Data Matching for the Maintenance of the Business Register of Statistics Austria

Abstract

The Business Register of Statistics Austria is the basic instrument for all surveys conducted in economic statistics. For the maintenance mainly four different administrative sources are used. Unfortunately, the units of the different registers do not agree exactly and there is no unique numerical key in the business register and the administrative registers. Each register uses its own key. The units of an administrative register belonging to a certain unit of the business register have to be found by comparing alphanumerical items like name and address. For that purpose we use the method of Ngrams after some parsing and standardising of the texts. With that method above 90% of the profit-oriented units of the business register could be linked with a corresponding unit of the tax register (these linked units account for 99% of total turnover). 80% of the links were found fully automatically, the rest was checked manually.