Estimating structural business statistics based on administrative data: the case of the Italian small and medium enterprises Luzi O., Rinaldi M., Seri G., Guarnera U., De Giorgi V. Italian National Statistical Institute (Istat) 1. Introduction Following the most recent EU Regulations on structural business statistics (SBS) establishing that, in order to estimate information on the structure of National production systems, Member States can integrate data available in different information sources (including administrative data), the Italian National Statistical Institute (Istat) has recently started a re-design project concerning the System of Enterprises Accounts Surveys with the main objective of reducing burden on enterprises, production costs and non response rates, while maintaining high data quality levels. The objective is widening the use of administrative data in this area and gradually move from a traditional stove pipe model for the statistical production to an integrated model where administrative data represent the information core, and statistical surveys are conducted to estimate specific sub-populations/variables which are not available in external archives. In Italy, SBS on enterprise accounts are currently obtained from two annual statistical surveys, partly integrated with administrative data. The main administrative sources used in this context are Financial statements (FS) and Tax Authority data, in particular the Sector Studies Survey data (SS). In this paper, besides the analysis of the usability of FS and SS in terms of completeness and coverage, experimental evaluations of the potential biasing effects of integrating these sources with SBS survey data are illustrated. Some results for key variables directly available in these administrative sources (Turnover, Purchases of goods and services, Personnel costs) are illustrated. Variables which are not directly available from external sources require a higher modelling effort: some experiments on the components of Change in stocks of goods and services (Casciano et al., 2011; Luzi et al., 2011) have been already performed in the context of the ESS-Net on the use of Administrative data for business statistics (http://essnet.admindata.eu/) 1 . The paper is structured as follows: in Section 2, the current Istat surveys on SBS and the administrative data useful in this context are illustrated. Section 3 contains some analyses on the quality of the external data sources, and the results of the performed experimental studies. Final remarks are reported in Section 4. 2. Structural Business Statistics in the Italian survey system In this paper we refer to the Italian system on SBS as to the set of annual surveys aiming at investigating mainly profit-and-loss accounts of enterprises. The SBS population (reference year 2008) is composed by about 4.5 million enterprises. The Italian enterprises' structure is characterised by a strong presence of small and very small units: in effect, the enterprises with 1-9 persons employed are about 4.2 million, and account for 33.2% of total value added and for 47.2% of total employment. On the opposite side, the enterprises with 100 or more persons employed are about 11,000: they account for 37.2% of total value added and 25.3% of total employment. In order to meet SBS regulation, Istat carries out two distinct annual surveys: 1) the Annual Survey on the Economic Accounts of Enterprises (SCI in the following), which is a total survey on the enterprises with 100 or more persons employed, and 2) the sample survey on Small and Medium Enterprises survey (SME in the following) upon enterprises with less than 100 persons employed. Both surveys involve units belonging to the industrial, construction, trade and services economic activities and collect information concerning profit-and-loss statements and balance sheets, as well as information regarding employment, investment, personnel costs and the regional breakdown of some variables. For both surveys the frame is represented by the Italian Business Register of active enterprises (BR in the following), resulting from the combination of both statistical and administrative information. Target parameters are estimated by publication domains in accordance with the SBS Regulation 2 . 1 funded by Eurostat in 2009 in the framework of the Modernisation of European Enterprises and Trade Statistics (MEETS) Program. 2 The data domains are: 1) class of economic activity (4 Nace-code digits); 2) economic activity (3 Nace-code digits) by size (classes of persons employed); 3) economic activity (2 Nace code digits) by regions (Nuts2 level).