Financial Crises and Government Interventions

We develop a model of financial crises with both a financial amplification mechanism, via frictional intermediation, and a role for sentiment, via time-varying beliefs about an illiquidity state. We confront the model with data on credit spreads, equity prices, credit, and output across the financial crisis cycle. In particular, we ask the model to match data on the frothy pre-crisis behavior of asset markets and credit, the sharp transition to a crisis where asset values fall, disintermediation occurs and output falls, and the post-crisis period characterized by a slow recovery in output. Our model with the frictional intermediation mechanism and fluctuations in beliefs provides a parsimonious account of the entire crisis cycle. The model with only the frictional intermediation mechanism misses the frothy pre-crisis behavior; fluctuations in beliefs resolve this problem. On the other hand, modeling the belief variation via either a Bayesian or diagnostic model match the broad patterns, with each missing some targets to different extents. We also show that a lean-against-the-wind policy has a quantitatively similar impact in both versions of the belief model, indicating that policy need not “get into the minds” of investors and condition on the true belief process.

This paper studies the equilibrium effect of public liquidity on financial crises. Banks borrow from households via insured deposits and partially runnable debt, and suffer endogenous funding withdrawals from households in crises. Holding public liquidity alleviates banks’ liquidity problems. In equilibrium, a larger public liquidity supply reduces crisis severity and expands bank lending, but it crowds bank deposits and increases bank vulnerability to real shocks. The model quantitatively explains 40% of Treasury liquidity premium variations. Counterfactual analyses reveal that QE1 significantly improves output, 20 times larger than QE3. However, QE policies raise bank fragility against non-financial shocks such as COVID-19.

A salient trend in crisis intervention has emerged in recent decades: government and central banks have offered funding directly to nonfinancial firms, bypassing banks and other credit intermediaries. We analyse the long-term consequences of such policies by focussing on firm quality dynamics. In a laissez-faire economy, firms with high productivity are more likely to survive crises than those with low productivity. The government funding support saves more firms but cannot be customized based on firm productivity, dampening the cleansing effect of crises. The policy distortion is self-perpetuating: a downward bias in the firm quality distribution necessitates larger interventions in future crises. Our mechanism is quantitatively important: we show that if policymakers ignore such distortionary effects on firm quality dynamics, the resultant credit intervention would almost double the optimal amount.

Deep learning has been shown to be an effective method for solving partial differential equations (PDEs) by embedding the PDE residual into the neural network loss function. In this paper, we design a methodology that utilizes deep learning to simultaneously solve and estimate canonical continuous-time general equilibrium models in financial economics, including (1) industrial dynamics of firms and (2) macroeconomic models with financial frictions. Through these applications, we illustrate the advantages of our method: generality, simultaneous solution and estimation, leveraging the state-of-art machine-learning techniques, and handling large state space.