Retail trade

Statistical Office of the Republic of Serbia (SORS)

Domestic Trade, Catering and Tourism Statistics Division

Danijela Mladenović

Statistician methodologist

Milana Rakića 5, Belgrade

danijela.mladenovic@stat.gov.rs

+381112412922

+381112411922

The main purpose of data collecting is providing the indicators on turnover in retail trade on monthly basis. Data are collected by statistical surveys Monthly Report on Retail Trade (Form TRG-10) and Quarterly Survey on Retail Trade (Form TRG-16) and by using administrative data (Tax Administration data from Value Added Tax).

The objective of the retail trade turnover index is to show the development of the consumer goods market and measure the change of retail sales. As an important short-term indicator, retail trade turnover index is used to analyse economic situation and as an indicator of individual consumption and input for National accounts statistics.

Indices on the retail trade turnover are presented in two ways:

·  for retail trade turnover including VAT (for national purposes);

·  for retail trade turnover excluding VAT (obligations according to Eurostat).

Indices on retail trade turnover are calculated and computed at current and constant prices. Indices at current prices are nominal indices that present trend of turnover value between the two periods. Indices at constant prices are real indices obtained by deflating indices at current prices with the corresponding consumer price indices.

Classification of activities CA (“Official Gazette of RS”, No 54/10), which is without any changes based on Statistical Classification of Economic Activities in EU – NACE Rev. 2.

National Nomenclature of Statistical Territorial Units.

Surveys listed in scope of section 3.1 involve:

- enterprises (legal entities) whose principal activity is classified in Retail trade, excluding trade of motor vehicles and motorcycles (division 47);

- smaller number of legal entities whose principal activity is classified in other CA sections, but they realize significant turnover in retail trade as secondary activity.

Administrative data source – included are unincorporated enterprises registered in value added tax system (VAT payers) within the section of Retail trade, excluding trade of motor vehicles and motorcycles (division 47 of CA), as well as bakeries and butchers.

Main statistical variable is turnover. Data collected refer to retail trade turnover including VAT and the amount of VAT which are calculated on invoiced turnover (included in the sales price to customers).

Turnover excluding VAT (for Eurostat needs) is derived from these two variables.

Retail trade turnover represents the sales of goods to final consumers (primarily to individuals for personal consumption, for use in the households, as well as to other consumers who are supplied at retail prices). The turnover of goods in retail trade also includes the sale of goods to individuals through consignments, whether performed from a warehouse or a store.

Retail trade turnover comprises the invoiced value from sales of goods during the reference period, no regardless it was paid or not. Turnover includes excise duties and excludes income from sales of fixed assets, as well as other operating income, financial income and other extraordinary income (rentals, subsidies, etc.).

Deflated turnover presents turnover at constant prices.

Indices in retail trade turnover are calculated with VAT (for national purposes) and without VAT (for Eurostat), in accordance with concepts and definitions from Eurostat Regulations that refers to short-term business statistics.

Reporting unit is an enterprise.

Statistical observation units are enterprises with principal activity classified under Division 47 of CA as well as KAU (kind of statistical unit) for enterprises not classified in CA division 47, but such units perform significant retail trade turnover.

Administrative data source - Tax Administration is the reporting unit; statistical observation units are VAT returns of unincorporated enterprises performing retail trade activity.

The basic population (target population) includes all economically active businesses entities with principal activity classified in Retail trade, excluding trade of motor vehicles and motorcycles (Division 47 of NACE Rev. 2). Also, statistical population includes certain number of business entities that perform retail trade, but according to principal activity they are classified in other activities.

Legal entities are surveyed in monthly and quarterly surveys, based on random samples. The goal of the monthly survey is to present trend tendency of retail trade turnover, for the level of statistical territorial units (Republic of Serbia, Srbija – Sever, Srbija – Jug) and activity branches by CA. The monthly sample is a subsample of the quarterly survey. The quarterly survey, conducted on a larger number of units in the sample provides final results.

