Italian National Institute of Statistics (ISTAT)
Directorate for Social Statistics and Population Census

Department for Statistical Production (DIPS)

Directorate for Social Statistics and Population Census  (DCSS)

Integrated System for Labour, Education and Training (SSE)

Istat - Italian National Institute of Statistics
Via Cesare Balbo, 16 - 00184 Rome – Italy

Indices of Wages and Salaries. Annex C - Retail trade.

These indicators are produced by the Oros survey, through the integration of Social Security administrative data (collected by the Social Security Institute-INPS), used mainly for small and medium enterprises (SMEs) and the survey data on labour input and labour costs on the large enterprises (LES) used for large enterprises (LEs).

NACE Rev. 2.

Annex C: Divisions 47 and 47 excluded 473.

Wages and Salaries are defined in coherence with Commission Regulation EC No 1503/2006.

Reporting and observation unit: Enterprise. For some large enterprises reporting and observation units are KAUs.

All enterprises with at least one employee which were active in the reference quarter in the STS economic activities. In 2016, these enterprises were, on average about 211,000 (source: BR-ASIA) for the Retail trade sector. 

Geographically the STS indicators on Wages and Salaries cover the whole country. Activities outside the country are excluded.

The indicators are available since Q1:2000.

The base year is 2015=100.

 Index.

Quarter.

Wages and Salaries index is produced according to the requirements of the Council Regulation (EC) No 1165/98, as amended by Regulation (EC) No 1158/2005 and the regulations implementing and amending these two instruments.
Furthermore, all statistics produced and published by the National Statistical Institute of Italy are subjected to:

- the Legislative Decree no. 322, of 6 September 1989 (and subsequent modifications and additions Decree of the President of the Republic (DPR) no. 166 of 7 september 2010), which is consistent with the U.N. Fundamental Principles of Official Statistics and places Istat at the center of the National Statistical System (SISTAN). Sistan is a network of public bodies and private agencies that provides official statistical information and covers the statistical offices of all levels of government, Chambers of commerce, industry, crafts industries, agriculture and other public bodies as well as private subjects having public functions;
- the Decree of the President of the Council of Ministers (DPCM) which every year approves the National Statistical Programme;
- moreover, the Committee for Directing and Coordinating Statistical Information (COMSTAT), over which Istat presides, defines and issues binding directives for executing the National Statistical Programme. Istat has a legal obligation to publish and disseminate data (Article 15, in particular paragraph 1[g] of the Legislative Decree no. 322, 6 September 1989).

The legal basis of the admin source used in the estimation process of Wages and salaries, is the Decreto Ministeriale 05.02.1969 and Decreto Ministeriale 24.02.1984 on the obligation of units to provide data: the enterprises are obliged to pay the social contributions and to submit the monthly declaration (DM10 form until December 2009, UniEmens since January 2010) to INPS within 30 days after the end of the reference month. All firms which do not meet those obligations can be condemned to administrative and penal sanctions.

None. Indices calculated on totals are not officially released in Italy. They are shared only internally at Istat.

According to article n.9 of the Legislative Decree n.322 of 6 September 1989 data collected by statistical offices within the statistical surveys included in the National Statistical Programme may not be disclosed other than in aggregated form such that no reference to identifiable people can be extracted. Furthermore, they may be used only for statistical purposes. Data may not be communicated or disseminated neither to any external subject, public or private, nor to any department of the public administration other than in aggregate form and using modalities which prevent the identification of the people involved. In any case, data cannot be used to identify again the people involved. The Code of Conduct annexed to the Legislative Decree n.196 of 30 June 2003 (Personal Data Protection Code) provides special rules concerning the processing of personal data for statistical purposes within Sistan. In order to make statistical secrecy and protection of personal data effective, Istat has taken appropriate organizational, logistical, methodological and statistical measures in accordance with internationally established standards. In accordance with the Legislative Decree n.196 of 30 June 2003 (Personal Data Protection Code) and subsequent modifications and additions, respondents are informed of their rights and obligations with regard to the provision of information, and they are assured that the information they provide will be used for statistical purposes only.

Links to relevant acts on statistics are presented on the website of Sistan - National Statistical System – (http://www.sistan.it/index.php?id=203).

In general Istat has special aggregation rules which have been developed to ensure that indirect disclosure of individual data does not occur when aggregations of data are presented. For instance, access to individual data is restricted to staff who require the information in the performance of their duties. Provisions are in place to supervise analysts that require access to disaggregated data.

