A unified framework of life cycle assessment

The dichotomy between the attributional approach and the consequential approach is one of the major unsettled questions in life cycle assessment (LCA). Debates continue on. Here, I suggest a different view that hopefully will help us move past the dichotomy and toward a unified framework for decision support.

I argue the dichotomy is unnecessary. Attributional LCA, as reflected in how the conventional models like process- and input-output (IO)-based LCA have been applied, is simply a linear consequential modeling that establishes cause and effect through product supply chain. This is how economists see IO analysis, namely, a linear model based on Leontief production functions. There are other consequential and causal models, such as computable general equilibrium (CGE) and system dynamics (SD), that may have different production functions or focus on other aspects of the economy. These models have been increasingly integrated into LCA to estimate the environmental effects, impacts, or consequences of products, which is at the core of LCA. I further argue that as a field, we may be better off eliminating both terms: attributional fails to capture the essence of LCA and consequential is redundant. Likewise, economists did not call IO attributional and CGE consequential because both are consequential models, nor did they use the term consequential as that would be stating the obvious.

I suggest LCA be unified around its goal to support decision-making, which requires estimating the impact of changes associated with a decision versus that without it (the counterfactual). This is the basic methodology adopted in many other fields. LCA then becomes an overarching framework that encompasses a suite of models, including our conventional IO/process-based LCA, to support different levels of decision-making related to products. Which distinguishes LCA from other fields of study is the focus on product systems. I also discuss, for the linear IO/process-based models, under what circumstances their estimates of the existing systems as an approximation of changes may be meaningful or misleading for decision-making. I further touch upon the importance of understanding the counterfactual, which has been largely neglected in LCA literature.

For an LCA to support decision-making, (1) make explicit what the potential changes are and clearly define the scale of change and (2) derive range estimates to capture the high uncertainties in modeling complex systems and be willing to admit inconclusiveness. For decisions with potentially large impacts across sectors, a multi-model approach may be helpful.