Hungarian Central Statistical Office

Education, Culture and Time Use Statistics Section

1525 Budapest, Pf. 51., Hungary

The Continuing Vocational Training Survey (CVTS) collects information on enterprises’ investment in the continuing vocational training of their staff. Continuing vocational training (CVT) refers to education or training measures or activities which are financed in total or at least partly by the enterprise (directly or indirectly). Part financing could include the use of work-time for the training activity as well as financing of training equipment.

Information available from the CVTS is grouped around the following topics:

- Provision of CVT courses and other forms of CVT (training/non-training enterprises)

- CVT strategies

- Participants in CVT courses

- Costs of CVT courses

- Time spent in CVT courses

- Characteristics of CVT courses

- Assessment of CVT activities

The CVTS also collects some information on initial vocational training (IVT).

For further information see the CVTS 6 legislation (http://ec.europa.eu/eurostat/web/education-and-training/legislation) and the CVTS 6 implementation manual (http://ec.europa.eu/eurostat/web/education-and-training/methodology).

The main groupings for enterprises are by economic activity (NACE), size group and training/non-training enterprises.

CVTS 6 covers all economic activities defined in sections B to N and R to S of NACE Rev. 2.

Definitions as well as the list of variables covered are available in the CVTS 6 implementation manual (http://ec.europa.eu/eurostat/web/education-and-training/methodology). 

Both the observational and the reporting unit for the CVTS 6 is the legal unit.

In Hungary, almost 95% of enterprises consist of only one legal unit, so for the vast majority of the population, the enterprise is identical to the legal unit.

Hence, the estimations calculated from the legal units' data are good approximations for the enterprise population.

CVTS 6 covers enterprises with 10 or more persons employed belonging to categories B to N and R to S of NACE Rev. 2.

The total number of enterprises in the target population: 36 567, see details in table 13.3.1 in annex "HU - QR tables CVTS 2020 (excel)".

Variable A2tot (persons employed) refers to 31 December 2020.

Hungary, all regions of the country are included.

Reference years for which CVTS data are available at national level: 1999, 2005, 2010, 2015, 2020

Not applicable.

Number, EUR.

The reference year for CVTS 6 is the calendar year 2020.

At European level

Basic legal act: Regulation (EC) No 1552/2005 of the European Parliament and the Council

Implementing act: Commission Regulation (EU) No 1153/2014, amending Commission Regulation (EC) No 198/2006

At national level

Act CLV of 2016 on Official Statistics

Government Decree 184/2017 (VII.15.) on implementing Act CLV on Official Statistics

Government Decree 388/2017 (XII.15.) on the Mandatory Reporting of the National Statistical Data Collection Program

Not applicable.

In the process of statistical data collection, processing and analysis and dissemination of statistical information, Hungarian Central Statistical Office fully guarantees the confidentiality of the data submitted by respondents (households, enterprises, institutions, organisations and other statistical units), as defined in the Confidentiality policy of the Hungarian Central Statistical Office. See https://www.ksh.hu/docs/bemutatkozas/eng/avpol_web_eng.pdf.

National legislation:

Government Decree 288/2009 (15 December)

Act CLV of 2016 on Official Statistics https://www.ksh.hu/docs/bemutatkozas/eng/act_no_clv_of_2016_on_official_statistics.pdf

Government Decree 184/2017 (VII. 5.) on implementing Act CLV of 2016 on Official Statistics

HCSO ensures confidentiality for all the data reported by data providers and the exclusive use of the data for statistical purposes. We disseminate only aggregated data in full compliance with the rules of confidentiality. Individual data, as well as aggregated data consisting of fewer than 3 enterprises are regarded as confidential and therefore not published. In tabular data the cell suppression statistical disclosure control (SDC) method consists of primary and complementary (secondary) suppression or reductions of detail. For anonymised micro data, top and bottom coding, removing and global recording are mainly used. Researchers have access to de-identified data sets and to anonymised micro data for scientific purposes with appropriate legal and methodological guaranties in place. As for the employees, they can work with datasets in their competence with registered and controlled access rights. For details see the information on confidentiality for data providers on the website of HCSO.

