Entropy, Input-Output Network, and Financial Crises

Abstract

This paper investigates the role of the input-output network of intermediate goods in recovery from financial crises. In networks, a node has a high value of entropy if a flow from this node stops with nearly equal probabilities at all other nodes. This paper shows that the distribution of entropy across industries in the input-output network is an important factor in explaining the variation in the aftermath of financial crises. The dynamic model with input-output linkages shows that the volatility of the aggregate output decreases when the entropy of the input-output network increases. Specifically, the volatility of the aggregate output in response to productivity shocks decreases if each industry has widely and equally distributed relationship with upstream industries. On the other hand, the volatility of the aggregate output in response to demand shocks decreases if each industry has widely and equally distributed relationship with downstream industries. Empirically, I examine the aftermath of financial crises for 24 advanced countries. The results show that the increase in entropy, especially the downstream entropy of global input-output networks, reduces the negative impact of financial shocks.