At the moment data on  Wages and Salaries expressed as totals are only transmitted to Eurostat but not disseminated nationally at any level of detail.

No calendar because no release.

Data are transmitted to Eurostat by teletransmission in confidential form. This kind of indices is not released at National level.

Quarterly (to Eurostat).

Indices on Wages and Salaries as required for STS purposes are not released at National level.

At the moment data are not released at National level.

Not available because data are not released at National level.

Not available because data are not released at National level.

Data are transmitted to Eurostat quarterly, within 60 days from the end of the reference quarter, in SDMX format.

Details on the LES and Oros surveys are available at the Information System for Survey documentation and Quality Control (Siqual), on Istat’s Internet website, respectively at the link http://siqual.istat.it/SIQual/visualizza.do?id=0026500 and http://siqual.istat.it/SIQual/visualizza.do?id=5000065

Other sources and methodologies are described in the following Istat methodological document: Istat 2019: "La rilevazione trimestrale Oros su occupazione e costo del lavoro: indicatori e metodologie" (link: https://www.istat.it/it/archivio/229033, in Italian). Information on the administrative data used as main source in the Oros survey can be found in the following document: Rapiti F.M., Ceccato F., Congia M.C., Pacini S. and Tuzi D. 2010. “What have we learned in almost 10-years experience in dealing with administrative data for short term employment and wages indicators?” avaliable at the link: http://www.ine.pt/filme_inst/essnet/papers/Session2/Paper2.4.pdf). 

An overview on the integration of sources and processes for the production of the main short term business indicators on labour market at Istat is described in the following document: Baldi C., Bellisai D., Ceccato F., Pacini S., Serbassi L., Sorrentino M., Tuzi D., 2011."The system of short term business statistics on labour in Italy. The challenges of data integration" available at the link: http://www.ine.es/e/essnetdi_ws2011/ppts/Baldi_et_al.pdf.

Not available specifically on the STS indicators production process.

Information on the quality management in the survey are descibed in the documents available at the following links:

Congia M.C., Rapiti F.M (2010), "Quality assessment and reporting in a short-term business survey based on administrative data", Documenti Istat, n.5. Link: https://www.istat.it/it/files//2018/07/doc_5_2010.pdf.

Congia M.C., Pacini S., Tuzi D. (2008a), “Quality Challenges in Processing Administrative Data to Produce Short-Term Labour Cost Statistics”, Proceedings of Q2008 European Conference on Quality in Official Statistics, Rome. Link: https://pdfs.semanticscholar.org/a0c9/6671ad72deb41b845261ada85674091d96aa.pdf.

Congia M.C., Pacini S., Tuzi D. (2008b), “The Editing Process in the Italian Short-Term Survey on Labour Cost based on Administrative Data”, paper presented to the UNECE - Conference of European Statisticians,Work Session on Statistical Data Editing, 21 – 23 April, Wien. Link: https://www.unece.org/fileadmin/DAM/stats/documents/ece/ces/2008/04/sde/wp.8.e.pdf.

Piras M.G., Tuzi D. (2018), “Revision Policy and Revision Analysis in Short-term Statistics at Istat: Recent Developments in Standardization and Communication Innovations”, Proceedings of Q2018 European Conference on Quality in Official Statistics, Krakov. Link: https://www.q2018.pl/papers-presentations/Speed_Talk_Session05.

Istat (2019), "La rilevazione trimestrale Oros su occupazione e costo del lavoro: indicatori e metodologie", Letture statistiche – Metodi, Istat. Link: https://www.istat.it/it/archivio/229033.

At the basis of the production of the STS indicators, the standard Istat systematic approach to quality following the International and European standards. Istat reference framework for quality policies relies on the European Statistics Code of Practice, adopted in 2005, revised in 2011 and, more recently, in November 2017 on Eurostat Quality Definition and on the recommendations of the LEG on Quality, approved by the Members States of the European Union in 2001. The Data Quality Assessment Framework, developed by the International Monetary Fund, also represents an important reference, especially for economic statistics and for National Accounts. Following the principles of the European Statistics Code of Practice, Italy has adopted the Italian Code of Official Statistics, O.G. n. 240 of 13/10/2010, in order to promote quality improvements of the statistics produced by the Italian National Statistical System (more details at the link: https://www.istat.it/en/organisation-and-activity/institutional-activities/quality-commitment/codes-of-statistics). The Quality Committee, set up in 2010, is a high level body in charge of quality monitoring and quality auditing of statistical processes and products. Quality auditing and self-assessment are aimed at verifying the compliance of statistical processes and outputs to the principles stated in the Quality Guidelines.

