Title: Risk in Network Economies

Abstract: Economic models with input-output networks assume that firm or sector (unit) growth is driven by a weighted sum of trade partners' growth and an independently-drawn idiosyncratic shock. I show that the idiosyncratic risk assumption in a broad class of network models implicitly generates restrictions on the network weights which are unrealistic. When allowing for correlated shocks, units are exposed to an additional risk term which captures the ability to substitute away from supply and demand shocks propagating through the network. I provide empirical evidence that changes in substitutability between trade partners are inversely related to changes in the panel of realized industry variance. Moreover, I find that supply-side (demand-side) substitutability is closely related to technological (product) dispersion of a unit's suppliers (customers). To synthesize these results, I propose a production-based asset pricing model in which supply chain substitutability is a function of dispersion in product/technology space and correlation in supply and demand shocks is driven by shared customers and suppliers between firms. The model predicts that assets which are positively exposed to average propagation of upstream and downstream shocks are useful hedges and thus earn lower average risk premia. Consistently, I find that estimated upstream (downstream) propagation factors earn return spreads of -11.4% (-4.2%) and are negatively associated with aggregate consumption, output, and dividend growth.