STATISTICS AUSTRIA

Directorate “Business Statistics” Trade and Services

Guglgasse 13
A-1110 Wien

Index of turnover.
The purpose of this indicator is to reflect the short term development of turnover in the relevant categories of NACE Rev. 2.
The indicator is based on absolute values in unit 1,000 Euro.

NACE Rev. 2

The indicators cover economic activities of NACE Rev. 2 division 47 in the breakdown required by STS-R.
There is no threshold in terms of size criteria.

Generally the definitions of turnover correspond to those provided for in the "Methodology of Short Term Business Statistics" produced by Eurostat. Turnover excludes VAT and is derived from monthly or quarterly VAT advance return for most of the enterprises; survey data is available for a subsample of enterprises.
The reference period is the turnover flow during the calendar month excluding VAT.
Deflated indicators represent the value of turnover in constant prices.

Enterprise (=legal unit in BR) is reporting unit and observation unit.

The target population are all active enterprises in BR in NACE Rev. 2 division 47 at the end of the relevant reference months. BR covers enterprises with at least one employee or 10,000 Euro yearly turnover. The frame is a file extracted from the BR with every base year change - the frame is updated with births and deaths of enterprises every month. Total population is about 45.000 enterprises.

The geographical area covered is Austria as a whole.

Time series available for all aggregates of NACE Rev. 2 required by STS-R:
- From 01/2018 time series according to the NACE Rev. 2 on base year 2015=100 are available backcasted till 2010
- Time series according to the NACE Rev. 2 on base year 2010=100 are available from 01/2005 till 12/2017
- Time series according to the NACE Rev. 2 on base year 2005=100 are available from 01/2000 till 12/2012

The base year for indices is 2015 since the first reference period 2018.

Indices of turnover. No absolute figures are published.

Monthly turnover.

The European legal base is the Council Regulation (EC) No 1165/98 concerning short-term statistics in the current version, the national legal base in Austria is the National Regulation concerning short term statistics in trade and services (BGBl. II Nr. 233/2003 in the current version).

No data are transmitted to other international organisations (except Eurostat).

The legal base for statistical confidentiality is:
— The general obligation to publish statistics and the strict provision on statistical confidentiality is regulated by the “Federal Statistics Act 2000” ("Bundesstatistikgesetz 2000") in the current version.
— The protection of personal data is covered by the “Data Protection Act 2000” ("Datenschutzgesetz 2000 (DSG 2000)", in the current version).
— National Regulation concerning short term statistics in trade and services (BGBl. II Nr. 233/2003 in the current version, BGBl. II Br. 259/2013).

Aggregated figures consisting of data of less than 4 enterprises are regarded confidential and therefore not published.
In general published STS aggregates are not confidential.

A release calendar covering release dates of the first/second half of the year is published on the website of Statistics Austria. News releases are announced in the release calendar as well.

http://www.statistik.at/web_en/

Release is simultaneous. Along with the national release on the website of Statistics Austria and the data transmission to Eurostat the most important national users (Austrian Central Bank, Austrian Institute of Economic Research etc.) are informed separately.
No prior access before official data release.
The „Federal Statistics Act 2000“ ("Bundesstatistikgesetz 2000" in the current version) explicitly states in §30 (3) the obligation of Statistics Austria to inform without delay the Federal Ministry responsible for the subject matter concerned about the results of statistical surveys and to publish them simultaneously.
According to the release calendar of Eurostat the data are transmitted to Eurostat.

The data are transmitted monthly to Eurostat and disseminated monthly to national users.

In general, for the publication of quarterly indices a news release is published featuring the key figures. News releases are issued at 9 a.m. CET online on the website of Statistics Austria.

National paper and electronic publications are:
- "Schnellberichte" (Quick Reports): "Schnellbericht 3.3: Konjunkturstatistik Handel und Dienstleistungen";
- "Statistische Übersichten" (Statistical Overviews -Supplement to "Statistische Nachrichten");
- "Statistisches Jahrbuch Österreichs 20xx" (Austrian Statistical Yearbook);
- Website of Statistics Austria: STATISTIK AUSTRIA - Trade, Services - Short Term Statistics;
Variables and breakdown of results as required by STS-R.

