Periodically, American economist Paul Krugman cherry-picks a few numbers to argue that economic growth is more sustainable than we think. While these short outbursts of optimism usually stay within social media (see, for example, these tweets from last Summer), it somehow made it to The New York Times last week (“Wonking Out: Why growth Can Be Green”). As someone who has been actively researching this very question for the last five years, I think Paul Krugman is wrong and that the argument he defends is not only baseless but also potentially dangerous.
Sustainability is a multi-faceted challenge
Paul Krugman brings two piece of evidence to defend his claim. The first is that air quality in New York is better than in Delhi even though GDP in the United States is higher than in India. This is no proof though and one can easily find opposite cases. According to the World Air Quality Index, air quality in Nairobi, Mexico City, and Jakarta is much better than in Milan, Seoul, or Paris, even though income levels are lower in the former countries than in the latter. So, this proves nothing. As Andrew Ahern remarks in his response to Krugman on Twitter: “the idea that an increase in GDP growth was responsible for higher air quality in New York City or anywhere else completely missed the mark. It is through regulations and prioritizing air quality that better air quality is reached – not in the hopes that once you become ‘rich’ you have better air quality.”
The fundamental point is this: sustainability is not only about air quality, carbon, or whatever your favourite indicator might be. It is about all of that together. Arguing that economic growth can be greened by showing evidence of one single impact going down is like bragging about being healthy by showing that one of your arms has lost weight. An economy that deserves to be called “sustainable” is one that respects the scale of its supporting ecology. Think about the economy as a giant metabolism that extracts resources from nature (water, metals, minerals, biomass, etc.) and generates a number of impacts (greenhouse gases, heavy metals, particulate matters, e-waste, excrements, etc.). To be sustainable, every single environmental pressure (both at the levels of extraction and pollution) must not transgress the carrying capacity of surrounding ecosystems.
It has become common to refer to the “planetary boundaries” as a simplified framework for sustainability. When the IPCC talks about “staying within planetary boundaries,” it means that humanity should not cross 9 quantitative thresholds (climate change, ocean acidification, ozone layer depletion, air pollution, biodiversity loss, land conversion, freshwater withdrawals, nitrogen and phosphorous loading, and chemical pollution). These can be downscaled at the level of a single nation. This recent study by Fanning et al. uses a simplified seven-boundary framework to evaluate the sustainability of different countries. As Figure 1 shows, the United States performs rather poorly, overshooting all of its ecological thresholds. For now, what is important is the fact that high-income countries like the US are no pinnacles of sustainability; it’s the opposite, the US is one of the most unsustainable economies in the world (see Figure 15).
If you want to show that growth is genuinely green, you must bring evidence that GDP has decoupled from all environmental pressures. After five years actively searching for this particular proof, I have never seen anyone managing to do that. The example of Spain is illustrative (Figure 2). Keeping track of GDP in relation to 16 indicators of environmental pressures, a report from May 2022 concludes that “overall and considering the weighted index of all environmental impacts, Spain’s consumption footprint per capita increased its value between 2010 and 2018 by 4.4%” (p. 28). In other words, GDP is going up but nature is going down, a situation we can observe in all high-income nations.
Figure 2: Spanish Ministry of Consumer Affairs/European Commission, May 2022
Don’t get fooled by per capita emissions
The second piece of evidence brought by Paul Krugman points to greenhouse gas emissions in the United Kingdom. He argues that British greenhouse gas emissions have reverted back to their 1850s level (yes, this is not a typo, 1850 as in the Victorian era). Let me start with the obvious. Emissions per capita are total emissions divided by population. There were 17.9 million people living in the UK in 1850 but there are 66.8 million today, so it is disingenuous to compare individual levels of emissions. What really matters for climate change is national emissions, the total number of tons being emitted into the atmosphere. This is Krugman’s first mistake: British emissions have not gone down but up, increasing from 124 Mt in 1850 to 380 Mt in 2018 (Figure 3).
Figure 3: PIK data, World Resources Institute online platform
There is a second problem with Krugman’s argument: the time span is way too long to be meaningful. Assessing the effectiveness of climate policies over centuries is as useful as measuring free diving records in weeks. What we really want to know is whether we will manage to reduce emissions by 2030 or 2050 (the deadlines usually given by climate scientists), and so the best we can do for reliable estimation is to look at what we already managed to achieve in the last 30 years. Figure 4 shows the evolution of British territorial emissions between 1990 and 2018, a decrease from 603 Mt to 380 Mt.
