The fallacies of GDP reductionism: A response to Warlenius

 

Rikard Warlenius has written an article on what he calls “the limits to degrowth”.  In it, he states that because degrowth researchers argue the climate crisis cannot be addressed by decoupling alone (which is true), the only conceivable alternative is to drastically reduce GDP (which is false). He also applies GDP reductions to developing countries. These are ill-informed approaches to climate mitigation that no serious academic would propose, and which scholarship on degrowth and climate justice explicitly rejects. 

The paper is based on false claims about the scientific literature, uses invalid methods, and yields cartoonish results: Warlenius says that to keep global warming to 1.5C, global GDP would have to decline by 86%.  In the UK, GDP per capita would have to decline by 97% (!). Indeed, he goes so far as to say that GDP per capita would have to decline for many lower-income countries too, including India by 39%, Indonesia by 68%, Brazil by 82%. 

These are bizarre figures that have nothing to do with what degrowth proposes. The scientific literature takes a categorically different approach to climate mitigation, with empirical evidence from degrowth scenarios that strongly contradicts Warlenius’ claims. Warlenius does not engage with this literature, or even cite it. This is not an acceptable approach to science.

Here I outline the five most serious flaws of the piece.  Each of these is enough to invalidate the work. Taken together, they indicate very serious lapses in scholarship.

 

1. DG scholarship does not propose GDP reduction as a direct climate mitigation lever. Instead, it proposes known efficiency and sufficiency-oriented mitigation policies.

Warlenius exhibits a very poor understanding of degrowth, which he conflates with GDP reduction even though this is contradicted in the very texts he cites (a fact he does not bother to acknowledge). Degrowth scholarship explicitly rejects the idea of reducing GDP as a direct climate mitigation lever.  Indeed, such an approach makes no sense.

In reality, degrowth calls for specific mitigation policies focused on efficiency, sufficiency and equity, which have been described in the published literature on energy demand reduction (including in the IPCC’s AR6 report), while ensuring decent living standards for all.  Crucially, degrowth does not propose to reduce all production and consumption (as Warlenius assumes). Rather, it focuses on reducing energy-intensive and socially less-necessary forms of production and consumption (e.g., SUVs, fast fashion, mansions, weapons, industrial meat, air travel, etc), while lengthening product lifespans and cutting the purchasing power of the rich. Meanwhile, some forms of production should be increased, like public transit and renewable energy. Other core policies include universal public services, a job guarantee, living wages, and public investment to speed up production of low-carbon infrastructure and efficient technology.

Warlenius does not explore (or even mention) these measures. Indeed, the paper suggests a limited understanding of climate mitigation policy in general. 

It is possible to assess how the policies listed above might affect energy and emissions (see point 5 below). But it is not straightforward to estimate what would happen to GDP in such a scenario.  GDP is an outcome, not a dial to be turned down. Yes, GDP in high-income countries is likely to decline as a result of socially and ecologically beneficial policy. But the specific GDP outcome depends on what sectors are reduced and increased, how provisioning systems are transformed, and how prices may be affected by these transitions and related processes such as decommodification, public investment, an increase in the bargaining power of labour, changes in the production boundary (e.g., compensation for work or production that is not presently compensated), as well as ending unequal exchange and other core-periphery inequalities (GDP represents prices, and prices are an artefact of property relations, which would change considerably under ecosocial policy).  To assess GDP outcomes would require sophisticated macroeconomic modelling. Warlenius does none of this work. Instead, he takes precisely the approach that scholarship in ecological economics rejects.

Furthermore, when it comes to the question of how income might change in a degrowth scenario, what matters is not income as such but rather whether people can access the real goods and services that are necessary for well-being. In a scenario with reduced inequality and expanded/decommodified universal public services, the welfare purchasing power of income increases.  In other words, people would have greater access to welfare-enhancing goods and services with any given quantity of income. To meaningfully model outcomes in a degrowth scenario, this dynamic must be accounted for. Warlenius does not.

 

2. DG scholarship does not posit a “maximum limit” to decoupling. In fact, decoupling can be accelerated through degrowth.

Warlenius’ paper is based on the claim that degrowth proposes a “maximum limit” to decoupling (he claims this is 4% per year). This is false.  In fact, the degrowth literature does not identify a maximum limit to decoupling. Warlenius claims to derive the 4% figure from a 2019 paper by Hickel/Kallis, but this paper does not claim that 4% is a maximum limit. It mentions the figure only as the rate of decoupling used in optimistic growth-oriented modelling scenarios, as a point of comparison for the much higher rates of decoupling required to limit warming to 1.5C or 2C if global economic growth continues at projected rates.

It is critically important to understand that the 4% figure is used for optimistic global decoupling under growth-oriented conditions, understanding that (a) decoupling in individual countries and sectors can and often does occur at faster rates, and (b) decoupling can be accelerated through degrowth policy (on which more below). Yet, Warlenius applies the 4% “limit” (which is not a limit) to all countries, even for “scenarios” that he claims to represent “degrowth” (which do not represent degrowth).

