Understanding Skill Gaps in Firms

Annex A. Design, development and implementation of the PIAAC Employer Module

Rationale for the PIAAC Employer Module

The Programme for the International Assessment of Adult Competencies (PIAAC) is a programme of assessment and analysis of adult skills. The best-known aspect of the programme is the Survey of Adult Skills, a household survey that assesses adults’ proficiency in key information processing skills – literacy, numeracy and problem-solving. It also collects information and data on how adults use their skills at home, at work and in the wider community. Data collection for the 2023 Survey of Adult Skills took place between 2022 and 2023, with the participation of more than 30 countries.

The PIAAC Employer Survey Module on Skill Gaps (hereafter referred to as the ‘Employer Module’, ‘the Module’ or ‘the Survey’) is an enterprise survey that complements the data collected by the Survey of Adult Skills. The Employer Module aims to understand employers’ perspectives on skill gaps at the enterprise level, the skills that employers believe they will need in the future, and the most common human resource practices that employers use to ensure access to these skills. Compared to existing national employer skills surveys, the Module collects information in a consistent way across OECD countries, using a common standardised questionnaire, in alignment with the concepts used in the 2023 Survey of Adult Skills.

Despite their importance for individual, business, and economic outcomes, skill gaps have received limited attention by the research and policy community compared to other forms of skill mismatch and shortages. Skill gaps occur when the skills of employees exceed or fall short of the requirements of their jobs under current market conditions. This includes qualification mismatch, where an employee’s qualifications do not meet the requirements of the job, and field-of-study mismatch, where an employee’s qualifications are in a different field to that required by the job. These skill gaps have significant economic consequences, including lower earnings and lower job satisfaction at the individual level, as well as lower productivity and slower growth at the level of the economy.

Relationship between Employer Module and the Survey of Adult Skills

The Survey of Adult Skills and the Employer Module are conceptually linked and provide complementary perspectives on skills mismatches from both the employer and employee perspectives, which enriches the evidence from each. The alignment between both surveys is achieved through the consistent use of key concepts and terminology related to skills and training in both questionnaires. There is a deliberate correspondence in the types of skills and macro trends measured (‘core questions’), as well as in the way data on key enterprise characteristics such as location, size of enterprise and economic activity are collected (‘essential questions’). In addition, ‘desirable’ questions extend the data collection in directions not covered by the other survey. For example, data on work organisation of employees from the 2023 Survey of Adult Skills can be interpreted alongside information on internal organisational structures, human resource practices and skill gaps from the Employer Module. For a more detailed description of the relationship and the mapping of survey items in both questionnaires see Marcolin and Quintini (2023[1]).

It is important to note that the data from the two surveys are not statistically linked at the individual level due to their independent administration. The Survey of Adult Skills has not specifically been administered on workers in enterprises included in the Employer Module or vice versa. Instead, a statistical link can be achieved through an intermediate level approach. This involves merging data from the Employer Module and the Survey of Adult Skills at a pre‑determined level of aggregation, typically by industry or enterprise size. By applying appropriate weights, data from both sources can be used together in a representative way for further analysis. Future iterations of the module may include modifications to facilitate the implementation of the surveys in a way that allows for individual linkage. This would allow a more detailed examination of how employers’ policies on training, staffing and addressing skill gaps directly affect employees’ skills, job satisfaction and wages.

Development and implementation

The development of the PIAAC Employer Module was a multi-year process, which included the development of the questionnaire, determining a suitable form of administration of the survey across countries and supporting participating countries in the implementation of the survey. So far, the module has been administered in five European countries: Hungary, Italy, the Netherlands, Portugal, and the Slovak Republic.

Questionnaire development

The questionnaire of the PIAAC Employer Module was developed by a group of international experts (see Annex D) between 2016 and 2018. During that time, the questionnaire underwent several iterations, with delegates of PIAAC participating countries providing valuable inputs during official OECD meetings between 2016 and 2020. In addition, a dedicated workshop with Eurostat delegates in December 2018 contributed to the refinement of the questionnaire.

The aim of the questionnaire development was to formulate questions that were easy to understand, that captured the underlying concepts accurately and that were conducive to cross-country comparability. Each question was designed to address a single concept whenever possible, and response options and scales were streamlined to simplify the questionnaire and minimise completion time.