Figure 4: PIK data, World Resources Institute online platform
Don’t forget imported emissions
These are territorial emissions, only counting whatever has been produced within the frontiers of a specific nation. But it’s an unsatisfactory way of estimating contributions to climate change since a country like the UK imports a large part of what it consumes. For a better assessment, we can calculate its carbon footprint using consumption-based indicators. The turquoise area in Figure 5 shows these “imported emissions,” which represented roughly 47% of the British carbon footprint in 2019, as compared to only 38% in 1996.
Figure 5: UK government, August 2022
This is important because it indicates that part of the reduction in territorial emissions can be explained by a delocalisation of polluting industries abroad. Only looking at inland emissions creates a local illusion of absolute decoupling where, in fact, the problem is just shifted elsewhere. This is like transferring one kilo from your left leg to your right leg and calling it weight loss. The structure of a carbon footprint should also inform the way we think about our future capacity to reduce emissions. I often hear that technological progress in rich countries will green production. Even if that were true (I have my doubts, see Decoupling Debunked), this would only cover a portion of emissions (a maximum of 53% in the UK).
On the same online platform used by Paul Krugman, you can compare production-based territorial emissions and consumption-based footprints (Figure 6). In 2018, British footprint emissions were at 535 Mt, 1.4 times larger than territorial emissions (380 Mt). This alone should tamper our optimism. Instead of the 5 tons per capita announced by Krugman on his tweet, the real carbon footprint of the average Brit in 2018 was closer to 8 tons. (Let’s also add here, as Genevieve Guenther did in her response on Twitter, that land use change is not included in these figures.)
Figure 6: Global Carbon Project data via Our World in Data
In the largest systematic review of the decoupling literature (835 articles), Haberl et al. (2020) synthesise all the studies looking at decoupling rates of PIB from greenhouse gases in high-income countries during the last decade. Using production-based indicators, there is indeed absolute decoupling: with 1 additional point of GDP comes -0.04 points of emissions. Minuscule but absolute. But if one uses consumption-based indicators instead, the absolute decoupling disappears: 1 additional point of GDP brings +0.22 points of emissions.
One final observation. Footprints are difficult to calculate as they require sophisticated global models. I can speak for France because that’s a country I know fairly well. Based on the Global Carbon Project data (the data used by Our world in data), France reduced its consumption-based emissions by 25% from 1995 (500 million tons, 8.7 ton per person) to 2021 (375 million tons, 5.8 tons per person). But this is far-away from reality. Looking at more refined and recent data from the French government, we get a slightly different picture. The carbon footprint in 1995 was 11 tons per person, 26% higher than the Our World in Data / Global Carbon Project estimation, and the one in 2021 was 8.9 tons, 53% higher. The difference is substantial. The -25% reduction announced by Our World in Data is actually only a -9% reduction, so 2,7 times smaller.
How much emission reduction is enough?
For now, let’s get back to the Global Carbon Project data. British production-based emissions in 2018 are 37% lower than they were in 1990, whereas consumption-based emissions are only 20% lower. To be fair to advocates of green growth, let us focus on the period with the fastest reduction rates. Le Quéré et al. (2019) estimate that British consumption-based emissions decreased by -2.1% per year between 2005 and 2015 with positive GDP rates around +1.1% (Figure 7). (If we look at a longer period, Lamb et al. (2021) find that, from 1970 to 2019, the yearly reduction was only of -0.9%.)
Figure 7: Le Quéré et al., 2019
Last time I checked, the British government had pledged to reduce territorial emissions by 5.1% per year. But to actually comply with the Paris Agreement without relying on not-yet-existing negative emission technologies, the UK must rather achieve a yearly 13% cut in emissions, starting now and continuing steadily until carbon neutrality is reached (Figure 8). The data on this is clear: there is a very large gap between what has been achieved until today and what remains to be done. The question is not whether or not growth is ‘green’ but rather how much have we managed to reduce footprints, concerning which impacts. And most importantly: is it enough relative to planetary boundaries?