Preliminary and final statistical information is published in accordance with an approved release calendar. Release calendar is publicly accessible on the website of HCSO.

https://www.ksh.hu/katalogus/#/tajekoztatasi-adatbazis/naptar/2022/10/en 

The data is disseminated to all users through the Dissemination database and Summary Tables on the website of the HCSO. The database and the summary tables are available in Hungarian and English. Users are informed about the date of release in the public release calendar.

Every 5 years.

There was no news release.

Summary Table (in English) https://www.ksh.hu/stadat_files/okt/en/okt0045.html 

Data warehouse (in English): https://statinfo.ksh.hu/Statinfo/themeSelector.jsp?&lang=en Education/Lifelong learning 

Not applicable.

In order to support scientific research, HCSO opens up data files for those requesting data/researchers; prior to this carries out an accreditation study. See also: https://www.ksh.hu/data_access_researcher_accreditation

For anonymised micro data, top and bottom coding, removing and global recording are mainly used. Researchers have access to de-identified data sets and to anonymised micro data for scientific purposes with appropriate legal and methodological guaranties in place. As for the employees, they can work with datasets in their competence with registered and controlled access rights. For details see the information on confidentiality for data providers on the website of HCSO.

As the purpose of statistical dissemination activity of HCSO is to disclose as many data of as good quality as possible to as high a number of users as possible, via the channel and in the format most suitable for them, summary tables and the on-line database are free available on the website of HCSO. Free of charge data releases and services:

a. HCSO fulfils all public interest related data/data compilation requests which: could be realised based on data made public prior to the request or could be produced based on non-public data available in the HCSO information system, indifferent of the applied methodological - IT solutions, in 1 hour.

b. The use of the constantly available files at the Safe Centre, previously compiled for research purposes, as well as their additional data protection verification are free of charge.

c. HCSO does not apply fees when fulfilling customer requests in relation with EU data and information.

d. HCSO offers free of charge assistance for customers in retrieving other – data collection related, methodological, technical – information.

Data of CVTS are used by other surveys (e.g. business income data collections) within the HCSO to validate their own data.

Compensation fee subjected data releases and services:

HCSO is charging fees for extra services related to individual requirement-based data compilations, services; the expert fee is HUF 7,500 / work hour + VAT.

In case of fee-related data releases and services the HCSO co-worker sends the customer a price quote in 5 workdays, defining precisely the topic of the requested service, the compensation fee and the delivery deadline.

Not applicable.

Dissemination of documentation on methodology and sources used in preparing lifelong learning statistics: the description is available via the website of the Hungarian Central Statistical Office for concepts and definitions; classifications, data production methods, data quality, data sources:

https://www.ksh.hu/docs/eng/modsz/okt_meth.html

https://www.ksh.hu/apps/meta.objektum?p_lang=EN&p_menu_id=110&p_ot_id=100&p_obj_id=ACCA

Not applicable.

User-oriented quality reports on statistical domains are prepared in the framework of methodological documentation and are published as meta-information on the HCSO website: http://www.ksh.hu/apps/meta.main?p_lang=EN.

Especially for CVTS: https://www.ksh.hu/apps/meta.objektum?p_lang=EN&p_menu_id=110&p_almenu_id=105&p_ot_id=100&p_obj_id=ACCA

An internal HCSO regulation is in place regarding the preparation of producer-oriented quality reports for each statistical domain on a yearly basis.

The HCSO Quality Policy lays out the principles and commitments related to the quality of statistics. The document is consistent with the goals set out in the Mission and Vision statements and with the principles of the European Statistics Code of Practice and is publicly available on the HCSO website.

The European Statistics Code of Practice is available on the website of the HCSO. Also, HCSO together with the member-organisations of the Hungarian Official Statistical Service created a National Statistics Code of Practice based on the European Statistics Code of Practice. (Currently the National Statistics Code of Practice is available only in Hungarian on the HCSO website.)

