Unemployment, Hysteresis and Transition

Abstract

Abstract:We quantify the degree of persistence in the unemployment rates of transition countries using a variety of methods benchmarked against the EU. In doing so, we will also characterize the dynamic behavior of unemployment in the CEECs during the past decade. In part of the paper, we will work with the concept of linear "Hysteresis" as described by the presence of unit roots in unemployment as in most empirical research on this area. Given that this is potentially a rather narrow definition, we will also take into account the existence of structural breaks and non-linear dynamics in unemployment in order to allow for a richer set of dynamics. Finally, we examine whether CEECs' unemployment presents features of multiple equilibria, that is, if it remains locked into a new level whenever a structural change occurs. Our findings show that, in general, we can reject the unit root hypothesis after controlling for structural changes and business cycle effects, but we can observe the presence of a high and low unemployment equilibria. The speed of adjustment is faster for CEECs than the EU, although CEECs tend to move more frequently between equilibria.

A further investigation of unemployment persistence in European transition economies

“…Our findings on a more recent sample period suggest that taking into account structural breaks is not enough to confirm stationarity. The different results may stem from the early transition period being included in samples of León-Ledesma and McAdam (2004) and Camarero et al (2005) but not in our more recent sample.…”

“…a substantial proportion of total unemployment in the early transition phase was structural in character (Garibaldi and Brixiova, 1998;León-Ledesma and McAdam, 2004). Active labour market policies were used at varying degrees in different countries but with success only in some cases (Boeri, 1997).…”

“…Studies such as León-Ledesma andMcAdam (2004) andCamarero et al (2005) stress the importance of structural breaks and find that the unemployment series are stationary when the estimation methodology takes into account these breaks. Our findings on a more recent sample period suggest that taking into account structural breaks is not enough to confirm stationarity.…”

The nature of regional unemployment in Italy

“…near unit root processes, in which the variable in question displays mean-reversion, albeit at a very slow speed [Bean (1992)]. The two phenomena have been referred to as, respectively, "pure hysteresis" and "partial hysteresis" [Layard et al (1991), León-Ledesma and McAdam (2004)]. …”

“…1 At a theoretical level the notion of a non-stationary unemployment rate is clearly problematic since, strictly speaking, a bounded variable cannot be a random walk. However, as León-Ledesma and McAdam (2004) point out, "hysteresis as a unit root should not necessarily be understood as a 'true' description of the data generating process but as local approximation over a sample period" (p. 384). Moreover, as argued by Brunello et al (2000), the time required for the series to manifest its stationarity may be quite long and "during this interval the unemployment rate is exactly equivalent to a standard unrestricted random walk" (p. 158).…”