The Oros Survey is an innovative case of short-term statistics produced with the help of administrative sources (Social Security data) in order to cover all size enterprises in the private sectors. The use of administrative data in short-term statistics implies paying attention to unusual statistical quality aspects. Statisticians cannot prevent or reduce no-sampling errors in raw administrative data capturing, and some ex-post traditional editing techniques, like questionnaire revision and enterprise recalling, are not applicable. The complexity of the production process is also caused by the huge number of records and the highly disaggregated level of raw data. In fact, given the short-time constraint in the releases, the Italian NSO was obliged to capture data from the Italian Social Security Institute without any previous process of aggregation and checking. So, the retrieval and translation of the administrative data into statistical information is one of the most critical aspect to be faced at the beginning of the process; and its effectiveness has a significant impact on the quality of the final indicators. On the other hand, the availability of very disaggregated data allows for the exploitation of a very rich informative source for different statistical aims, and for a more direct control on the overall translation phase. When the statistical variables have been made available, a more traditional micro level check procedure is applied. Editing on outliers and anomalous values, and imputation of unit non responses may consequently be needed, with a particular attention to influential observations. Considerations linked to the quality of data suggest a micro-level integration between the administrative source and the Large Enterprises Survey data. This integration involves a record-linkage aspect and the computation of harmonized variables. When macro data are available, validation implies, among other aspects, time series analysis, macro-level comparisons with other statistical sources and analysis of revisions.

Finally, the standardization and documentation of the whole check and editing process is a fundamental target of the Oros quality procedure (for further details on the quality management in the Oros survey see: Congia M.C. and Rapiti F.M. 2008 “Quality assessment and reporting in a short-term business survey based on administrative data” available at the link: https://www.istat.it/it/files/2018/07/doc_5_2010.pdf and Congia M.C., Pacini S., Tuzi D. 2008 “The Editing Process in the Italian Short-Term Survey on Labour Cost based on Administrative Data” available at the link: https://www.unece.org/fileadmin/DAM/stats/documents/ece/ces/2008/04/sde/wp.8.e.pdf.

Recently Istat worked on the set-up of a shared framework for internal governance of the quality in data and underlying processes and the development of an easily accessible platform where information on practices and measures of revisions is made available to stakeholders (Istat website at link: http://www.istat.it/it/congiuntura/revisioni) (Piras M.G., Tuzi D. 2018).

Eurostat is the main user. Data produced are coherent with the requests of the STS Regulation.

The data are considered satisfying the STS Regulation requests by Eurostat.

All STS requirements are fulfilled.

Assessing accuracy on the indicator of Wages and Salaries, compiled using admin data integrated at micro level with survey data implies taking into account only non-sampling errors. Infact, these data sources refer to the census of the target units.

The massive quantity of administrative micro data (used for SMEs) covers almost the 95% in terms of number of jobs. By the way it requires, however, a very careful processing phase of E&I, since if the non-response for the total economy is not relevant, it is not the same for the different economic sector.  At this stage a micro level model-based process of imputation is performed, so for an estimate of the accuracy the main components to take into account are the estimates of the parameters, and the residual bias for some sectors for which the model does not fit well, for which a macro adjustment  based on the time series analysis is performed. Another source of bias comes from the definition of the statistical population. If from one hand the use of auxiliary information from the Business Register and from the more updated Tax Register helps defining the list of target units in the Social Security Register reducing problems of over-coverage, on the other hand the delay of the Business Register's updating date implies an additional bias due to the misclassification of the units. Survey data (used for LEs) refer to enterprises with more than 500 employees at the base year 2015, this group adds up to about 1.3 thousand enterprises covering about 24% of total employees in Italy in the STS sectors. Each one of these firms has a considerable influence on the estimates. Editing and imputation on this data are global (all units are checked) and performed by very expert personnel, assuring very high quality data and fast management of changes in units legal asset, non-responses, errors, etc. When considering accuracy, the micro level integration between the administrative records and the Large Enterprises Survey data must also be considered. Record linkage and computation of harmonized variables are the main processes to be taken into account.

Estimations are not based on samples.

Preliminary: about 60 days after the end of the reference quarter, widely anticipating the Regulation requirements, that is 90 days after the reference quarter. Final: about 1 year and 60 days after the end of the reference quarter.

Punctuality always achieved.

Wages and Salaries indicators are defined in coherence with Commission Regulation (EC)  No 1503/2006. The data cover the entire national territory.

The methodology at the basis of the estimation of the Wages and Salaries has been subjected to several innovations over the time, implying discontinuities in the time series. The same effect occurs in the occasion of the transition to a new reference base, when the most relevant innovations are introduced, or when non-ordinary interventions are performed. In all these cases, to guarantee coherence of all the occurrences of the time series, linking factors are normally calculated on overlapping periods and applied to the quarters estimated in old methodology. The last methodological changes and innovations have been introduced in data released on May 2018, in the occasione of the transition of the STS indicators to the new base 2015=100. Link coefficients have been calculated as ratios between the totals referred to year 2015 in old and new basis' criteria, at the level of 2-digit Nace. These coefficients have been applied to the totals of the same variables for the quarters from Q1:2000 to Q4:2014.

The quarterly and annual dynamic of Wages and Salaries indicators are constantly compared with figures drawn from the National Accounts and with National contractual (agreed) wages' figures.

Comparisons with Structural Business Data on annual basis are performed, too. Level of coherence is good. Differences can be attributed to the different methodologies used to estimate the considered aggregates and to the different levels of coverage, concepts, definitions and classifications.

Good coherence between indicators, known the differences in methodology, concepts, definitions etc.

For total Wages and Salaries, the discrepancy between the preliminary estimate and later ones depends on the revisions of the Oros-LES indicators. In a standard practice of revisions, figures that contribute to this aggregate’s estimate are revised four times before they become final, that occurs after one year from their first publication. The main reasons of revision are:

• the final version of the administrative micro data which are checked by INPS on reporting units, substitutes completely the preliminary version (which is checked and edited only by Istat);
• non reporting units in the preliminary data are present in the final version;
• the annual revision of the LES data referred to the previous year, included in the Oros-LES estimates yearly, in the delivery of the first quarter (May);
• non-standard revisions (es. transition to a new base year).

As a synthesis, each quarter the last four quarters of the time series of Wages and Salaries indices are revised.

An internal database of vintages exists and can be made available at request.

In the release of May 2018, with the first transmission of Q1:2018, the time series of Wages and Salaries indices were released for the first time in base 2015=100. This implied a revision of the entire time series. For the sectors of Retail trade the average revisions of the the year on year growth rates were, with respect to the previous data transmission (March 2018), the following:

• aggregate 47: MAR=0.3%;   MR= 0.0%; RMAR=0.1%.
  

 In the release of September 2018, with the first transmission of Q2:2018, average revisions of the year on year growth reates for the Retail trade sectors' estimates were:

• aggregate 47: MAR=0.1%;   MR=-0.1%; RMAR=0.0%.

affecting only the last 4 quarters (see §17.1) and mainly due to revsions in the administrative data used to estimate SME's employees.

Being the average revision statistics calculated as follows:

MAR= n^-1 Σ_t=1,n|L_t- P_t| = n^-1 Σ_t=1,n|R_t|

MR= n^-1 Σ_t=1,n(L_t- P_t) = n^-1 Σ_t=1,nR^t

RMAR= (Σ_t=1,n|R_t| ) / (Σ_t=1,n|P_t|)

Where L_t and P_t are respectively the last estimate and the first one.

Wages and Salaries are estimated on the basis of the integration of administrative data, used for SMEs and survey data, for LEs. Administrative data are stored by the Social Security Institute (INPS) in electronic format and delivered to Istat using inter-institutional electronic transmission. Data referring to the large enterprise survey are collected monthly by questionnaire, via website.

Analysis on non responses and outliers. Corrections (imputation) at micro (based on selective criteria) and macro level. Checks are carried out via both automated procedures and experts’ analyses on data.

For the large enterprises sub population, reporting units may also be contacted again in order to validate or correct the data.

The files that are sent to Eurostat are produced from data stored in an Oracle database via a generalised Istat software. After their production, they are not checked with any further software or specialized tool.

Once admin data on DMs are acquired (about 10 million records per month) the monthly basic variables are calculated at unit level through a complex pre-treatment procedure based on a data base of metadata on rules, laws and regulations on social security contributions declaration. At the end of the process information on the main target variables referred to each unit identification is summarized in a single record (about 1.4 million records per month). Finally, quarterly variables are calculated as simple means of the related monthly statistical variables retrieved as above mentioned.  Once the main quarterly variables have been derived from DMs, the survey target population has to be outlined, excluding the out-of-scope units  (public sector or whose economic activity is not included in the target Nace). This operation is performed using the auxiliary information drawn from the Business Register (ASIA) and from the Tax Register. 