Figure 8: Anderson et al., 2020, personal adaptation
Here is another way to show the same point, namely that the necessary cuts in emissions we need before 2030 are disproportionate to the ones we had, even during the fastest decoupling years. Using the online 1.5°C national pathway explorer, we see that UK must reduce its territorial emissions by 75% compared to 1990 (Figure 9), which is a bit more than the government’s latest climate target (-68%). As a reminder: UK reduced territorial emissions by 38% in the last three decades (1990-2018) and so is expected to do that again within the next 7 years, and probably much more if we include a commitment to reduce imported emission as well (and probably even much more if we take into account responsibility over historical emissions and issues of equity).
Figure 9: 1.5°C national pathway explorer, consulted Feb 2023
Economic growth makes it harder to reduce emissions
Krugman’s graph showed values until 2020 and so it includes the reduction of emissions caused by the Covid lockdown (and the drop following the 2008 financial crisis as well). This leads to a strange argument. We are expecting faster economic growth to accelerate decoupling even though a large portion of historically achieved decoupling has happened because of slower growth. This is quite a twist: the data supposedly demonstrating that economic growth can be green actually shows the precise opposite. There lies another crucial point: economic growth makes it harder to reduce emissions compared to a no-growth or negative growth scenario.
Take France, for example. It reduced its consumption-based emissions by 125 Mt between 1995 and 2021 (I have just argued these data are not precise enough, but this doesn’t matter for this particular point). Looking at the yearly change, the biggest reductions happened during the 2008 financial crisis (-47 Mt) and the pandemic (-52 Mt). These two events account for 79.2% of the carbon reduction during the period (Figure 10).
Figure 10: Our World in Data; red areas when emissions increase, yellow when stable, green when they decrease
Back to the Le Quéré et al. (2019) study. The authors acknowledge that this period is nothing extraordinary: “These reductions in the energy intensity of GDP in 2005-2015 do no stand out compared to similar reductions observed since the 1970s, indicating that decreases in energy use in the peak-and-decline group could be explained at least in part by the lower growth in GDP.” It comes as no surprise then that, using simulations to forecast future changes in emissions, the authors estimate that “if GDP returns to strong growth in the peak-and-decline group, reductions in energy usemay weaken or be reversedunless strong climate and energy policies are implemented.”
This is something so obvious that it is bizarre that one needs to restate it. Check out the three pathways from the “Scenarios for the 21st century” platform (I ran the model for France). The business-as-usual of +2% yearly GDP growth with -4.7% reduction in carbon intensity leads to an overshoot of the 1.5°C carbon budget before 2030 (the red area in the graph). In order to achieve carbon neutrality before that budget runs out, we can either maintain GDP growth at 2% (but then we have to accelerate decarbonation to 16% per year), or we can lower production and consumption in order to accelerate the curbing of emissions (see the graph at the bottom right in Figure 11). Don’t take these numbers too seriously, they’re only here to show the point that the reduction of emission we are trying to achieve can be accelerated through degrowth.
Figure 11: IFSO online platform, Scenarios for the 21st century for France
In which sectors is decoupling happening?
Ultimately, an economy doesn’t have a GDP button we can push up or down. Economic growth and degrowth are aggregated, ex post phenomena, each the result of myriads of production and consumption decisions within a complex economy. While looking at GDP decoupling can give us a general idea of the overall footprint of an economy, it doesn’t really tell us which sectors have been driving that decoupling. To see precisely that, let’s look at one of the few decoupling studies which disaggregates emissions at the sectoral level. Lamb et al. (2021) find that most emissions reductions between 1970 and 2018 were achieved in the energy systems sector, specifically in electricity and heat generation, even though emissions in other sectors like transport were either stable or increasing.
Figure 12 shows a decomposition of emissions in France between 1970 and 2018. Industrial emissions (metals, chemical, and other industries) were lower in 2018 than during the country’s peak year sometime in the early 1970s. Same case for non-residential buildings and energy (electricity and heat). On the other hand, road emissions increased by +73.8% over the same period, and so did residential buildings (+7.9%), fuel combustion in agriculture (+46.5%), and domestic aviation (+1519%).
Figure 12: Lamb et al. (2021)
What this tells us is that we should apply a diversity of strategies to reduce emissions: avoid, shift, and improve. Avoid consists in consuming less of something; shift means substituting one type of consumption for another; and improve is the greening of an existing type of consumption. If we’re talking about transport, we should avoid unnecessary travels, shift to low-carbon modes of transportation, and improve the efficiency of all the modes of transport we cannot avoid using. There is pure sufficiency (avoid), pure efficiency (improve), and a mix of the two (shift).
