Institut national de la statistique et des études économiques (STATEC)

ENT3 - Structural Business Statistics

STATEC
B.P. 304
L-2013 Luxembourg

Data on the Information and Communication Technologies (ICT) usage and e-commerce in enterprises are survey data. They are collected by the National Statistical Institutes or Ministries and are in principle based on Eurostat's annual model questionnaires on ICT usage and e-commerce in enterprises.

Large part of the data collected is used to measure the progress in the implementation of one of the main political priorities of the European Commission for 2019 to 2024 – A Europe fit for the digital age. Part of this is the "European strategy for data", envisioning a single market for data to ensure the EU's global competitiveness and data sovereignty, in which context a comprehensive set of new rules for all digital services was proposed: the Digital Services Act and the Digital Markets Act, which are centrepieces of the EU digital strategy. Furthermore, the Commission and the High Representative of the Union for Foreign Affairs and Security Policy presented a new “EU cybersecurity strategy”, which is intended to bolster the EU's collective resilience against cyber threats, safeguard a global and open internet and protect EU values and the fundamental rights of its people. Furthermore, data will allow monitoring the progress towards  A Europe fit for the digital age, one of the six priorities for the period 2019-2024 of the von der Leyen European Commission.

The aim of the European survey on ICT usage and e-commerce in enterprises is to collect and disseminate harmonised and comparable information at European level.

 NACE Rev.2 2008

All economic activities in the scope of Annex I of the Commission Regulation are intended to be included in the general survey, covering enterprises with 10 or more employees and self-employed persons. These activities are: NACE Rev. 2 sections C, D, E, F, G, H, I, J, L, M and N, division 95.1.

For micro-enterprises see the sub-concepts below.

Micro-enterprises are not covered by the survey in Luxembourg.

The model questionnaire on ICT usage and e-commerce in enterprises provides a large variety of variables covering among others the following areas:

-          Access to and use of the Internet

-          E-commerce and e-business

-          Use of cloud computing services

-          Artificial Intelligence

-          Other topics: Data utilisation, sharing, analytics and trading, Invoicing.

The annual model questionnaires and the European businesses statistics compliers’ manual for ICT usage and e-commerce in enterprises comprise definitions and explanations regarding the topics of the survey.

The enterprise unit was used.

Target Population

As required by Annex of the Commission Implementing Regulation, enterprises with 10 or more employees and self-employed persons shall be covered by the survey.

For micro-enterprises see the sub-concepts below.

Micro-enterprises are not covered by the survey in Luxembourg.

The whole territory was covered.

Years 2022 and 2023.

Not applicable

Percentages of enterprises, Percentages of turnover, Percentages of employees and self-employed persons.

The reference periods defined in the model questionnaire were followed in the national survey: “Where not otherwise specified, the reference period is the current situation (year 2023). The reference period for the percentages of sales/orders data is the financial year 2022.”

Complementary national legislation constituting the legal basis for the survey on the use of ICT in enterprises:

Loi du 10 juillet 2011 portant sur l’établissement de l’Institut national de la statistique et des études économiques (STATEC).

Not applicable for this survey

Regulation (EC) No 223/2009 on European statistics (recital 24 and Article 20(4)) of 11 March 2009 (OJ L 87, p. 164), stipulates the need to establish common principles and guidelines ensuring the confidentiality of data used for the production of European statistics and the access to those confidential data with due account for technical developments and the requirements of users in a democratic society.

At national level : 

Quantitative variables
For quantitative variables, the same policy as in SBS is applied.
Primary confidentiality, if:
• n units dominate the total turnover by at least k% ;
• the cell contains less than n units.

Secondary confidentiality if:
• protection is needed to address primary confidentiality or linked table risks (historical data, subdivisions, etc.) ;
• there is a link with the SBS suppression pattern.

Every cell linked to the variable "turnover" is confidential if the cell is confidential for the variable "turnover".
A link can be established as:
• direct link: the value of the variable of interest depends on the value of turnover (e.g. value-added, cost of sales, gross operating margin, etc.) ;
• indirect link: the value of turnover provides an idea of the dimension of the variable of interest (e.g. data calibrated with the variable « turnover », etc.).

In addition to the link inherited via the variable "turnover", variables relating to e-commerce turnover ("awsval", "axsval", etc.) are also checked for primary confidentiality individually.

