Close this search box.

An input–output sensitivity analysis of climate–economy integrated assessment models




Key takeaways

  • For multivariate models, such as those used by the NGFS, there are key differences in the underlying assumptions in IAM models when projecting key climate and economic variables used by financial institutions, such as the amount of carbon sequestration and the capital costs of renewable energy.


This paper analyses the underlying interactions between input and output variables of three process–based climate–economy integrated assessment models according to a standardised sub- set of social and economic input variables, and how they contribute to variability in scenario outputs.

This paper compares two key climate scenario output variables of transition risk, including carbon sequestration and technology costs, modelled according to the three IAMs used by the Network for Greening the Financial System: GCAM, REMIND–MAgPIE, and MESSAGEix–GLOBIOM. Evidence shows variability in the outputs of NGFS under the same policy ambition, the same temperature target, and the same social and economic input data.

Using a cumulative distribution sensitivity analysis (PAWN) to compare IAM input data to output trends, this paper observes the sensitivity of model input variables to driving differ- ences in scenario outputs according to underlying interactions and assumptions between social and economic variables.

Findings demonstrate that differences in climate scenario trajectories under the same narrative pathways can be distinguished according to sensitivities of a standard subset of input variables from the SSPs. Understanding the sensitivity of climate scenario trajectories to a standard set of social and economic input variables can help financial institutions in the selection, categorisation, and application of different IAMs, scenarios, and pathways for financial analysis.


Header image by Nina Mercado via Unsplash