Aggregate material footprint as a proxy for environmental pressures: A response to Max Roser

by Lorenz Keyßer (Twitter: @LorenzClimate), Post-Growth Zürich (

On 2nd February, Max Roser tweeted the following text: “Material Footprint’ is a terrible metric, but it is unfortunately used as one of the indicators of the UN’s Sustainable Development Goals. A metric that allows you to offset the use of a ton of coal by using one less ton of sand should really not have that status.” (1) In the following discussion he said that we should: “stop reporting environmental impact in this way.” (2) and that he finds “Equal weights for fossil fuels and stones […] ethically absolutely horrible.” (3) To make this point, he referred to a recent publication, comparing and disaggregating the EU’s overall material footprint (MF) between 1995 and 2011, and more specifically, to one apparent substitution of clay and sand (decreased) with coal use (increased by roughly the same amount) (see appendix 5 from ref. 4).

In the following, I briefly lay out why I believe his reasoning to be not only wrong but in contradiction with his stated goals of understanding and alleviating pressures on the environment (5). To begin with, his interpretation of the initial paper is incorrect: coal use did increase, as did the MF, but fossil fuels in total were the only broader category that actually decreased (4). This points to the usefulness of aggregate material flow indicators: their purpose is to study the social metabolism, which is the material scale, composition, and pattern of the human-environment interaction (6). It is clear that the sole use of MF for all these aspects has serious limitations, along the lines that Roser voices (e.g. giving equal weight to very different materials), because it is only an indicator of scale. However, this rather seems to be a problem of careful treatment and interpretation of MF alongside and complementary to other and more detailed/fine-grained indicators. In the end, no single indicator can adequately capture all environmental pressures in all contexts. But instead of such a differentiated treatment, Roser completely discards the MF indicator. This is dangerous, given its demonstrated usefulness in indicating aggregate environmental pressures.


If we look beyond Roser’s blanket rejection of MF and at available scientific publications, we find that MF is actually a useful proxy for aggregate environmental pressures. This is simply because all material extraction (be it biomass, metals, fossil fuels or non-metal minerals) has some impact (7). So, if we measure MF and environmental impacts separately, we would expect to find a correlation between the two. This is exactly what happens. MF is highly correlated with other indicators for environmental pressure, such as carbon dioxide and ecological footprint (6). Also, as Krausmann et al. state (8, see also 9), material use (here domestic) correlates well with more complex indicators such as environmentally-weighted material use, accounting for different impacts of the respective material categories. Another reason why MF is a useful indicator is its simplicity while being representative. Indeed, Steinmann et al. (10) assessed 976 products and found that (p. 1) “the resource footprints [here material (excl. energy and biomass), energy, land, and water] accounted for >90% of the variation in the damage footprints. […] Our results indicate that relatively simple resource footprints are highly representative of damage to human health and biodiversity”. Finally, Voet et al. (11) reach similar conclusions (p. 130): “if we compare environmental impact with DMC [Domestic Material Consumption], we can conclude that the contribution to the environmental pressure and the contribution to the DMC is not so different for these [resource] categories”. Thus, MF captures important connections between human extractive activity and environmental pressures quite well on an aggregate level, and this in a simple and understandable way.


Again, it is clear that it is problematic to solely rely on MF to assess environmental impacts on a detailed level. But instead of completely rejecting MF as an indicator, it simply needs to be acknowledged that MF should be complemented by other indicators and differences between materials need to be considered (which is anyway already mostly acknowledged). I can not understand Roser’s anger about using MF and he does not provide a valid alternative to his >240’000 followers. The problem with the Sustainable Development Goals rather seems to be that they currently have a bias towards measuring environmental impacts relative to GDP (another “terrible” aggregate indicator) and not absolutely (12) and rely on further overall GDP growth (12, 13). The latter is highly likely to be at odds with environmental sustainability (6, 14, 15), which Roser stated to advocate. Strangely, this relative assessment and reliance on GDP as a goal is not criticised at all by Roser.


Edit, 24.02.21: I am thankful that Max Roser has replied to this article in the following Twitter threads (, Especially in the second one, he clarifies his argument, pointing out that his alternative to MF is to only study environmental pressures separately. He lists some of his work on these pressures, e.g. meat consumption and land use, here ( To me, his arguments in favor of his approach rather are arguments in favor of supplementing MF with more detailed indicators, not completely discarding it (because of its demonstrated usefulness described above), leaving the core argument of my article intact.





I am grateful for helpful comments on an earlier draft of this text by Viktoria Cologna and Giulia Fontana. The remaining errors are mine


One Comment on “Aggregate material footprint as a proxy for environmental pressures: A response to Max Roser”

  1. Nicely made argument. I saw the tweet and was thinking the same way. The data show that material footprint increases as a power law of GDP, much like Brown et al. (2011) showed for energy. I agree that no single indicator is perfect, but aggregate MF is rich with information if thought about carefully. And while a tone of sand may not equal a ton of coal, it takes a lot of coal to extract and process the sand for use in production. I truly love “Our World In Data” but Max is clearly wrong here.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.