiting and despoiling nature as well as workers, what little Marx had to say about inputs from nature appears mostly in Part VI at the very end of Volume III of Capital. However, Marx’s discussion of “differential” and “ground” rent there differs little from the theory of rent Ricardo had elaborated before him.
But that was then, and now is now. Not only has population grown five-fold since Marx lived, human impact per capita on the natural environment has increased even more. Today a number of essential ecosystems are seriously imperiled, and we are perhaps less than a decade away from triggering cataclysmic climate change.
Lacking a formal framework that facilitates accounting for inputs from the natural environment and focuses entirely on how much labor time it takes to produce things, some Marxists today have scoured Marx’s voluminous writings to find a few passages where he conjectured that capitalism would cause what they have labeled a “metabolic rift” between humanity and nature, and treat environmental problems as one of the “crises” that inevitably plague capitalist economies.1 Citing the master himself – “Accumulate, accumulate! That is Moses and the prophets!” (Marx 1967a: 595) – these “ecological Marxists” argue that because capitalism is all about accelerating economic growth it must be incompatible with environmental sustainability, as if this were obvious.
As discussed in chapter 4, prior to the 1990s Marxists argued that “accumulate, accumulate” leads to what they called “internal contradictions” that render capitalist growth economically unsustainable. Some argued that capital deepening would eliminate the source of profits. Others argued that when wage increases failed to keep pace with increases in labor productivity, and capitalists’ propensity to save increased, crises would result when aggregate supply outstripped demand. However, more recently ecological Marxists have argued that an even bigger problem arises when capitalism does sustain economic growth sufficiently to surpass critical environmental thresholds and become environmentally unsustainable. Six American Marxists who have written extensively on capitalism and the environment are David Harvey (1996), James O’Connor (1998), Joel Kovel (2002), John Bellamy Foster (1994, 2000, 2002, 2009), Paul Burkett (2006), and Jason Moore (2015).
In a recent example, John Bellamy Foster and Fred Magdoff (2010) begin their essay “What Every Environmentalist Needs to Know About Capitalism” with an excellent summary of evidence suggesting that we are experiencing a “planetary ecological crisis.” They begin a subsequent section titled “Capitalism is a System that Must Continually Expand” as follows: “No-growth capitalism is an oxymoron…. Capitalism’s basic driving force and its whole reason for existence is the amassing of profits and wealth through the accumulation (savings and investment) process. It recognizes no limits to its own self-expansion.” Foster and Magdoff go on to assert that because capitalism recognizes no limits to its self-expansion, environmental crises can only worsen unless capitalism is replaced by socialism. Before examining whether this argument holds up under scrutiny, or simply assumes its conclusion, let’s see what Sraffian theory has to offer.
One of the convenient properties of the Sraffian framework is how easily it allows us to incorporate inputs from nature in production and rents paid to their owners into our theory of income and price determination. Suppose in addition to labor and produced inputs production requires a “primary” input from nature. All we need to do is add its input coefficient to our “recipes” for production. Suppose 0.3 units of nature must be present to produce a unit of good 1, 0.2 units of nature must be present to produce 1 unit of good 2, u is the rent per unit of nature, and capitalists must pay for all inputs – produced inputs, labor, and nature – in advance. Now our recipes look like this:
a(11) = .3 a(12) = .2
a(21) = .2 a(22) = .4
L(1) = 1.0 L(2) = .5
N(1) = .3 N(2) = .2
And our price equations look like this:
(5) (1+r)[p(1)a(11)+p(2)a(21)+wL(1)+uN(1)] = p(1)
(6) (1+r)[p(1)a(12)+p(2)a(22)+wL(2)+uN(2)] = p(2)
Substituting in values for the coefficients and setting p(2) equal to 1 we have:
(1+r)[.3p(1) + .2p(2) + 1.0w + .3u)] = p(1)
(1+r)[.2p(1) + .4p(2) + .5w + .2u] = 1
The four unknowns we need to solve for are p(1), w, r, and u. If we set two of the distributive variables equal to zero, we can solve for p(1) and the maximum possible value of the other distributive variable. Doing this we find:
w(max) = .691 and p(1) = 1.273 when r = u = 0
r(max) = .798 and p(1) = .781 when w = u = 0
u(max) = 1.900 and p(1) = 1.100 when w = r = 0
We can also discover what the effect of increasing the value of any distributive variable is on the other distributive variables. Suppose, for example, instead of setting w = u = 0 we increase w to 0.100, keep u = 0 and solve for r. We find that r falls from its maximum value, 0.798, when both of the other distributive variables are zero, to 0.584. Also if we raise r from zero to 0.100 and keep w = 0 we discover that u falls from its maximum of 1.9 to 1.485.
