Nonlinear vs. nonstationary of hysteresis in unemployment: evidence from OECD economies

Abstract

This study examines the lower and higher boundaries for the threshold value to be considered an indicator of unemployment in a specific country. Specially, the objective is to conduct the critical moment of hysteresis effects happening in unemployment rate using a group of 16 OECD countries. The methodological strategy applies a developed tool of threshold tests involving unit root against stationary but nonlinear alternative by Caner and Hansen (2001). A significant contribution of this study is identifying a trigger point from the nonstationary of time series process for the first time in the literature. Our empirical results finds strong evidence of the existence of nonlinear stationary in Australia, Canada, Finland, France, Germany, Ireland, Japan, Netherlands and the USA when the threshold effect holds. The hysteresis hypothesis is further confirmed by the fact that the unemployment rate exceeds the boundaries of the band, for Australia, Finland, France, Germany, Japan and the USA when the threshold unit root test of Caner and Hansen is rigorously implemented.

Notes

¹  denotes the Wald statistic of the null of θ₁ equal to θ₂ for fixed λ and is the residual variances from OLS estimation of the null linear model. Then since W_T (λ) is a decreasing function of , thus the Wald statistic for H ₀ is often called the ‘Sup–Wald’ statistic.