An Introduction to Causal Inference: Concepts and an Application in Financial Extremes

Abstract

This thesis introduces the foundations of causal inference, covering counterfactuals, identification assumptions, and causal graphs to distinguish causation from association. It then extends causal analysis to extreme events, combining Granger causality with tail‑based methods that remain robust to hidden confounding. Finally, these techniques are applied to real financial data—cryptocurrencies and S&P 500 companies—revealing causal relationships that emerge specifically during extreme market movements.