In short, integrating rent for primary inputs from nature along with wages and profits as distributive variables poses no difficulties for a Sraffian determination of income distribution and prices. For proof that there is a negative relationship between all distributive variables, just as there was in the system with only the two distributive variables, w and r, and that this negative relationship holds even when we expand the system to account for multiple different “primary” inputs from nature, such as iron ore, oil, and fresh water, in addition to land of different qualities, each with its own rental rate, as well as multiple different kinds of labor, such as welding, carpentry, and computer programming labor, each with its own wage rate, see theorem 19 in chapter 1 of Hahnel (2017).
Recently it has also been demonstrated how the Sraffian framework can help establish rigorous sufficient conditions for environmental sustainability.2 Ecological economists provide the key concept which they call environmental throughput. Throughput is material matter from the natural environment’s stocks of “natural resources” used up in production processes, and material matter released into “environmental sinks” by human economic activity which diminish their remaining storage capacities.
Some components of environmental throughput are already so large that they threaten to exhaust a crucial “environmental service.” For example, scientists tell us if annual global greenhouse gas emissions do not decrease by 80% or more by 2050 we run an unacceptable risk of triggering cataclysmic climate change. Present levels of other kinds of throughput have reached the point where if they increase further they will threaten some key environmental service. Global fresh water throughput may have reached this level. Present levels of other kinds of throughput are still sustainable, either because they do not exceed the regenerative capacity of the natural resource, or because at present levels of throughput the natural resource would still remain abundant for a very, very long time.
Some ecological economists have called for giving up on the goal of increasing production, and instead focusing on achieving a steady-state, i.e. zero economic growth, to prevent further growth of throughput. Others in the de-growth movement go even further and insist that production must decrease in order to save the environment. But is it true that we must stop increasing production of goods in order to stop throughput from increasing? Or is it possible to produce more goods without increasing throughput? Fortunately, the Sraffian framework allows us to model both production of diverse goods and throughput rigorously so we can become clear when growth of output is compatible with sustainable levels of throughput and when it is not.
But first it is important to notice that the appropriate coefficients for inputs from the natural environment when deriving prices, wages, and rents are not always the appropriate coefficients when measuring throughput. For example, suppose the input from the natural environment is land, measured in acres, and suppose that after a farmer uses the land it remains in exactly the same condition as it was initially. The farmer may well pay rent to a landowner to use the land even though it does not diminish in size, or by hypothesis deteriorate in any way. In which case, in our example if the farmer produces one unit of good 1 on this land she will need to have 0.3 acres available and will have to pay 0.3u in rent. But notice that in this case throughput is zero by hypothesis. On the other hand suppose the input from the natural environment is fresh water drawn from an aquifer. In this case the farmer will diminish the aquifer by 0.3 gallons when she produces a unit of good 1. So in this case water throughput is 0.3 gallons. When calculating environmental throughput we need the coefficients in our production technologies to represent the amount a stock of a natural resource, or the storage capacity of a natural sink, is diminished, which may sometimes correspond with the amount which must be on hand and available for use, but in some cases, as just explained, may not.3
To make this distinction we will use the letter T to represent input coefficients for throughput from the natural environment, whereas we use the letter N to represent the quantity of a primary resource from nature that must be on hand. For simplicity we begin by assuming that nature is “homogeneous,” i.e. there is only one input from “nature,” just as we often begin by abstracting from differences between carpentry, welding, and computer programming labor and assume labor is homogeneous.4 Suppose our production technologies, which now have throughput consequences, are those below:
a(11) = .3 a(12) = .2
a(21) = .2 a(22) = .4
L(1) = 1.0 L(2) = .5
T(1) = .2 T(2) = .1
We can calculate the amount of nature throughput used up, both directly and indirectly, when we produce a unit of each good in exactly the same way we calculated how many hours of labor it takes, both directly and indirectly, to produce a unit of each good. In short, we must account for the fact that to produce a unit of good 1, for example, it also requires some throughput to produce the a(11) and a(21) we need, just as we took into account that it requires some labor to produce a(11) and a(21) when we calculated the labor value, V(1). Define H(1) as the amount of throughput used, both directly and indirectly, when we produce a unit of good 1, and H(2) similarly. Our “throughput equations” are:
(7) H(1) = H(1)a(11) + H(2)a(21) + T(1)
(8) H(2) = H(1)a(12) + H(2)a(22) + T(2)
Substituting in the coefficients in our example:
H(1) = .3H(1) + .2H(2) + .2
H(2) = .2H(1) + .4H(2) + .1
Which can be solved to give H(1) = 0.368 and H(2) = 0.289, the amounts of throughput from the environment “used up” both directly and indirectly when we produce one unit of output of good 1 and good 2 respectively. And if during a year we produce x(1) units of good 1 and x(2) units of good 2, environmental throughput will be: H(1)x(1) + H(2)x(2).