Mind carbon inequality
There is also something else going on. In the same way that the concept of macroeconomic decoupling hides different sectoral dynamics, total and per capita emissions disguise inequalities among different classes of people. Let’s look at data from the United States. If the average carbon footprint in the US in 2019 was 41.7 tCO2e per capita, the footprints of the richest households were much higher: 250 tons for the top 1% and 98 tons for the top 10%. In comparison, the average footprint of the poorest decile of the US population is only 16.8 tons, which is still massive compared to the average human footprint (6 tons). Overall, the top 30% wealthiest Americans are responsible for half of all national emissions, almost twice the volume of emissions of the poorest bottom half of the population (Figure 13).
Figure 13: Starr et al., 2023, March 2023, personal adaptation
These studies are precious because they hint that reducing inequality can deliver faster emission reduction than aiming for an abstract decoupling at the level of the nation (this was one of the key points of the latest IPCC report). In the 2023 Climate Inequality Report, Lucas Chancel and others put into context the terrifying magnitude of global carbon inequality. The richest 10% of humans owns 76% of all financial wealth and are responsible for roughly half of emission on the planet while half of humanity owns less than 2% of world wealth and only causes 12% of total emissions (Figure 14).
Figure 14: Chancel et al., 2023, January 2023
Decouple what really matters
Here is my last point: decoupling GDP from environmental degradation is an obsolete goal. Recently, there has been a shift in the literature from GDP to more holistic indicators of wellbeing. If GDP is a flawed indicator of prosperity, why are we trying so hard to green it? Perhaps a novel, more exciting objective should be to decouple wellbeing from environmental pressures instead.
My favourite piece of science on the matter is this comparative study from Fanning et al. (2022). Figure 14 tracks how countries evolved from 1992 to 2015 in terms of social performances (the 11 indicators of Kate Raworth’s doughnut displayed on the vertical axis) and levels of ecological overshoot (a selection of 7 planetary boundaries measured on the horizontal axis). A truly sustainable development should be as vertical as possible: an increase in life expectancy, education levels, subjective wellbeing etc. rise without blowing up ecological limits.
This is where the ideal of green growth starts to feel outdated. A swathe of recent empirical studies have found that past a certain level, GDP per capita loses all correlation with indicators of quality of life. For instance, this study from Vogel et al. (Figure 16) reports that after a certain level of affluence, economic growth become socio-ecologically detrimental – Herman Daly would have called this “uneconomic growth,” additional production that bears more costs than benefits. In such situation, a planned, selective downscaling of reduction and consumption (degrowth) can in fact improve quality of life.
Figure 16: Vogel et al., 2021, personal adaptation
Paul Krugman’s piece makes a poor defence of green growth. Picking a random graph on Our World in Data mixed with hearsay about air quality in Delhi is not science, it’s banter. There are almost 1,000 empirical studies on decoupling with the latest IPCC report attempting a synthesis of them; I recommend people who want to participate to that debate to read at least some of them.
When it comes to the decoupling we need to effectively address the multiple biocrises of today, all scientific evidence confirm that it hasn’t happened yet. Being precautious, we should not expect much more than what we already had, that is relative decoupling with rare situations of often local and temporary – and in any case meagre – absolute decoupling of a few isolated resources or impacts. I say this without a celebratory grin. I actually wish green growth existed. Of course, there are also social issues linked to economic growth, but taking ecosystems out of the picture would simplify the problem, or at least give us more time to solve it. I have no sentimental prejudice here, it just feels so irresponsibly foolish to bet the survival of humanity on a highly improbable miracle. My worry is that we’re losing precious time arguing that maybe, one day, perhaps, if-this-if-that, decoupling could happen. In the meantime, we are merely tinkering with a system that should be radically transformed.
The story of decoupling is reassuring; it’s a don’t worry, everything is fine, everything is going to be okay kind of thing to say. And this is precisely why that story is dangerous. As ecosystems are getting nightmarishly worse, the fable of green growth is acting as a kind of macroeconomic greenwashing, especially when mobilised to discredit other, more radical solutions to the ecological crisis.
 We can look at the World Inequality Database for numbers in the UK: the average Brit emits 9.9 tCO2e. In comparison, the top 1% of British people emits 76.7 tons, the top 10% emits 27.7 tons, and the average footprint of the poorest half of the population is 5.6 tons.