Noisy Global Value Chains

We study international propagation of both fundamental and non-fundamental shocks in a global production network model with information frictions. Producers in a sector do not perfectly observe other country-sector fundamentals, and their production decisions depend on their beliefs about worldwide exogenous states as well as other producers’ behavior. In this environment, “noise” shocks – errors in the public signals about fundamentals – propagate internationally and generate aggregate fluctuations. Using a novel panel dataset containing the frequencies of country-industry-specific economic news reports by 11 leading newspapers in the G7 plus Spain, we show that greater news coverage is associated with both smaller GDP forecast errors, and less disagreement among forecasters. We use these empirical regularities to discipline the parameters governing the severity of information frictions. We find that noise shocks are a quantitatively important source of international fluctuations. Noise shocks propagate relatively more powerfully to the more distant parts of the network, while TFP shocks propagate less powerfully to the more distant sectors in the presence of informational frictions.