Online Program

A risk of using administrative data for statistical purposes – a case study in the Swedish survey Newly Started Enterprises

Keywords: Administrative data, newly started enterprises,

The use of administrative data for producing business statistics has grown extensively the last decade at Statistics Sweden. The discussion about administrative data for statistical purposes is often focusing on the positive aspects and all the possibilities the registers provide. There is less focus on the risks.

In this paper we study the effects of using administrative data from the Swedish Business Register for producing short term statistics in the survey Newly Started Enterprises. The newly started enterprises starting time was before collected in the survey, but has from 2010 been removed from the questionnaire in effort to reduce the response burden. The study shows that using start-up date from the Swedish Business Register in order to estimate the total number of newly started enterprises on short term basis gives significantly different estimates compared to estimates based on the starting time in the questionnaire.

A test was conducted in order to transform the administrative variable start-up date to a variable that better suits the statistical requirements in the survey. The estimates based on the transformed variable were clearly improved.