HCSO Quality Policy: http://www.ksh.hu/docs/bemutatkozas/eng/minosegi_iranyelvek_eng.pdf

National Statistics Code of Practice: https://www.ksh.hu/docs/bemutatkozas/hun/gyakorlati_kodex.pdf

European Statistics Code of Practice: https://ec.europa.eu/eurostat/documents/4031688/8971242/KS-02-18-142-EN-N.pdf/e7f85f07-91db-4312-8118-f729c75878c7

The HCSO has developed quality guidelines to ensure the quality of the statistical processes. The document has been in place since 2007 (1st revision in 2009, 2nd revision in 2014 and 3rd revision is currently ongoing). The latest version (2014) is available on the HCSO website. As HCSO has been carrying out its core activity as a member of the European Statistical System (ESS) since 2004, organisational improvement in quality matters is also based on EU pillars. It relies on the quality guidelines of international statistical institutes with a proven track record and excellent results of quality management (e.g. Statistics Canada, Statistics Finland, the US Federal Statistical Agencies, UK Office for National Statistics and the Italian National Institute of Statistics) and on the Code of Practice adopted in 2005 and revised in 2011 by the EU Statistical Programme Committee.

For more information see HCSO's website: https://www.ksh.hu/docs/bemutatkozas/eng/minosegi_iranyelvek_eng.pdf

The Hungarian survey includes all the questions of the European Standard Questionnaire to meet the demands of Eurostat. In response to the epidemic situation, Hungary added one question and one item to the national questionnaire in the sixth wave, based on the recommendation of Eurostat. Also reacting on the OECD's needs, some questions of the PIAAC Employer Module were built in into the wave 6.

As for relevance, the main weakness of CVTS in Hungary is that no user satisfaction survey was undertaken.

As regards accuracy and reliability, external experts of adult education were involved in the survey preparation process of the first wave in Hungary.

Based on the experiences gained from field work, the wording of the question regarding paid working hours spent on CVT courses seems to be necessary.

The overall accuracy of the key estimates is considered good as the sampling errors (measured by the CV) stay below a reasonable level for the majority of the size and NACE categories. The main sources of non-sampling error are the non-response (measured by the unit non-response rate) and the coverage errors (measured by the over-coverage rate). Both of them have moderate values, hence their effect on the estimates and on the bias is considered small.

Timelines and punctuality: data files were sent according to the timetable set by Eurostat.

Geographical comparability can be considered complete, as concepts and definitions used in the Hungarian questionnaire comply with the Eurostat standards.

The Hungarian questionnaire of CVTS 6 includes some extra variables and in case of the question regarding the reasons for non-training one extra item, see table 15.1 in annex "HU - QR tables CVTS 2020 (excel)".

The survey is organized on the basis of the methodology elaborated by Eurostat, so the international data provision obligation as described in the CVTS legislation can be fulfilled completely. This survey supports the preparation of educational policy decisions, educational and vocational training measure and the design of middle- and long-term workforce forecasts. Potential domestic users (Ministry responsible for human resources, for education, for labour force, and for vocational training, Ministry of Finance and research institutes) take part in finalizing CVTS questionnaire, this ensures that every important user need can be satisfied. The most important users are government organizations (ministries), the Hungarian Academy of Science, higher educational institutions, enterprises, journalists, market researchers, researchers dealing with vocational training and workforce forecast, university students writing their thesis.

The Hungarian survey includes all the questions of the European Standard Questionnaire to meet the demands of Eurostat. In response to the epidemic situation, Hungary added one question and one item to the national questionnaire in the sixth wave, based on the recommendation of Eurostat. Also reacting on the OECD's needs, some questions of the PIAAC Employer Module were built in into the wave 6.

No user satisfaction survey has been done so far.

There is no deviation from CVTS 6 legislation in terms of covered NACE sectors, enterprise size groups and variables.

Not applicable.

The overall accuracy of the key estimates is considered good as the sampling error (measured by the CV values) is quite low for the whole population and stay below a reasonable level for the main size and NACE breakdowns.

