Information, Production Networks and Optimal Taxation 

Abstract

This paper studies optimal taxation in an economy with information frictions and a production network across industries. I show that when all industries share the same information structure, production efficiency holds and optimal policy features no taxes on intermediate goods. Deviating from this benchmark, I characterize the optimal policy when information structure is heterogeneous across industries: The government optimally imposes higher revenue taxes on industries during economic downturns (i) it has greater information rigidity, (ii) its downstream industries have smaller information rigidity, and (iii) its input goods are also used by less informed industries. I quantify information heterogeneity across industries with a standard text analysis method. Industries exhibit varying degrees of attention to economic outcomes correlated with their exposure to business cycle shocks. The calibrated model indicates that, in response to the COVID-19 shock, the optimal taxation leads to a welfare increase of 0.7% for the U.S. and 1.23% for China in terms of consumption, compared to the equilibrium that assumes a homogeneous information structure.

Keywords: optimal policy, production networks, informational frictions, business cycles