International Industrial Ecology Day 2021

Price information plays a crucial role in Hybrid LCA by linking physical process data to monetary input-output models. Prices however, are subject to market dynamics and a multitude of trading relations, leading to different prices for the same commodity across supply chains, or even between the different layers of the same supply chain. Given the linear relation between the inputs into a process’ supply chain from the input-output model and the price of it’s reference product, the variability of commodity prices directly results in added uncertainty of the product’s environmental footprint. So far the impact of this variability has only been assessed using theoretical price distributions that had not been based on real world data. This is perhaps not so surprising given that accurate price data for commodities are hard to come by, making it practically impossible to gather enough data to accurately capture the price distributions representing the myriad trading relations a process is typically involved in.
To be able to model commodity price distributions based on real world data we therefore use trade data from BACI/UN COMTRADE, which captures bilateral trade flows of 5000+ commodity groups in both physical and monetary units. These flows are then mapped to the processes’ reference products based on the commodity type and geographic resolution of the process to obtain volume weighted price distributions. These price price distributions are then directly sampled in a Monte Carlo analysis to obtain footprint probability distributions. In this session I would like to discuss the main challenges of modelling variance with limited and/or aggregate data, such as how to deal with the absence of covariance information, low sample sizes in combination with direct sampling, and the communication of uncertainties within the IE community in order to make results comparable across studies.