The main sources of non-sampling error are the non-response (measured by the unit non-response rate) and the coverage errors (measured by the over-coverage rate). Both of them have moderate values:

• the unweighted non-response rate ranges between 7.1% and 23.7% for the different size and NACE categories and has a value of 16.1% for the whole sample
• the over coverage rate is 7.6%

Since 75% of non-respondents, and 90% of data providers subject to over-coverage belong to the smallest size category, non-sampling error and the resulting bias mostly affect the smallest units of the population. Hence the effect of non-sampling errors on the estimates and on the bias is negligible in case of estimates measuring the different costs of CVT courses, and is considered small in case of other total and ratio type estimates. To minimise unit non-response those enterprises that failed to return the questionnaire by the deadline, written reminders are mailed and are reminded by phone about the expiry of the deadline.

The sample selection is done by using stratified random sampling. The dimensions used for stratification are 1) NACE Rev.2 activity groups, 2) size of enterprises (according to the number of persons employed), 3) regional breakdown. The sample size and the distribution of the sample among the above defined strata are determined by the well-known Neyman-type optimal allocation scheme.

For estimation and variance calculation, the Horwitz-Thompson formulas are used.

The non-respondent units are assumed to resemble those who have responded to the survey and are treated as non-selected units. For this, the weights used for estimation are adjusted: the design weight Nh / nh is replaced by Nh / mh where Nh is the size of stratum h, nh is the sample size in stratum h and mh is the number of respondents in stratum h.

Outlier detection is not applied for CVTS 6.

See table 13.2.1 "Sampling errors - indicators" in annex "HU - QR tables CVTS 2020 (excel)".

Non-sampling errors are covered by items 13.3.1 - 13.3.5 below.

At the Hungarian Central Statistical Office, the Business Register is used to create the sampling frame for the CVTS 6. The Business Register is updated at the beginning of November each year with the most recent information available at that moment. The sampling selection takes place shortly after the updating of the Business Register, in the middle of November. Since the information used to update the Register comes from many sources with different timeliness, the sampling frame does not cover perfectly the target population. This causes over- and under-coverage. Furthermore, the timing of the field work was set too early, some data were not available for the businesses that time. This caused under-coverage. The over-coverage is measured by the rate of enterprises that reported less than 10 as the total number of employees. Since the Business Register is updated once in a year, and there is no other reliable source, under-coverage and the resulting bias cannot be measured. Multiple listing is very rare.

See table 13.3.1 "Coverage error" in annex "HU - QR tables CVTS 2020 (excel)".

See table 13.3.1.1 "Over-coverage - rate" in annex "HU - QR tables CVTS 2020 (excel)". 

Not applicable.

There were two main sources of measurement error as follows.

In case of questions regarding the amount of money, respondents sometimes oversighted that the sums were asked in thousand HUF, not in HUF. These data were cross-checked based on external data sources as LFS and tax return of companies. During the editing phase, consistency checks were performed.

The concept of paid working time spent on internal CVT seemed to be hard to understand, because the majority of respondents considered it as the part of the working hours. A similar problem occurred in case of internal cost of CVT courses (it was observed in smaller degree). To minimise these problems, the explanatory notes contained guidance, and the staff of the help centre also got instructions on it. In the wave 6, in case of paid working time spent on CVT courses and direct cost of CVT courses, automatic warning popped up in the questionnaire, if the number of hours/costs was left empty. In case of insufficient reasoning for value 0, businesses were asked to give explanation for it. If the respondents were not eligible, missing data were imputed (in the former waves there was no imputation).

Compared to the previous wave, the level of unit non-response has increased (16.1% vs. 13.6%); the growth was the highest among the medium-sized enterprises. The drop in response rate is partly due to the COVID19 pandemic. To reduce nonresponse, at the beginning of the reference year, an awareness-raising e-mail was sent to the large companies. During the field work, reminders were sent by e-mail to non-respondents two times. After sending the reminders, contact attempts were made by phone. At the estimation stage, the data received was weighted according to the sample.

