Les méthodes d'estimation de la précision de l'Enquête Emploi en Continu

The Labour Force Survey is one of the major household surveys conducted by INSEE as well as a centrepiece of the French employment and unemployment statistical system. It results from a complex sample design, which makes variance estimation particularly difficult. This working paper constitutes the outcome of years of work on this subject. The first part, titled “Compendium†, draws the methodological framework of the Labour Force Survey and presents on the one hand the main methodological choices made for the variance estimation, boss for cross-sectional and longitudinal estimations, and on the other hand the main results for 2012. The following sections set out the different formulas, their justification and the different results obtained with more details. Nota bene : the estimates – unemployment rates, quarterly variation of the number of unemployed – presented in this paper correspond to raw results from direct estimates performed on an ad hoc basis for the variance estimation. They do not tally with the official estimates released by INSEE, neither at the national level – in particular due to the lack of seasonal adjustment – not at the regional level – for which official estimates (the “localised unemployment rates†), also seasonally adjusted, are computed according to specific estimators mobilizing the ongoing Labour Force Survey and allocation keys based on end-of-month jobseekers. The results refer to the period 2003-2012, before the updating of the questionnaire at the beginning of 2013.