Qualitative variables
For qualitative variables in frequency tables, a distinction has to be made between "Higher level confidentiality" and "Variable-specific rules":

Higher level confidentiality
Higher level confidentiality is checked only for cells broken down exclusively by variables which are available for every unit in the target population.
Higher level primary confidentiality is applicable if a cell in the target population contains less than n units for a given subdivision or breakdown ;
Higher level secondary confidentiality is applicable for any cells which are
• either a subdivision or a breakdown of the primary confidential cell, including cells with zero units ;
• or needed to protect primary confidential cells.
On each level of breakdown, all the cells have to be checked for both primary and secondary confidentiality. This rule is applied on the highest level for the total number of units, e.g. units per economic activity, units per size class, units per economic activity and size class, i.e. no additional variable of interest is used to subdivide the population (e.g. enterprises using computers, enterprises selling online, etc.). Each variable of interest inherits the higher-level suppression pattern.

Variable-specific rules
Each variable needs to be checked a priori for the following risk potentials:
• disclosure risk (none/low/high): the potential of identification of the variable ;
• sensitivity (none/low/high): the potential damage dealt by the information in case of the disclosure ;
• group sensitivity (none/low/high): the potential damage dealt by the information in case of group disclosure.

 
One should be aware of the consequences of a « none » risk, meaning that no confidentiality is applied at all.

Data are transmitted via eDamis (encrypted) and delivered to a secure environment where they are treated. Flags are added for confidentiality in case results must not be disclosed.

At national level :

Data are transmitted via eDamis (encrypted) and delivered to a secure environment where they are treated. National Statistical Institutes are requested to add flags for confidentiality in case results must not be disclosed.

Quantitative variables

The basis for any suppression pattern is the software package tau-Argus. However, the process also involves manual procedures, i.e. checking the tau-Argus output, comparing the historical data series and addressing linked table disclosure risks (see secondary confidentiality for further details).

The statistical disclosure control procedures are not performed for every variable individually but only for a primary shadow variable, i.e. "Turnover". If a given cell is confidential for that variable (no matter if primary or secondary), the same cell will be suppressed for all the other available quantitative variables. Variables relating to e-commerce turnover are also checked individually, an pass their flags on to related qualitative variables as well.

Primary confidentiality rules

 a) Sensitivity rule: We apply the (n,k)-dominance rule, i.e. a cell is suppressed if n units separately or jointly dominate the total value of a cell by at least k% .

 b) Minimum frequency rule: For any cells that are left after applying the sensitivity rule, a minimum frequency is applied. A cell is suppressed if there are less than n units in a given cell.

Secondary confidentiality rules

The secondary suppression is calculated by tau-Argus using the ‘Modular’ algorithm. Manual suppressions or cost adjustments are performed using the tau-Argus ‘apriori’ file facility.

 a) Secondary suppression within a table

 - A cell is suppressed for secondary confidentiality if n units dominate jointly or separately the confidential total value by at least k% ;

 - special attention is paid to the impact of singletons, a risk which is in most cases directly addressed by the tau-Argus Modular algorithm ;

 - tau-Argus is set to minimise the cost when determining the secondary suppressed cells. However, we also want to provide the user with useful data, whether it is in terms of interpretation and/or availability of time series. Consequently, the cost minimisation can be overridden for economic and/or historical reasons.

b) Secondary suppression due to linked tables disclosure risks

 - historical disclosure: in conformity with the SDC handbook, we ensure that no historical cell is compromised by disclosing the same cell for the current reference year. As long as there is any significant link with prior year data, a cell may not be disclosed for the current reference year.

 - Links to any other statistics: Turnover for the ICT survey is compared to the SBS preliminary series of the same reference year (e.g. T-1 for the survey carried out in year T).

Not available.

Not available.

Not available.

Annual

National dissemination of results

None available at the time of this report.

National dissemination of results

Selected results will also appear in the annual statistical yearbook, “Luxembourg in figures” and in “Un portrait chiffré des entreprises au Luxembourg”.

See detailed section 10.3.1.

Results for selected variables collected in the framework of this survey are available for all participating countries on Digital economy and society of Eurostat website.

At national level :

Results will be published on https://statistiques.public.lu/fr/donnees/themes/entreprises/sciences-technologies.html

Not applicable

Not requested

Not requested

Not available

Not requested

Not available

The European businesses statistics compliers’ manual for ICT usage and e-commerce in enterprises provides guidelines and standards for the implementation of the surveys. It is updated every year according to the changed contents of the model questionnaires.

