Does Unemployment Hysteresis Falsify the Natural Rate Hypothesis? A Meta-Regression Analysis

A quantitative survey of 24 studies containing 99 national estimates of unemployment persistence reinstates unemployment hysteresis as a viable falsifying hypothesis to the natural rate hypothesis. Empirical evidence to the contrary may be attributed to small-sample, misspecification and publication biases. Larger estimates of unemployment persistence are produced by models that use more information (t= 9.03; P< 0.0001) and are better specified. A theme of bias and misspecification among studies that are more supportive of natural rate hypothesis emerges in several independent ways. The nonstationarity of the unemployment rate is confirmed both by the observed rate of convergence of persistence estimates across the empirical literature and by the point towards which they converge. The natural rate hypothesis may be regarded as 'falsified' should we choose to do so