Abstract
The Phillips curve plays a central role in the macroeconomics literature. However, there is little consensus on the forcing variable that drives inflation in the model, i.e., on the appropriate measure of “slack” in the economy. In this work, we systematically assess the ability of variables commonly used in the literature to (i) predict and (ii) explain inflation fluctuations over time and across U.S. metropolitan areas. In particular, we exploit a newly constructed panel dataset with job openings and vacancy filling cost proxies covering 1982-2022. We find that the vacancy-unemployment (V/U) ratio and vacancy filling cost proxies outperform other slack measures, in particular the unemployment rate. Beveridge curve shifts—notably, movements in matching efficiency—are responsible for the superior performance of the V/U ratio over unemployment.
