Abstract
During the COVID period, the vacancy-to-unemployment ratio (V/U) emerged as a more accurate measure of labor market tightness than the unemployment rate (U). However, some argue that this claim may be over-fitting the COVID-19 pandemic episode. This paper addresses this critique by using pre-pandemic time-series data and exploits better identification from Metropolitan Statistical Area (MSA)-level panel data. I construct state-space models and apply the Kalman filter to MSA-level panel data, thereby jointly estimating a non-linear Phillips curve and time-varying natural rates of U and V/U. Time series data alone cannot distinguish between the two measures on pre-pandemic data, but panel data indicates that V/U is superior. These findings show that V/U was a better measure of economic slack even before the pandemic, and suggest a greater role for it in economic forecasting and monetary policy.