National Statistical Institute of Spain

Labour Market Statistics Directorate

Paseo de la Castellana 183
28046 Madrid
Spain

In Spain,five Structural Earnings Surveys were undertaken, all in collaboration with the Statistical Office of the European Communities (Eurostat). The first was the Structural Earnings Survey for the period referring to 1995, which covered units with ten or more employees in the activities of industry, building, commerce, hotels and restaurants, transport, communications, finance institutions and insurance. The second, referred to 2002, broadened the coverage to include the activities outlined in sections M, N and O of NACE Rev.1. The third, with 2006 as a reference year, included the small units (those with less than 10 employees) in the same activities than in 2002. The fourth and fifth, referred to 2010 and 2014 respectively, has as a main characteristic to include economic activities regarding public administration.

The Community Regulations used as the bases for producing the last survey were as follows:

• Council Regulation (EC) No. 530/99 of 9 March 1999 concerning structural statistics on earnings and labour costs.
• Commission Regulation (EC) No. 1738/2005 of 21 October 2005 amending Regulation (EC) No. 1916/2000 as regards the definition and transmission of information on the structure of earnings.
• Commission Regulation (EC) No. 698/2006 of 5 May 2006 Implementing Council Regulation (EC) No. 530/99 as regards quality evaluation of structural statistics on labour costs and earnings.

 The Structure of Earnings Survey (SES) is a 4-yearly survey which provides structural data on gross earnings, hours paid, annual days of paid holiday leave and personal characteristics of employees (sex, age, occupation, length of service, highest educational level attained, etc.) and their employer (economic activity, size and economic control of the enterprise).

Regional data is also available at NUT 1 level and regional metadata is identical to that provided for national data.

The aim of this document is to be used to evaluate the quality of the survey. The structure of this report follows the content of Commission Regulation (EC) No. 698/2006 of 5 May 2006 Implementing Council Regulation (EC) No. 530/99 as regards quality evaluation of structural statistics on labour costs and earnings.

The "economic activity" is coded in NACE Rev. 2 (General industrial classification of economic activities within the European Communities) at two-digit level.

The "occupation" is coded according to the International Standard Classification of Occupations (ISCO-08) at two-digit level.

Information on the 'Highest successfully completed level of education and training' variable is classified using the International Standard Classification of Education, 2011 version (ISCED11), only the four main  groups are provided to Eurostat.

Regional breakdown is based on the Nomenclature of Territorial Units for Statistics (NUTS) reference year 2013 at NUTS 1 level.

The survey covers all economic activities defined in NACE Rev. 2 sections B to S.

The units included are all units with at least 1 employee.

Employees are all persons who have a direct employment contract with the enterprise or local unit and receive remuneration, irrespective of the type of work performed, the number of hours worked (full or part-time) and the duration of the contract (fixed or indefinite).

Annual gross earnings  include regular payments and also cover all 'non-standard payments', i.e. payments not occurring in each pay period, such as: 13th or 14th month payments, holiday bonuses, quarterly or annual company bonuses and annual payments in kind.

Monthly gross earnings in the reference month cover remuneration in cash paid before any tax deductions and social security contributions payable by wage earners and retained by the employer, and are restricted to gross earnings which are paid in each pay period during the reference month.

Hourly gross earnings are defined as gross earnings in the reference month divided by the number of hours paid during the same period.

Number of hours paid includes all normal and overtime hours worked and remunerated by the employer during the reference month. Hours not worked but nevertheless paid are counted as 'paid hours' (e.g. for annual leave, public holidays, paid sick leave, paid vocational training, paid special leave, etc.).

The survey is based on local units. Using as statistical unit the Account of Social Security Contribution, an administrative concept used by the Social Security Institution in its management of the Social Security System. Each Account relates to a set of employees working for an employer in one or several establishments of the company, within the same province and, generally, under one economic activity. In practice, each company tends to open an account for each establishment (local unit).

The Register of Employers (named Register of Accounts of Social Security Contributions) is used as a business register in the Structure of Earnings Surveys performed.

The populationis formed by all local units with independence of their size (accounts of social security contributio included in the General Regime of the social security) in the areas of economic activities defined by NACE Rev. 2 sections B to S.