To achieve this goal, the questionnaire drew on survey items from tried-and-tested existing surveys that had a track record of multiple administrations. In the final version of the questionnaire, four of the five ‘core’ questions on skill gaps were inspired by established national and international surveys, including the UK Employer Skills Survey, the Eurofound European Company Survey, the Cedefop Employer Survey Pilot, the Australian NCVER Survey of Employer Use and Views of the VET System and the MEADOW framework. In addition, some questions were aligned with the background questionnaire used in the second cycle of the PIAAC Adult Skills Survey administered to households.

The final questionnaire design is presented in Box A A.1 and the full questionnaire is available in Annex C.

Box A A.1. Questionnaire design

The module’s questionnaire has two parts: 1) core items, focusing on how enterprises encounter and address skill gaps, and 2) optional – or desirable – items, which enable countries to collect additional information in line with their policy priorities:

• The core consists of five items on skill gaps, the actions put in place by the enterprise to address them and the changes at enterprise level that took place recently (Q1 to Q5), and five ‘essential’ questions on the background of the enterprise (Q6 to Q10). These background questions on sector, location, size, and age are needed to contextualise the module and link it to other data sources.

• The optional/desirable questions (QA1 to QE2) explore other policy-relevant characteristics of the enterprise, such as its ability to innovate, its work and organisational practices, its difficulties in recruiting and retaining staff and its employees’ participation in training.

The full questionnaire can be found in Annex C.

Translation

Careful translation to maintain conceptual fidelity was crucial to ensure the comparability of results across countries. The questionnaire was translated into national languages by each participating country, with the support of the OECD for consistency and review. The translation benefited from synergies with the background questionnaire of the PIAAC household survey and translations of the items inspired by other existing surveys.

The translation process typically involved different government ministries and agencies. In Italy, a first draft of the translation was produced by ANPAL (National Agency for Active Labour Market Policies) and then validated by INAPP (National Institute for Public Policy Analysis) and ISTAT (National Statistical Institute). In the Netherlands, the translation of the questionnaire was carried out by CBS (National Statistical Institute) and cross-validated both internally and by the Ministry of Education. In Portugal, the translation was carried out by the same service responsible for the PIAAC household study, was validated by the PIAAC national co‑ordination group and benefited from the comments of GEP/MTSSS (Office of Strategy and Planning of the Ministry of Labour, Solidarity and Social Security), which administers the CVTS on behalf of INE (National Statistical Institute). In the Slovak Republic, the questionnaire was translated by the Ministry of Education and sent to the National Statistical Office, which administers the CVTS. In Hungary, the translation of the questionnaire was carried out by the Hungarian Central Statistical Office (HCSO), which was also responsible for the fieldwork of the PIAAC household study.

Host survey

While the module was developed as a stand-alone survey, the first iteration of the survey was administered as an add-on to the 2020 wave of the European Continuing Vocational Training Survey (CVTS 6). The CVTS is an enterprise survey and a suitable host survey for the Employer Module as there is some overlap and complementarity between the two questionnaires.

Administering the Employer Module together with the CVTS offered considerable cost savings and provided the opportunity to include a wider range of data collected through the main CVTS data collection in the analysis. The advantages and disadvantages of using the CVTS as a vehicle for the OECD Employer Module were discussed in detail at the meeting of experts on continuing vocational training statistics in Brussels on 6‑7 December 2018 (see (Marcolin and Quintini, 2023[1]) and European Commission/Eurostat (2019[2])). The use of the CVTS as a host survey impacted several parameters for the data collection of the module, including the:

• reference area: whole country,

• sector coverage: NACE Rev. 2 categories B, C, D, E, F, G, H, I, J, K, L, M, N, R, S,

• unit of observation: legal units (enterprises),

• enterprise type and size: private enterprises, with 10 persons employed or more,

• time of data collection: 2021 and 2022,

• main reference period: 2020,

• unit of measure: all countries used numbers, percentages, and euros, where relevant.

The reference period being 2020 meant that employers responded to the questionnaire based on their firm’s performance for 2020, making 2020 the baseline year for all future iterations of the Module.

Box A A.2. The pros and cons of the CVTS as the host survey of the Employer Module

The main advantages of administering the Module as an add-on to the CVTS were:

• Consistency in theme and content: The CVTS was aligned with the theme and content of the PIAAC Employer Module, ensuring a seamless integration into the existing survey framework.

• Efficiency: Using CVTS as a host survey streamlined the process, making it less resource intensive and more efficient for participating countries due to the overlap in variables. Implementing the employer module within CVTS reduced the administrative burden on countries, as they can seamlessly integrate the additional questions.

