This paper investigates the nature of regional unemployment in Italy. Taking as a starting point the evidence of growing disparities in the 1977-2003 years, we assess whether regional unemployment rates are diverging from each other in a non-stationary way, i.e. as a result of pure hysteresis. Relying both on univariate and panel unit-root tests, we show that the finding of a unit root in the Italian regional unemployment rates, which previous studies provided evidence of, is largely dependent on the use of low power tests. Exploiting the greater power of panel unit root tests allows us to confidently reject the unit root and, thus, the pure hysteresis hypothesis. The implication of this result is that, however persistent, shocks to regional unemployment will be temporary, in the sense that unemployment will return to its natural rate or NAIRU in the long-run. We, then, proceed to estimate the NAIRU for each of the 20 Italian regions. Our estimates of the regional NAIRUs turn out to be fairly precise, at least if compared to similar studies in the literature, and allow us to draw two interesting conclusions. Firstly, the hypothesis of a constant NAIRU between 1977 and 2003 is supported by the data for all of the Italian regions, with the possible exception of Abruzzo. Secondly, we find that there is a significant degree of heterogeneity among the regional NAIRUs. Thus, long-term regional unemployment disparities do seem to reflect structural or equilibrium unemployment differences across regions, as implied by the results of the unit root tests. This suggests that economic policy intervention aimed at reducing long-term regional unemployment differentials in Italy should, as much as possible, be region-specific in character. National policies, homogenous across regions, are likely to have diverse effects on the regional NAIRUs.