Within each unit, all the employees associated with the account dutinr October 2014 are investigated with independence of the type of contract and the length of the working day.

The geographic area includes the whole national territory.

2014

Not applicable.

Not requested

Not requested

Not requested

Not requested

Not requested

Not requested

Quality assurance framework for the INE statistics is based on the ESSCoP, the European Statistics Code of Practice made by EUROSTAT. The ESSCoP is made up of 15 principles, gathered in three areas: Institutional Environment, Processes and Products. Each principle is associated with some indicators which make possible to measure it. In order to evaluate quality, EUROSTAT provides different tools: the indicators mentioned above, Self-assessment based on the DESAP model, peer review, user satisfaction surveys and other proceedings for evaluation.

The aim of this document is to be used to evaluate the quality of the survey. The structure of this report follows the content of Commission Regulation (EC) No. 698/2006 of 5 May 2006 Implementing Council Regulation (EC) No. 530/99 as regards quality evaluation of structural statistics on labour costs and earnings.

Not requested

The main users could be classified in the following groups:

• International Organisations: European Union Institutions, OECD, International Monetary Fund, International Labour Organisation, etc.
• Public Organisms: different Ministries such as the Ministry of Economy, the Ministry of Labour and Social Affairs, etc.; the National Statistical Institute itself for several of its units, such as National Accounts; the Bank of Spain; Regional Institutions, etc.
• Social Institutions such as trade unions, employers’ organisations, political parties,...
• Research Centres and Universities
• The media

Each one of these users has different needs, according to the purpose and use they will make of the information they require.

The INE has carried out general user satisfaction surveys in 2007, 2010, 2013 and 2016 and it plans to continue doing so every three years. The purpose of these surveys is to find out what users think about the quality of the information of the INE statistics and the extent to which their needs of information are covered. In addition, additional surveys are carried out in order to acknowledge better other fields such as dissemination of the information, quality of some publications...

On the INE website, in its section Methods and Projects / Quality and Code of Practice / INE quality management / User surveys are available surveys conducted to date.(Click  next link)

No specific user satisfaction survey related to the SES has been carried out among users to know their needs of information and whether they are satisfied or not with the published results. This may be accounted for by the lack of contact with most users since the remittance of results is often impersonal, and by the fact that the information is looked for in INTERNET.

It is known only the opinion of users who have received information on request or who have asked for methodological details.

In general, these users are satisfied. Nevertheless they consider that the survey should offer more detailed breakdown of some variables (in particular of regions, branch of activity and wage components) and also include more variables related to the employee (family situation, etc).

Not requested

Not requested

Not requested

The calculation of the coefficients of variation of key variables is performed in each survey implementation and disseminated in the publication of their results and the successive quality reports.

The estimators used for the survey are separate ratio estimators, the number of employees in the register being used as an auxiliary variable.

The estimators for economic data of the employee j in the unit i classified in the activity r, size h and region t are formed.

The grossing up factors of first and second stage are respectively:

  and      

where, D_i is the number of employees in the register for the unit i, B_i is the number of employees registered during the whole month of October 2010 in the Social Security and b_i is the number of employees in the sample.

Thus:

  and 

are the hourly earnings and the earnings per employee in any cell C of the table (by activity, occupation, sex and regions)

Being:

     Total earnings (monthly or annual)

     Hours

                Employees

j makes reference to the employees included in the cell C.

The coefficient of variation is defined:

where

(h is referred to the cross of variables region, activity and size)

and

being

Please see the attached document Coefficients of variation.

The general rule to publish a cell in a multidimensional table used in the national publication is that, at least, 100 observations support the estimation. All figures estimated with less than 100 observations have been erased; cells estimated with a number of observations between 100 to 500 have been marked to indicate this situation.

A control of non-sampling errors is performed in every statistical process. It also has specific information on the non-response rate.

The framework of the survey is obtained from the Social Security General Register of Contributions Accounts. Employers that hire employees for the first time should request their own registration as a company, at the Social Security General Treasury. They should do this before commencing work activities. Registration is an administrative act by which the Social Security General Treasury gives an identification and control number to the employer. This establishes what is known as the Account of Social Security Contribution.

The list of Accounts is used as a business register in all the Labour Cost and Structure of Earnings Surveys performed. The procedure for random selection of units corresponds to a stratified sampling with optimal allocation, in which the sampling units are the accounts. The stratification criterion is accomplished using three variables: Autonomous Community, economic activity (NACE rev.2 from B to S) and unit size (in terms of number of employees).

When the Register is received from the Social Security, a first debugging is made prior to the selection of the sample, which implies several stages:

• To eliminate economic activities regarding agricultural activities, livestock, fishery, households with domestic employees and extra-territorial organisms since these are not part of the survey.
• To eliminate the units that belong to the special regime of Social Security sales agents, whose main compensation consists in commissions on sales and who, consequently cannot be surveyed either.

Afterwards, the sample is drawn and the list of numbers of accounts of social security contributions selected is sent to the Social Security General Treasury again. It provides the list of all the employees, identified by their affiliation numbers, included in these units during the reference year. Specifically, the employees to be targeted for the SES are those employed in the observation unit during the whole reference month. A simple random sample of employees is taken within each of the selected local units according to the size of the unit. In SES-2014 the number of employees selected was:

• All employees in units with 1-4 employees
• 4 employees in units with 5-9 employees
• 5 employees in units with 10-19 employees
• 7 employees in units with 20-49 employees
• 10 employees in units with 50-99 employees
• 16 employees in units with 100-199 employees
• 22 employees in units with 200-499 employees
• 25 employees in units with 500 or more employees
• There were 52 local units with 50 employees selected due to their special location.

 The advantage of this method is that the respondent does not choose the employees, because the employees to be included in the questionnaire are identified by their affiliation number.

The main difference between the reference population and the study population is that the first does not include the apprentices. The labour legislation on apprentices in Spain establishes very low labour costs (both wages and social contributions). As a consequence, the number of apprentices is very small. Thus, at the end of Septermber of 2014 the number of apprentices was 116.946 from a total of 12.127.757 employees in the activities included in the survey (it represents 0,96,% of the total).

Moreover, the problem with apprentices is that, due to their particular type of contract, the Social Security General Treasury registers them, for control, in a different affiliation file, with different characteristics, that make difficult to use it jointly with the general file. On the other hand, because of it is such a small group, a random selection does not assure to obtain representative separate figures for this collective. This fact makes necessary a great effort carrying out a specific survey for the apprentices to assure the results. This effort is not corresponded with the small figures obtained as was showed in the experience from the 2002 SES and 2004 LCS. On the other hand, a significant proportion (one third approximately) of apprentices estimated by the 2004 LCS survey were, in practice, scholarship employees, circumstance not known before selecting the sample, and their inclusion in the final figures distorts the apprentices’ figures. Consequently the apprentices are not included since the SES 2006.

Section O of NACE Rev.2. are partly covered. There are two kinds of employees working in public administration: The first group comes under the Social Security General Scheme and is subject to the general rules of employment law, just like all other employees in other NACE Rev 2 sections included in the SES 2014.

The other group comes under the Special Scheme for civil and military public servants. Career civil servants in Central Government, armed forces personnel and justice and Parliament officials come under the State Employee Pension Scheme, whereby the State assures them protection against the risks of old age, disability, death and survival. For healthcare, temporary disability benefits, etc., such employees are covered by public-employee mutual societies. This group is not covered by the Social Security General Scheme, and therefore are excluded from the survey.

Once the questionnaires are sent to the selected units, the data collection and debugging reveal the errors in the surveyed units.

The sample was composed by 27.339 selected units. 26.534 units were surveyed: 25.087 units answered and 1.447 not.

Data collection showed that 722 units were not located, 39 units were inactive or closed down in 2014 and 44 units were erroneously classified.

0,09%

Not requested

Before sending the questionnaires to the units, the telephone numbers and addresses for the units were checked and updated.

The filled questionnaires were given back to the statistical office by mail, in the enclosed postage paid envelope, or electronically, to which purpose a registration and transmission format on Internet was designed.

It was also possible to fill in the questionnaire by Internet using an identification number provided in the questionnaire.

Debugging errors

After receiving the questionnaires, the statistical office recorded them, using an ad hoc computer application, which at the same time made a first debugging for the questionnaire’s internal consistency.

This first debugging consists in using filters that allow to separate valid questionnaires from those with inconsistencies to be revised.

The filters are of two kinds: those detecting type I and type II errors.

Type I errors:

If they are not thoroughly corrected, the questionnaire cannot be considered as valid.

Type II errors:

They involve norms regarding the coherence of the data. The non-satisfaction of these norms does not necessarily mean that the questionnaire is not valid, but the apparent incoherence must be explained. In case of doubt, a telephone call is made to the respondent to elucidate the question.

The questionnaires are filtered a first time during the recording and a second time by the team responsible for the results of the survey (this team is different from the recording one).

There are more than four hundred rules checked in each employee data. They assure:

• not missing data (partial non-response is not allowed)
• coherence among individual characteristics: age, length of service, level of education, type of contract, occupation, and so on.
• coherence among economic data: monthly and annual earnings, between themselves, and both related to the hours paid, to the economic activities, occupations, etc.
• the codes assigned for the level of education and the occupation exist in the classification used (ISCED-11 and ISCO-08) and are coherent with the variables, as economic activity, age, etc.

The final sample size has been of 25087 units and 209436 employees. The overall response rate is 90,2% of local units and 91,9% of employees. The tables with the response rates brokendown by economic activity and by NUTS are in the file annexed.

Partial non-response is not allowed. When there was no response or an incidence in the sample, the value of the analysis variables for each ‘empty’ sampling unit or unit without information was imputed using the information obtained for the stratum to which the unit belonged. This form of imputation only requires replacing the raising factors obtained with the selected sample with the ones that result from the effective sample.

There is only one exception in the item non-response: the level of education of the employee. It has been the most difficult variable to obtain, mainly in large units or groups of units, with a great number of employees selected in occupations where the level of education is not an important requirement in the job (eg.: Major Group 9 of ISCO-08).

Most of this kind of units needed a lot of time to answer the questionnaire, because they had to ask to the employees their level of education. At the end, this item was empty for 1.497 employees (0,7% of the sample).

The level of education of these employees was imputed using the software application IVEware (Imputation and Variance estimation Software). This software performs imputations of missing values using the Sequential Regression Imputation Method [1]. This method has two main advantages: it takes into account the structure of correlations of the whole set of variables in the sample and it is built on the SAS Macro Language, that is the software used for the rest of processes.

The effect in the final results of the imputation is negligible due to the small number of missing values imputed.

[1] This method is described in the article "A multivariate technique for multiply imputing missing values using a sequence of regression models" by Raghunthan, Lepkowski, Van Hoewyk and Solenberger (Survey Methodology, June 2001).

Not requested

Not requested

The variables ‘level of education’ and ‘occupation’ were codified at the time of recording the questionnaires. Rules to assure that the code assigned exists in the classification were established. Moreover, the sample was divided in portions. Random subsamples were selected from each portion and the codification in it was revised. If the errors in the codification were higher than the 3% of the total number of employees in the subsample, the whole portion was recodified. This process was repeated until this percentage of errors was achieved.

 The processing, grossing up and tabulation of the data have been programmed and supervised by two different teams. After the tabulation, the results obtained were analysed in order to know whether they were coherent with the available short-term statistics on labour and wage costs.

The imputation rate for level of education is 0,7%

to ensure that a representative month is selected:

The monthly questions have as reference October 2014 in the questionnaire. It is not possible to answer for other month.

Practices in the companies in Spain suggest using October because it is a month without seasonal payments and absences (like Christmas pay or summer holidays). On the other hand, October was the month used in the previous SES, so it is the most suitable month to keep comparability over time.

to adjust the accounting or fiscal year to the calendar year

The accounting or fiscal year coincides with the calendar year in Spain.

to ensure that NACE Rev.2 sections are fully covered

The register used to select the sample has the economic activity as a variable of classification. The design of the sample takes into account this variable in the stratification process jointly with the unit size and the region.