For simplicity assume that the number of units of nature used up by production during some initial year is exactly equal to the number of units which regenerate naturally each year. If the labor force does not grow, and the number of hours worked in each industry does not change, output and throughput will remain constant, throughput will continue to equal regeneration year after year, and the economy will continue to be environmentally sustainable. But what if we discover and adopt new technologies which increase labor productivity?
This will increase economic wellbeing because we will get more goods for the same amount of work, but it will also increase output and therefore environmental throughput. In this scenario the only way the economy can remain environmentally sustainable is if technological change also increases throughput efficiency. In other words, if the ‘x’s in our above expression for annual throughput increase because of increases in labor productivity, the ‘H’s in the expression must decrease to the same extent to preserve sustainability.5
But we can be even more precise. In chapter 3 we discovered it is possible to rigorously measure the rate of increase of overall labor productivity from any technical change in any industry in a Sraffian framework. It is also possible to rigorously measure the rate of increase of overall throughput efficiency from any technological change in any industry. Theorem 20 proved in Hahnel (2017) provides a way to calculate the size of the change in overall throughput efficiency stemming from any technological change in any industry. The change in throughput efficiency is: ρ(n) = (1−α’), where the initial technology for the economy is defined by {A, T}, the technology for the economy after some technical change is defined by {A’, T’}, d̃ is chosen so dom[A+d̃T] = α = 1, and α’ = dom[A’+d̃T’].
Which means that for any technological change, in any industry, we can calculate both how much it changes labor productivity, ρ(l), and how much it changes throughput efficiency, ρ(n). As long as all technical changes introduced in the economy during a year collectively increase throughput efficiency, ρ(n), by as much as they increase labor productivity, ρ(l), environmental throughput will remain constant. Which means that ∑(i) ρ(n)(i) ≥ ∑(i)ρ(l)(i) is a sufficient condition for preventing environmental throughput from growing, which in turn means:
Even if we continue to work the same number of hours from year to year, i.e. we take none of any increase in labor productivity in the form of leisure, it is possible to increase labor productivity and economic wellbeing without putting greater strain on the environment as long as throughput efficiency grows as fast as labor productivity.
Let’s see how this works out in the example we have been using. In chapter 3 we analyzed a particular CU-LS technical change in industry 1 which increased labor productivity by ρ(l) = (1.00−0.984166) = .015834, or 1.5834%. While it may seem that this change has nothing to do with throughput efficiency, unfortunately this is not the case. Even though the change does not affect T(1), because it increases a(21) it will have the unfortunate effect of increasing H(1), which will indirectly increase H(2) as well. The expression ρ(n) = (1−α’) allows us to calculate the decrease in throughput efficiency for the economy as a whole caused by the CU-LS change in industry 1 in our example. In this case it turns out that when a(21) is increased from 0.2 to 0.3, α’ = 1.0419, and therefore ρ(n) = − 0.0419, indicating that while the CU-LS change adopted by capitalists in industry 1 increases labor productivity by 1.5834% it also has the unfortunate effect of simultaneously decreasing throughput efficiency by 4.19%.6
To prevent putting more pressure on the environment we would need some other technical changes that increase throughput efficiency. Such changes could take place in any industry, but let’s explore a single capital-using, nature-saving (CU-NS) change in industry 2 which increases throughput efficiency.
Notice that such a change in industry 2 will necessarily decrease labor productivity somewhat because it is capital-using. So, as in the case of the CU-LS change in industry 1, we will have to calculate both ρ(n)and ρ(l) for this CU-NS change in industry 2 if we want to know its effects on labor productivity as well as throughput efficiency.
