Mind the Stance Gap: State-Dependent Filtering in the Federal Reserve Communications

The Federal Reserve's public statements do not directly reflect the FOMC's internal deliberations. The committee filters the message before releasing it, and the filter behaves differently when the policy rate is constrained. We compare meeting-level monetary policy stance indices built from FOMC transcripts and post-meeting statements across 128 meetings between 2004 and 2019, and measure the gap between them. In unconstrained periods, the public message leans modestly more hawkish than the deliberation behind it, consistent with the Fed protecting its credibility while conventional policy still has room to operate. At the effective lower bound, the sign flips, and statements turn substantially more dovish than the internal discussion. The reversal is not an artifact of forward guidance or quantitative easing language, and communication takes on a more direct role when the rate instrument is constrained. Using staff economic projections as an intermediate benchmark, we show that the regime-dependent shift comes from deliberation, not from public communication. We also show that measured stance gap predicts later revisions in the Fed's statements, since the FOMC initially communicates with a bias but the statement gradually catches up to the internal view. Statement-based text measures are not transparent proxies for internal policy preferences, particularly across policy regimes.