Not requested

Not requested

Not requested

Not requested

The Structural Earnings Surveys are published t+22 months after the reference year.

Not requested

Not requested

Key data collection dates:

The fieldwork took place between April - December 2015. The stages of the collection period are the following:

 Remittance of the material to the respondent units. This first stage took 3 days. In general terms, each mailing contains the following documents:

• A questionnaire that must be remitted in a delay not surpassing 20 days after it is received.
• A list with the number of Social Security of the employees selected in the unit.
• A letter from the General Director indicating the purpose of the survey and informing on the laws that oblige to complete the questionnaires and on those regarding Statistical Confidentiality. It also contains the username and the password to access to the web site and then it could fill in the questionnaire directly field by field or upload the XML file.
• A postage paid envelope bearing the address where the respondent has to send the filled in questionnaire.

 Location: The length of this stage depends on the number of respondent units to be located and lasted about one or two weeks.

The work consisted in finding the telephone numbers and/or the addresses of those units for which no contact telephone number was available or whose envelopes with the documents were returned.

The most frequent steps to locate a unit were the following:

• To phone or e-mail the information services of the telephone firms
• To consult telephone directories: White Pages and Yellow Pages (manual search or electronic search )
• To contact municipalities (Tax Department)
• To contact Social Security Treasure
• To contact enterprises of the same sector in the same municipality
• To contact Tax Agencies
• To search INTERNET: White Pages, Yellow Pages, Camerdata

Contacts and claims: This stage is essential for a fluent and efficient collection, to get a high percentage of success.

At this point, the calls to enterprises were started and the questionnaires claimed. The most useful tool for this activity was the telephone.

Telephone contacts may occur in both directions. To foster the respondents’ willingness to call the NSI, they are provided, whenever possible (in some of the documents forwarded to them), with a free telephone number. The calls are preferably answered by the interviewer in charge of obtaining their questionnaires. If this is not possible, any person tasked with the collection will resolve the respondent’s doubt or duly take the message (indicating the enterprise’s National Register Number, its address, name of the person who calls, contact telephone number, identification number in the survey and other comments).

There is also a free fax number to receive questionnaires and written communications.

This stage lasted approximately two or three weeks and each interviewer must contact the enterprises assigned to him and request their questionnaire.

Claim with acknowledgement of receipt: All the respondent units which had not remitted their filled in questionnaire by the end of the above stage, received by registered mail and with an acknowledgement of receipt, a second questionnaire with the mention “Claim of compulsory statistical data” (document PS2)

Location of non-found units: At the end of the first round of calls (to all the units in the survey), it turned out that a percentage of them could not be touched. They all belonged to a special queue of units: the QUEUE OF NON TOUCHED UNITS. Despite the implementation of all the available means, it was impossible to touch some of them. However, the above mentioned telephone information web pages continued to be looked through.

Stage following the collection of questionnaires

One of the interviewer’s tasks is the recording and debugging of all the incoming questionnaires.
The general rule is that the questionnaires must be recorded at the latest from 3 to 5 days following their arrival, to facilitate consultations with the enterprise as soon as possible after they were filled in. At the outset of the collection period, the location and first contact with respondent units have the priority over the recording and debugging of questionnaires.

After recording the questionnaires, they pass to the codification team. The occupation and the education level are codified at this point.

To eliminate the errors, all computer applications classify the errors in two large blocks: type I or big errors and type II or small errors. Type I errors are so important that they invalidate the questionnaire. Type II errors may arise from specific circumstances of the enterprise’s activity, from its activity during the data reference period or from any specific event of the respondent unit.

The first debugging should be carried out at the latest from 8 to 10 days after the recording, that is, 10-15 days after the questionnaires are received.

The recording and the first debugging stage finished by the end of January 2016.

The second debugging of all the units was completed in April 2016. The tabulation was prepared at the end of May and in June.

The first remittance of data to Eurostat was done at the 24 June 2016; several errors were found and corrected so the final version was sent on 29 June 2016.

Publication dates:

On 28 October 2016, the detailed results were disseminated. This publication is composed by a document with the comment of the main results and large set of tables. The information is available on INTERNET and on electronic support at request. 