The Database STATcube can be found on the Website of Statistics Austria: STATISTIK AUSTRIA - Trade, Services - Short Term Statistics - Stat. Databases (http://www.statistik.at/web_en/statistics/trade_services/short_term_statistics/index.html#reiter_dblinks).
This Database is updated with every publication and the NACE Rev. 2. levels as presented in point 3.3 are published. The time series goes back till 2005; all indices, including calendar or seasonally adjusted indices are provided. Monthly, quarterly and yearly results are available.

No microdata are available.

Data are sent to Eurostat.

Metadata and quality reports:
- Statistical News 10/2013: "Short Term Statistics Trade and Services: Base year 2010 and new variables" (only available in German);
- Standardised metadata and quality report for Short Term Statistics in Trade and Services - (only available in German);
- Conceptual overview available on the website: STATISTIK AUSTRIA - Trade, Services - Short Term Statistics - Further information.

Metadata and quality reports:
- Statistical News 10/2013: "Short Term Statistics Trade and Services: Base year 2010 and new variables" (only available in German);
- Standardised metadata and quality report for Short Term Statistics in Trade and Services - (only available in German).

Quality checks and validation of data are done throughout the whole compilation process. This is described in the standardised metadata and quality report for Short Term Statistics in Trade and Services on the website of Statistics Austria.
The chapter “Quality” of this report contains the following points:
— Relevance;
— Accuracy (Sampling Errors, Non-Sampling-Errors, revisions,…);
— Timeliness and punctuality in dissemination;
— Accessibility and clarity of the information;
— Comparability;
— Coherence with other statistics.
Following relevant changes discussions with national experts and the Quality Committee of the Statistical Council of Statistics Austria are held for feedback and modification suggestions.

Indices are based on reliable data, regular meetings with decision making bodies and users take place. Publication of results is almost punctual.
Further methodological details and other information can be found in point 13.
Coverage of the indices is the whole Austrian territory.
Multiple plausibility checks with different sources are made.

Short Term Statistics are used by different national and international institutions and non-institutional users like media, enterprises, educational and research institutions and the public.
Examples for national users are federal ministries, political institutions, interest groups and tourism associations but there is also an internal use by other statistics produced by Statistics Austria.
International Institutions are e.g. European Commission, OECD, and the European Central Bank.

No index for user satisfaction is available, but there are regular meetings with national experts and essential users to react to new requirements, if these are not in contrast with legal necessities or constraints. In addition a user’s satisfaction survey takes place in periodic intervals.
The survey 2013 has covered trade and services as a whole, ratings were from "very good" to "unsatisfactory" according to Austria's grade system in school.
Actuality was rated mostly "very good" or "good" by users (both clear above 30%, N=24), and periodicity was rated "very good" by more than 70% of the users (N=20). Scope and level of detail was rated "very good" by more of 50% and good by more than 20% (N=20). The availability of data was rated "very good" by more than 70% of the users (N=21), the comprehensibility of presentation and of metadata were rated "very good" by approx. 80% of the users (N=20, N=12). Accuracy was rated "very good" by over 60% and "good" by over 20% (N=22). Comparability was rated "very good" by nearly 60% and "good" by over 25% (N=22).
Every year an advisory board for enterprise statistics is held where concepts and new developments are presented and discussed.

This index is complete, there are no gaps or missing data parts.

Impact of sampling error: In the first half of the year 2018 the sampling error was approximately 1.7%. (95% confidence level)
Impact of non sampling error: approx. 20% of the data are checked.
accuracy of first estimates: in the first half of 2018 the preliminary results differ from the revisioned data in the range from 0.1 to 1.4 percentage points.
Detailed information for the sampling and non-sampling errors see points 13.2 and 13.3. First release of preliminary results with t+30, final results with t+60.

