(with Jonathan Brogaard)

Credit Default Swap (CDS) spreads exhibit network effects due to firms' default interdependence. This paper employs Graph Neural Networks (GNNs) to predict CDS spreads by modeling firms as nodes and idiosyncratic volatility spillover measures as directed edges. GNNs capture inter-firm network dynamics, improving prediction accuracy by over 50% compared to traditional models without edge features. We enhance the GNN with node- and edge-attention layers, identifying key nodes (e.g., manufacturing and intermediary firms) and edges (e.g., connections between intermediary, retail trade, or information firms and other firms) as critical to CDS spread prediction.

(with Tim Johnson)

This paper examines the microfoundation of macroeconomic tail risk, defined as the frequency of extremely negative GDP downturns relative to what is predicted by a normal distribution. We demonstrate that firm-level productivity shocks, even when normally distributed, can be amplified into large macroeconomic tail events through a dynamic production-based input-output network, challenging the traditional view that only sufficiently large firm-level shocks impact the aggregate market. The evolving network structure, driven by technological shocks, hinders risk diversification across firms, delays the mean reversion of aggregate volatility, and causes volatility to cluster from local to global scales over time. In bad times, when idiosyncratic risks are high and firms treat all inputs as complementary, these risk spillovers can accumulate, resulting in severe GDP downturns.

(with Mahyar Kargar, Sébastien Plante, John Shim, and Karamfil Todorov)

(Draft to be updated)