We solve a rational expectations model of price formation with nominal rigidity and information frictions analytically to study how dispersed information impacts inflation and inflation expectations, without imposing strong structural assumptions on the aggregate marginal cost. The closed-form solution implies an ARMA(1,1) nowcast error and an ARMA(1,2) one-step-ahead forecast error of inflation, both supported by the Survey of Professional Forecasters data. We derive a Phillips curve that links inflation, average inflation forecast (i.e., a first-order expectation), and the net effect of higher-order expectations (HOEs). We quantify the essential role of dispersed information in generating predictable average nowcast and forecast errors and estimate the net effect of HOEs on the Phillips curve.