• Adherence to high survey standards: Data collection, validation and transmission have been carried out according to Eurostat standards and in compliance with EU legislation, with significant quality assurance and quality control.

The main disadvantages of administering the Module as an add-on to the CVTS were:

• Length: As the CVTS was already a demanding survey for respondents, many countries participating in the Employer Module chose to implement only a limited number of desirable questions beyond the core items.

• Coverage: The implementation of the Employer Module together with the CVTS imposed significant limitations on several parameters of data collection. For example, the CVTS targets enterprises with 10 or more employees in specific NACE sectors, which limited the scope of the PIAAC employer module.

• Legal considerations: Some countries had established legal frameworks for the CVTS and the addition of OECD questions raised legal and compliance issues.

Source: European Commission/Eurostat (2019[2]), The meeting of experts on continuing vocational training statistics on 6 and 7 December 2018. Conclusions

It should be noted that Portugal was the only country that did not implement the OECD Employer Module at the same time as the CVTS. Data collection for the module started nine months after data collection for the CVTS. This is reflected in the fact that the overall unit response rate of the CVTS was 60%, while it was lower at 52% for the Employer Module. Despite this difference, the Employer Module was administered to all enterprises that were in the CVTS sample and all other parameters for the data collection remained aligned with the CVTS.

Sampling approach

Eurostat issues strict recommendations on the sampling strategy and methods for the CVTS, which countries all adhere to, and which ensures a consistent and harmonised approach. These include firstly the selection of enterprises from business registers using random sampling techniques. Secondly, to capture the diverse business landscape in each country, countries use a stratified sampling approach. This involves categorising enterprises according to size and economic activity in line with Eurostat guidelines. Finally, Eurostat guidelines play a key role in determining the gross sample size. Factors such as the expected proportion of enterprises involved in vocational training are considered, in line with the overall objective of obtaining a statistically significant and robust data set.

Table A A.1. below provides an overview of the target population, gross sample size and actual achieved sample for those countries where data is available.

Table A A.1. Statistical target population, sample sizes and final number of respondents

	Target population	Gross sample size	Achieved sample size
Hungary	36 567	8 607	6 715
Italy	207 295	31 622	17 617*
The Netherlands	58 249	5 370	4 450
Portugal	41 566	6 583	3 419
Slovak Republic	14 649	2 612	2 387

Note: The target population included all companies in the relevant sectors that employed more than 10 people in the year under review. The gross sample size is number of enterprises initially selected from the sampling frame.

*The sampling plan of the Italian CVTS 6 was implemented by considering the statistical unit ‘Enterprise’, that is the statistical unit for the observation and analysis implemented within the Italian system of business registers and economic accounts, according to the Council Regulation (EEC) No 696/93 of 15 March 1993, which applies on the Structural Business Statistics (SBS). The Enterprise is namely “the smallest combination of legal units that is an organisational unit producing goods or services, which benefits from a certain degree of autonomy of decision-making, especially for the allocation of its current resources”. Consequently, the enterprise may correspond to a single legal unit or a group of legal units under common control. The reference population of the Italian CVTS 6 carried out consisted of Enterprises active and residing in Italy, and included in the ENT Business Register, according to CVTS coverage criteria (10 or more persons employed in NACE Rev. 2 sections).

Source: Meta-data submitted by countries; analysis of the micro-data.

Mode of administration

Data collection, recording, checking, and processing were carried out by the statistical offices of the participating countries. For the CVTS, the data collection method was decided at national level, i.e. countries conducted the survey using the method that was best suited to achieve a sufficiently high response rate. While in previous waves countries used multi-mode data collection methods to ensure the participation of enterprises in the survey, in 2020 Computer Assisted Web Interviewing (CAWI) was the main mode used.

CAWI is an easily accessible and cost-effective mode of data collection as it simplifies the routing of questions and question blocks according to the choices made by the enterprise during the survey and minimises errors in data entry and post-processing. CAWI also allows respondents to answer at their own pace, possibly consulting additional sources before answering and returning to previous sections of the questionnaire to complete unanswered questions or amend previous entries.

Countries participating in the employer module were generally positive about the mode of administration.

Optional questions

The module was administered in five European countries: Hungary, Italy, the Netherlands, Portugal, and the Slovak Republic. The questionnaire administered for the module included “core” and “essential” questions, as well as the “desirable” items on training strategies (items QB1 to QB4, 5, Annex C), which are included in the official CVTS questionnaire. Each participating country extended the questionnaire of the module with further “desirable” questions according to the country’s policy priorities (Table A A.2. ).