Since our primary focus is on how much this change in industry 2 increases throughput efficiency we calculate this effect first, using the formula ρ(n) = (1–α’), and find that α’ = 0.8632, ρ(n) = 0.1368, and therefore this change increases throughput efficiency by 13.68%.7 But since this change is CU, and therefore adversely affects labor productivity, we must also calculate ρ(l) = (1−ß’) for this change in industry 2. We find that ß’ = 1.0044, ρ(l) = − 0.0044, and therefore this change decreases labor productivity by 0.44%.8 We have now calculated both effects, of both technical changes – the first a CU-LS change in industry 1 which increased labor productivity but decreased throughput efficiency, and the second a CU-NS change in industry 2 which increased throughput efficiency but decreased labor productivity.
Since ∑(i) ρ(l)(i) = + .015834 − .0044 = .011434 > 0, the effect of the two changes combined increases labor productivity by 1.1434%.
Since ∑(i) ρ(n)(i) = − .0419 + .1368 = .0949 > 0, the effect of the two changes combined increases throughput efficiency by 9.49%.9
Since ∑(i) ρ(n)(i) = .094900 > .011434 = ∑ρ(l) the overall increase in throughput efficiency is greater than the overall increase in labor productivity, and therefore annual environmental throughput would be reduced if both changes were adopted.
To summarize:
• A change in a labor coefficient only affects overall labor productivity (and not throughput efficiency), and a change in a throughput coefficient only affects overall throughput efficiency (and not labor productivity). However, any change in a capital coefficient will change both labor and throughput efficiency, as is apparent from inspection of the equations for labor values, ‘V’s, and throughput “values,” ‘H’s.10
• Therefore, any technical change that changes a capital input coefficient will affect both labor productivity, ρ(l), and throughput efficiency, ρ(n), and a full evaluation of its effects requires calculating both.
• When multiple technical changes are introduced during a year all we need to do is: (1) sum their individual ρ(l)s to calculate the economy’s overall change in labor productivity, (2) sum their individual ρ(n)s to calculate the economy’s overall change in throughput efficiency, and finally (3) compare ∑(i) ρ(n)(i) with ∑(i) ρ(l)(i) to determine whether environmental throughput has increased, decreased, or remained constant.
Let’s return to: “Accumulate, accumulate! That is Moses and the prophets!” What Marx was intent on conveying in this phrase, and for that matter in the whole of section 3, Chapter XXIV, Part VII of Volume I of Capital where it appears, is that competition forces capitalists to be relentless accumulators, as no organizers of production before them. But what exactly was it that Marx claimed capitalists are so hard-driven to accumulate? The very next sentence gives his answer: “Therefore, save, save, i.e., reconvert the greatest possible portion of surplus-value … into capital!” (Marx 1967a: 595). So it is surplus value Marx insisted capitalists are driven to “accumulate, accumulate.” But surplus value – which is measured in some unit of labor time – is not the same as the physical mass of goods produced – which is measured in appropriate units of weight and volume.
Marx defined surplus value as the difference between the number of hours of labor needed to produce all goods and services minus the number of hours needed to produce the intermediate goods used up in the production process and the consumption goods purchased by the workers with their wages. In other words, the accumulation Marx referred to above is an accumulation of hours of labor expended, which he called exchange value, and the growth of surplus value is limited only by the total number of hours worked, and how many of those hours capitalists can manage to appropriate. To be precise, Marx’s argument was that competition would drive capitalists to seek to accumulate an ever larger percentage of exchange value as surplus value. In theory, this percentage could continue to increase indefinitely up to 100%,11 but in any case, it is in no way limited by the availability of physical matter. So just because the planet has physical limits does not mean that capitalist accumulation of surplus value cannot increase indefinitely. And conversely, even if capitalist accumulation of surplus value approached 100% of total exchange value, this implies nothing about breaching physical planetary limits.
In conclusion, the latest version of “inevitable collapse Marxism” by a group of “ecological Marxists” who claim that environmental disaster is unavoidable unless capitalism is replaced by eco-socialism because a continual increase in the growth of capitalist accumulation of surplus value is impossible on a finite planet does not survive the sniff test. Those who make this claim fail to realize that value is not throughput, carelessly applying reasoning to value as if it were throughput, and, in effect, are guilty of assuming their conclusion.
