An article led by Aaron Paris from the Institute of Environmental Sciences at Leiden University, and with contributions from ForestPaths partner Amelie Müller, offers practice-oriented recommendations for the informed use of Integrated Assessment Models (IAMs) in prospective Life Cycle Assessment (pLCA) and raises awareness of their limitations. 

IAMs are increasingly used to generate prospective Life Cycle Inventory (pLCI) databases for pLCA. While this approach offers advantages, studies often show limited awareness of its limitations. Drawing on their own experience, the literature from the IAM and LCA communities, and recent discussions on the topic, the authors outline nine characteristics of IAMs linked to potential limitations, and offer practical recommendations for using IAM-based pLCI databases responsibly. 

Nine characteristics of IAMs for pLCI databases 

1. Sectoral perspective: IAMs model the economy sector by sector rather than tracking full life cycles, and their technological detail is often too coarse for LCA purposes. This leaves gaps in IAM-based pLCI databases and adds uncertainty when translating IAM data into LCI data. 

2. Focus on climate change: IAMs are built around climate-relevant sectors and emission-reducing technologies. This means pLCA built on IAM scenarios is less useful for assessing environmental impacts beyond climate change. 

3. Weak representation of material cycles: IAMs track energy flows carefully but largely ignore material flows, recycling, and circular economy strategies. As a result, material-efficiency-based mitigation strategies are missing from IAM-based pLCI databases. 

4. IAMs are not neutral: The assumptions built into IAMs reflect the backgrounds of their developers and predetermine what futures are considered possible. This means IAM-based pLCI databases only reflect a narrow slice of possible futures, especially since social and economic structures are assumed to stay largely the same. 

5. Narrow economic paradigms: Most IAMs rely on a specific "green growth" economic model with well-functioning, balanced markets. This limits the range of scenarios available. 

6. Techno-optimism: IAMs tend to underrepresent lifestyle and demand-side changes, leaning instead on optimistic assumptions about fast technology rollout. This carries over into the pLCI databases built on them. 

7. Structural modelling choices shape technology adoption: IAMs tend to over-rely on negative emissions technologies and underestimate renewables. Even scenarios that avoid negative emissions technologies rely on their own strong assumptions, like very fast reductions in energy intensity. 

8. Limited transparency and specialised expertise required: IAM documentation is often incomplete and hard to follow, and even well-documented models require a lot of expertise. This complexity makes pLCI results based on IAMs harder to interpret. 

9. Additional assumptions in translating IAM scenarios into pLCI databases: Turning IAM data into detailed LCI data requires extra assumptions and external data. Certain tools might reduce error here, but frequent updates make results version-dependent and hard for non-experts to trace. 

Based on these considerations, the authors offer the following recommendations: 

• Use IAMs with purpose: Before using an IAM-based pLCI database, weigh a few factors: whether the study's timeframe actually needs it, where impacts are likely to occur, and how concentrated those impacts are in a few key processes. Custom, expert-built scenarios remain a good alternative when more control is needed.  

• Focus on a limited set of distinct and diverse future scenarios: Pick a manageable set of scenarios that still capture a wide range of uncertainty. Avoid excluding "unrealistic" scenarios without strong justification – instead, be clear about which conditions drive which results, and keep the foreground and background scenario assumptions consistent with each other. 

• Interpret critically: Treat results as conditional and exploratory, not predictions. Use contribution analysis to trace what's driving your results, and treat unusual differences between scenarios as a possible error worth reporting. Be especially cautious with impacts beyond climate change, where IAMs offer weaker coverage. Above all, treat the model as a reasoning tool – pair results with their assumptions, limitations, and uncertainty. 

• Ensure transparency and reproducibility: Clearly report which IAM, SSP, and climate target were used, along with software versions. Document how scenarios were built, and share scripts/notebooks with package versions for full reproducibility. 

Read the full study.