Variables with higher item nonresponse rate were the paid working hours spent on CVT courses, especially in case of the internal ones. Furthermore, costs of internal courses, particularly the labour costs of internal trainers, were more subject to item non response. The main reason for the lack of data was that these kinds of information were not registered by the businesses. To reduce nonresponse, pop-up warnings were built-in the questionnaire and was asked about the reason for the lack of data. In case of inefficient response, the enterprise was contacted by phone. When the enterprise could not or has not given information, the missing data was imputed primarily using LFS data, if not available, using a donor enterprise. Donor enterprises were selected by the number of persons employed and by principal economic activity. Another variable was the amount CVT contribution, because this data was not available yet during the field work. This variable was imputed using LFS and tax return data of companies.

See table 13.3.3.1 "Unit non-response - rate" in annex "HU - QR tables CVTS 2020 (excel)".

See table 13.3.3.2 "Item non-response - rate" in annex "HU - QR tables CVTS 2020 (excel)".

The main source of processing errors was the coding of the national variables into the Eurostat variables. This issue was treated by double checking of the transcoded variables. No manual coding was used.

No model was used to handle errors or for estimation.

National data deliveries to the ESS (average for CVTS 4, CVTS 5 and CVTS 6): preliminary data 512 days, final data 520 days.

TP1 Time lag - first results: The number of days (or weeks or months) from the last day of the reference period to the day of publication of first results. = 390 days.

TP2 time lag - final results: The number of days (or weeks or months) from the last day of the reference period to the day of publication of complete and final results = 650 days.

All national data were delivered on time to Eurostat; countries should transmit data to Eurostat no later than 18 months after the end of the reference year. 

TP3 indicator: number of days separating a previously announced date of publication and the actual date. In case of CVTS, TP3 = 260 days.

See table 14.2 "Project phases - dates" in annex "HU - QR tables CVTS 2020 (excel)".

Not applicable.

National concepts and definitions are in line with the legislation and in the implementation manual, so the geographical comparability is complete.

See table 15.1 "Comparability - geographical" in annex "HU - QR tables CVTS 2020 (excel)".

Some additional variables/information related to COVID-19 were collected, see table also 15.1.

Not applicable.

There are changes in the national questionnaire CVTS 6 that have an impact on the comparability with CVTS 5.

• Some questions of the PIAAC Employer Module were added to the questionnaire, after Section A.
• Added question at the end of Section D, regarding the effect of COVID-19 pandemic on the provision of CVT courses or other forms of CVT.
• In case of non-training enterprises, one additional item regarding COVID-19 pandemic was added to the list of reasons for non-training, before other (question E1).

For details see table 15.1 "Comparability - geographical" in annex "HU - QR tables CVTS 2020 (excel)", section 'CVTS 6 variables'.

Improvements in the questionnaire:

• Paid working time spent on CVT courses (C3I, C3E, which have affected C3TOT and PAC): In this wave, automatic warning popped up in the questionnaire, if the number of hours was left empty. Missing data were imputed (in the former wave there was no imputation).
• Direct cost of CVT courses (C7sub, which affects C7TOT): automatic warning plus imputation.

The improvements above might have increased the response rate in case in these questions.

The modification of the questionnaire might have influenced the comparability over time only in case of the question regarding the reasons of non-training.

See also table 15.2 "Comparability - over time" in annex "HU - QR tables CVTS 2020 (excel)".

Not applicable.

See table 15.3 "Coherence - cross-domain" in annex "HU - QR tables CVTS 2020 (excel)".

Not applicable.

Not applicable.

Not applicable.

Summary of costs and burden for production of statistical data was calculated based on the working hours, which the enterprises needed for filling in the questionnaire. The sum of the working hours for production of data was 7 778 hours; the average was 1.1 hour. Calculating with the average total labour cost per hour, it made 1 917 871 Euros. The operational costs of the CVTS project is not available for the HCSO.

ICT was used efficiently: data collection was conducted via computer assisted web-interview (self-administered), with built-in data checking.

The range of data collected was broader than necessary, because certain questions of the PIAAC Employer module and some COVID19-related questions were embedded in the standard CVTS questionnaire. Furthermore, in case of some questions, more detailed data were collected in Hungary. HCSO plans to review on these data for the next wave.