Table A A.2. Optional and desirable questions selected by participating countries

section	section denomination	question number	Hungary	Italy	Netherlands	Portugal	Slovak Republic
2A	optional to core	QA1	x	x		x
		QA2		x	x	x
		QA3		x	x	x
		QA4
		QA5
2B	desirables on training	QB1	x	x	x	x	x
		QB2	x	x	x	x	x
		QB3	x	x	x	x	x
		QB4	x	x	x	x	x
2C	desirables on HR	QC1				x
		QC2				x
2D	desirables on recruitment	QD1				x
		QD2				x
		QD3			x
		QD4			x	x	x
		QD5				x
2E	desirables on firm structure	QE1		x	x	x
		QE2				x

Note: For further information and the exact wording of the questions, please see Annex C.

Source: Information provided by countries and micro-data.

Changes to questions and their order

The order of the questions in the module was carefully considered as it could affect respondents’ understanding, speed of response and the overall coherence of the survey. While it was recommended that participating countries maintained the suggested order of questions within their surveys, most countries chose to place the questions of the employer module where they felt most appropriate in the CVTS, rather than simply adding them all at the end of the CVTS questionnaire:

• The Netherlands integrated the questions of the employer module throughout the CVTS questionnaire and placed them where they were most closely related to the topics of the CVTS.

• Portugal made some minor changes to the order of the desirable questions, moving the module “2E: desirable questions on the structure and business model of the enterprise” forward immediately after the questions of the core module.

• Italy also made some minor changes to the order of the questions. Q4 and Q5 from the core were moved to the introductory block of the survey, which collects data on general enterprise characteristics. QA1, a core optional question, was moved forward and asked after Q2. The selected desirable questions were placed where they were most closely related to the topics of the CVTS.

• Hungary made one change to the order of questions, moving QA1 after Q1, where it was most closely related.

• Slovak Republic did not report any changes in the order of the questions. However, the order of the categorical response options for Q1, a core question, were asked in reversed order. This was largely due to difficulties translating the question into Slovak, which resulted in asking “Thinking about the skills of people employed in your enterprise and the skills needed to do their current job, how many do you think have the skills needed to do their job to the required level?” instead of “Thinking about the skills of people employed in your enterprise and the skills needed to do their current job, how many do you think do not have the skills needed to do their job to the required level?”. This meant that response options were reversed such that “All” became “None” and “Most” became “Some”.

Data transmission, treatment, and validation

All field data collection was completed by September 2022. Prior to submission to the OECD, countries undertook a process of data cleaning and validation at the national level. Countries then securely transmitted a dataset to the OECD that included firm-level responses to the employer module, a country code, reference year and enterprise identifier, and sampling weights. The exception was the Netherlands, where data were not transmitted but had to be accessed remotely by the OECD secretariat. Countries also provided the metadata used in the preparation of this technical report. The OECD then carried out a second international validation of the data, producing cross-country statistics and checking the validity of the constructs.

National quality assurance

Countries participating in the PIAAC employer module showed a collective commitment to ensuring the quality of the data collected. Approaches to quality assurance were varied, with an emphasis on different strategies tailored to each country’s specific context, but some common approaches were followed:

• ISO certification and compliance with the Code of Practice: Several countries demonstrated their commitment to data quality through ISO certification, indicating compliance with international standards. The adoption of the European Statistics Code of Practice further underlined the shared commitment to sound statistical practices.

• Use of web-based questionnaires: The use of web-based questionnaires ensured consistency of responses and minimised problems with routing. This approach not only improved data quality, but also facilitated timely checks and corrections.

• Engagement with enterprises: In the Netherlands, the survey was announced well in advance (one year in advance) to large enterprises (>=250 employees) to ensure their inclusion in the survey and to allow them to collect the necessary information during the year. Large enterprises were selected as these were to be integrally included in the survey and the exact CVTS sample was not drawn at that time. Moreover, several countries set-up phone and e‑mail help desks to answer any questions that enterprises might have about the survey. Participation in the survey was mandatory for enterprises in the Slovak Republic and Hungary, thus achieving high response rates. Moreover, the survey was announced to large enterprises in Hungary one year in advance.

Countries were generally positive about the quality assurance mechanisms and – as a result – the quality of the data.

