Comparing expert elicitation and model-based probabilistic technology cost forecasts for the energy transition

We conduct a systematic comparison of technology cost forecasts produced by expert elicitation methods and model-based meth-ods. Our focus is on energy technologies due to their importance for energy and climate policy. We assess the performance of sev-eral forecasting methods by generating probabilistic technology cost forecasts rooted at various years in the past and then com-paring these with observed costs in 2019. We do this for six tech-nologies for which both observed and elicited data are available. The model-based methods use either deployment (Wright's law) or time (Moore's law) to forecast costs. We show that, overall, model-based forecasting methods outperformed elicitation meth-ods. Their 2019 cost forecast ranges contained the observed values much more often than elicitations, and their forecast medians were closer to observed costs. However, all methods underestimated technological progress in almost all technologies, likely as a result of structural change across the energy sector due to widespread policies and social and market forces. We also produce forecasts of 2030 costs using the two types of methods for 10 energy tech-nologies. We find that elicitations generally yield narrower uncer-tainty ranges than model-based methods. Model-based 2030 forecasts are lower for more modular technologies and higher for less modular ones. Future research should focus on further method development and validation to better reflect structural changes in the market and correlations across technologies.

Automated summary

This study systematically compares technology cost forecasts generated by expert elicitation methods and model-based methods, finding that model-based methods generally outperform elicitation methods. However, all methods tend to underestimate technological progress. Elicitations typically yield narrower uncertainty ranges than model-based methods. Model-based forecasts perform differently depending on the modularity of the technologies. Future research should focus on further method development to better reflect market structural changes and correlations across technologies.