We compare the causal effects of forward guidance communication about future interest rates on households' expectations of inflation, mortgage rates, and unemployment to the effects of communication about future inflation in a randomized controlled trial using more than 25,000 U.S. individuals in the Nielsen Homescan panel. We elicit individuals' expectations and then provide 22 different forms of information regarding past, current and/or future inflation and interest rates. Information treatments about current and next year's interest rates have a strong effect on household expectations but treatments beyond one year do not have any additional impact on forecasts. Exogenous variation in inflation expectations transmits into other expectations. The richness of our survey allows us to better understand how individuals form expectations about macroeconomic variables jointly and the non-response to long-run forward guidance is consistent with models in which agents have constrained capacity to collect and process information.