Communication without Consensus: Forward Guidance, Central Bank Projections, and Private Sector Expectations

Abstract

Monetary policy committees use both collective and decentralized forms of communication. At the zero lower bound, the Federal Open Market Committee (FOMC) used forward guidance (collective) and the Summary of Economic Projections (decentralized) to influence expectations. The projections have been criticized for revealing divisions among FOMC participants and undermining the credibility of forward guidance. Comparing the projections and private forecasts, I show that in the calendar-based forward guidance period, large shares of both private forecasters and FOMC participants projected "liftoff" before the guidance indicated, and FOMC participants' disagreement was roughly as large as private forecasters'. But by late 2012, private forecasters became increasingly more pessimistic than the FOMC about the rate at which unemployment would decline, and were less likely to believe the threshold-based forward guidance introduced in December 2012. This can be explained by differences in private and FOMC models and beliefs about model parameters, particularly related to the natural rate hypothesis. I document systematic differences between the expectations of private forecasters who report using versus not using the natural rate of unemployment to forecast, and show that forecasts from natural rate users are consistent with the accelerationist hypothesis. Private estimates of the natural rate of unemployment increased more than FOMC estimates from 2009 to 2012, when forecasters who underestimated inflation during the "missing disinflation" revised their estimates upwards. FOMC transcripts reveal that FOMC participants did not revise their natural rate estimates up as much because their staffs' research found little evidence for the skill mismatch narrative that dominated the financial press.

Keywords: Central Bank Communication, Federal Reserve, Summary of Economic Projections, Forward Guidance, Expectations, Natural Rate of Unemployment, Missing Disinflation, Survey Data

JEL Classification: E52, E58, E43, D83, D84