Structural breaks and relative price convergence among US cities

Abstract

This paper examines whether inclusion of structural breaks helps resolve the puzzling result of excessively slow speed of convergence in relative prices across US cities while using long time series data on CPI. With an endogenously determined single break in 1985 in annual CPI data for 17 major US cities between 1918 and 2010, the panel unit root test results provide overwhelming evidence of convergence of relative prices across cities, which is consistent with the existing literature. Most importantly, this study finds that the speed of convergence with structural break is faster than that reported by previous panel studies with no structural break that use similar long time series data. Furthermore, correcting for small-sample bias (the so-called “Nickell Bias”) and time aggregation bias generates a half-life of 3.9years, which is 64% shorter than the half-life estimate with no structural break and no bias correction. The breakpoint in 1985 coincides with the beginning of the period of Great Moderation during which inflation, along with other macro variables, has become more stable and its impact on relative prices across US cities seems to have waned.