Role of sampling: Stratified random sample of approximately 2,800 enterprises.
For calculation of t+30 results not enough VAT data is available - so a direct inquiry is made with a subsample of enterprises. From the 1,200 inquiries there are approximately 600 enterprises (unit-non-response-rate of first estimates) who respond to this in principal voluntary survey.
multiple use of sampling in production: This sample is used for calculation of turnover and deflated turnover.
coverage of main aggregates: In the first half of the year 2020 the sampling error was approximately 1.75%. (95% confidence level)
The coefficient of Variation for Retail Trade (G47) is 0.9% in January, 0.9% in February, 0.9% in March, 0.8% in April, 0.9% in May, 0.9% in June, 1% in July, 0.9% in August, 1% in September, 0.9% in October, 0.9% in November and 0.9% in December.

coverage errors: Approximately 90% of the active enterprises in the whole BR can be linked to VAT data. The missing 10% represent only about 2% of total turnover. After t+60 days approx. 88% of total turnover data is available for enterprises in the coverage of STS. The response rate of the sampled enterprises in VAT advance return is about 85%. (mainly small enterprises are missing)
data collection or access errors: In the case of group taxation VAT declarations are available for the group head - distribution to different enterprises of the group has to be carried out manually. Dependence on the quality of BR and the legal frame for timely VAT data has to be considered as well.
response rate (unit non-response)/ item non-response: Unit non response of VAT advance return- it is assumed that enterprises with a positive development report earlier; as available data is used as basis for imputation of missing data a more positive result is possible.
In the year 2020 the unit non response rate was 18.1%, this units will be substituted based on the value of the year before (The share regarding the volume of substituted units is smaller).
editing, coding and imputation errors: In NACE activities with many small enterprises the index can be influenced by imputation based on the design of the BR (minimum 10.000 Euro turnover per year or at least one employee).
modelling errors: Slight differences in the definitions of the VAT and turnover from company accounts.
Plausibility checks (approx. 20% of the data are checked) in the course of data editing minimise these effects.
Changes of NACE classification, new enterprises and inactive enterprises are checked for plausibility and impact on results as well.

In general about 30 days after the end of the reference month for the aggregates 47; 47 (without 47.3); 47.11 and 47.2 and 47.19; 47.4 - 47.9.
60 days after the end of the reference month for all levels of detail according to Annex C of STS-R.
Final data are not revised after their first publication; the revision for the preliminary data is the publication of final data.

In general, Statistics Austria provides t+30 results two days before EUROSTATs publication date. An earlier transmission is not possible for every month. This is based on the fact that sufficient administrative data is not available - so a survey among 1200 enterprises (response rate about 50%) is done to provide the data required for index calculations. Final data at t+60 were delivered on time.

For the whole Austrian territory the same statistical concepts are applied.

Monthly time series with base year 2015 and NACE Rev. 2: from 01/2015
Monthly time series with base year 2010 and NACE Rev. 2: 01/2005-12/2017
Monthly time series with base year 2005 and NACE Rev. 2: 01/2000-12/2012

Coherence has to be described in the standardised meta information and quality reports for each statistical product of Statistics Austria.
The most important sources for cross checks are structural business statistics and other statistical products of Statistics Austria e.g. foreign trade statistics, price statistics, VAT statistics, statistics of registration of new and used vehicles, consumer price indices, wholesale price indices, statistics of overnight stays, social security statistics, national accounts; studies from economic research institutes, news releases.
All indices (turnover, employment, hours worked and wage and salaries) are compared to assure consistency of developments.

Internal coherence is assured.

Burden on enterprises is zero because the data is mainly withdrawn from administrative sources and the statistical survey for large enterprises is on a voluntary basis, which is excluded from the response burden barometer in Austria.

In the course of routine revisions preliminary results are replaced with final results at t+60.
Major Revisions like base year changes are done according to Eurostat's requirements.
The revision calendar correlates to the publication of the monthly indices.

Data sent to Eurostat is the same which is published nationally.

Schedule is based on Eurostat's release calendar.
Benchmarking for the most important aggregates is done in regular intervals.
Results are compared to other relevant internal and external statistics. Differences in definitions and concepts have to be considered in the course of interpretation of results.
Revisions and methodological changes are discussed with all involved decision making bodies in advance and announced simultaneously.
Data are stable.
MAR and MR of growth rates for unadjusted and calendar adjusted data series (1/18 till 12/20):
MAR for G47 TOVT (0.48%) and G47 TOVV (0.50%) unadjusted and for seasonal adjusted TS (0.29% for TOVT and 0.28% for TOVV).
MR for G47 TOVT (0.31%) and G47 TOVV (0.32%) unadjusted and for seasonal adjusted TS (0.11% for TOVT and 0.12% for TOVV).
These figures were calculated starting 1/18 due to the base year change and are based on the difference between the rates of change of preliminary and revisioned indices.