Here’s the key point: scholarship in ecological economics indicates that decoupling can be accelerated through degrowth policy. Remember, degrowth does not propose to reduce all production and consumption, but specific energy-intensive and/or emissions-intensive sectors. If we reduce e.g. SUVs, mansions, weapons, air travel, industrial meat, etc. this would reduce the overall carbon intensity of the economy, meaning faster decoupling. It also liberates productive capacity (factories, labour, materials) that can be mobilized to speed up decarbonization efforts, such as accelerating renewable energy rollout, green tech innovation, insulating buildings, etc. In both of these ways, degrowth can enable faster decoupling than could be achieved in a growth-oriented scenario.  I develop this point further here.

  

3. DG scholarship does not target developing countries. In fact, it calls for faster and more effective development in the global South through industrial policy.  

The "contraction and convergence" scenario that Warlenius deploys bears no resemblance to how contraction and convergence has been coneptualised in the literature.  Warlenius assumes that all countries would converge to a specific GDP/capita, regardless of the level, even if it means impoverishing people (including with large reductions in GDP for India and Indonesia, etc). This is ethically unacceptable, and inconsistent with the relevant literature.

In reality, degrowth scholarship focuses on demand-reduction in high-income countries, and calls for faster and more effective development in the global South through industrial policy to meet human needs.

For climate mitigation, which is the focus of Warlenius’ paper, the proper objective of contraction and convergence is to limit emissions in line with internationally agreed climate targets and equity commitments (thus to share the carbon budget fairly), while achieving decent living for all.  With such an approach, countries with high per-capita emissions need to immediately and rapidly reduce their emissions, while countries with low emissions per capita (such as India and Indonesia) can increase their emissions in the near-term before eventually reducing them, or mitigate at a slower pace. This makes it possible for lower-income countries to increase aggregate production to achieve human development objectives. Warlenius’ GDP reductions in lower-income countries are strongly at odds with the literature on degrowth and climate justice. 

Recent literature on degrowth has defined convergence as bringing energy and material use to levels that are (a) compatible with sufficiently rapid decarbonization and ecological stability, and (b) sufficient for high levels of human well-being.  Warlenius’s approach ignores this specificity.  It is impossible to ascertain what his results mean for energy and material use (which is a problem in itself), but it is clear that the large reductions in GDP per capita (especially in developing countries!) very likely violate the latter condition.

In sum, the scenario that Warlenius explores is not a degrowth scenario (it does not align with any of the published literature on degrowth and indeed violates core degrowth principles). Rather, it is a scenario of massive and prolonged recession and impoverishment.

4. Real degrowth policy scenarios have been modelled in the scientific literature and the results contradict Warlenius’ claims.

In ecological economics, degrowth scenarios have been modelled by focusing on energy demand-reduction, and the energy requirements of decent living (i.e., required to provision specific necessary goods and services). Remarkably, Warlenius does not engage with this literature, which is obviously relevant to the question he seeks to address.

Keysser and Lenzen (2021) find that decarbonization consistent with 1.5C can be achieved without negative emissions and with feasible efficiency improvements if annual global final energy demand is reduced by 25%, from 400EJ to 300 EJ by 2050, with reductions occurring primarily in the global North. Even assuming a population of 10 billion in 2050 (which is higher than the probable peak), per capita annual energy use would be 30 GJ. Research by Millward-Hopkins et al (2020) indicates that, if we transform the economy to prioritize well-being, this level of energy would be more than enough to ensure universal access to decent living – a standard that is presently not attained by large portions of the human population (Kikstra et al 2022).  In other words, this scenario would represent a dramatic improvement in the living standards of billions of people. These results clearly contradict Warlenius’ claim that global GDP would need to decline by 86% and that people in countries like India and Indonesia would need to accept massive reductions in consumption.

Barrett et al (2022) find that degrowth-aligned policy in the UK can reduce energy use by 52%, through a combination of efficiency and sufficiency measures, which would make it possible to reduce emissions fast enough to stay within fair-shares of Paris-compliant carbon budgets “without compromising citizens’ quality of life”. The authors do not model what this would mean for GDP (again, such a study would be highly complex), but, given efficiency improvements, any GDP reduction would certainly be less than the 52% energy reduction. This clearly does not accord with Warlenius’ claim that the UK’s GDP per capita must decline by 97%.

Similar modelling has been done for France and Germany, with similar results. There are many other studies that have modelled post-growth and degrowth scenarios in empirical terms (including the CLEVER report that finds efficiency+sufficiency measures can reduce energy use by 55% across 30 European countries by 2050, making it possible for them to decarbonize fast enough to stay within 1.5C fair-shares).  None of them are discussed in Warlenius’ paper.  Warlenius also fails to engage with any of the relevant empirical literature on sufficiency or demand-side mitigation, which is now a large field of research and is substantially represented in the IPCC’s AR6.

