Identifying the Phillips Curve from Shifts in Demand

Abstract

Despite decades of research, the parameters of the Phillips curve (old or new) remain uncertain, because their estimation is fraught with endogeneity issues: confounding from supply shocks, unobserved inflation expectations and an unobserved output gap. In this work, we use sequences of past aggregate demand shocks (notably monetary shocks) as instrumental variables to address these endogeneity issues and obtain consistent Phillips curve estimates. Our approach can be interpreted as a regression in impulse response space, i.e., to a Phillips curve regression where the actual variables are replaced by their impulse responses to aggregate demand shocks. Once instrumented with monetary shocks, the estimated slope of the Phillips curve is twice as large as with standard methods.

Keywords: Phillips curve, instrumental variables, impulse response regression

JEL Classification: C14, C32, E32, E52