Administrative and survey data are used for the enterprises in the sample. (Monthly and quarterly VAT advance return from tax authorities; statistical survey for a small number of large enterprises (maximum 10% of the enterprises in the sample)).
The BR is the frame on which the source is based.
Sample design:
- Stratification of population is conducted by 44 economic activities and four groupings of turnover size classes.
- The total sample size (for Division 47) comprises about 2,800 enterprises. This amounts to about 7% of the population.
- The sampling procedure applied was methodologically in accordance with a stratified random sampling in the course of which the sampling units were arranged within the strata. The selection of new enterprises is conducted monthly in an analogous manner.
- Slight update with births and deaths of enterprises is done every month. A new sample frame is used with the change of the base year at five year intervals.

Data are collected monthly. The reference period for the turnover is the whole month.

Data sets with monthly or quarterly VAT advance returns are transmitted from the Tax Authority electronically and linked with the enterprises of the BR. Large enterprises report their monthly turnover using a Web form for data transmission. For purposes of short term statistics these two data sets are combined for further calculation of turnover indices in the relevant categories of NACE.

Validation level 0: Mainly administrative data sources are used so structure of files and format of data are not checked especially. In the web data the dimension is checked.
Validation level 1: Every change of NACE Rev. 2. in the most detailed level, new enterprises and inactive enterprises are controlled for plausibility and impact. Furthermore considerable value-changes are checked.
Validation level 2: The actual reporting period is compared with the last and the corresponding reporting period of the last year. If there are any unexpected developments, the enterprise data are checked and possibly changed.
Validation level 3: The indices are compared to comprehend plausible and identify implausible developments. Furthermore the indices are crosschecked with the number of persons employed of the same NACE Rev. 2. category.
Validation level 4: The results are compared with the annual structural business statistics and other statistical products of Statistics Austria e.g. foreign trade statistics, price statistics, VAT statistics.
Validation level 5: The results are cross-checked with publications of other external institutions. Further comparisons are made with studies from economic research institutes and news releases.

Data transmitted from the tax authorities are linked to enterprises of the BR and extracted for purposes of STS; non-responses are estimated in the following way: missing data for the month t are estimated by multiplying the data of the previous year in the same month (t-12) by a stratum-specific factor. If a value for the same month of the previous year is not available the value of the previous month (t-1) is taken into account. So gross data can be calculated for each category of NACE.
To get an estimator of total turnover the value of each enterprise is multiplied with a stratum-specific grossing up factor (this is the ratio of population size and sample size in each strata). The values are summed up to each relevant category of NACE. The grossed up results are used for the compilation of indices and are not disseminated.
The indices are calculated from the grossed up totals. The indices measure the turnover for each reference period and each subgroup as a percentage of the average value of the base year. Weights are updated in 5-year intervals with every base year change.
The average STS turnover for 2010 is used as weight for index calculation.
Data is converted to GESMES/TS and sent to Eurostat.

The turnover index is deflated using an appropriate price index from the consumer price index.
For seasonal and working day adjustment X13-ARIMA Version 1.0 is used. NACE aggregates are calculated by using the indirect method. The model and regressors are fixed in the program once a year, coefficients are estimated whenever new data becomes available.
The calendar adjustment was based on a country-specific calendar. The adjustment was performed within the RegARIMA part of the program, including regressors for easter effects and leap year. Outlier identification was performed automatically within the program (additive Outliers, Level Shifts and Transitory Changes).
Models/Filters are selected automatically, if the automatic choice is not considerd adequate it is corrected manually; models are fixed for one year, parameters are estimated every time a new value becomes available; the whole series is revised every time a new value becomes available; the seasonal adjustment decomposition (additive vs. multiplicative) is done dependent on the series' characteristics; the series is checked for outliers automatically (level shifts, transitory changes and additive outliers) - if the automatic choice is considered unadequate (visual inspection of the series) or if too many outliers are identified the model is rechecked and/or critical values for outlier identification are adepted; the filter length is chosen automatically, eventually changed manually after inspection of si-ratios; adjustment is done directly at the level of publication with all higher aggregates adjusted indirectly.

No comment.