Presentation is loading. Please wait.

Presentation is loading. Please wait.

Philippe Brion Johara Khélif Insee 28/04/2014 Questions raised by the implementation of the data editing device for French structural business statistics.

Similar presentations


Presentation on theme: "Philippe Brion Johara Khélif Insee 28/04/2014 Questions raised by the implementation of the data editing device for French structural business statistics."— Presentation transcript:

1 Philippe Brion Johara Khélif Insee 28/04/2014 Questions raised by the implementation of the data editing device for French structural business statistics Work session on statistical data editing, Paris, April 2014 (topic iii)

2 28/04/2014 Conference on European Statisticians – Work Session on Statistical Data Editing 222 Outlines of the presentation  A new production process for French Structural Business Statistics has been working for 5 years  What have we learnt from it ?

3 28/04/2014 Conference on European Statisticians – Work Session on Statistical Data Editing 33 1. A few elements on the production process  The Esane process was put into place in 2009 –Produces Structural business statistics every year ; –Is the main yearly statistical source on firms for Eurostat and French National Accounts ; –The old system used two different sources : a survey and administrative data (tax declarations) ;  A new system under constraint –A change under constraint : budget and staff were called to diminish ; –Methodology : keep high standards on statistical quality  Insee’s answer through the Esane process –Produce only one set of statistics, instead of two before ; –Put selective editing methods into place. 3

4 28/04/2014 Conference on European Statisticians – Work Session on Statistical Data Editing 44 2. From theory to practice (1/2)  In theory, the data editing system was based on the following idea : –Implementation of local scores reflecting the “influence” of a firm on the final result (based on temporal and contemporary drop-out scores) ; –Synthetic global priority scores pointing problematic firms out.  In practice, three main problems –The scores were sometimes unsatisfactory ( unstable, not adapted to some variables, problem with measuring the potential error) ; –The method led to too much work for the staff (thresholds calculated « a priori » had to be adjusted) ; –The first intended calendar was unrealistic. 4

5 28/04/2014 Conference on European Statisticians – Work Session on Statistical Data Editing 55 2. From theory to practice (2/2)  Since 2009, the methods have been fine-tuned –The computation of local score aggregates is more robust (based on a yearly median growth rate estimate for each sector) ; –Computation of the global score, taking into account the relative importance of the results of each local score –But the data editing method still lacks efficiency for some variables  The calendar for the production process has been reviewed –Instead of publishing one final result in December of year N+1… –…We publish  Semi-final results in December N+1 by implementing only part of the administrative data (75% of the firms, 85% of the final value added) ;  Final results in June N+2 after implementing all the data. 5

6 28/04/2014 Conference on European Statisticians – Work Session on Statistical Data Editing 66 3. Lessons learnt (1)  Lesson one : Selective editing is better when there is feedback from the producers –we changed our methods thanks to feedback from the editing staff; –We plan on keeping very close to them :  Yearly reviews on Esane process’ help us improve;  The editing staff is more satisfied than in the beginning –Working closely prevents the “black box” effect 6

7 28/04/2014 Conference on European Statisticians – Work Session on Statistical Data Editing 77 3. Lessons learnt (2)  Lesson two : feedback from the users is also helpful –It helped us change our calendar –Led us to invest on output editing… –Helped us to improve our editing methods  Lesson three : we need to keep working on the methods –Keep robust output editing methods (as a safeguard) –Having a continuous improvement of the methods through the use of metadata –But we don’t think it would be efficient to have globally standardized selective editing tools for all Insee  Could be useful for structural firm surveys  And administrative data. 7

8 28/04/2014 Conference on European Statisticians – Work Session on Statistical Data Editing 8 Thank you for your attention ! Contact M. Philippe Brion Courriel : philippe.brion@insee.fr Insee 18 bd Adolphe-Pinard 75675 Paris Cedex 14 www.insee.fr Informations statistiques : www.insee.fr / Contacter l’Insee 09 72 72 4000 (coût d’un appel local) du lundi au vendredi de 9h00 à 17h00 Questions raised by the implementation of the data editing device for French structural business statistics


Download ppt "Philippe Brion Johara Khélif Insee 28/04/2014 Questions raised by the implementation of the data editing device for French structural business statistics."

Similar presentations


Ads by Google