Abstract
Large partisan gaps in inflation expectations are well-documented, but whether they reflect genuine beliefs or cheap talk remains an open question. We address this using a two-wave experiment conducted around the 2024 U.S. presidential election, in which we randomly assign monetary incentives for forecast accuracy. Without incentives, partisan gaps are large and reverse with political control of the presidency. Introducing incentives compresses the Democrat–Republican distributional gap by roughly 70 percent, tightens within-party dispersion, produces better-anchored expectations, and sharply attenuates partisan differences in belief updating in response to identical FOMC signals. Residual polarization is concentrated among pessimists and non-copartisans. Overall, most partisan disagreement in inflation expectations reflects cheap talk; the remaining gap is consistent with sentiment-driven motivated reasoning.