As Sraffian theory demonstrates, if hours worked in every industry remain constant, environmental sustainability reduces to whether or not increases in throughput efficiency keep pace with increases in labor productivity, or, as environmental economists put it, on whether or not we can sufficiently “de-couple” growth of output from growth of throughput. It is easy to motivate the impression that this is impossible – that when output of goods increases environmentally damaging throughput must increase as well – by pointing out that this has, in fact, historically been the case under capitalism to date. But that is (a) obvious to anyone who recognizes that we have been exhausting the environment, but (b) completely irrelevant to whether or not this must necessarily continue to be the case.12 What is at dispute is whether or not throughput efficiency can grow indefinitely. Because if it can, no matter how slowly, then labor productivity, and the increases in economic wellbeing this increase in output of goods brings, can also grow indefinitely at that same rate without any increase in environmental throughput, as has now been proved using a Sraffian framework.
Nothing said here should be interpreted as denying that there is an unhealthy and environmentally destructive growth imperative in today’s capitalist economies, and perhaps in any capitalist economy. It just means we must go beyond facile arguments which, upon inspection, prove not to be compelling, to explain why this is the case.13
1 The passage giving rise to a whole school of “ecological Marxism” appears at the very end of section V, “Metayage [share cropping] and Peasant Proprietorship of Land Parcels,” in Chapter XLVII, “Genesis of Capitalist Ground-Rent,” in Part VI, “Transformation of Surplus Profit into Ground Rent,” in Volume III of Capital: “On the other hand, large landed property reduces the agricultural population to a constantly falling minimum, and confronts it with a constantly growing industrial population crowded together in large cities. It thereby creates the conditions which cause an irreparable break [‘rift’ in some translations] in the coherence of social interchange prescribed by the natural laws of life. As a result the vitality of the soil is squandered, and this prodigality is carried by commerce far beyond the borders of a particular state…. Large-scale industry and large-scale mechanized agriculture work together. If originally distinguished by the fact that the former lays waste and destroys principally labour-power, hence the natural force of human beings, whereas the latter more directly exhausts the natural vitality of the soil, they join hands in the further course of development in that the industrial system in the country-side also enervates the labourers, and industry and commerce on their part supply agriculture with the means for exhausting the soil” (Marx 1967c: 813). Marx also commented on how nineteenth-century “modern” agriculture was destroying soil fertility in section 10, “Modern Industry and Agriculture,” in Chapter XV, “Machinery and Modern Industry,” in Part IV, “Production of Surplus Value,” in Volume I of Capital. “Capitalist production, by collecting the population in great centres, and causing an ever-increasing preponderance of town population, on the one hand concentrates the historical motive force of society; on the other hand, it disturbs the circulation of matter between man and the soil, i.e. prevents the return to the soil of its elements consumed by man in the form of food and clothing; it therefore violates the conditions necessary for the lasting fertility of the soil…. Moreover, all progress in capitalist agriculture is a progress in the art, not only of robbing the labourer, but of robbing the soil; all progress in increasing the fertility of the soil for a given time is a progress toward ruining the lasting sources of that fertility…. Capitalist production, therefore, develops technology, and the combining together of various processes into a social whole, only by sapping the original sources of all wealth – the soil and the labourer” (Marx 1967a: 505–506). Whether or not Marx’s observations about the deterioration of soil fertility based largely on the work of a nineteenth-century German agricultural chemist named Justus von Liebig can be extrapolated into a compelling theory of how and why capitalism must irreparably destroy the natural environment, as a new generation of “ecological Marxists” claim, is what we examine in this chapter.
2 See chapter 2 in Hahnel 2017 for a more extensive treatment, including proofs of key theorems.
3 In truth what we care about is how much any environmental service deteriorates when goods are produced, but it is common to think of this in terms of diminishing scarce stocks of natural resources or depleting the storage capacity of environmental sinks.
4 See chapter 2 in Hahnel 2017 for discussion of what we can still conclude about sustainability even when nature is heterogeneous, and even when some components of nature do not regenerate.
5 Another way to think about increases in labor productivity is as decreases in labor values, ‘V’s. If we think about increases in throughput efficiency as decreases in throughput “values,” ‘H’s, then if hours worked remain constant, as long as the ‘H’s are shrinking as fast as the ‘V’s, throughput will remain unchanged.
