Are business cycles all alike? Evidence from long-run international data

Most recent empirical analyses of aggregate economic fluctuations rely on the implicit assumption that business cycles are all alike. This paper investigates this assumption for the long-run international data set of Backus and Kehoe and presents the results of statistical tests of the 'representative cycle' hypothesis. Findings show that there is substantial heterogeneity across individual cycles and phases in terms of duration, amplitude, and co-movements between variables, and that such heterogeneity is generally statistically significant. These results underline, at an empirical level, the limitations of measuring business cycles in terms of sample second moments, and, at a theoretical level, the difficulties of developing theories providing a unified explanation of the business cycle.