PhD-candidate: Stan Koobs
Start: Fall 2022
Modern economies are deeply interconnected. A financial shock in one country can spread to others through channels that researchers cannot directly observe, and the pathways that matter in calm times often differ from those that activate during a crisis. My research develops statistical methods to uncover these hidden networks of influence from data, and to capture how they shift between tranquil and turbulent periods.
The challenge is that the connections between economic units are not given in advance, and the number of potential links grows rapidly with the number of units involved. I address this by combining recent advances in high-dimensional regression with machine learning techniques, producing a framework that identifies the relevant connections, allows them to differ across market conditions, and delivers reliable uncertainty statements about the effects of interest, all with formal statistical guarantees.
Applied to two decades of sovereign credit risk data covering economies across multiple continents, the method reveals connections that geographical proximity alone cannot explain. Some of the strongest links arise between countries far apart on the map, reflecting deep trade relationships and strategic partnerships. These connections are often directional, meaning that one country may transmit risk to another without being equally exposed in return, and they intensify during periods of financial stress.
Understanding how risk travels between economies matters well beyond academic research. Investors rely on such information to construct more resilient portfolios, regulators to detect vulnerabilities before they spread, and policymakers to anticipate how disturbances elsewhere may reach their own economy. As globalisation forges connections that conventional indicators fail to capture, methods of this kind play an increasingly important role in safeguarding financial stability.
