Inferring the Aggregate: Information Frictions and Fiscal Policy
Working paper 2026
Countercyclical fiscal transfers increase households’ resources but may also signal weak economic conditions. To quantify this information channel, I develop a heterogeneous-agent model with dispersed information and a solution method combining Bayesian filtering with approximate aggregation. Households do not observe aggregate conditions directly, but rationally infer them from their own cash flows, which reflect aggregate and idiosyncratic movements. The information conveyed by transfers largely offsets their stimulus effects, reducing the initial aggregate consumption response from 38 cents per dollar under full information to 8 cents. Households with high cash on hand are most sensitive to the information channel and even reduce consumption upon receiving a transfer. Evidence from a randomized information experiment supports the mechanism: information about a fiscal stimulus lowers GDP forecasts and reduces spending intentions, with the decline in the latter strongest among high-income households. Targeting transfers toward low-income households mitigates the information channel, and the gains from targeting are larger than under full information.