When a central bank buys lower inflation with unemployment, does it get fewer price increases or more price cuts? We answer this for the United States by decomposing the Phillips multiplier — the cumulative inflation response to a monetary shock that raises unemployment by one percentage point — into its two directional margins, using 245 disaggregated PCE sub-indices and identified monetary shocks. Before 1990, a one-point rise in the cumulative unemployment gap raises deflationary pressure by 0.4pp: the trade-off runs through price cuts. After 1990, excluding the zero lower bound, weak instruments identify neither the multiplier nor either margin.
