Methodological questions raised by the combined use of administrative and survey data for the French structural business statistics Work session on statistical data editing, Oslo, September 2012 (session iii)
Outlines of the presentation The project of Insee’s new device of production of the French Structural Business Statistics (ESANE) has been presented in former data editing work sessions Mainly, it is based on administrative data (in particular, annual income statements sent by enterprises to the tax authorities), combined to a statistical survey This presentation focuses on two topics: Field issues were revisited when implementing the new system The consequences of the use of multiple sources on data editing 2
1. Field issues (1) The theoretical definition of the field : market-oriented enterprises belonging to some sectors The practical implementation : two options Using the referential of the tax authorities Defining the field within the business register by using some codes Here the second option was chosen, but some problems did appear Some fiscal data are missing within the field Fiscal data arriving oustide the field 3
1. Field issues (2) First problem : missing fiscal data within the field Classic question of handling of non-responses But what was observed gave elements on the quality of some codes of the register Second problem : fiscal data outside the field This question raised problems linked to the definition of the enterprise : since the legal units are used, some of them had been wrongly considered as not participating to the field Particulary, some juridical societies do exist in the register for specific purposes, for big groups (for example they do own the assets of these groups but have no production) ; it is difficult to isolate them using just the codes of the register The methodological choices concerning the definition of the field had to be adjusted 4
2. The use of a composite material (administrative and survey data) (1) As presented in former data editing work sessions, each source has its separate data editing process A specific work for the fiscal files (particularly in case of multiple tax declarations for the same entreprise, and in case of non calendar year for the accounting period) A step of comparison of individual data made for the enterprises belonging to the sample of the statistical survey 5
2. The use of a composite material (administrative and survey data) (2) To produce statistics, it was decided to use statistical combined estimates, rather than to use mass imputation Main principle : use of a difference estimator This gives a specific role to the enterprises of the sample of the statistical survey, especially those changing of NACE code 6
Conclusion When combining administrative and survey data, each step plays its own role Field issues have been revisited, with consequences on the aggregates Each source has its own process of control The step of comparison of individual data between survey and administrative data leads to a mutual improvement of each source The use of combined estimates has consequences on the data editing 7
Thank you for your attention ! Methodological questions raised by the combined use of administrative and survey data for the French structural business statistics Thank you for your attention ! 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 Contact M. Philippe Brion Courriel : philippe.brion@insee.fr