At the scheduled time for the acquisition of the provisional population, it may happens that some DM are missing due to delays depending on firms liability or administrative system flaws. These missing units (late reporters) usually correspond to about 2-5% of the final population. That means that in the admin data an almost complete coverage of the target population occurs. Nevertheless, late reporting has a non-negligible impact in the estimates of variables expressed as totals, like Wages and salaries: even small differences from the final values could give misleading signals on the short dynamic of the target variables. In the Oros process, the jobs level estimate is massively adjusted to correct for the incompleteness of the preliminary data file due to the late reporters and Wages and salaries are adjusted proportionally under the hypothesis that the per-capita wages are not affected by late reporting. Micro imputation is the approach used in this phase of the process. The micro imputation procedure consists of two steps: a) the identification stage, that is the definition of a list of non-reporting units to impute; b) the imputation stage, that is the assignment of imputed values to the list of units above defined. Due to the absence of a theoretical list of active units for the admin data, a list of non-reporting active units has to be predicted. It’s derived on the basis of the structure of the preliminary admin data itself: based on the observation of the reporting patterns of the units. In the second imputation phase, the value of the units identified as active (expected late reporters) is imputed. The availability of a large quantity of micro data at longitudinal level on non-reporting units and considering the inertial trend of employment, a natural choice to reconstruct the missing values of jobs is to use a general regression model, where only the lagged values of the same variable are used as auxiliary variables, in detail, job values on the previous month and on the same month of the previous year. The adjusted estimates on jobs are applied to the per-capita wages calculated before imputation to get total wages.

Oros quarterly indicators derives from the integration of administrative and Large Enterprises survey (LES) data. The integration process aims at producing quarterly micro data by replacing the admin source with the large firms survey data, for the overlapping enterprises. This integration improves the estimates’ quality given both the higher quality of survey data and their adherence to statistical contents and of the overall admin-LES (Oros) estimates in case of missing response (MR), considered the relevant influence of large enterprises especially in some economic activity sectors. Producing quarterly integrated admin-LES microdata implies identifying and excluding from the admin source the enterprises belonging to the LEs survey and estimating coherent variables from data collected for different purposes. This process implies taking into account linkage and variables harmonization issues. The main phase of the integration process is the definition of quarterly lists of common units in both admin source and LES. To  identify correctly this sub-population avoiding double counting, record linkage and micro-integration are implemented. For units belonging to the defined quarterly list, OROS economic variables are replaced with the LES ones. The higher detail of LES variables on employment allows to perform coherent estimates from the two sources at the basis of Oros. To this scope an harmonization procedure according to the statistical requirements is carried out. Before the aggregation of the main variables, units not belonging to LES (already including E&I operations from Les survey) undergo an editing and imputation procedure. This operation is aimed at identifying and correcting missing response and/or outliers that could generate bias in the estimates both in preliminary and final data. It is based on selective criteria: influent units (less than 50 units a quarter) are automatically detected through functional relations on information on the same quarter of previous year (t-4), according to established cut-off thresholds. Specifically, the quarter-on-quarter variations are the target variables of the selective editing procedure,  to better detect outlier incorporating seasonality, contributing to reduce the number of influent units. The most anomalous values or missing response are interactively analyzed and, if necessary, imputed by an automatic deterministic procedure.

For furter details on the methodology at the basis of the Oros process see:

Istat, 2019. La Rilevazione trimestrale Oros su occupazione e costo del lavoro: indicatori e metodologie. Letture Statistiche. Metodi. Rome. https://www.istat.it/it/archivio/229033.

Congia M.C., S. Pacini e D. Tuzi. 2008. Quality Challenges in Processing Administrative Data to Produce Short-Term Labour Cost Statistics. Paper presented at the Conference: the European Conference on Quality in Official Statistics, Q2008. Rome, 8-11 July. https://pdfs.semanticscholar.org/a0c9/6671ad72deb41b845261ada85674091d96aa.pdf.

Rapiti F.M., F. Ceccato, M.C. Congia, S. Pacini e D. Tuzi. 2010. What have we learned in almost 10-years experience in dealing with administrative data for short term employment and wages indicators? Paper presented at the seminar: Using administrative data in the production of business statistics”. Roma, 18-19 marzo. http://www.ine.pt/filme_inst/essnet/papers/Session2/Paper2.4.pdf.

The indices on Wages and Salaries are available only in unadjusted form.

None