Feb 3, 2020
Following the 2008 financial crisis, significant attention was paid to systemic risk within heavily interconnected financial networks, but little attention was paid to systemic risks in the economy at large. The authors of this report address that gap, detailing efforts to gather large data sets on observed interfirm linkages, procedures to infer missing data on network linkages, and an approach to calibrate likely connections to financial data.
In the years following the 2008 financial crisis, significant attention was paid to systemic risk within heavily interconnected financial networks. The academic discussions on interbank network structure, market stability, and contagion gave rise to a policy debate about whether major banks had become both too big and too interconnected to fail. However, despite the focus on systemic risk—the risk of market collapse resulting from firm-level risks—within the financial sector, little attention was paid to systemic risks in the economy at large. The authors of this report address that gap in research. To begin to measure the potential magnitude of systemic risk in the broad economy, the authors estimated firm-to-firm connections across sectors of the U.S. economy. Using network analysis on observed firm-level networks to elucidate heavily interconnected firms and areas of centrality (i.e., firms of significant network importance), statistical inference, and network calibration, the authors provide a new approach to modeling the economy at the firm level that expands on the traditional sector-level input-output modeling by estimating firm-level input-output flows. The result allows one to use traditional input-output modeling to estimate the size of potential idiosyncratic shocks and to use economically weighted measures of centrality to reveal systemically important firms. The approach is a contribution to the growing literature on the microfoundations of economic risk, with the potential for use across a wide range of applications from financial stability to natural disasters.
The Model of Firm-Level Linkages
Observed Interfirm Production Network
Estimating Production Networks Through Inference
Contagion Across Firm Networks
Conclusions and Implications