Do material efficiency improvements backfire?: Insights from an index decomposition analysis about the link between CO2 emissions and material use for Austria

Abstract To keep global heating and other negative consequences of socioeconomic activities within manageable boundaries, industrialized countries must undergo substantial decarbonization, requiring the exploitation of synergies with other environmental endeavors. Improving resource efficiency—that is, reducing the resources required to generate a unit of economic output—is a prominent goal pursued across levels of scale. How does resource efficiency relate to decarbonization? Do economies decrease their emissions as they become more efficient? We examine this relationship for Austria from 2000 to 2015 by conducting an index decomposition analysis at the sectoral level by using consumption‐based indicators from the multi‐regional input–output model Exiobase. Our analysis shows that for Austria, the currently popular pursuit of material efficiency appears to run the risk of coinciding with higher emissions, suggesting that the opportunities to achieve both decarbonization and dematerialization are limited. The Austrian service sectors could contribute to a reduction of the CO2 footprint via material efficiency improvements, but strong economic growth foils this possibility coming to fruition. The Austrian economy would do well to either curb demand for goods and services driving global CO2 emissions or to produce imported goods and services domestically in an environmentally more benign manner.


Methodological approach
We analyzed changes in material and CO2 footprints and derived efficiencies using the index decomposition method LMDI I. Index decomposition analysis (IDA) is a tool to study the impacts of sectoral intensity changes and other economic structural changes on trends in e.g. emissions and energy use in industries (Ang, 2004;Ang et al., 2009Ang et al., , 2010 and is widely accepted by policy-makers. A decomposition analysis dissects the underlying factors determining the development of a certain endogenous variable (e.g. CO2 emissions or other pollutants) (Dietzenbacher & Los, 1998). It assumes that there is a functional dependency between the exogenous underlying factors and the endogenous variable, which can be decomposed into the changes between two points in time determined by each factor using differential calculus (Hoekstra & van den Bergh, 2002).
We started from the basic equation of environmentally extended input-output analysis (EEIOA) to calculate Austria's material (MF) and CO2 footprint (CF). For the derivation of the equation and more information on EEIOA see, e.g., Miller and Blair (2009).
with the elements of f representing the materials or CO2 necessary for one Euro production of sector j in each country, L representing the production of sector j in each country that is necessary for one Euro of consumption of sector i in a country (Leontief inverse) and y representing the Austrian final demand (i.e. private and government consumption, investments and changes in inventories). We furthermore distinguished between the part of the footprint that is processed domestically, i.e., when materials and CO2 occur due to requirements of Austrian production sectors (that then ultimately deliver to Austrian final demand), and the part that is internationally appropriated via imports, i.e., when material extraction and CO2 emissions are connected to products delivered to Austrian final demand by foreign production sectors. A schematic representation of the calculation of the footprints and the distinction between domestic and imported footprints is given in figure S1-1.

S1-3
Figure S1-1. Scheme of the calculation of the domestic and the imported footprint in a multiregional input-output model. Country A represents Austria in our study; orange indicates the domestic (dom), blue the imported (imp) fraction of the Austrian footprint.
Our decomposition equations are based on the following underlying functional form, distinguishing four explicit determinants: (S2) • A -economic growth effect: changes in Austrian GDP • S -value added structure effect: changes in gross value added structure (GVAi/GDP) per sector i • ME -material footprint intensity effect: changes in material footprint per value added of sector i (MFi/GVAi) • EI -emission-to-resource ratio effect: consumption-based CO2 emissions per material use of sector i (CFi/MFi) In comparison to other IDA methods, the logarithmic mean Divisia index method (LMDI) is often preferred as it is easy to use and highly adaptive to different study designs (Ang, 2004).
The basic formula for LMDI I in the additive form for the k th factor x for the aggregate variable V is: Where function L(a,b) is the logarithmic average of two positive numbers a and b (a  b) given by (S4) When we apply the LMDI additive decomposition to the four-factor decomposition we defined earlier, we arrive at following decomposition equations: In addition to the IDA for the total Austrian footprint, we decomposed the domestic and the imported fraction of the material and the CO2 footprint separately. The respective LMDI I equations have been changed to only include the domestic or the imported fractions of CO2 as well as material footprint, resulting in two decompositions, which were as well calculated on the economy and on the sector level: