National Institute for Statistics and Economic Studies (INSEE)

DIRECTORATE OF DEMOGRAPHIC AND SOCIAL STATISTICS

Department for Employment and Working Income

Earnings and Working Income Division

TIMBRE DG75-F240 - 18 bd Adolphe Pinard - 75675 PARIS CEDEX 14 - FRANCE

As for the ESS 2006, the ESS 2010 file delivered by the Insee to Eurostat in the summer of 2012 is the result of the concatenation of the data collected by the annual surveys on the cost of labour and the earnings structure (Ecmoss) in 2009 and 2010 (earnings structure version), after the data collected were updated in 2009. The situation of each employee in the 2009 survey is updated by comparing employees with the same characteristics in 2010. Compared to 2006, the field for the annual surveys was extended to employees of the local and regional authorities. Apart from these two concatenated examples, the ESS 2010 file also includes government staff for whom an additional survey was conducted in order to supplement the field in accordance with the Regulation.

Not available.
New concept added with the migration to SIMS 2.0.
Information (content) will be available after the next collection.

Not available.
New concept added with the migration to SIMS 2.0.
Information (content) will be available after the next collection.

Not available.
New concept added with the migration to SIMS 2.0.
Information (content) will be available after the next collection.

Not available.
New concept added with the migration to SIMS 2.0.
Information (content) will be available after the next collection.

Not available.
New concept added with the migration to SIMS 2.0.
Information (content) will be available after the next collection.

Not available.
New concept added with the migration to SIMS 2.0.
Information (content) will be available after the next collection.

Not available.
New concept added with the migration to SIMS 2.0.
Information (content) will be available after the next collection.

Not available.
New concept added with the migration to SIMS 2.0.
Information (content) will be available after the next collection.

1) Sampling of the 2009 and 2010 Ecmoss and of the additional survey for the public sector

2009 and 2010 Ecmoss surveys

The population comprises all employees in companies with more than ten employees, established in mainland France and working in the private, semi-public or regional civil service sector (sections B to N, Q, R and S of the NACE, as well as P Education on a market basis and the Regional Public Administration part of Section O).

The method used to take the 2009 and 2010 Ecmoss samples is the same as for the 2005 and 2006 Ecmoss which were used to deliver the 2006 ESS. It is a sampling plan stratified into two stages, the first stage comprising establishments and the second

comprising employees in these establishments. The aim of that stratification is to improve the accuracy of the estimate by creating population subsets which differ enormously and which are relatively uniform in themselves in terms of the variable of interest.

The Ecmoss sampling plan was optimised in order to minimize the variance of the hourly wage estimator. This variable is indeed available in the sampling frame and is strongly correlated to the premium rate, which would be the ideal stratification variable in Ecmoss, in its ESS dimension. However, this variable is not available in the sampling base.

In order to produce this sample plan, we need to know the dispersion of the hourly wage over the population sub-domains. If the hourly wage is highly dispersed because of outliers, the probability of identifying the real average value is low, since we may have used individuals with hourly wages which differ substantially from each other, thereby making the average very inaccurate. Conversely, if there is little dispersion, a small group of individuals can be used without overly undermining quality.

The sampling frame used is the file of the 2008 Annual Declarations of Social Data (DADS) for the 2009 ESS (in fact, the survey took place in 2010 and the DADS focus on those present as at 31/12/2008, which suggests that they will be present at least for part of the 2009 in the establishment in question). Likewise, the 2009 DADS are used for the 2010 ESS.

The strata chosen in the sampling plan are developed from the crossing of four variables (sector of activity, firm size, establishment size and location). The combination of these different classes leads to a total of 910 strata.

Sector of activity

"BB" = Extracting industries (section B)

"AT" = Food, beverages, tobacco (sub-section CA)

"TH" = Textiles, clothing, wood, paper (CB+CC)

"RC" = Refining, chemicals, pharmaceutics, plastic, other minerals (CD+CE+CF+CG)

"MM" = Metals, machines-electronics, electrics, IT, automobile (CH+CI+CJ+CK+CL)

"MR" = Furniture, repairs (CM)

"ED" = Electricity, gas, water, sanitation, waste (D+E no distinction is made here between sanitation and depollution of the remaining waste)

"FF" = Construction (F)

"GG" = Trade (G)

"HH" = Post and transport (no communication - H)

"II" = Accommodation, catering (I)

"JJ" = Information, communication (J)

"KK" = Finance (K)

"LL" = Property (L)

"MA" = Legal, accounting, engineering and technical activities (MA + part of MC)

"MB" = Scientific research and development, specialised and veterinary scientific activities (MB + part of MC)

"NN" = administrative and support service activities (N)

"OO" = administration (O)

"PP" = education (P)

"QQ" = Human health and social action (Q)

"SR" = Arts, entertainment (R)

"RS" = Other services (S)

Region of location

"BAPA" = Paris region and Paris basin (Z1 and Z2)

"NEST" = North and East (Z3 and Z4)

"OUES" = West and South-West (Z5 and Z7)

"SUDE" = Centre East and Mediterranean (Z8 and Z9)

Some modalities were grouped into "TTZT" = all ZEAT, or with the sole distinction Paris basin ("BAPA") / Other ZEAT ("AUTZ").

Firm size

"TEG1" = from 10 to 49 employees

"TEG2" = from 50 to 499 employees

"TEG3" = 500 employees or more

Establishment size

"A1" = from 1 to 9 employees

"A2" = from 10 to 19 employees

"A3" = from 20 to 99 employees

"A4" = from 100 to 499 employees

"A5" = 500 employees or more

The sample was divided between the strata in such a way as to minimize the variance in the estimator of the hourly wage in the strata, with a minimum number of establishments per stratum (Neyman allocation subject to restriction). The procedure is therefore, successively, a calculation of the minimum number of establishments in each stratum with a given precision objective, a Neyman allocation of the number of establishments (1^st stage), a calculation of the minimum number of employees to achieve the accuracy objective for each stratum and, lastly, a Neyman allocation subject to restrictions on the number of employees (2^nd stage).  At the employee level, the allocation makes a distinction in each stratum defined previously between management and non-management.

Supplementary survey for the public sector

In order to cover the entire field requested by Eurostat, the Insee performed a survey in 2011 of the civil servants present in 2010. The population concerned comprises all civil servants, irrespective of the Nace section. They are concentrated in particular in non-market education services (section P) and administration (section O). The information collected in the survey, expanded by administrative information (earnings, premia, etc.), were used to build the variables requested by Eurostat.

The collection method differs from that chosen for the ESS survey. On the one hand, the civil servants are contacted directly but not the establishments in which they are working, as is the case for the ESS. On the other hand, information is first collected on-line. Next, the possibility of replying by paper questionnaire was offered to officials who had not replied on-line.

The sample is composed of 30 000 officials. It was created using proportionate selection by stratum. The strata are defined by the following variables: gender; age range; civil service category; status; ministry for budgetary purposes.

2) Variables of administrative sources

In order to lighten the response burden on establishments, the data from the Annual declarations of social data (DADS) for private sector employees and from the Public Service Employees Information System (Siasp) for civil servants were used. These data can be used to characterize the employing establishments, their employees (age, gender), the characteristics of the jobs and their corresponding earnings.

The DADS and SIASP datasets are the reference sources created by Insee to establish statistics on employment and wages. These data are obtained from the mandatory declarations to social security. They are exhaustive in their field.

The information on employees of use for the "structure of earnings" survey is as follows:

- age, gender, duration of pay in days (not requested in ESS, but taken from the DADS).

Other variables, included in the questionnaire, are used to process the partial non-response in the adjustment phase:

- social category and profession

- total gross remuneration in the year

- type of employment, employment contract.

For civil servants, in the database sent to Eurostat, the variables on earnings and on paid hours come from the Siasp administrative source.

[Not requested]

[Not requested]

[Not requested]

Processing operations for total non-response

Ecmoss surveys 2009 and 2010

After the concatenation of the 2009 and 2010 Ecmoss surveys, the final weights of employees were obtained from the initial sampling weights correcting for total non-response (the weights are recalculated by stratum according to non-respondents: by dividing the initial weights by the rate of response in the stratum), by adapting the structure of the sample of employees to the margins of total population.

The data are then re-weighted according to the margins from the 2010 DADS, so that from the key variables available in the administrative sources, the total obtained in Ecmoss is representative of the known total for the 2010 exhaustive survey, year of validity of the table sent to Eurostat.

The chosen margins are:

• Socioprofessional category * gender;
• firm size * sector of activity * location (Ile-de-France, Other regions of the Paris basin, other regions);
• full-time / part-time;

The margins in the population are calculated using the DADS file corresponding to the year of survey. They count all employees in firms which employed at least ten employees in 2010, for the field corresponding to the 2010 Ecmoss. The adaptation variables are:

• the unit variable (to be adapted in terms of employees)
• the total gross compensation over the year
• the job duration, in days
• the number of paid hours

Supplementary investigation with civil servants

As for the Ecmoss survey, the final weighting occurs in two stages. First of all, we recalculate corrected weights for non-response by post-stratification (the weights are recalculated by stratum according to the non-respondents: by dividing the initial weights by the rate of response in the stratum). The strata used are obtained by comparing the status, category, gender, age (same brackets as for the adaptation) and the employing ministry. Second, the adaptation changes these weights so that the sample is representative of the population of government officials (source: Siasp) over certain variables. The adaptation is made to obtain a range of weights which ensure that the weighted sums of some of these variables are equal to the totals obtained in Siasp for these same variables.

The workforce, hours, gross earnings and post durations are thus adapted according to the following variables:

• Category (A, B or C) * Status (Official, non-official) * gender
• Ministry * location
• Age bracket (under 35, 35 to 49, 50 and +)
• Full-time / part-time

Processing operations for partial non-response

Eurostat  asked the Member States to provide complete files, with no missing variables for any of the requested detail levels and therefore, by extension, for any employees. Also, so as not to eliminate too many individuals, a certain number of variables were imputed. Without describing the imputation processes here in detail, a hot-deck based on a division of the characteristics of the employee or of the employer establishment was generally used (age, gender, level of income, sector of activity, size of establishment).

The imputed variables are non-central variables which have no impact on the hourly wage, except for employees on a flat-rate for whom the compensated hours were recalculated based on the number of flat-rate days declared in the survey. For these employees with a flat rate, it is not an imputation linked to a partial non-response. Indeed, for these employees, the number of hours is irrelevant because they are not paid at an hourly rate but according to their job duration measured in days. We therefore estimate a number of hours depending on the number of days in the flat rate. Furthermore, the number of days in the flat rate is never imputed.

[Not requested]

Not available.
New concept added with the migration to SIMS 2.0.
Information (content) will be available after the next collection.

[Not requested]

SES and LCS survey match Eurostat request concerning french salary structure and labor cost.

[Not requested]

[Not requested]

[Not requested]

[Not requested]

Refer to 'Sampling error indicator'.

Accuracy of the 2010 ESS: estimated variance of monthly and hourly earnings

Ultimately, the 2010 ESS is an amalgamation of the two Ecmoss and of the supplementary survey for the public sector. The variance of the interest variable - hourly earnings – was estimated empirically by bootstrap, with a selection of 1000 replications . The "bootstrap weight" of each observation was recalculated from the initial weight (the initial weight corresponds to variable B52 in table B provided to Eurostat) depending on how many times it was selected in the samples. The empirical estimate of the variance is presented in appendix, with the detail per sector of activity (NACE sections), the firm size and regions (where cv = coefficient of variation and p_2_5 and p_97_5 are the lower and upper limits of the confidence intervals, respectively).

The tables in the attached document Sampling errors present the further information obtained by sector, firm size, region, age, profession and distinction by gender, full time/part time for the variables B42 (gross hourly earnings in the reference month) and B43 (average gross hourly earnings in the reference month).

Refer to sub-part of non sampling error for more information.

Firms with less than 10 employees are not covered by the survey.

nothing to declare

[Not requested]

To avoid measurement and processing errors, Insee has integrated a data control software program into the IT application for monitoring the survey. This software program ensures  consistency between different sources, so as to quickly detect outliers. It can be checked automatically, for example, that the elements included in a total amount are below that amount. Large-scale orders can also be checked using the data distributions from the previous survey. This procedure makes it possible to detect entry and scanning errors, confusion between firm or establishment levels, or calculation errors.

Adjustment

Two adjustment procedures were implemented successively. The aim of the first procedure was to check that the responses were consistent with each other and with the external reference source, the DADS. For each survey year, this work makes it possible to obtain the adjusted base which becomes the annual national release base, after benchmarking. The second adjustment relates specifically to Eurostat submissions. Indeed, some controls are performed a posteriori by Eurostat (number of monthly hours excluding overtime hours in full-time equivalent associated with hours, quota and post duration; number of holidays; valuation of overtime hours in relation to basic earnings, etc.). In order to keep individuals who were outside the limits set by Eurostat, we chose to make a number of adjustments. These adjustments did not concern the main variables of the survey, namely earnings, premia and hours. The adjustment variables were the quota, the number of overtime hours (and the corresponding compensation) and the job duration.

These adjustments made specifically for Eurostat resulted in particular in an artificial increase in the percentage of part-time by around 2 points (20.3% compared to 18.6% without adjustment). Nevertheless, these are employees for whom the working hours are fewer than the statutory working time (35 hours per week).

By construction, all the 2009 earnings variables were subject to another "adjustment", since they were updated. The quantitative compensation variables are adjusted by category, using the corresponding data in the DADS source.

We end up with the following adjustment rate for each variable (not including the 2009 data update):

Response rate

In total, out of the two Ecmoss surveys in 2009 and 2010, 30 243 establishments have received questionnaires, corresponding to 281 917 employees. 235 245 responses were exploitable, which in fine gives a response rate of 83.4%.

For the survey on civil servants, the response rate was 39.2%.

[Not requested]

[Not requested]

To avoid measurement and processing errors, the Insee has integrated a data control software program into the IT application for monitoring the survey. This software program ensures, on the one hand, consistency between data, so as to quickly detect outliers. It can be checked automatically, for example, that the elements included in a total amount are lower than that amount. Large-scale orders can also be checked using the data distributions from the previous survey. This procedure makes it possible to detect entry and scanning errors, confusion between the business or establishment levels, or calculation errors.

The imputation rate, by Eurostat variable, is as follows:

The imputation rate is very high for variable B33 (number of days of leave) in the additional survey for civil servants. Indeed, for teachers entitled to take all school holidays, the number of days of leave is not directly requested. It is recalculated a posteriori: the holidays which the teacher declares to have spent on class preparation are deducted from the school holidays. Since the number of days of leave is not directly requested, this calculation is assimilated to an imputation.

Firms with less than 10 employees are not covered by the survey.

[Not requested]

[Not requested]

[Not requested]

[Not requested]

Data were send to Eurostat on July 2012. And publish by the national institut of statistic on January 2013.

[Not requested]

[Not requested]

For the 2010 Ecmoss, the survey schedule is as follows:

• the letters of announcement were sent on 28 March 2011;
• the questionnaires were sent on 23 May 2011; firms which had not responded received a reminder on 8 July 2011;
• firms which had still not responded received a formal request to reply within ten days, sent on 12 September 2011, followed by acknowledgement of non-response on 7 October 2011.

For the 2009 Ecmoss, the survey schedule has been very close to that described below for the 2010 Ecmoss, with a one-year gap.

From the dispatch of the questionnaires until collection, managers (accounting for a total of 25 full-time equivalents) were present in the regional directorates of Insee to ensure contact with firms (answers to questions, setting a response deadline) and check the questionnaires, including a reminder for firms about most frequent errors. So as not to call firms back too long after their response, entries were made on an ongoing basis from the first responses from establishments.

The data codification and adjustment phase continued until the data were sent to Eurostat in July 2012.

[Not requested]

The NUTS (region in which the establishment is located), TAILLE (firm size) and NACE (economic activity at division level) variables, and others, come directly from the firm register.

For the employees concerned by the standard ESS survey (excluding civil servants), age, gender and job duration in days come from the DADS. For civil servants, apart from this information, the variables relating to compensation and hours come from Siasp.

The other variables come directly from the establishments' responses to the questionnaires on employees.

[Not requested]

The last edition of the survey on earnings structure (ESS) was in 2006. The field was extended in 2010 to regional civil servants (part of Section O selected for the legal category of the firm). Furthermore, civil servants were consulted in 2010 and included in the 2010 ESS sent to Eurostat.

Some methods have evolved between 2006 and 2010 to prepare the 2010 ESS Eurostat file.

a) Adaptation/Imputations

In 2006, for employees for whom the number of full time equivalent hours associated with duration, quota and number of hours was lower than 1 560 or higher than 2 580, the hours were changed so that the FTU (full-time unit) was equal to the limit accepted by Eurostat. In 2010, a decision was made to change hours as little as possible so as not to modify the hourly wage. The adjustment variables were the job duration and the quota.

No imputation was made in 2006. The 2006 base comprised around 60% of employees interviewed, compared to 90% this year.

b) Adaptation

The full-time/part-time dummy was added in 2010 for adaptation purposes.

Furthermore, the field of the adaptation base (on which the margins to be respected for sampled employees are calculated) was slightly modified. Employees with zero hours, usually excluded from the scope of Insee publications, were removed from the adaptation base. The impact is relatively low for employees in the ESS field because only 0.3% of employees in this field were withdrawn from the adaptation base.

c) Construction of the Eurostat variables

We have modified the implementation of basic earnings with respect to 2006 in order to be closer to the Eurostat definition. Indeed, Eurostat requests that a distinction be made between total compensation according to frequency of payment with, on the one hand, earnings paid per period and, on the other hand, its supplement, exceptional earnings, paid on an irregular basis. This change was made to respond as accurately as possible to the request and to treat all employees in the same way, which was not the case in 2006 (since flat-rate days were processed separately).

In the 2006 ESS, exceptional earnings were composed of the premia corresponding to collective or individual performance (for all employees), and of the allowances, profit-shares, membership of and contribution to the employee savings plan and PERCO (collective pensions savings plan) (for employees who were not on a flat rate).

In the 2010 ESS, exceptional earnings are composed of the premia and salary bonuses which do not correspond to premia for seniority, job or performance. The 13th month, holiday or end-of-year premia and benefits in kind are thus included. Allowances and sums paid for profit-sharing, membership and contribution to the employee savings plan and PERCO are also considered to be exceptional.

Furthermore, Eurostat requests paid hours for all employees, including those on a daily flat-rate. In 2010, the same approach was taken here as in the 2008 ECMO: a number of hours which had been remunerated and actually worked was imputed for employees on the daily flat-rate, based on the working day established by the Employment survey (8.7 hours).

Compensated hours = (no of flat-rate days + days of leave + public holidays included in the flat-rate) * working day

where the number of public holidays included in the flat rate is 7 (average number of public holidays).

This method leads to compensated durations that are much higher than those observed for an employee working 35 hours per week. But it ensures consistency with the ECMO 2008 and with the number of work hours.

This calculation devoted to paid hours concerns 24 051 employees, i.e. 11.5% of the base.

In 2006, the hours for employees on a flat rate were imputed using the average hours of employees belonging to the same stratum defined by the comparison gender*cs*age*sector.

[Not requested]

The consistency of the survey results was ensured thanks to the DADS.. which are also used as an auxiliary variable for the adjustment phases for gross compensation and paid hours, and provide adaptation margins.