The sampling frame for both surveys is based on statistical business register (SBR) which includes about 6500 enterprises in retail trade activity that have reported annual financial statements (for the year t-2, where t is the reference year). The sampling frame was constructed from this list of enterprises by excluding the smallest enterprises. Annual turnover and size class of the enterprise are used as size measure. Included are all large and medium enterprises, as well as small enterprises with annual turnover of more than 2 million RSD. Annual turnover and size classes (large, medium, small and micro) for enterprises are taken from financial statements for the year t-2, where t is the current year. Size of the enterprise is defined according to the annual turnover, number of employees and assets of the enterprise. Excluded enterprises make up less than 1% of turnover. The sampling frame includes about 4300 enterprises out of which 50 of them are not registered in retail trade as their principal activity. About 600 enterprises are surveyed every month. The sample size for the quarterly survey is larger, amounting to approximately 1900 enterprises.

Coverage of unincorporated enterprises performing retail trade activity is based on VAT returns received from Tax Administration.

The average number of monthly VAT returns obtained for the retail trade activity (division 47 of NACE Rev. 2) and bakery and butcher shops involves about 6000 unincorporated enterprises, while the average number of quarterly VAT returns accounted about 11000 unincorporated enterprises.

Republic of Serbia

Note: Starting from 1999 the Statistical Office of the Republic of Serbia has not at disposal and may not provide available certain data relative to AP Kosovo and Metohija and therefore these data are not included in the coverage for the Republic of Serbia (total).

Comparable time series of retail trade indices are available from January 2000 onwards. Historical time series for the period 2000-2010 were recalculated in accordance with NACE Rev. 2 classification, so the time series can be considered continuous.

In accordance with Eurostat Regulation for short-term indicators, indices are computed on 2015 base year.

Monthly indices of retail trade turnover are issued relative to “the same month of the previous year”, “previous month” and “the same period of the previous year”.

Moreover, calculated and issued are the corresponding indices on quarterly and annual level.

Combination of statistical surveys (for legal entities) and administrative data source (for unincorporated enterprises) is used.

Enterprises (legal entities)

For statistical surveys, the sampling frame is based on the data of the Statistical Business Register, as of 31^st of December of the previous year. Stratification of the sampling frame is done according to economic activity of enterprise, size classes according to financial statements and amount of annual turnover (see 3.3 and 3.6 above).

Stratification according to economic activity of enterprises is into 2 classes (47.11, 47.19), 8 groups (47.2, 47.3, 47.4, 47.5, 47.6, 47.7, 47.8, 47.9), and into one special class reserved for those enterprises whose principal activity is not retail trade.

Stratification according to size of the enterprise is into large, medium, small and micro. Size classes are taken from financial statements and they are defined according to the annual turnover, number of employees and assets of the enterprise.

Within each class defined by cross classifying size class and economic activity group/class, enterprises are further stratified into 2 classes according to the annual turnover: larger and smaller. Units with largest turnover; large and medium units by size and those with businesses related to retail trade, but with the principal activity outside retail trade, presented census units. A random sample is selected from strata marked as ‘smaller’, all other strata are census strata, their units are completely enumerated. Allocation of sample was done using the Bethel algorithm (Bethel, J. “Sample allocation in multivariate surveys”, Survey methodology, 15, 1989: 47–57). Turnover was used as the auxiliary information. In this way, a sample of 1822 units was allocated for quarterly survey and sub-sample of 595 units for monthly survey.

Unincorporated enterprises

In order to provide data on retail trade turnover from unincorporated enterprises, administrative source are used, i.e. VAT returns obtained from Tax Administration.

For statistical surveys, periodicity of data collection is monthly and quarterly.

Data from the Tax Administration (VAT returns) are available in monthly periodicity for large taxpayers and in the quarterly periodicity for small taxpayers.

In statistical surveys, data are collected by post (questionnaires), web questionnaires, mail and fax.

The questionnaires used in the surveys are Monthly Retail Trade Report (TRG-10 form) and Quarterly Retail Trade Report (TRG-16 form). Deadline for delivery of the completed questionnaire is the 15^th calendar day following the end of the reference month (monthly survey), i.e. 15^th calendar day following the end of the reference quarter (quarterly survey).

