Temporal disaggregation for unemployment rates

Abstract

This thesis deals with the estimation of monthly unemployment rates for the Maltese population which are officially supplied with quarterly frequency. Standardization of estimation methodologies over statistics bureaus from various countries has become a major research goal within international official statistics. One important and popular such method, the temporal disaggregation method as developed by Chow and Lin (1971), was studied. Noting some similarities between the statistical agendas catered for by this method and the generalized estimating equations method, it was decided to look into the latter as well. These two methodologies were compared, and particular features of each where highlighted. Finally, these techniques were applied to estimate the Maltese monthly unemployment rates given quarterly figures, using data provided by the National Statistics Office and monthly figures provided by the Employment Training Cooperation. The results obtained using the Chow-Lin methodology were very close to those issued by Eurostat. The estimates obtained using the Generalized Estimating Equations methodology, were conservatively staying close to the overall means, thus downplaying local variations in time. The estimates presented by Eurostat turned out to run closer to the Chow-Lin estimates in fact. This confirms that this methodology should be applied for temporal disaggregation situations at least within the Maltese context as was the case in this dissertation.