National-level validation and weighting

To ensure the reliability of the data in the PIAAC Employer module, validation procedures were implemented in a consistent manner across participating countries. All countries implemented continuous monitoring of response rates, regular global and plausibility checks to validate the completeness of information, correct coding, and formats. The electronic mode of administration meant that automated tools and checks built into the electronic questionnaire were used to maintain data completeness and accuracy. Some countries cross-checked source data against other surveys and administrative data sources (such as tax return data). Non-respondents were actively followed up, for example by phone and email, and corrective action was taken promptly to address errors and inconsistencies. Errors were corrected by editing or imputing as needed, using general correction rules. Several countries used results from previous waves of the CVTS to check the plausibility of results.

Countries participating in the PIAAC employer module shared a commitment to robust weighting procedures. Final weights were usually calculated considering the stratification methods used and adjusting for non-response to ensure representative estimation of key variables. Primary design weights were modified based on inconsistencies in the responses. Several countries used the General Regression Estimator (GREG) or Horwitz-Thompson formulae for weighting, with an emphasis on precision in the estimation of target variables. In some cases, calibration procedures were used to minimise the weighted distances between direct and final weights, thus helping to converge to known population totals.

Anonymisation and transmission

To ensure confidentiality, countries anonymised employer information and, where necessary, further manipulated elements of the micro-data to ensure that the surveyed enterprises were not identifiable. Any data manipulation that had to be carried out at national level was communicated to the OECD in advance, with the OECD team ensuring that any anonymisation techniques did not compromise the usefulness of the data.

Portugal, for example, followed CVTS convention by reporting enterprise size in categories (0‑49 persons employed, 50‑249 persons employed, 250+ persons employed) rather than a numerical size for each enterprise to avoid the risk of identifying enterprises. Hungary and Italy chose to report enterprise in six categories. See Table A A.3. for the full list of variables which were suppressed or modified by countries for the purpose of anonymisation.

Once the data had been made confidential, countries submitted them to the OECD for processing and analysis. In some of the participating countries, a legal framework was in place to allow the transfer of the module’s micro-data to the OECD. In the absence of such a framework, the OECD and the participating country signed ad hoc legal agreements, setting out the conditions for the secure transfer, storage, and use of the data. Anonymised data were securely transferred via an OECD tested transfer facility and then stored on a secure server at the OECD premises in Paris, France. In one case, the Netherlands, data was not transferred to the OECD, but only accessed remotely, due to restrictions imposed by the Dutch legal framework.

Table A A.3. Missing data and changes made to variables by countries

Hungary (N = 6 715)	Italy (N = 17 617)	Netherlands (N = 4 450)	Portugal (N = 3 419)	Slovak Republic (N = 2 387)
No postcode. Region is set to missing in 920 cases to preserve anonymity.	No postcode or region	No postcode	No postcode or region	No region
No enterprise size. Categorical variables provided instead: 1) 10‑19 persons 2) 20‑49 persons 3) 50‑249 persons 4) 250‑499 persons 5) 500‑999 persons 6) 1 000+ persons	No enterprise size. Categorical variable provided instead: 1) 10‑19 persons 2) 20‑49 persons 3) 50‑249 persons 4) 250‑499 persons 5) 500‑999 persons 6) 1 000+ persons		No enterprise size. Categorical variable provided instead: 1) 10‑49 persons 2) 50‑249 persons 3) 250+ persons
No birth year of firm		No birth year of firm	No birth year of firm. Categorical variable provided instead: 1) Up to 2000 2) 2000 and 2020
NACE 20 reported	NACE 20 reported		NACE 5 reported

Source: OECD validation of data submitted by countries.

International validation and quality control

After receiving the anonymised data from the countries, the OECD carried out a series of data validation and quality control checks. Most countries complied with the codebook provided by the OECD, although there were minor discrepancies, e.g. in variable names. In these cases, the OECD renamed variables to comply with the international codebook. After some minor editing, the countries’ datasets appeared as in the codebook, except for Italy, the Netherlands, and the Slovak Republic.

For Italy and the Netherlands, data were received in a coarser form and some data cleaning and manipulation was required to bring the data in line with international standards. Firstly, some variables had to be renamed to ensure that they conformed to the naming conventions set out in the codebook. Then, some new variables were created. Standard validation was then carried out, as for the other three countries, to ensure that all the values variables took were plausible. After this process, the datasets resembled the other countries and corresponded exactly to the standards of the codebook.

