Descomposición Estructural de las Series de Desempleo: Una Aplicación para las Ciudades de la Región del Bío Bío

In this paper we characterize the unemployment rate for the main cities of the Bío Bío Region using the state-space multivariate approach, in order to isolate and evaluate their tendency and seasonal components. We present theoretical expressions for each structural model for monthly unemployment rate data which spans between I.1996 and XII.2005. We use an additive multivariate structural model for the cities of Chillán, Concepción, Gran Concepción and Talcahuano, an additive bivariant structural model for Coronel and Lota, and an additive univariate structural model for Arauco and Los Angeles. The main results show, firstly, that the estimated state-space approach models present a good explanation power (MAPE is less than 8 per cent for four cities); secondly, the hypothesis which states that seasonal peak levels for unemployment rate are related to cities strongly linked with the primary sector (raw materials) is verified, thirdly, quantifying the seasonal component for unemployment series in each city gives useful information to local policy-makers, because it can determine the optimal length for emergency employment plans.