 

5. Warlenius makes claims about future decoupling that are not supported by any empirical evidence.

Warlenius seems to believe high-income countries should continue pursuing increased aggregate production.  He acknowledges that existing decoupling rates are far too low to reconcile this vision with ecology. To deal with this problem, he implies that a massive and more or less immediate 10-fold increase in decoupling may be possible to achieve. However he does not provide evidence for this.  Instead, he says it can be done with “structural changes” to the economy (which he says may affect GDP “in contingent ways”, without attempting to specify further, which is odd given that he is otherwise preoccupied with this question). Incidentally, scholarship in ecological economics also argues that the rate of decarbonization can be accelerated with structural changes to the economy, such as the degrowth policies described above, but – unlike Warlenius - provides empirical data to demonstrate this (which, again, Warlenius does not engage).

Warlenius claims to “sketch out” a “dynamic theory of decoupling” as an alternative. This sketch has close to zero content, and mainly hinges on a reference that has no bibliographic information, making it impossible for readers (and, presumably, reviewers?) to engage with or scrutinise the proposed theory.

Warlenius does not consider the time dimension of decoupling. In the way he conceptualises decoupling rates, they relate to constant exponential decay rates of emissions. Slower decoupling rates in the near term would require (much) faster decoupling rates later on to remain within the same carbon budget. This point is crucial because there are very real physical limits to how fast decoupling can be accelerated in the near term (at least under growth-oriented conditions) given that it requires replacement of existing infrastructure (e.g. energy infrastructure) and technology (e.g. existing car fleet) at an enormous scale, which cannot occur at just any desired speed, and cannot be accelerated by an order of magnitude overnight.

If we recognize this material reality, it makes sense to take a multi-pronged approach.  Yes, we must mobilize to improve efficiency and build out renewable energy infrastructure. But such efforts on their own are extremely unlikely to achieve sufficiently rapid decarbonization. So, we should also introduce policy for sufficiency and equity, and scale down less-necessary forms of production - because we know that this approach would dramatically accelerate decarbonization and make the crucial difference.  Every fraction of a degree counts.

Addendum

As it happens, we have a new paper in The Lancet Planetary Health on GDP-CO2 decoupling that further establishes many of these points (“Is ‘green growth’ happening?”). Here are some excerpts that are directly relevant to the question at hand:

-”Decoupling can certainly be accelerated. However, there are real physical limits to how much and how fast decoupling can be sped up within a growth-based approach. Under growth-oriented conditions, decoupling (indeed mitigation) relies mainly on replacing existing infrastructure and technology (eg, energy infrastructure and the car fleet) with low-carbon or low-energy alternatives. This type of transition cannot be done at just any desired speed, nor promptly accelerated at any desired rate, given available production facilities, know how, labour, material resources, existing infrastructure, and so on. And slower decoupling rates in the near term would require much faster decoupling rates later to remain within a given carbon budget. The large, near-instantaneous acceleration of decoupling that would be required for high-income countries to achieve green growth is thus very unlikely to be feasible.”

-”In decoupling terms, the [post-growth and degrowth mitigation] measures described in this paper substantially and rapidly reduce the overall carbon intensity of the economy, and thus accelerate decoupling beyond what can be achieved in a growth-oriented scenario through replacement of infrastructure and technology.”

-”Debates about green growth relate to high-income countries. Lower-income countries typically have much lower emissions per capita, which makes the mitigation and decoupling rates required for them to stay within their fair-share carbon budgets more modest and therefore more achievable. Countries such as Uruguay and Mexico are already making strides in this direction. With adequate access to the necessary finance and technology, freedom to use industrial policy, and a development strategy focused on human needs, lower-income countries should be able to stay within their fair-share carbon budgets even while increasing production and consumption to achieve decent living standards for all. Indeed, post-growth transitions in high-income countries are crucial for enabling and creating space for sovereign development in lower-income countries.”

-”We want to emphasise that post-growth climate mitigation scenarios cannot be modelled by assuming some decoupling rate and simply reducing GDP. Indeed, post-growth scholarship explicitly rejects the idea of reducing GDP as a lever for climate mitigation, focusing instead on specific sufficiency and efficiency policies (as described above), along with public investment to accelerate decarbonisation. Crucially, post-growth proposals do not seek to reduce all production and consumption, but primarily carbon or energy intensive and less-necessary forms of production and consumption, while also increasing necessary forms of provisioning as needed. Whereas the energy and emissions impacts of key post-growth climate-mitigation policies have been modelled, what would happen to GDP in a postgrowth scenario depends on various factors, including what sectors are reduced or expanded, how provisioning systems and income distributions are transformed, to what extent provisioning gets decommodified, to what extent currently unpaid work or production gets remunerated, and what happens to prices. Clearly, changes in GDP cannot be simply deduced from an assumed emissions pathway and decoupling rate (see the Methods section). It is quite possible that GDP could decline in a post-growth scenario, but post-growth labour and welfare policy can secure livelihoods and improve wellbeing independently of what happens to GDP.”

-”Pathways of GDP cannot reasonably be inferred from assumed emissions pathways and decoupling rates because emissions are the outcome of economic activity, not the other way around, and because decoupling reflects both physical and monetary changes in the economy.”