Open assessment of Swiss economy and society

Background

There have been multiple reports looking into environmental impacts and critical resources in the Swiss economy. These reports were ad hoc and difficult to reproduce or compare. A more accurate and complete accounting is of critical interest to a broad set of stakeholders across Switzerland. Such accounting must represent both Swiss and global supply chains in detail, must have quantified uncertainty, and must be reproducible and easy to update in the future.

Aim

1) Develop a new approach that combines a top-down with a bottom-up database of how resources and energy flow through the economy, in order to produce a detailed and complete picture of these supply chains and their resulting environmental impacts; 2) Add social impacts to this picture; 3) Add critical resources and potential supply chain disruptions to this picture; 4) Develop a global sensitivity analysis framework that can efficiently and accurately process these data and indicators.

Results

Supply chain disruption impacts

A new approach has been developed that allows assessing country-specific supply disruption impacts along a supply chain of technologies. The approach has been applied to the case study of electric vehicles sold in Switzerland, for which hotspots as well as the overall impact of supply disruptions of the cobalt and aluminium supply chains were assessed. Although some materials or steps were already broadly known, such as extraction of cobalt in the DRC, new critical steps were identified, such as the manufacture of aluminium wiring harnesses in Morocco. The disruption to global supply chains of seemingly common resources during the Covid-19 pandemic shows the value of our approach. 

An open source approach 

An open source approach to merging life-cycle inventory databases has been developed and applied to ecoinvent and exiobase. Combining a complete but highly-aggregated database with a detailed set of process models covering some parts of the economy has produced our best picture yet of the environmental impacts of Swiss household consumption of resources, and allowed us to prioritise future data improvement efforts. 

As our approach is available as open source software, and uses open supplemental data, it can continue to be updated in the future by researchers and other stakeholders. The geographic resolution of our input data proved incompatible with the social indicators database, and no robust analysis of social impacts could be carried out. 

Global sensitivity analysis 

We developed a global sensitivity analysis procedure that can efficiently handle large databases with hundreds of thousands of uncertainty input parameters, and can include correlated or grouped parameters, and uncertainties that cannot be made to fit probability distribution functions. One key innovation was the use of iterative validation during the screening steps which identify and exclude noninfluential parameters. 

This validation step allows us to quantify how well our reduced model replicates the full model results, and is computationally efficient as it allows already calculated model runs to be reused. A test for linearity of the reduced model allows for quick calculation of correlation coefficients where possible, or the use of a machine-learning algorithm to calculate sensitivity indices if necessary. The procedure is implemented in open source software building on the Brightway open source life-cycle assessment framework.

Implications for research

The development of a criticality indicator based on detailed supply chains found in BACI trade data opens up new possibilities for life-cycle assessment database development. Tightly coupling finished databases such as ecoinvent and exiobase proved infeasible, but combining the input data behind these databases could be fruitful. Global sensitivity analysis shows the urgent need for more realistic and complex inventory models and high-quality data on uncertainty.

Implications for practice

Our research revealed that existing database quality and modelling poses severe questions for robust life-cycle sustainability analysis (LCA). Sensitivity analysis should be applied to all LCA studies, and data quality iteratively improved through revised statistical models and geographic disaggregation. Multiple databases with different data foundations should be used whenever possible. Resource criticality indicators using open data can be calculated for detailed supply chains.