The Transmission of Shocks in Endogenous Financial Networks: A Structural Approach
The paper uses bank- and instrument-level data on asset holdings and liabilities to identify and estimate a general equilibrium model of trade in financial instruments. Bilateral ties are formed as each bank selects the size and the diversification of its assets and liabilities. Shocks propagate due to the response, rather than the size, of bilateral ties to such shocks. This general equilibrium propagation of shocks reveals a financial network where the strength of a tie is determined by the sensitivity of an instrument’s return to other instruments’ returns. General equilibrium analysis predicts the propagation of real, financial and policy shocks. The network’s shape adjusts endogenously in response to shocks, to either amplify or mitigate partial equilibrium shocks. The network exhibits key theoretical properties: (i) more connected networks lead to less amplification of partial equilibrium shocks, (ii) the influence of a bank’s equity is independent of the size of its holdings; (ii) more risk-averse banks are more diversified, lowering their own volatility but increasing their influence on other banks. The general equilibrium based network model is structurally estimated on disaggregated data for the universe of French banks. We used the estimated network to assess the effects of ECB quantitative easing policy on asset prices, balance-sheets, individual bank distress risk, and networks systemicness.
We would like to thank Andy Atkeson, Celso Brunetti, Giovanni Dell’ariccia, Bernard Dumas, Carlos Iván Cañón Salazar, Jeffrey Frankel, Ralph Koijen, Augustin Landier, Andrei Levchenko, Stavros Panageas, Hashem Pesaran, Hélène Rey, David Thesmar, Aaron Tornell, Pierre-Olivier Weil, as well as the audience of the NBER International Seminar on Macroeconomics, the North American Meeting of the Econometric Society at UC Davis, the International Association for Applied Econometrics annual conference in Montreal, the Tokyo Macroeconomics workshop, CREST, the Tsinghua Workshop in International Finance, Banque de France, the Paris Financial Management Conference, UCLA, UC Riverside, IMF, USC/INET, the SCOR conference at the Paris School of Economics, the Bank of England for fruitful comments on previous versions of the paper. The standard disclaimers apply. We thank CEPREMAP for financial support, and the Banque de France for providing us access to bank-level data through the Open Data Room initiative. The views expressed herein are those of the authors and do not necessarily reflect the views of the National Bureau of Economic Research.