Understanding the European energy crisis through structural causal models

Area of Science:

• Energy Economics
• Causal Inference
• Statistical Modeling

Background:

• European energy markets faced unprecedented volatility following Russia's invasion of Ukraine.
• France experienced disproportionately high electricity price increases despite a low natural gas share in its energy mix.
• Traditional correlation studies yielded paradoxical results in understanding these market dynamics.

Purpose of the Study:

• To apply causal statistical methods to analyze French and Spanish electricity markets.
• To identify key factors influencing electricity prices and net exports.
• To demonstrate the superiority of causal inference over simple correlation for complex energy market analysis.

Main Methods:

• Development and application of causal graphs to model market interdependencies.
• Implementation of a linear structural causal model.
• Utilizing non-linear tree-based machine learning with Shapley Flow values for detailed analysis.

Main Results:

• Causal models successfully resolved paradoxical findings from correlation analyses.
• Identified the interplay between natural gas prices and nuclear power plant unavailability as critical factors.
• Quantified indirect causal effects and enabled 'what-if' scenario analysis.

Conclusions:

• Causal inference is crucial for accurately understanding energy market dynamics and price fluctuations.
• Nuclear power plant availability significantly impacts electricity prices and import dependency, particularly in France.
• The study provides a robust framework for analyzing energy crises and informing policy decisions.