Financial network analysis : causal inference approach
PhD: Università della Svizzera italiana
English
This doctoral dissertation contributes to the field of financial network analysis by exploring contagion risk estimation through a causal inference framework, specifically within the Forex market. By integrating causal inference, network contagion analysis, and machine learning techniques, this research introduces innovative metrics and frameworks that surpass traditional risk management methodologies. It offers a nuanced understanding of contagion dynamics among individual currencies, detailing the pathways of contagion and evaluating the systemic risk impact of major events, such as the COVID-19 pandemic. Building on traditional methods, this research introduces a distinctive measure of contagion in the Forex market through a causal network approach. This stands in contrast to the widely used instrumental variable and Granger causality methods. Employing causal inference theory – already validated in fields such as genetics, medicine, and climate science, to distinguish causality from correlation in observational data – contagion pathways are identified. The resulting Network Contagion (NECO) measure evaluates both the market as a whole and individual currencies in terms of diversification and exposure to systemic risk. Further advancing the domain, the dissertation provides a unique framework for Value at Risk. This framework takes into account the intricate dynamics of contagion, improving its dependability in volatile market situations. This new method specifically addresses long-standing challenges related to non-normality, non-stationarity, and unforeseen market shocks. When compared with conventional VaR measures, the superior accuracy of this approach is evident.
-
Collections
-
-
Language
-
-
Classification
-
Economics
-
License
-
License undefined
-
Open access status
-
green
-
Identifiers
-
-
Persistent URL
-
https://n2t.net/ark:/12658/srd1326651