During the early analysis phase, it became apparent that the key variable on skill gaps (Q1) for the Slovak Republic had not been coded in accordance with the codebook. This issue was identified through deviating response patterns compared to other countries (issue also described above). Upon request, an updated dataset was provided, and standard validation procedures were carried out. After validation, the dataset aligned with those of the other countries and adhered fully to the codebook standards.

Also, during this phase, it became apparent that some variables required further recoding before analysis could be undertaken; for instance, for the Netherlands, the coding of the key variable on skill gaps (Q1) and the variable on how often a firm innovates (QA2) both required reverting the scale. Some additional minor recoding was made also to variables for Italy and Portugal, after which all variable labels aligned with the original codebook sent to countries.

Results of the international validation

After ensuring all five datasets were in line with the codebook, the OECD conducted several checks to ensure the quality and integrity of the data. For each of the country datasets, the OECD:

• Validated responses to each question and checked which set of questions countries chose to ask;

• Ensured there were no duplicate responses from the same employer;

• Checked total sample size, and for each variable, checked the number of valid, invalid and missing values;

• Checked logical consistency between variables.

The above checks confirmed that countries asked questions as set-out in Table A A.1. and that each enterprise was uniquely identified. Except for Italy, all countries provided a variable containing a random identification number to identify each enterprise in the dataset. To rectify this for Italy, the OECD confirmed with the relevant Italian authorities that each row was indeed a different enterprise and subsequently assigned their own random identifiers for each enterprise.

As mentioned above, some countries have chosen not to report certain variables in order to preserve anonymity – the full list of missing variables for each country can be found in Table A A.3. above. Most countries have some missing geographical information. Some countries chose to aggregate some variables in order to reduce the risk of identification. For example, Portugal reported the economic activity of the enterprise at the NACE 1‑digit level rather than at the most detailed 4‑digit level. Italy, Portugal and Hungary also chose to report the size of the enterprise as a categorical variable rather than a numerical one for confidentiality reasons. These cases of non-reporting are not expected to cause significant problems for current and future analysis, as the categories used are standard categories also used in the PIAAC household study and as it is unlikely that research will be carried out at detailed geographical or sectoral levels.

Finally, the logical consistency of variables was checked to further validate the integrity of the data. Table A A.4 details examples of the most important logical consistencies that were checked. Aside from a few instances, most countries passed these checks. In three instances, when data took on implausible or unexpected values, the check is marked as failed. Future analysis of the data may require some minor recoding of variables before conducting any analyses. Researchers also confirmed that variables took values that were logically consistent and as defined in the codebook – that is, variables took on values that were plausible and values across variables were consistent.

Table A A.4. Example of logical consistencies checked by OECD researchers

	Hungary (N = 6 714)	Italy (N = 17 617)	Netherlands (N = 4 450)	Portugal (N = 3 419)	Slovak Republic (N = 2 387)
Region is not missing if postcode is missing	Unable to calculate (see missing data above)	Unable to calculate (see missing data above)	Unable to calculate (see missing data above)	Unable to calculate (see missing data above)	Unable to calculate (see missing data above)
Number in training is less than number employed1	Unable to calculate (see missing data above)	Unable to calculate (see missing data above)	Pass	Pass	Fail 11 cases
A solution to under-skilling is reported only if under-skilling is reported as an issue	Pass	Pass	Fail (small number of cases*)	Pass	Pass
Training is provided to support staff through organisational changes only if organisational change is reported to have affected the firm in the reference year	Pass	Pass	Pass	Pass	Pass
An outcome of under-skilling is only reported if under-skilling is identified (optional question)	Pass	Pass	Pass	Pass	Pass

Note: (*) Unable to reveal exact number due to data privacy reasons.

1. It is a valid occurrence for the number of persons in training to be more than the number employed as the number in training refers to participants across the year while enterprise size is a reference to the number of persons employed on a single date (31 December 2020). Therefore, it can be the case that the total number of persons in training across the reference year is above the number of persons employed on 31 December 2020.

Source: OECD validation of data submitted by countries.

References

[2] Commission/Eurostat, E. (2019), The meeting of experts on continuing vocational training statistics on 6 and 7 December 2018. Conclusions.

[1] Marcolin, L. and G. Quintini (2023), “Measuring skill gaps in firms: the PIAAC Employer Module”, OECD Social, Employment and Migration Working Papers, No. 292, OECD Publishing, Paris, https://doi.org/10.1787/903c19c9-en.