Model confirmation in climate economics

Area of Science:

• Climate Science
• Environmental Economics
• Integrated Assessment Modeling

Background:

• Benefit-cost integrated assessment models (BC-IAMs) couple climate and economic systems to evaluate greenhouse gas abatement strategies.
• These models provide qualitative insights but are increasingly used for real-world policy selection, necessitating confidence in their quantitative outputs.
• Confidence in climate models stems from empirical testing and confirmed physical principles, unlike the often untested economic components of BC-IAMs.

Purpose of the Study:

• To assess the empirical validity of economic components within BC-IAMs by applying methods similar to climate science model confirmation.
• To investigate the potential benefits of model confirmation exercises for enhancing the reliability of BC-IAMs for policy applications.

Main Methods:

• Conducted a long-run hindcasting experiment using a leading BC-IAM.
• Focused on validating the model's representation of long-run economic growth, a critical component.

Main Results:

• The hindcasting experiment demonstrated that the BC-IAM's model of long-run economic growth exhibited questionable predictive power over the 20th century.
• This highlights a potential weakness in the empirical validity of key economic assumptions within BC-IAMs.

Conclusions:

• Model confirmation exercises, particularly hindcasting, are valuable for building confidence in the economic components of BC-IAMs.
• Refinement of economic components in BC-IAMs is necessary to improve their reliability for informing climate policy selection.