Central Statistical Office of Poland
Eurostat project "Use of administrative data for business statistics"
Starting date: 1 November2009
Duration: 14 months
Project leader: Ms Grażyna Dehnel
Description of the action
Introduction
Economic transformation of Poland implied a growth of demand for information on economic statistics at a low level of aggregation. However, the European Commission pays its account on essential burdens of economic entities acting in Poland, resulting among others from the statistical reporting.
It concerns especially such enterprises, in which the number of employees exceeds 9 persons, i.e. medium and large. They are subject of a monthly survey. Results of the survey are a basis of estimation of basic characteristics included in short–term statistics, such as: revenues from sale of products (goods and services), number of employed persons, gross wages and salaries, value of wholesale and retail sale, excise tax and entity subventions. An obligation of monthly preparation of report laid upon economic entities is connected with their significant burdens. Due to this fact, there exists a necessity to modify the information system on economic statistics in relation to medium and large economic entities. On the one hand, this modification will concern a wide usage of administrative data in the enterprise statistics. On the other hand, an application of modern estimation methodology will be based on the „borrowing strength” from outside of a sample by, among others, an usage of so–called auxiliary variables. Such approach
is aimed at an improvement of estimation quality and extension of scope of distributions in which results of the survey are presented.
Objective of the project
The objective of the project is:
–investigation of an utility of available sources for yearly and short–term enterprise statistics
in relation to medium and large economic entities (i.e. with the number of employees exceeds
9 persons),
–reduction of burdens for medium and large economic entities resulting from statistical reporting,
–increasing of efficiency of conducted estimation,
–application of calibration methods for estimation of lacking data,
–application of new indirect estimation methods (small area statistics) in order to improve
of estimation precision,
–extension of collection of profiles, in which results of the survey are presented.
Detailed description of action
In relation of tasks performed within the project, one can distinguish the following areas of action:
- Conduction of the survey concerning an utility of available information sources of the public administration towards recognition of possibilities of their application in creation of official short–term and yearly economic statistics of the medium and large enterprise sector. In the survey, variables collected within labour market, business register and tax system will be taken into account. Actions connected with realisation of the task will be conducted by the Statistical Office in Katowice.
The results of the survey will be as follows:
–separation from the administrative register a set of variables which could be used in the enterprise statistics both as target variables and as auxiliary variables (or covariates) applied to the calibration and the indirect estimation,
–elaboration of a conception of modified system of creation of information regarding economic statistics towards as wide as possible usage of data coming from administrative registers. In relation to the exhaustive surveys it will affect an improvement of completeness and precision of statistical data. In the case of surveys conducted using sample method it will improve the quality and timelines both sample frame and estimation.
2.Conduction of the survey concerning possibilities of an usage of information coming from various administrative sources to the calibration estimation in the situation of incomplete data for the short–term and yearly enterprise statistics.
A connection of information from various sources will increase subject content of the database. It will enable to apply modern methods of estimation of lacking data, resulting among others from non–response or refusals of participation in the survey; an example can be here the calibration estimation.
Calibration is a method of estimation parameters in population which can be applied to surveys affected by non–response problem. This is a method in which so called calibration weights are found by minimizing a distance measure between the sampling weights and the new weights in order to satisfy certain calibration constraints. When the new weights are applied to the auxiliary variables available in the sample, they reproduce the known parameters in the population exactly. In the presence of powerful auxiliary information, it meets the double objective of reducing the sampling error and the non–response error. This method is very general and can be applied for any of the common sampling designs and for any auxiliary vector.
Using calibration and all available information describing activity of enterprises will meet the double objective of reducing bias and variance of estimators what is a consequence of negative influence of non–response and outliers indirectly.
Actions connected with realisation of this task will be performed by the Statistical Office in Poznań. The person responsible for methodological correctness of the proceedings will be the employee of the University of Economics in Poznań, specialist of the calibration estimation, Dr. Marcin Szymkowiak.
3.Conduction of survey related to an application of methods suggested by the small area statistics, which is based on the indirect estimation using auxiliary variables (covariates) coming from administrative registers.
A wide application of administrative data in creation of the enterprises statistics doesn’t solve all problems connected with the economic statistics. Distributions of entities by analysed variables are characterised, however, by very strong right–side asymmetry, large diversification and strong concentration. Therefore there exists a necessity to use new non–classical estimation methods concerning the small area statistics.
The small area statistics is a „tool”, which enables – on the basis of small sample of units and by maximum usage of all available additional auxiliary information – to provide the necessary information in a short term and with not high costs. Additionally, as a result of an application of the small area statistics, one can do an estimation at the level, for which obtaining any information was impossible till now. At present, the results of the survey of medium and large entities are published using the spatial (voivodships) or branch (NACE sections) distributions. An application of the small area statistics in relation to action included in the project will enable to take a trial to estimate of data at lower level of aggregation. It would be a connection of the branch distribution, which is a type of conducted economic activity with a spatial profile, represented by the voivodships. Actions connected with realisation of this task will be performed by the Statistical Office in Poznań. The person responsible for methodological correctness of the proceedings will be the employee of the University of Economics in Poznań, specialist of the small area statistics, Prof. Dr. Elżbieta Gołata.
Expected results of the action
Planned results of the project are as follows:
1. Improvement yearly and of short–term economic statistics in relation to the medium and large enterprise sector by:
–application of data collected in administrative registers to the survey,
–improvement of estimation precision by enlargement of level of completeness of collected information,
–application of non–classical methods of indirect estimation,
–extension of scope of distributions, in which results of the survey will be presented.
2.Enhancement of efficiency of statistical survey by:
–reduction of burdens of economic entities resulting from the statistical reporting,
–reduction of data collection costs,
–improvement of data precision,
–changes in the profile of exploitation of the staff in the statistical office by an extension of team of employees involved in data analysis together with a simultaneous reduction of number of persons responsible for data collection.
Planned deadlines of implementation of the results of the project:
improvement of quality of the economic statistics: from 2011
improvement of efficiency of statistic al survey: 2011 – 2012.
1