Actions taken to speed up or increase the rate of response are written reminders sent by mail and e-mail, usually 1-2 days after the time given for the replies. There is also re-contacts by phone. In the monthly survey, data collection is closed 25 days after the end of the reference month.

Data on VAT returns are gained from Tax Administration on CD.

There are the following stages of data validation: control of individual data (at micro level) and control of aggregated data and indices.

In order to ensure data quality, simultaneously with collecting and coding, data are visually checked by statisticians for missing, invalid or inconsistent questionnaires or values, and if needed, the reporting units are contacted for additional clarifications. Afterwards, checking is done during the process of data entry and during data processing. Data checking is performed in accordance with the defined criteria of logical and computer control within IST application (statistical information system fully developed in SORS) and error corrections are carried out.

Data are compared with the previous period of the same year, as well as with same period of previous year. If necessary, the respondents are contacted in order to clarify the deviation and correct recognized errors. There is the remark field in the questionnaire for any remark and note with the explanation of the unit (especially in case of large increase or decrease in turnover relative to the previous period).

Possible deviations, observed during the estimation of totals by the sample experts are also checked subsequently.

Statistical results (aggregated data) for the reference month are also compared to the results of the previous month and with the corresponding month of the previous year at each level of aggregation.

The prepared indices are also checked in detail before publication and data transmission to Eurostat.

The methods used to treat non-response:

imputation for large non-respondents is based on the previous values for the same enterprise or value of the previous periods adjusted by the growth rate of the equivalent period of the previous year; the other units are treated by re-weighting (weights are adjusted to take account of non-response).

The estimations of totals are calculated pursuant to standard procedure for stratified random sample (Horvitz-Thompson estimates). This means that estimates of totals are calculated by summing the weighted values from the sample. Weight for the units of the same stratum is equal to the quotient of the number of units in the stratum and the number of units in the realized sample in the stratum. The realized sample consists of the units that filled out the questionnaire, for which the data were evaluated and for a part of the over-covered units for which the response information is one of the following: not working; has an activity out-of-scope or bankruptcy. The rest of the units are classified into the unrealized sample. Unrealized sample also includes part of over-coverage units (for reasons of "closed" or "in liquidation") in order to compensate, by number, newly created units that a sampling frame does not contain.

Enterprises that were detected as outliers with respect to turnover are put into special census strata. The weights of the remaining units in the strata from which they originated were corrected.

The retail trade turnover index is compiled as a simple index (value relative). The indices are calculated directly from the data (obtained estimates). The retail trade turnover indices are obtained by deflating the turnover index at current process with the appropriate consumer price index, which exclude: water (from public utility systems), electricity and motor vehicles, motorcycles and parts thereof.

Seasonally adjusted indices of retail trade turnover are calculated. More information on time – series adjustment is provided in the Annex.

Following the model of the European Statistical System, and in accordance with the mission and vision, as well as the Quality Policy, the Statistical Office of the Republic of Serbia strives to harmonize statistical production with the European Code of Practice (CoP).

The quality management system of the Statistical Office of the Republic of Serbia is based on the European Statistics Code of Practice and is fully adapted to statistical needs, because the quality of statistical processes is being systematically improved, as well as the final results, i.e. data, and the provision of services to users.

The document Quality Policy is available on SORS website:

http://www.stat.gov.rs/media/2485/quality-policy.pdf

Data quality and surveys’ conducting are monitored throughout all stages of data production, by the responsible statisticians, sample and time series experts, according to specifications for concrete surveys, and quality is at a very satisfactory level.

Statistical activities are directed to monitoring of the main quality aspects in scope of statistical surveys’ organization (data collection, coverage, data processing, punctuality and timeliness).

The main data users are state administration bodies (Government, Ministry of Trade, Tourism and Telecommunications, Ministry of Finance), Chamber of Commerce, business entities, economic analysts, media representatives, researchers, students, internal users (National accounts statistics, Prompt and complex reporting and public policies support division), international users (Eurostat, IMF).

The Statistical Office of the Republic of Serbia conducted in the period October 10-24, 2017, for the fourth time, the User satisfaction survey; the aim of the Survey is to gain the information on users’ needs, their satisfaction with SORS data and services, as well as providing information on quality of data and services offered by SORS.

The questionnaire consisted of six segments: general information on the respondents, general aspects of data access and use, statistical data quality, data dissemination, communication with SORS employees and users’ recommendations.

Overall quality of statistical data was estimated with the average of 4.1. The time of data publishing relative to the referent period of the data (timeliness) was estimated with average grade of 3.78. The level of coherency/comparability of the statistical data was estimated with 3.79.

When the total quality estimate of the data provided by SORS is compared with the previous survey from 2015, it is apparent that satisfaction with total quality has increased by 0.21 p.p.

The results of the User satisfaction survey, 2017 are available on SORS website:

http://www.stat.gov.rs/media/2490/rezultati-istrazivanja-o-zadovoljstvu-korisnika-2017en.pdf

The survey on users’ satisfaction which separately relates to short-term indicators has not been conducted yet.

All statistics according to Eurostat STS Requirements – Annex C, variables C120 and C123 are available.

Compilation and dissemination of Retail trade turnover indices is harmonized with the relevant Eurostat STS Regulation, thus providing 100% data completeness.

The retail trade turnover is computed by combining estimates on turnover from statistical surveys and administrative data.

Enterprises (legal entities)

The main error sources in statistical surveys (mentioned under point 3.1.) are of sampling and non- sampling nature.

The surveys are based on random sample. Sampling frame is constructed using enterprise data from the Statistical Business Register (SBR).

The sample design is such that the estimates for the main domains are of satisfactory precision.

The following non-sampling errors are present:

· coverage errors due to not completely updated SBR and existing time lag in registering the changes of enterprise units;

· errors that are a consequence of cut-off sampling, smallest enterprises are excluded from the sampling frame;

· measurement and non-response errors that are inevitable, but the instruments of data collection, organization and data entry are such that they are reduced to the lowest possible level.

The general assessment is that the type and design of survey and data collection methods ensure sufficient overall accuracy.

Unincorporated enterprises (Entrepreneurs)

Since data on turnover of unincorporated enterprises are derived from VAT returns, obtained from Tax Administration, unincorporated enterprises (entrepreneurs) which are not obliged to VAT calculating are not covered.

For quarterly VAT payers, monthly turnover values are estimated (for each month of the quarter).

The data (results) are final 60 days after the reference quarter for the months of that quarter, after quarterly data revision, when the preliminary results obtained on the basis of the monthly survey are replaced with final data from quarterly survey (t+60) that is carried out on a larger number of units in the sample and when data from Tax Administration for quarterly taxpayers are available.

Enterprises (legal entities)

Sampling error is estimated for monthly turnover and chain index in retail trade and is presented by the corresponding coefficients of variation for the estimation domains. Sampling weights that are used in estimation are corrected for non-response. For total monthly turnover, estimates of the coefficients of variation are in the interval 1.1 -1.4%. The standard error of chain index is less than 1%, monthly average equals 0.4%.

In the Annex referring to Accuracy is a table with monthly estimates of coefficients of variation by main economic classes in retail trade.

The average coefficient of variation for the estimated total monthly turnover achieved in NACE Rev.2, CA division 47 at total level of the Republic of Serbia is 1.2%.

In the Annex is a table with monthly estimates of coefficients of variation by statistical territorial units in retail trade.

Enterprises (legal entities)

Non-sampling errors can arise in stages of survey activities. Non-sampling errors can arise due to: coverage errors, non-response, processing, etc.

The magnitude of the under-coverage error is not computed as it is not considered significant. These errors are mainly caused by not updated Statistical Business Register that is used for constructing the sampling frame.

The smallest enterprises according to turnover in annual financial statements for year t-2, where t is the survey reference year, are excluded (cut-off) from the sampling frame.