At national level :

Not available.

European level :

At European level, the recommended use of the annual Eurostat model questionnaire aims at improving comparability of the results among the countries that conduct the survey on ICT usage and e-commerce in enterprises. Moreover, the Methodological Manual provides guidelines and clarifications for the implementation of the surveys.

National level :

The online questionnaire was maintained as the main survey vehicle with no paper questionnaires sent out after the previous survey showed sharp increases in online response rates using this procedure. Alternatively, a PDF version of the questionnaire was available for downloaded and print out.

Experiences gained with past questionnaires as well as from other domains (R&D, CIS) helped to improve the quality of the online questionnaire while trying to reduce the response burden as much as possible.

The 2023 data collection is the tenthconsecutive edition completely managed in-house. This allows for a better overview of the status of the survey, the problems encountered by respondents, as well as an improved follow-up with 2 reminders, as well as a planned 3rd reminder to high-impact enterprises sent as a registered letter.

European level : 

At European level, European Commission users (e.g. DG CNECT, DG GROW, DG JUST, DG REGIO, DG JRC) are the principal users of the data on ICT usage and e-commerce in enterprises and contribute in identifying/defining the topics to be covered. Hence, main users are consulted regularly (at hearings, task forces, ad hoc meetings) for their needs and are involved in the process of the development of the model questionnaires at a very early stage.

User needs are considered throughout the whole discussion process of the model questionnaires aiming at providing relevant statistical data for monitoring and benchmarking of European policies.

National level :

STATEC's research unit is routinely asked to provide feedback on new modules in the model questionnaire, as well as to indicate any variables used in research projects that might be missing from the model questionnaire. These additional variables, the pursued goal and their value-added are generally discussed between both units, in order to reach a consensus that allows to keep the questionnaire relatively short, while not unnecessarily reducing its usefulness to the research community.

European level : 

At European level, contacts within the Commission, the OECD and other stakeholders give a clear picture about the key users' satisfaction as to the following data quality aspects: accuracy and reliability of results, timeliness, satisfactory accessibility, clarity and comparability over time and between countries, completeness and relevance. Overall users have evaluated positively (good, very good) the data quality on the ICT usage and e-commerce in enterprises.

National level :

There is no survey led at the national level to assess the user's satisfaction on the data quality on the ICT usage and e-commerce in enterprises.

Detailed information is available in “ Annex I _ Completeness “ excel file - related to questionnaire, coverage, additional questions.

Not requested. 

Comments on reliability and representativeness of results and completeness of dataset

These comments reflect overall standard errors reported for the indicators and breakdowns in section 13.2.1 (Sampling error - indicators) and the rest of the breakdowns for national and European aggregates, as well as other accuracy measurements. The estimated standard error should not exceed 2pp for the overall proportions and should not exceed 5pp for the proportions related to the different subgroups of the population (for those NACE aggregates for the calculation and dissemination of national aggregates). If problems were found, these could have implications for future surveys (e.g. need to improve sampling design, to increase sample sizes, to increase the response rates).

More detailed information is available in “ Annex II. _ Accuracy “ excel file - related to European aggregates, comments on reliability and use of flag.

For calculation of the standard error see 13.2.1.1.

Standard error (for selected indicators and breakdowns)

Precision measures related to variability due to sampling, unit non-response (the size of the subset of respondents is smaller than the size of the original sample) and other (imputation for item non-response, calibration etc.) are not (yet) required from the Member states for all indicators.  Eurostat will make basic assumptions to compute these measures for all indicators produced (e.g. stratified random sampling assuming as strata the crossing of the variables “Number of employees and self-employed persons” and “Economic Activity” as it was defined in the 3 tables of section 18.1).

More detailed information is available in“ Sample and standard error tables 2023 “ excel file – worksheets starting with “Standard error".

Calculation of the standard error

Various methods can be used for the calculation of the standard error for an estimated proportion. The aim is to incorporate into the standard error the sampling variability but also variability due to unit non-response, item non-response (imputation), calibration etc. In case of census / take-all strata, the aim is to calculate the standard errors comprising the variability due to unit non-response and item non-response.

See detailed sections below.

See concept 18.1.1. A) Description of  frame population.

0.4%

Not requested

No measurement errors detected.

See detailed sections below.

See detailed sub-concepts below.