Usage of administrative data was not possible. Data from other survey resources (tax return data, LFS) was used for validation. Usage of data from other survey resources is limited; in case of LFS, because of the non-respondent businesses.

The data sought was only partially available for businesses from their account. The fieldwork should be shifted to a later date, because in the time of fieldwork the total labour costs was not available. Some data was not gathered by the businesses, like the number of hours spent on CVT (especially in case of internal courses), CVT costs especially the labour costs of internal trainers and the costs of training centre, training premises or specific training rooms of the enterprise.

Best estimates and approximations are accepted when exact details are not readily available.

Reporting burden on individual respondents is limited to the extent possible by minimizing the overlap with other surveys: HCSO will review the possibility of using some data from LFS and tax return statistics.

Not applicable.

Not applicable.

Not applicable.

The CVTS 6 is based on a sample survey at the Hungarian Central Statistical Office. The sample is selected by stratified random sampling, where the strata are compiled as the combination of the enterprises' size, NACE category and region code.

The Gross sample size is 8607, whereas the net sample size is 6715.

See also table 18.1 "Source data and data collection" in annex "HU - QR tables CVTS 2020 (excel)".

Every 5 years.

The method used for data gathering was CAWI (computer assisted web-interview, self-administered). In case of large enterprises (businesses with 250 or more persons employed), a comprehensive sample was used, while in case of medium-sized and small-sized ones a representative sample was applied. The population of interest for CVTS 6 are the enterprises with 10 or more persons employed belonging to the 20 NACE Rev. 2 categories [B, C10-C12, C13-C15, C17-C18, C19-C23, C24-C25, C26-C28+C33, C29-C30, C16+C31-C32, D-E, F, G45, G46, G47, H, I, J, K64-K65, K66, L+M+N+R+S]. As data collecting method was CAWI, there was no interviewer, the data gathering staff (persons who contact the enterprises to ask them to report) got a training on the survey. The nonresponse was monitored via an electronic feedback system. Questionnaire was available only in Hungarian. One company demanded an English version of questionnaire, so we translated the Hungarian to English.

See also table 18.1 "Source data and data collection" in annex "HU - QR tables CVTS 2020 (excel)".

Monitoring of population coverage and response rates was continuous during the field work; non-respondents were contacted by e-mail first, than by phone, where it was possible. Large companies (number of persons employed over 249) were dealt with extra care. Feedbacks and questions of respondents were collected.

The first step of checking the source data was the CAWI questionnaire with its built-in checks and warnings. Where the reason given in the questionnaire was not sufficient, the respondent was contacted by phone.

Source data were cross-checked both with internal (LFS) and external (tax return data, database of tenders) resources. Outliers and inconsistencies were detected using rate calculations (e.g. total labour cost per number of persons employed). Micro data was edited based on these procedures. Aggregated output data was compared with previous wave.

The CVTS 6 is based on a sample survey, the sample is selected by stratified random sampling, where the strata are based on the population units' size, NACE category and region code.

Both the observational and the reporting unit is the legal unit. In Hungary, almost 95% of enterprises consist of only one legal unit, so for the vast majority of the population, the enterprise is identical to the legal unit. Hence, the estimations calculated from the legal units' data are good approximations for the enterprise population.

Variables with high item nonresponse rate were imputed. The imputation is based on administrative data (LFS and tax data), or if no other data sources is available, donor imputation is used.

The non-respondent units are assumed to resemble those who have responded to the survey and are treated as non-selected units. For this, the weights used for estimation are adjusted: the design weight Nh / nh is replaced by Nh / mh where Nh is the size of stratum h, nh is the sample size in stratum h and mh is the number of respondents in stratum h in the net sample. Both item- and unit non-response increase the variability of the estimates and introduce bias. As the non-response rates are not significantly high, their effects on the estimates are considered small.

Weight calibration is not used as there is no other reliable source that has a strong correlation with the key variables of the CVTS 6.

The final weights are the non-response corrected design weights for the net sample.

See table 18.5.1 "Imputation - rate" in annex "HU - QR tables CVTS 2020 (excel)".

Not applicable.

Not applicable.

No comment.