Reconsidering the natural rate hypothesis

The natural rate hypothesis states that there exists an unemployment rate at which inflation is stable, and that this unemployment rate is independent of aggregate demand shocks. The hysteresis hypothesis, in contrast, states that the long run unemployment rate can be affected by aggregate demand shocks. While policy makers have warned of the risk of hysteresis since the 2008 financial crash, hysteresis effects are not incorporated into the macroeconometric models used by policy making institutions. This paper presents Bayesian estimates of hysteresis effects using unobserved components models of the type used by the European Commission and OECD. We demonstrate that the posterior probability of the natural rate hypothesis holding in Germany, France, and the UK is very low, lending empirical support to the hysteresis hypothesis. We suggest that the models used by the European Commission and OECD should be amended to reflect policy makers' views on hysteresis.