Skewed Fluctuations and Propagation Through Production Networks

Skewness in macroeconomic time series may arise exogenously if shocks are asymmetrically distributed, or endogenously, as shocks propagate through production networks. These two possibilities are often considered in isolation by previous theoretical work. To evaluate the relative importance of these channels, we nest all possible sources of skewness in an empirical model where output has a network, a common, and an idiosyncratic component. In this model, skewness can arise not only from the three components, but also from coskewness due to their higher order covariation. An analysis of output growth in 43 U.S. sectors shows that coskewness, especially between the network and the common component, is a key source of asymmetry in the data and constitutes a connectivity channel not previously explored in the literature. To help interpret our results, we construct and estimate a micro-founded multi-sector general equilibrium model and show that it can generate skewness and coskewness consistent with the data.