Systemic risk in banking, fire sales, and macroeconomic disasters
构建了一个银行系统动态计算网络模型，其中甩卖放大金融冲击，发现银行资产增长率分布存在左尾厚尾，对应罕见经济灾难，且初始冲击分布不能完全预测经济活动。
We develop a dynamic computational network model of the banking system where fire sales provide the amplification mechanism of financial shocks. Each period a finite number of banks offers a large, but finite, number of loans to households. Banks with excess liquidity also offer loans to other banks with insufficient liquidity. Thus, each period an interbank loan market is endogenously formed. Bank assets are hit by idiosyncratic shocks drawn from a thin tailed distribution. The uneven distribution of shocks across banks implies that each period there are banks that become insolvent. If insolvent banks happen also to be heavily indebted to other banks, their liquidation can trigger other bank failures. We find that the distribution across time of the growth rate of banking assets has a ‘fat left tail’ that corresponds to rare economic disasters. We also find that the distribution of initial shocks is not a perfect predictor of economic activity; that is some of the uncertainty is endogenous and related to the structure of the interbank network.