Heterogeneous beliefs and the Phillips curve

We establish a set of novel empirical facts concerning cross-section distributions of inflation expectations reported in surveys. Almost all the variation in expectations about their mean may be summarized via three factors we call disagreement, skew, and shape. We adopt a functional principal component regression approach to estimating forward-looking models of inflation that exploits the heterogeneity present in individual-level data. By using survey information more effectively, our approach reveals an enhanced role for expectations in inflation dynamics that is robust to lagged inflation, trend inflation, and supply factors. Our findings hold in similar form across two major economies.