Productivity Shocks, Unemployment Persistence, and the Adjustment of Real Wages

This paper applies a set of unit root and cointegration tests with non-linear error-correction mechanisms to a subset of the OECD countries to investigate the empirical conclusions of some of the labor market models in the literature. I generally find that the unemployment rate, productivity, and real wages have a unit root even if one controls for threshold effects. This finding justifies the use of a cointegration approach to assess the existence of a long-run equilibrium among the variables of interest. For roughly half of the OECD countries in the sample, the unemployment rate, real wages, and productivity trend together over time. For four countries (i.e, Germany, Japan, Sweden, and the US) the adjustment to the long run relationship appears mostly asymmetric. Also, an impulse response function analysis suggests that real wages and productivity adjust faster to the long-run equilibrium, while shocks to unemployment take longer to extinguish. Also, according to the sign of the shocks, the unemployment rates respond differently. These findings suggest that a proper analysis of the behavior of productivity, real wages, and unemployment should consider non-linear adjustment mechanisms to long-run equilibrium since a liner approach would be biased