Equilibrium Heterogeneous-Agent Models as Measurement Tools: some Monte Carlo Evidence

We study how the heterogeneity of information impacts the efficiency of the business cycle and the design of optimal fiscal and monetary policy. We do so within a model that features a standard Dixit-Stiglitz demand structure, introduces dispersed private information about the underlying aggregate productivity shock, and allows this information to be imperfectly aggregated through certain prices and macroeconomic indicators. Our key findings are the following: (i) When information is exogenous to the agents' actions, the response of the economy to either fundamentals or noise is efficient along the flexible-price equilibrium. (ii) The endogeneity of learning renders the business cycle inefficient: there is too little learning and too much noise in the business cycle. (iii) Both state-contingent taxes and monetary policy can boost learning over the business cycle. (iv) Typically, this implies that the optimal tax is countercyclical, while the optimal monetary policy is less accommodative than what is consistent with replicating the flexible-price equilibrium. (v) Even if monetary policy were to replicate the flexible-price equilibrium, this would not mean targeting price stability. Rather, the optimal monetary policy has the nominal interest rate increase, and the price level fall, in response to a positive innovation in productivity.