Not requested

There are no differences between national and European concepts regarding statistical units, definition of variables and classifications. The only difference in the reference population is that apprentices are not included since SES 2006 as explained above.

Most of the effort made by the unit responsible for Labour Cost Statistics went on the detailed study of the variables contained in Commission Regulation No. 1738/2005 and its comparison with labour laws and forms of retribution in force in Spain in the year 2014. From this comparative study we obtained a version of the questionnaire adapted to the reality of the country, which allowed us to obtain the variables as defined in the above-mentioned regulation.

The use of a common methodology and common procceses of collection, validation, editing and grossing up in all  geographical scope, guarantees the comparability of the results between the different Spanish regions

Not requested

Since the first Structural Earnings Survey was conducted the coverage of the following surveys has been extended including different groups of units.

Thus, in first SES 1995 units with ten or more employees in the activities of industry, building, commerce, hotels and restaurants, transport, communications, finance institutions and insurance were included. The second, which referred to the 2002, broadened the coverage to include the activities outlined in sections M, N and O of NACE Rev.1. The third with 2006 as a reference year, has as a main characteristic to include the small units (those with less than 10 employees) in the same activities than in 2002. And finally, SES 2010 uses NACE Rev.2 and ISCO-08 as new classifications and includes partially section O. SES 2014 has the same scope and coverage as SES 2010.

As a consequence of the inclusion of the small units since SES 2006, there is a decrease of the average earnings compared with the general SES 2002 results. It is necessary to eliminate the size 1-9 employees from SES 2006 to compare homogeneous results with SES 2002.

The main difficulty to compare SES 2010 with the previous surveys is the change in the classifications used in last one. So, it is not possible to compare the results by economic activity or by occupation.

SES 2010 and SES 2014 are fully comparable.

Not requested

According with the Regulation a comparison should be made between the variable “gross annual earnings in the reference year”, expressed per employee, and the variable “wages and salaries”, per employee, of the National Accounts.

The figures available from National Accounts correspond to the 2014 provisional data about Wages and salaries and the occupied posts of employees from the Annual National Accounts 1995-2015. Base year 2010” published on 31 October 2016.

Data from SES are the estimations on average gross annual earnings per employee.

The table is attached as an annex.

The differences between the two sources of information are very uneven between economic activities. They vary between 0.0% (Section E) and 40% (section I). In general, figures are higher in NA data than in the SES data. The main reason for this, is that NA data include all the employees regardless of how their remuneration is (for example, commissions), of their type of contract and of their being incorporated or not in the Social Security, and the data are also adjusted by what is known as the exhaustivity of the GDP.

This last adjustment affects specially to services activities (sections I,J and R). The NA compensation of employees includes tips that are not in the SES and that are considerable, mainly, for sections I and J. Sales representatives and other persons who are wholly remunerated by way of fees or commission may be important in sections G, K and L,  and they are included in NA but not in SES.

Sections P-Q are formed by units belonging to private and public sectors but public sector has an important weight.  It is possibly a problem of different classification in the register used for public units. Salaries may be paid by a unit belonging to section O (administration) while the real activity may belong to another NACE section, for example P (education). Moreover, in these sections, as in the case of section O, are two kinds of employees:  those under Social Security General Scheme that is the frame of the survey and career civil servants in Central Government, armed forces personnel and jus-tice and Parliament officials under the State Employee Pension Scheme that are exclude from the survey. Mainly in education this kind of employees could be of some importance.

Not requested

Not requested

Not requested

Not requested

The tables, the document on the results and the methodological document are available for free on the INE web site.

The release was sent to the main official users.

The tables, the document on the results and the methodological document are available for free on the INE web site.

Not requested

A microdata standard anonymised file using a similar methodology as presented in the LAMAS Working Group on March 2008 is available for free at the web site.

Moreover, it is possible to prepare customised anonymous survey files studying the variables requested and also, based on the basic statistical operations files, crosses other than those published may be carried out between variables, according to the needs of the user.

The results are not remitted to the respondents.

The tables, the document on the results and the methodological document are available for free on the INE web site.

Not requested

Not requested

Not requested

Not requested

Not requested

Not requested

More information on