The over-coverage and non-response errors (or the rate of response) are computed using the coded information on response that is classified in 9 categories:

1. Runs business, submits report;

2. Runs business, refuses to submit a report;

3. Not found (undeliverable questionnaire);

4. Closed, extinguished;

5. Blocked, not working;

6. Outside the scope of survey (does not perform selected activity);

7. Bankruptcy;

8. Liquidation;

9. Other reasons (e.g. unknown reason of non-response).

Over-coverage units, determined not to belong to the target universe for the survey, are those for which the coded information on response is 4, 5, 6, 7 or 8. The average rate of over-coverage was 9.1% in 2017. The main reasons for over-coverage are ‘not working’ and ‘outside the scope of the survey’.

Non-response units are in-scope units for which data were not collected. Non-response units are those for which the response code is 2, 3, or 9. The average non-response rate was 8.2% in 2017.

In order to reduce non-response and other errors, the following measures are undertaken: web questionnaire is introduced, written and e-mail remainders are sent, units are contacted by phone, collected data are entered in the application that represents the so-called "controlled entry". During data collection and data entry phase, the program does not allow the entry of false data, it is checked whether the questionnaires are filled out completely. The next step refers to computer logical control. Further steps in data editing are based on the analysis of statistical experts so as to discover and correct possible errors. If necessary, reporting unit is contacted in order to get correct data. Final data validation is realized during analysis of aggregates, horizontal logical dependences, vertical relations, minimum/maximum values etc.

In the Annex referring to Accuracy are tables with monthly un-weighted over-coverage and non-response rates by main economic classes in retail trade.

Unincorporated enterprises (Entrepreneurs)

VAT returns obtained from Tax Administration are used for estimation of entrepreneurs’ turnover.

Under-coverage is caused by small units which are not VAT payers, but it is expected that it does not have a significant impact on total retail trade turnover index.

Analysis of administrative data is focused on the quality of data from the tax return of value added tax. The tax data are thoroughly checked, both for individual months/quarters as well as over months and quarters. Turnover and VAT values of entrepreneurs are sorted in the decreasing order to detect values that are suspicious for errors. For those values that are suspected to be with errors, further checks are conducted by comparison with the data from the previous months/quarters.

Over-coverage errors occur when the sampling frame contains units determined not to belong to the target universe of the survey. Over-coverage units are those for which the coded information on response is 'Closed, extinguished'; 'Blocked, not working'; 'Outside the scope of survey (does not perform the selected activity)'; or 'Bankruptcy'. The average rate of over-coverage on the level of the Republic of Serbia was 9.1% in 2017. The main reasons for over-coverage are ‘Blocked, not working’ and ‘outside the scope of the survey’.

The under-coverage error was not computed.

The following table shows the average over-coverage rates:

In the Annex is a table with monthly estimates of over-coverage rates by statistical territorial units in retail trade.

Not applied.

Not calculated.

Non-response units are those for which the response code is 'Runs business, refuses to submit a report'; 'Not found (undeliverable questionnaire)' or 'Other reasons (e.g. organizational changes)'.

The non-response rate on the level of Republic of Serbia is 8.2%.

In order to reduce the non-response, the following measures are undertaken: introduction of Web questionnaire, written and e-mail remainders, telephone contacts with the reporting units. Once a year, a new coordinated rotating sample is selected so that some non-responses are replaced. Every effort is made to reduce non-sampling error to a minimum.

The following table shows the average non-response rates:

In the Annex is a table with monthly estimates of non-response rates by statistical territorial units in retail trade.

Not applied.

Processing errors are corrected during the editing process and process of data validation.

Not calculated.

Not applied.

General Revision Policy is available on the website of SORS:

http://www.stat.gov.rs/media/2332/general-revision-policy.docx

Routine revisions - all published Retail trade turnover indices in Monthly statistical release are treated as preliminary, meaning that certain corrections can appear on the basis of the estimated results, obtained from the regular quarterly survey that are conducted on a larger number of units in the sample. Also, every detected error is corrected on that occasion, as well as any correction, initiated by data providers is performed. Routine revisions are made when preliminary results from monthly survey are replaced with final data from quarterly survey and when the Tax Administration database is completed with data for quarterly tax payers. Retail trade turnover are usually revised four times a year (each time for the months which belong to the specific quarter).

Regarding seasonally adjusted indices, by adding new data, the whole series is updated.

