Empirical tests of the New Keynesian Phillips Curve have provided results often inconsistent with microeconomic evidence. To overcome the pitfalls of standard estimations on aggregate data, a Full Information Partial Equilibrium approach is developed to exploit sectoral level data. A model featuring sectoral NKPCs subject to a rich set of shocks is constructed. Necessary and sufficient conditions on the structural parameters are provided to allow sectoral idiosyncratic components to be linearly extracted. Estimation biases are corrected using the model's restrictions on the partial equilibrium propagation of idiosyncratic shocks. An application to the US, Japan and the UK rejects the purely forward looking, labor cost-based NKPC.