6 In our example a “real rent” consumption vector for owners of nature that would reduce the initial rate of profit to zero even when wages are zero is 1.66 units of good 1 and 1.33 units of good 2. Using this to formulate our 2×2 matrix [A’+ d̃T] yields: a(11)’ + d(1)T(1) = .3 + (1.66)(.2) = .6333, a(12)’ + d(1)T(2) = .2 + (1.66)(.1) = .3666, a(21)’ + d(2)T(1) = .3 + (1.33)(.2) = .5666, and a(22)’ + d(2)T(2) = .4 + (1.33)(.1) = .5333. Entering these four coefficients of our 2×2 matrix in the Comnuan calculator at http://comnuan.com/cmnn01002/ yields a dominant eigenvalue of 1.0419, which gives ρ(n) = − .0419.
7 Again, in our example a “real rent” consumption vector for owners of nature that would reduce the initial rate of profit to zero even when wages are zero is 1.66 units of good 1 and 1.33 units of good 2. Using this to formulate our 2×2 matrix [A’+ d̃T’] yields: a(11)’ + d(1)T(1)’ = .3 + (1.66)(.2) = .6333, a(12)’ + d(1)T(2)’ = .2 + (1.66)(.01) = .2166, a(21)’ + d(2)T(1)’ = .2 + (1.33)(.2) = .4666, and a(22)’ + d(2)T(2) = .41 + (1.33)(.01) = .4233. Entering these four coefficients of our 2×2 matrix in the Comnuan calculator at http://comnuan.com/cmnn01002/ yields a dominant eigenvalue of .8632, which gives ρ(n) = + .1368.
8 In our example a real hourly wage of .33 units of good 1 and .266 units of good 2 would reduce the initial rate of profit to zero even when rent is zero. Using this to formulate our 2×2 matrix [A’+d̃L’] yields: a(11)’ + b(1)L(1) = .3 + (.33)(1) = .6333, a(12)’ + b(1)L(2) = .2 + (.33)(.5) = .3666, a(21)’ + b(2)L(1) = .2 + (.266)(1) = .4666, and a(22)’ + b(2)L(2) = .41 + (.266)(.5) = 5433. Entering these four coefficients of our 2×2 matrix in the Comnuan calculator at http://comnuan.com/cmnn01002/ yields a dominant eigenvalue of β’ = 1.0044 and ρ(n) = − .0044.
9 We can confirm the conclusion that in combination the two changes – our CU-LS change in industry 1, and our CU-NS change in industry 2 – increase both labor productivity and environmental throughput efficiency by calculating the new labor values and throughput “values” for the new coefficients in both industries and comparing them to the original values before either change was introduced. The new labor values are V(1)’ = 1.762 which is less than the original V(1) = 1.842, and V(2)’ = 1.445 which is less than V(2) = 1.447. The new throughput “values” are H(1)’ = .343 which is less than H(1) = .368, and H(2)’ = .133 which is less than H(1) = .289.
10 From the perspective of the environment a CU-LS change must increase throughput because it is CU without being NS. From the perspective of labor productivity a CU-NS change must reduce labor productivity because it is CU without being LS. In effect CU-LS changes substitute more nature for less labor, while CU-NS changes substitute more labor for less nature.
11 Whether or not Marx was correct in anticipating that capitalists would necessarily succeed in appropriating an ever larger share of exchange value is also questionable. If labor’s bargaining power did not deteriorate there is no reason to believe this prediction would prove true. But this is beside the point, since even if capitalists’ share of total exchange value approached 100%, the conclusion that environmental throughput must necessarily increase does not follow.
12 Take GHG emissions for example: For the period from 1960 to 2000 real global GDP grew at 2.7% per year but global GHG emissions grew only at 1.3% per year because GHG throughput efficiency increased by 1.4% per year. However, what scientists are telling us is that unless GHG throughput falls dramatically, i.e. unless increases in GHG throughput efficiency outstrip increases in real GDP considerably, we are headed for climate disaster by mid-century. Which is why the fact that global GHG throughput efficiency only increased by 0.7% from 2000 to 2014 is worrisome, but the fact that it has started to rise again in the past few years is somewhat encouraging. Nonetheless, increasing GHG throughput efficiency is clearly possible. And if we can increase it enough it is perfectly consistent with continued increases in global GDP. We just need much, much greater increases in GHG throughput efficiency. We need a crash program to reduce fossil fuels in the energy mix and increase energy efficiency.
13 See Hahnel 2013 for discussion of specific ways in which capitalism tends to generate an unhealthy and environmentally unsustainable “growth imperative.”