Anticipated Financial Contagion

We revisit the seminal model of financial contagion of Allen and Gale (2000) by allowing aggregate liquidity shocks to occur with positive probability. We study how an ex-post shock’s size and probability affect ex-ante portfolio choices and risk sharing across states and over time. We characterize a numerically approximate symmetric Nash equilibrium in the non-cooperative game between two regional banks. We describe parameter regions where contagion does and does not occur for positive probability of the aggregate liquidity shock. Our solution fully characterizes banks’ ex-ante optimal portfolio choices. Additionally, we present novel benchmarks where optimal risk sharing with observable types involves (i) full default after a large but sufficiently unlikely aggregate liquidity shock; (ii) holding excess liquidity when the shock is relatively likely; (iii) partial liquidation of investment after a small and unlikely shock; and (iv) both excess liquidity and partial liquidation for shocks of intermediate size and probability. We obtain a number of additional benchmark results numerically to show the robustness of our numerical approach and highlight its usefulness for the literature.