🤖 AI Summary
This study investigates how firm-level idiosyncratic shocks propagate through production networks to generate aggregate macroeconomic fluctuations and tail risks. To this end, the authors develop a dynamic production network model incorporating an overlapping adjustment mechanism and introduce the concept of “productivity waves” to capture the superposition and interference of shocks across time as they diffuse through the network. By integrating dynamic network theory, spectral analysis, and stochastic process methods, the paper identifies an endogenous source of macroeconomic volatility: the dominant transient eigenvalue of the network—not the tail of the degree distribution—plays the pivotal role. The findings suggest that once the temporal averaging effect of shocks is accounted for, the contribution of granular shocks to actual macroeconomic volatility may be substantially smaller than predicted by static equilibrium models, thereby revising conventional wisdom on the origins of aggregate fluctuations.
📝 Abstract
We study how idiosyncratic firm-level shocks generate aggregate volatility and tail risk when they propagate through a production network under overlapping adjustment: new productivity draws arrive before the economy reaches the static equilibrium associated with earlier draws. Each innovation generates a `productivity wave'that mixes and dissipates over time as it travels through the production network. Macroeconomic fluctuations emerge from the interference between these waves of different vintages. The interference between these waves is governed by the dominant transient eigenvalue of the production network, and therefore so is the macroeconomic fluctuations they generate. In such a dynamic regime, the tail of the degree distribution is a markedly weaker determinant of macro fluctuations than in the fully adjusted static benchmark. And the macroeconomic significance of the degree-heterogeneity of production networks cannot be known without knowing the rate at which the economy converges to equilibrium or equivalently the spectral properties of the production network. More concretely, once we permit the time-averaging of shocks, granular shocks may account for only a small fraction of the empirically observed aggregate volatility.