Selective Editing Techniques and Seasonal Adjustment of STS

Selective Editing Techniques and Seasonal Adjustment of STS Antonia Bertin, Marie Cordier-Villoing INSEE – STS department 28/04/2014

Outline Short presentation on the use of selective editing techniques in monthly production of short-term economic indicators Three main points : 1. General Process of selective-editing for STS (time series) 2. Score function for Seasonnal Adjustement Quality measures 3. Next steps : improve the process 28/04/2014

Elaboration of a Short Term Indicator 28/04/2014

Score function for SA Quality Measures The score function is a combination of various criteria : · Number and concentration of outliers; · Statistical properties of the Reg-ARIMA residuals ; · Lack of residual seasonality and trading-day effects in the irregular component and in the seasonally adjusted series; · Similar to M and Q-statistics (Statistics Canada); · Revision analysis (stability of the model, stability of the components). The global score is simple (-1, 0, +1). The component scores show where the problem is. The “alarm matrix” 28/04/2014

An example of “alarm matrix” 28/04/2014

Comparing the adjustments (1/2)

Comparing the adjustments (2/2)

Next steps Improving the imputation of missing data using the dynamic of the business answers. For the big companies which tend to be late, => forecast their answer using an ARIMA modeling of their past answers, possibly benchmarked on the estimated growth rate of the sector. Improving the score function used to check the quality of the seasonal adjustment. A principal component analysis could for example give a better set of criteria to check and a better linear combination of them. 28/04/2014

Thanks for your attention! Insee 18 bd Adolphe-Pinard 75675 Paris Cedex 14 www. insee. fr Contact Dominique Ladiray Tél. 01 41 17 50 18 Courriel : dominique. ladiray@insee. fr Informations statistiques : www. insee. fr / Contacter l’Insee 09 72 72 4000 (coût d’un appel local) du lundi au vendredi de 9 h 00 à 17 h 00