Major revisions that are performed due to changes in CA (2010) or changed base year, performed is so-called back casting on order to obtain comparable data series. Explanations are provided at the time of data publishing or when data dissemination is performed.

The same revision policy is applied nationally and in transmissions of the data to Eurostat.

Preliminary results of retail trade are published on the last day in a month for the previous (referent) month. Preliminary data are corrected by using the final monthly data of the quarterly survey (t+60) that is conducted on a larger number of units in the sample.

There is no time lag between the actual delivery of the data and the target date when it should have been delivered. Deadlines are respected and data are published and delivered on time, according to the pre-announced Release calendar.

The same statistical concepts are applied for the entire territory.

Not applicable.

The time series of the presented indicators are comparable over time.

Length of the comparable time series in retail trade turnover can be observed from January 2000 until the last reporting month.

Data from Monthly report on retail trade are compared with the data from Quarterly Survey on retail trade and completely coherent.

Comparison with VAT data does not indicate major discrepancies between available data from administrative sources and results of the survey.

Occasionally, comparisons with other statistical sources covering turnover, quarterly Structural Business Statistics data, are carried out, but it should be noted that there are certain methodological differences and some differences in definitions.

Results of retail trade are used for creation of Quarterly National Accounts.

Not applicable.

Not applicable.

Retail trade data are internally coherent. In order to check the internal coherence, comparison is carried out between the data of the previous month and data of the reference month, as well as comparison of the current data with the data of the corresponding period of the previous year.

All aggregated data are derived on the basis of individual data.

Data on retail trade turnover are published in monthly periodicity, in Statistical release ‘’Retail trade turnover’’ and in Press release, on SORS website

http://www.stat.gov.rs/en-us/oblasti/unutrasnja-trgovina/trgovina-na-malo/ simultaneously, in Serbian and in English, according to Release calendar.

Domestic trade turnover data are published in Statistical release on a quarterly basis. The data sets are published for Divisions 45, 46 and 47 of the Classification of Activities.

“Monthly Statistical Bulletin” - statistical data are presented in quarterly and monthly series, by the territorial level and by Classification of Activities. Monthly series comprise period of the last 13 months;

“Trends” - a quarterly publication providing information on current trends in economy;

“Statistical Yearbook of the Republic of Serbia” - an comprehensive annual publication compiles all final data and series of the most important indicators from different areas of economic and social life of the country;

“Statistical Pocketbook of the RS” - an edition that offers the principal statistical data on national social and economic development.

Retail trade data (monthly, quarterly and annual indicators) are available in the Statistical Database in scope of the statistical area of Domestic Trade:

http://data.stat.gov.rs/?caller=SDDB&languageCode=en-US

Information not available.

Micro-data are not disseminated to users.

Monthly indices on retail trade at current and at constant prices, seasonally adjusted and working-day adjusted indices are transmitted monthly to Eurostat, starting from the first month of 2000, in accordance with the requested level of details, deadlines and required form.

On users’ request, SORS may provide the data and information obtained by special processing.

Methodological explanations and statistical questionnaires on retail trade are available on the website:

http://www.stat.gov.rs/en-us/istrazivanja/methodology-and-documents/?a=21&s=2101

http://www.stat.gov.rs/en-us/o-nama/sistem-upravljanja-kvalitetom/

Not calculated.

Not applicable.

Standardized information on cost and burden is not systematically collected.

A system of coordination of random samples of business entities has been introduced, reducing the burden of small units.

National legislation: statistical confidentiality is covered by the Law on Official statistics (Official Gazette of RS, No 104/2009) and the Rulebook on Statistical Data Protection in SORS.

European legislation: Regulation (EC) No 223/2009 of the European Parliament and of the Council of 11 March 2009 on European statistics.

Confidential data may be used for statistical purposes only. Confidential statistical data are individual data as well as aggregates on which the data holder may be recognized in any way.

Confidential data are considered official secret and cannot be published or communicated, that is, they cannot be part of aggregated data from which individual data can be identified.

When processing confidential data for users’ needs, data protection must be provided so that individual data cannot be identified on either level of aggregated data.

No comment.