tion schemes). Since all of these determinants show little variation through time, regional
disparities can be viewed as an equilibrium outcome.
Alternatively, regional unemployment disparities may reflect disequilibrium in the re-
gional labour markets. All regions would converge to the same competitive equilibrium
but, due to sluggish adjustment mechanisms in the regional labour markets, adverse shocks
have persistent after-effects. Being initially at equilibrium, a one-off shock generates dif-
ferences in the regional unemployment rates. Regional unemployment rates respond by
moving slowly towards equilibrium but, before their adjustment process is completed,
they are hit by new shocks. As a result, full adjustment is never achieved and disparities
in the regional unemployment rates are even greater, resulting in a polarization effect.
Blanchard and Katz (1992) show that in the US regional unemployment disparities
are not persistent due to high labour and firm mobility. Workers move from high to low
unemployment regions in search for better labour market prospects, while firms move to
high unemployment regions to benefit from lower labour costs.3 The Blanchard and Katz
model focuses exclusively on idiosyncratic shocks in a perfect labour mobility framework.
On the other hand, Decressin and Fatas (1995) show that the adjustment of European
regional unemployment rates to shocks is driven by participation rather than by migration.
Jimeno and Bentolila (1998) find that Spanish migration does not play any significant
role in regional labour market adjustment, and the lack of wage flexibility reinforces the
persistence of disparities in regional unemployment rates.
Overman and Puga (2002) show that a polarisation (clustering) process of European
regional unemployment rates has taken place since 1986 as a result of changes in relative
labour demand which have been similar across geographical neighbours. Idyosyncratic
characteristics, national or regional, can only partly explain this neighbour effect which
is strong both within and across national borders.
Bande et al. (2005) find that Spain is a differential country within the EU as regards
regional unemployment disparities and their relationship with the business cycle. They
conclude that this result is related to changes in the degree of wage bargain centraliza-
tion/coordination, which introduces wage imitation effects in the sectoral bargainings,
imposing high unit labour costs in the less dynamic industries of the less dynamic re-
gions, allowing thus for less employment creation during booms and more employment
destruction during recessions.
In this paper we explain the evolution of Spanish regional unemployment rates by
applying the Chain Reaction Theory (CRT) of unemployment.4 Following the CRT ap-
proach, we use a dynamic multi-equation labour market system to model regional dispar-
ities. Our model consists of three equations: labour demand, wage setting and labour
supply. Arguments in favour of such a multi-equation model, as opposed to a single equa-
tion one, can be found in the survey by Elhorst (2003) and in Karanassou, Sala, and
Snower (2003).
3 This is because the large fraction of unemployed workers puts downward pressure on wages.
4The CRT was developed by Karanassou and Snower (1996). See also Karanassou (1998), Karanassou
and Snower (1998, 2000), and Henry, Karanassou and Snower (2000).