Comparison with the data from the national accounts (Comptes nationaux, CN)

The figures presented above correspond to the average annual gross earnings in the reference year (B41 average weighted by B52) for the 2010 ESS and to the relationship between "wages and salaries" and "employees" for the national accounts.

The difference between the first two columns comes essentially from a difference regarding the concept of earnings. Indeed, for the national accounts (column 1), the relationship between payroll and the average employees is closer to the concept of annualised earnings. The variable B41 (column 2) corresponds to annual earnings.  Thus, for the sake of comparison with the national accounts, the earnings in the ESS must be annualised. The product of annual earnings (B41) by the inverse of the job duration (52/B31, where B31 is the number of weeks to which annual earnings refer) corresponds to annualised earnings. Column 3 thus presents the average annualised earnings based on the variables present in the ESS.

Other sources of discrepancy exist:

• The field ESS is limited to firms with at least 10 employees in mainland France, whereas the Comptabilité Nationale (national accounts) include firms with less than 10 employees and those from the overseas departments.
• In the national accounts, the activity classification is based on the concept of branch while the ESS relies on that of sector. A branch of activity covers units of economic activity at a local level performing an identical or similar economic activity; groups by branch therefore "outline" firms according to their various activities, to create groups of units with uniform production. A sector of activity covers all firms with an identical or similar activity. This difference can therefore explain any discrepancies on the level of the various activities but not on a global level.
• The payroll calculated by the Comptabilité Nationale comprises more remuneration details not taken into account in the ESS, such as tips (generally in accommodation and catering), the remuneration of majority shareholders in companies (particularly numerous in real estate activities, accommodation and catering, and construction) and expenditure in the paid leave funds (very common practice in construction and transport).

Comparison according to changes in hourly earnings

As we saw previously, some methods changed between 2006 and 2010, in particular for the calculation of the hours of employees on a flat rate. By adopting the method used in 2010 for 2006, we obtain the following developments. We can compare them to those of the DADS, the national accounts or the "earnings only" ICT.

[1] What we refer to here as the complete field relates to the main entries for the sectors of the Ecmoss field with no restriction in terms of size (firms with less than 10 employees are included) or location (overseas departments are included).

[Not requested]

[Not requested]

[Not requested]

[Not requested]

Files made available:

Each annual ECMOSS survey is subject to two release files made available to French employees of the Service statistique public (SSP, Public Statistical Service): an "establishment" level file and an "employee" level file. Furthermore, the "establishment" level file is available on request for persons external to the SSP from the Comité du secret (Committee on confidentiality).

A fichier de production et de recherche (FPR, production and research file) for employees on the ESS version of the ECMOSS survey is also produced for researchers outside the SSM and made available via the Quételet network.

These files will be available by the end of 2012.

Research:

Muller L. (2012), "Les écarts de salaire entre hommes et femmes en 2009 :le salaire horaire des femmes est inférieur de 14 % à celui des hommes », Dares Analyse No 16, Dares ( http://travail-emploi.gouv.fr/IMG/pdf/2012-016-2.pdf)

Data available on Eurostat database.

[Not requested]

[Not requested]

[Not requested]

The annual ECMOSS survey has a description sheet on the Insee website. The questionnaires from the last two surveys are also available on the Insee website.

(http://www.insee.fr/fr/methodes/default.asp?page=sources/sou-enq-ecmoss.htm)

Furthermore, the dictionary of variables accompanying each file (release or production and search files) presents the survey (background and methodology) and states for each variable whether it was subject to an adjustment.

[Not requested]

[Not requested]

[Not requested]

In order to lighten the response burden on establishments, the data from the Annual declarations of social data (DADS) for private sector employees and from the Public Service Employees Information System (Siasp) for civil servants were used. These data can be used to characterize the employing establishments, their employees (age, gender), the characteristics of the jobs and their corresponding earnings.

[Not requested]

[Not requested]

[Not requested]