The following table contains the number of units (i.e. enterprises), by type of response to the survey and by the percentage of these values in relation to the gross sample size.

In order to reduce the non-response, 3 reminders are sent (with a registered letter to high-impact enterprises, on the 3rd reminder).

Not available

There are no questions with response rates lower than 90%.

Other comments relating to the item non-response

No processing errors were detected.

Not requested

See detailed section below.

Not applicable

European level : 

Data are to be delivered to Eurostat in the fourth quarter of the reference year (due date for the finalised dataset is 5th October). European results are released before the end of the survey year or in the beginning of the year following the survey year (T=reference year, T+0 for indicators referring to the current year, T+12 months for other indicators referring to the previous year e.g. e-commerce).

At national level : 

A first complete set of data has been sent to Eurostat on 05/10/2023. Final data, including a complete SDC pattern, has been sent to Eurostat on 27/10/2023.

 See detailed section below.

A first complete set of data has been sent to Eurostat on 05/10/2023. Final data, including a complete SDC pattern, has been sent to Eurostat on 27/10/2023.

The model questionnaire is generally used by the countries that conduct the survey on ICT usage and e-commerce in enterprises. Due to (small) differences in translation, in the used survey vehicle, in non-response treatment or different routing through the questionnaire, some results for some countries may be of reduced comparability. In these cases, notes are added in the data.

Detailed information on differences in the wording of the questions in the national questionnaires is available in “ Annex I _ Completeness “ excel file - related to questionnaire, coverage, additional questions.

Not applicable

See section below.

The length of comparable time series depends on the module and the variable considered within each survey module. Additional information is available in annexes attached to the European metadata.

No methodological changes occurred compared to the 2022 survey.

Not applicable

 Not applicable

Not applicable

Not applicable

Currently, no revision of the ICT data is foreseen.

ICT data are currently not revised.

 Not requested

A) Frame population description and distribution

For more information see concept 18.1.1.

B) Sampling design - Sampling method

Description of the sampling method used (e.g. stratified random sample, quota sampling, cluster sampling; one-stage or two-stage sampling) and information which variables were used to stratify, the categories of those variables, in particular for the NACE Rev. 2 categories related to the "possible calculation of European aggregates", and the final number of strata: 

This section includes a description of the sampling method used (e.g. stratified random sample, quota sampling, cluster sampling; one-stage or two-stage sampling) and information which variables were used to stratify, the categories of those variables, in particular for the NACE categories related to the "possible calculation of European aggregates", and the final number of strata.

 The method used for sampling was a stratified random sample, with varying sampling rates depending on size class:
• For the two size classes 50-249 and 250+, the sampling rate was 100% (i.e. a census);
• For the size class 10-49, the sampling rate was generally fixed at 60%, the only exceptions being F41_43, G45-46, G47, and I56 with a sampling rate of 35%. For small strata where the sample would have consisted of 3 units or less, a minimum of 3 units were sampled.

C) Gross sample distribution

More detailed information is available in “ Sample and standard error tables 2023  “ excel file (Worksheet: GROSS SAMPLE)

D) Net sample distribution

More detailed information is available in “ Sample and standard error tables 2023  “ excel file (Worksheet: NET SAMPLE)

A) Description of frame population

 B) Frame population distribution

More detailed information is available in “ Sample and standard error tables 2023  “ excel file (Worksheet: FRAME POPULATION)

Annual

See detailed sections below.

Possibility for respondents to choose between an online questionnaire or printing out and filling in a PDF version of the questionnaire.

Data has been validated by server-based EDIT validation.

Additional checks are performed on the microdata and aggregated levels.

Grossing-up procedures

To treat non-response, the initial sampling weight is first adjusted using the response rate for each stratum. Strata are defined by crossing the following size classes and NACE groupings.

In order to obtain reliable results for quantitative variables (that are in line with SBS totals) the corrected weights are calibrated using to the number of units, the total turnover and the total employment per stratum as auxiliary information.

Calibration is carried out in R, using the “calib” method of the “sampling” package with a “logit” distance function.

 Please note that due to the small number of observations leading to co-linearity problems, some strata cannot be calibrated over all size classes. For these strata, several size classes were combined.

 The strata used for calibration consist of the strata listed in 13.2.1.1.  (e).

The imputation rate ranges from 0,05% to 1.50% depending on the question.

Not applicable

Not applicable

Problems encountered and lessons to be learnt: