Report of the Task Force on International Data Cooperation Presented to the Meeting of the Inter-Agency Group on Economic and Financial Statistics September 23 and 24, 2013 1 Table of Contents List of Acronyms ............................................................................................................................ 2 Executive Summary ........................................................................................................................ 3 I. Introduction ................................................................................................................................. 4 II. International Data Cooperation: General Principles .................................................................. 4 III. Data Cooperation for GDP Main Aggregates and Population, and Sectoral Accounts ............ 5 A. Background ............................................................................................................................ 5 B. Mandate and Objectives ......................................................................................................... 6 C. Deliverables and Timing ........................................................................................................ 6 IV. Pilot Exercises .......................................................................................................................... 7 A. Scope of the Pilot Exercises ................................................................................................... 7 Datasets ................................................................................................................................... 8 Pilot Countries ........................................................................................................................ 9 Expected results and timeframes .......................................................................................... 10 Criteria for evaluation ........................................................................................................... 10 B. Templates ............................................................................................................................. 10 C. SDMX Infrastructure in Member Agencies ......................................................................... 12 Review of the Infrastructure ................................................................................................. 12 Implications for Pilot Exercises ............................................................................................ 16 V. Current Workflow Practices and Workflow for the pilots ....................................................... 16 A. Data transmission from country to international agencies ................................................... 17 B. Sharing of validated data among international agencies ...................................................... 17 ANNEX I: Members of the TFIDC .............................................................................................. 22 ANNEX II: Country coverage for the Pilot Exercise (by group of countries) ............................ 23 ANNEX III: Template for GDP Main Aggregates and Population .............................................. 25 ANNEX IV: SDMX Survey Questionnaire .................................................................................. 27 ANNEX V: Summary of SDMX Survey Results ......................................................................... 30 ANNEX VI: Data Workflow Practices ......................................................................................... 35 ANNEX VII: Eurostat-OECD Protocol: Agreed Minimum Checking Rules .............................. 42 2 LIST OF ACRONYMS 2008 SNA API BIS DSD ECB ECB SDW EDAMIS ESA Eurostat GPG IAG MSD OECD PGI SDMX SDMX-EDI SDMX-ML SDMX-NA TFIDC UNECE UNSD System of National Accounts, 2008 Application Programming Interface Bank for International Settlements Data Structure Definitions European Central Bank ECB Statistical Data Warehouse Electronic Dataflow Administration and Management Information System European System of Accounts Statistical Office of the European Union GNU Privacy Guard Inter-Agency Group on Economic and Financial Statistics Metadata Structure Definitions Organisation for Economic Cooperation and Development Principal Global Indicators Statistical Data and Metadata Exchange SDMX using EDIFACT Syntax SDMX using XML Syntax SDMX in National Accounts Task Force for International Data Cooperation United Nations Economic Commission for Europe United Nations Statistics Division 3 EXECUTIVE SUMMARY The Inter-Agency Group on Economic and Financial Statistics (IAG) has emphasized the need to improve cooperation among international and supranational agencies in terms of collecting, validating, and disseminating public official statistics from national and international/supranational sources. In this regard, the IAG has established the Task Force on International Data Cooperation (TFIDC) to examine the elements and undertake pilot exercises on a framework that would allow member countries of the agencies represented on the IAG to submit data only once, and for these data to be shared among the member agencies. The overall objective of the TFIDC is to determine the procedures that could be applied for a successful data cooperation arrangement across international agencies that would streamline and improve the efficiency of data collection, sharing, and dissemination. To meet this objective, the TFIDC will oversee two pilot exercises for (i) GDP main aggregates and population, and (ii) sectoral accounts. The pilot exercises will begin in September 2013 and will run for about two years. In preparation for the exercises, the TFIDC has agreed on data collection templates for each of the datasets. The template for GDP main aggregates and population comprises a set of main national accounts aggregates and auxiliary indicators that are widely used and are available for a large number of countries. The TFIDC will use the templates on sectoral accounts that have been developed by the Working Group on Sectoral Accounts and endorsed by the IAG. The cooperation with regard to the agreed datasets should demonstrate that the exchange of data among international agencies is technically feasible for both sending and receiving agencies. This should result in the availability of consistent data across international agencies and efficiency gains for all parties. In general, participating international agencies are expected to use the pilot exercises to evaluate the benefits and costs associated with international data cooperation. In particular, the pilot exercises will determine the following: (i) whether the current data collection practices of the respective agencies meet the requirements of the templates; (ii) where gaps exist in terms of country coverage; (iii) the responsibilities for the validation of the data, the validation principles to be used, and the timeframe for validation (iv) a suitable schedule for the transmission of data from national agencies to the collecting international agencies that will satisfy the timetables of all receiving international agencies (with the implication that the strictest schedule currently in place among all the agencies may have to be followed; and (v) the IT infrastructure in terms of the ability of the agencies to receive and send (where appropriate) national accounts data using SDMX-ML with the appropriate 2008 SNA data structure definition. Overall, the TFIDC has determined that the technical systems of the participating agencies are capable of handling the SDMX formats for data exchange and are generally ready for the pilot exercises. In the medium-term, data cooperation among the agencies may render unnecessary, multiple collection and validation of data from countries. These agencies may focus on having the data reported to only one international agency. 4 I. INTRODUCTION 1. The Inter-Agency Group on Economic and Financial Statistics ((IAG1) plays a key monitoring and coordinating role in the implementation of the recommendations made in the Report to the G20 on Data Gaps and the Financial Crisis by the International Monetary Fund and the Financial Stability Board2. In addition to addressing the recommendations made in the report, the IAG has focused its attention to improve the practical cooperation between international and supranational agencies in terms of collecting, validating and disseminating public official statistics from national and international/supranational sources. 2. The Task Force on International Data Cooperation (TFIDC) was established by the IAG and began its work in early 2013. The purpose of the TFIDC is to examine the elements and undertake pilot exercises on a framework that would allow member countries of the agencies represented on the IAG to submit data only once, and for these data to be shared among the member agencies. 3. This report provides an update on the work of the TFIDC and outlines the details for undertaking the pilot exercises. Section II outlines the general principles underlying international data cooperation; Section III reviews the background and mandate of the data cooperation arrangements for GDP main aggregates and population, and sectoral accounts; Section IV reviews the scope of the pilot exercise, the templates for the datasets, and SDMX infrastructure existing in international agencies; and Section V reviews the workflow practices for the pilot exercises as well as presents the timing of the pilot exercises. II. INTERNATIONAL DATA COOPERATION: GENERAL PRINCIPLES 4. The IAG has emphasized the need to improve cooperation among international and supranational agencies in terms of collecting, validating, and disseminating public official statistics from national and international/supranational sources. 5. The general objectives of the improved cooperation are as follows: reduce the reporting burden on national authorities; make more efficient use of resources at the national and international/supranational agencies; 1 The IAG includes the Bank for International Settlements (BIS), the European Central Bank (ECB), the Statistical Office of the European Union (Eurostat), the International Monetary Fund (IMF), the Organization for Economic Co-operation and Development (OECD), the United Nations (UN) and the World Bank. The IMF chairs the IAG and provides its secretariat. 2 See http://www.imf.org/external/np/g20/pdf/102909.pdf. 5 ensure that the economic and financial data and related metadata in the databases of international/supranational agencies are identical for the same statistical concepts and are of the highest quality, including in terms of frequency and timeliness; and improve the dissemination to users globally of a more consistent set of economic and financial data. 6. The IAG has started to develop a set of cooperation guidelines for the international and supranational agencies that are willing to participate in the improvement of data flows amongst themselves. The guidelines are based on the following: the implementation of more aligned statistical concepts and definitions; the harmonization of reporting templates; the efficient exchange of data between international and supranational agencies, in particular the use of SDMX technical standards and data structure definitions; the allocation of responsibilities among these agencies with respect to data validation and quality assurance; making use, to the extent possible, of existing data cooperation arrangements already in place; the implementation of more efficient and consistent data flows to and from various data hubs, both across agencies and on public websites; and more formal arrangements between international agencies where appropriate. III. DATA COOPERATION FOR GDP MAIN AGGREGATES AND POPULATION, AND SECTORAL ACCOUNTS A. Background 7. The TFIDC organized its first physical meeting in April 2013. The meeting discussed a framework for its activities leading up to the implementation of the pilot exercises on the data cooperation arrangements for sectoral accounts, GDP and population statistics. Before and after the April 2013 meeting, the TFIDC arranged its work via electronic means (e-mails, teleconferences, and video conference). 8. The IAG has mandated that the TFIDC undertake two pilot exercises to test the cooperation of national accounts datasets in line with the draft general principles outlined above. The goal of the pilot exercises is to develop a set of commonly shared principles and working arrangements for data cooperation that could be implemented by the international agencies. One pilot exercise will test the arrangements for a set of GDP main aggregates and population data with the widest possible representation of countries and agencies. The other pilot exercise will test arrangements relating to institutional sector accounts within the scope of the G20 Recommendations (Recommendation #15 of the G20 Data Gaps Initiative). 6 B. Mandate and Objectives 9. The overall objective of the TFIDC is to determine the procedures that could be applied for a successful data cooperation arrangement across international agencies that would streamline and improve the efficiency of data collection, sharing, and dissemination. Thus, the TFIDC is expected to develop the framework for the data cooperation process—with the work processes clearly outlined—and recommend a feasible program for its implementation based on the outcomes the pilot exercises. 10. The IAG has mandated the TFIDC to prepare, by September 2013, a report outlining the practical workflows among the international agencies and between the data-providing countries and international agencies in terms of collecting the relevant time series for each of the datasets covered in the pilots, the procedures to validate the collected data, sharing, and dissemination. 11. At the April 2013 meeting, The TFIDC agreed on a basic timeline for its work, including the schedule and anticipated outcomes of the pilot exercises. The mandate of the TFIDC has been revised to reflect the anticipated workflows and expected outcomes. Following the completion and review of the pilots, the initial informal guidelines outlined in the general principles will be reviewed and, depending on the outcome of the pilots, these principles may be applied to other datasets. C. Deliverables and Timing 12. The TFIDC is expected to deliver the following: the templates to be used for the pilots on data cooperation; the agreement about which international agency covers which countries in which pilot; the data flows from countries to international agencies; the workflows among the agencies; the clarification and establishment of the IT environment for the data cooperation exercise; transmission process to be used in case SDMX cannot be used directly by some agencies for a transitional period; feasibility of exchanging non-validated data; and processes beyond the successful pilots. The first six items are addressed in this first progress report, as they are instrumental in getting started with the pilot exercises. The two remaining items will be addressed at a later stage. 13. The TFIDC will function for an initial period of two years while undertaking the pilot exercises. A first review of the exercises will be undertaken nine months after the initiation of the pilot. 7 IV. PILOT EXERCISES A. Scope of the Pilot Exercises 14. The TFIDC will oversee pilot exercises covering (i) GDP main aggregates and population, and (ii) sectoral accounts. They will be launched in September 2013 and will run for about two years. 15. The pilot exercises will determine whether the current data collection practices by the respective agencies meet the requirements of the templates developed for the data cooperation (see Section IV.B for a description of the templates). It will therefore review the collection practices currently being followed by the agencies and will identify their data requirements in terms of the items presented in the templates. The exercises will also determine where gaps exist in terms of country coverage and the completeness of the templates. In this regard, the TFIDC will use the results of the pilot exercises to propose mechanisms to address the country coverage and data gaps. 16. In preparation for the pilots, the TFIDC conducted a review of the data collection and workflow practices in member agencies as they relate to the datasets under consideration. Thus, the review was conducted separately for GDP main aggregates, population, sectoral accounts. 17. The review determined that currently, the country coverage, range of data collected, and collection practices vary widely across the agencies (see Annex VI). The agencies focus on collecting data for their members and, in some cases, a few additional countries of interest. Therefore, country coverage ranges from currently 36 for ECB and Eurostat (covering the 28 European Union member countries, EFTA countries and Candidate countries) to 235 for the UNSD (covering all UN member countries as well as other areas and territories.) Whereas all the agencies collect data on GDP main aggregates, five collect data on sectoral accounts, these being the BIS, ECB, Eurostat, OECD and UNSD. Further, the number of countries covered by data collection for the sectoral accounts is significantly lower than the number of countries covered for GDP main aggregates. 18. While data specified in the proposed templates for GDP main aggregates and sectoral accounts are covered by the ESA Transmission Programme, also used in the context of OECD data reporting requirements and quarterly data reporting requirements for the ECB as regards sector accounts, the full dataset is not yet provided by all countries covered by these three institutions due to derogations or non-compliance issues. On the other hand, the World Bank has no data reporting requirements and collects the data from the national agencies through visiting World Bank missions. As regards the level of detail, the ECB, Eurostat and OECD collect a broad range of data by institutional sector and subsector, but the BIS and UNSD collect a limited number of items for the main institutional sectors. 19. The agencies follow various formal and informal data cooperation arrangements. Most of the arrangements involve the ECB, Eurostat or OECD as a partner. The OECD currently receives non-validated data from Eurostat but an agreement has been reached for the transmission of 8 validated data from Eurostat, as is currently the case with the ECB. The OECD also receives sectoral accounts data from Eurostat and the ECB. Sectoral accounts data received by the OECD from Eurostat and ECB and remaining OECD countries are transmitted to the IMF by the OECD for publication on the Principal Global Indicators (PGI) website. The IMF receives the GDP main aggregates data for most EU member countries from Eurostat in a bulk file (using SDMXEDI) on a monthly basis. It also gets additional data for non-EU countries from OECD.Stat. The agencies are currently working on a data exchange process where the IMF would directly query Eurostat's databases (“Pull” technology). The UNSD receives annual GDP data from OECD and other non-IAG regional agencies and collects data from other UN member states. 20. The data collection schedules of the agencies follow closely the national release calendars of countries but are designed to meet the publication schedules of the agencies. Therefore, the agencies may collect the data some time after these data are released by the national statistical agencies (usually on their websites) but a cutoff data for data transmission is usually prescribed by each IAG. The data collection schedules for the ECB and Eurostat are based on specific regulations although national release calendars allow the countries to submit the data to these agencies ahead of the prescribed timeframes. The pilot exercises will evaluate the existing processes for handling of data revisions which will lead to proposals for approaches to streamline processes for data revisions. 21. During the pilot exercises, the agencies will determine the time lapse between the release of the data by the national agencies and the transmission of the data by the receiving agencies. There are two aspects of this flow of data that will be monitored: 1. The lapse of time between the release of data by a national agency and the receipt by the collecting international agency; and 2. The amount of time taken to validate the data before transmission to other agencies. Datasets GDP Main Aggregates and Population 22. For GDP main aggregates and population, all agencies will continue to collect the data as they currently do, which implies that the same data may be collected more than once for some economies. In principle, only the primary validating international agency will push these data to the other interested agencies using SDMX, where feasible. However, for countries beyond the validation responsibilities of Eurostat, ECB and OECD, transmission from multiple international organizations may be possible during the pilot, primarily to be able to assess the allocation of the validating role for these countries. The exercise will be evaluated after the first nine months and if agencies are satisfied with the progress, then they may choose to implement the process for data cooperation before the end of the two-year period. 9 23. During the pilot exercises, agencies will disseminate the data only after the validation process is complete. The timeliness of the dissemination of the validated data will then be compared with the current timeframe for data collection. The agencies will also explore the possibility to allow for the dissemination of non-validated data (where the data will be pushed to the receiving agencies as soon as they are collected from the countries). 24. Costs are mainly related to handling the reception and transmission of data. While it was agreed that the pilot exercise on GDP and population data should be started on the basis of the current technical environment involving varying technical infrastructure and manual interventions, the exchange of data between international agencies will be more efficient once SDMX is operational. Sectoral Accounts 25. For the sectoral accounts data, the pilot exercise will continue until data are available on a 2008 SNA/ESA 2010 basis for most of the participating countries. The data collection process during the pilot exercise will follow the practices for submitting sectoral accounts data that have been adopted by the IAG Working Group on Sectoral Accounts. 26. Under the existing arrangements, Euro area and EU member countries submit the data to the ECB and Eurostat. EFTA members and EU candidate and associated countries also already submit partial datasets to Eurostat. These organizations then transmit the data to the OECD. The OECD collects data for OECD member countries that are not members of the EU and for selected OECD Key Partner countries where the data are available. The OECD submits the entire data to the IMF for publication on the PGI website. For assessing the feasibility of data cooperation, the pilot could test the transmission of sector accounts data to the other participating agencies as well. In the future, the IMF may probably need to assume the responsibility for collecting data from countries that are not covered by the OECD, subject to data availability. The scope for any such collection could be broadened by the inclusion of annual data in the pilot. Pilot Countries 27. Building on already agreed responsibilities among the ECB, Eurostat and the OECD, the TFIDC proposes that the transmission of data for the euro area, EU and remaining OECD and Key Partner countries to other agencies be organized by these three agencies, while the IMF agreed to send existing data it collects for all remaining countries. The BIS and the World Bank will only participate as a recipient of pilot project data. 28. Annex II summarizes the country list for the two pilot exercises and outlines the possible arrangements for country by country sender responsibilities and receiving agencies. It is expected that the pilots will review the list in the Annex II and propose a feasible arrangement for future cooperation. 10 Expected results and timeframes 29. The cooperation with regard to the agreed datasets should demonstrate that the exchange of data among international agencies is technically feasible for both sending and receiving agencies. This will result in the availability of consistent data across international agencies and efficiency gains for all parties. In the medium-term, this will also demonstrate that multiple data collection and validation from national authorities may no longer be necessary and that agencies may start working on reporting of data by countries to only one international agency. Criteria for evaluation 30. Participating international agencies should use the pilot exercises to evaluate the benefits and costs associated with international data cooperation. Eurostat, the ECB and the OECD would benefit by gaining additional country data that could be used to analyze the economies of main trading and financial partners. The BIS, IMF, WB and the UN would need to assess the timeliness and quality of data in comparison to current arrangements, and evaluate the savings from possibility of reducing data collection and validation routines. 31. The documentation that will be developed as a basis for the evaluation of the pilots will include the following: 1. Agencies sending data, will document, country by country, when the data was pushed. 2. Agencies receiving data will document, country by country, the following: the time they receive data through existing channels and when they publish the data; the time they receive the data through the pilot exercises and when they could publish these data (note: during the pilot the data will not be made public) any quality problems found in the data received and notify the sender of them without delay; and resources spent on data validation in person days, for the data received via own channels and via the pilot exercises. B. Templates 32. The TFIDC has agreed on two sets of templates for (i) GDP main aggregates and population and (ii) sectoral accounts. Regarding GDP main aggregates and population, the TFIDC agreed on a set of main national accounts aggregates and auxiliary indicators that are widely used and are available for a large number of countries (see Annex III). 11 GDP main aggregates and population 33. The proposed set of 36 indicators presented in template covers GDP (including statistical discrepancies), main GDP aggregates relating to output, expenditure and income, as well as data on gross national income, saving and net lending. These indicators are complemented by population and employment figures, which are important auxiliary indicators to derive indicators per inhabitant or monitor productivity. The TFIDC has included four key components in the templates for GDP and aggregates. These are as follows: Seasonally adjusted and (original) non-seasonally adjusted data Quarterly and annual data Nominal and volume measures Inclusion of flash estimates (as received by countries) 34. In terms of the published data, the TFIDC has decided that seasonally adjusted data submitted by the countries should be used for dissemination. If these seasonally adjusted series are not available, then the agencies may conduct the seasonal adjustment and transmit these data to other agencies. However, the seasonally adjusted data represent an additional series that should be clearly presented as an estimate of the relevant international agency and it would be the responsibility of the agencies to get back to the countries to inform them of the seasonal adjustment. Population Statistics 35. The TFIDC agreed that the definition of population used should preferably be based on the national accounts concept of residence as this definition is more relevant an indicator for per capita measures. While EU countries report population statistics based on the national accounts concepts, other countries may only have data based on demographic statistics. Therefore, where the definition of population based on the national accounts concept is not available, countries may report the data based on the demographic statistics. Sectoral accounts 36. The templates on sectoral accounts that have been developed by the Working Group on Sectoral Accounts and endorsed by the IAG will be used for the pilot3. The distinction between the minimum and encouraged items stipulated in the templates will not be applied for the pilot exercise. Therefore,countries are expected to complete the templates as they deem feasible. 37. The data currently being compiled by countries are already being provided for publication on the PGI website, thus to a central location. It is expected that the sector accounts templates will be applied as of end 2014, and data will be collected using the new SDMX-ML data standards definitions for national accounts. A formal data cooperation pilot exercise will not 3 The templates are available on the IMF website: http://www.imf.org/external/np/sta/templates/sectacct/index.htm 12 be established until then. The current data cooperation exercise involves the provision of data, mostly from European countries to OECD and IMF. Currently, not all agencies receive these data from the primary validating/disseminating agencies. 38. The TFIDC will focus on the collection of non-seasonally adjusted quarterly data (seasonally adjusted data may need to be considered in cases where non-seasonally adjusted data are not available). The agencies will also collect annual data including also those countries that compile quarterly sectoral accounts data. 39. The TFIDC will investigate further whether it is possible to make the currently collected sectoral accounts data available directly to other participating international agencies. 40. For the future, the following issues need to be clarified: Whether the current cooperation arrangements regarding quarterly sectoral accounts data should be extended to include as recipients the other members of the TFIDC Whether the data collection should include annual data. This raises issues of consistency between quarterly and annual data; Whether the agencies should collect and share seasonally adjusted data; and Whether the collection should be expanded beyond the G-20 economies. C. SDMX Infrastructure in Member Agencies Review of the Infrastructure 41. The TFIDC conducted a survey in July 2013 to determine the status of their preparations regarding their SDMX infrastructure data / metadata transmission and reception, and the readiness of their technical environment for exchanging data (see Annex IV for survey questionnaire and summary results). The infrastructure assessment is in the context of the conclusions by the TFIDC that the pilot should test the agencies’ ability to receive and send (where appropriate) national accounts data using SDMX-ML, using the appropriate 2008 SNA DSD. 42. The overall assessment, based on the answers received to the questionnaire, is that the participating agencies’ technical systems for data exchange are prepared to handle SDMX formats for data exchange and are generally ready for the pilot exercise. In order to continue with the next stage of the project, the following common features prevail: SDMX-ML Compact 2.0 format is the preferred format for the data exchange; the SDMX-EDI format would provide problems with OECD, IMF and WB. For the BIS and the ECB, who initiated SDMX-EDI already 15 years ago and based their system on EDI specifications, the SDMX-ML messages should therefore also be EDI compliant for the purpose of the pilot exercise (e.g. Key Family/Data Set Identifier information to be 13 43. present in the header of the data file as agreed in the BPM6 and 2008 SNA DSD technical WG), in order to benefit from automated processes. DSD messages (over MSD messages) are preferred SDMX messages for processing; Automated creation, loading and reception of SDMX messages is preferred; Current/existing/testing environment systems will be used; Automated data transmission via SDMX messages is preferred; Manual intervention, i.e. data exchange via E-mail to be used only as last resort; The pilot phase is intended to test data exchanges and synchronization of data vintages between IO’s, as such the exchanged data is experimental. The data cannot be publicly disseminated with the 2008 SNA DSD until the latter is officially used for data transmission. Different level of systems development and usage include: New SDMX artifacts like hierarchical code lists cannot be handled by all participating organizations; Use of SDMX-NA Sandbox versus existing systems; SDMX-EDI format is not supported by all organizations, see also a) above in common features; SDMX transmission format The table 1 highlights the results of the survey on SDMX transmission formats. 14 Table 1: Summary of SDMX Transmission Format Messaging format (Q2) Type of messages (Q3) SDMX-ML 2.0 SDMX-ML 2.1 Eurostat IMF OECD (only compact) ECB (only compact) SDMXEDI (not preferred) Hierarchica l code lists Attributes at any group level DSD New SDMX artefacts (Q5) Creation of SDMX messages from database (Q6) MSD (with extra effort) System rely on SDMXEDI dependences (Q4) (only structure specific) Automatic data loading (Q7) Reception of SDMX messages (Q8) WB (only compact) (currently not used; if needed will be built upon) (no data post on the web) (possible data post on .Stat) Automatic transmission channels in place on the dissemination side (Q10) (EDAMIS) (secured channel with OECD) (only for SDMXEDI and EDI compliant ML messages) (direct data exchange with BIS, EUROSTAT and IMF; with OECD via Eurostat; secure e-mail with UN and WB) Monitoring of receipt of SDMX messages/post data during the pilot (Q9) (for most data) (internal mapping may be needed) (no data post on the web) (under evaluation) BIS (only for SDMXEDI and EDI compliant ML messages) (no data post on the web) This table highlights the following aspects: The transmission format used in the pilot should be SDMX-ML 2.0 Compact, EDI SDMX compliant, if all agencies manage to implement EDI compliance in due course. The pilot should involve sharing of data, not metadata Hierarchical code lists and attributes at group level would be avoided. 15 Transmission methods 44. Based on the review, several possible transmission methods have emerged as described below. Option 1: Agency to agency transmission of data using existing infrastructure: All six responding agencies consider this feasible, when it occurs with SDMX-ML 2.0 format [ideally EDI compliant ML messages]. This would allow data coordination involving the transmission of data using existing infrastructure from the receiving agencies to receiving agencies directly, without manual intervention and without intermediate data storage. Agencies that are equipped for automated transmission and reception could make use of their existing infrastructure, and actual turnaround times could be validated. Option 2: Use of the technical pilot sandbox: Four out of five agencies consider this option viable. The ECB is the only agency that does not support this option (the BIS did not express an opinion). The main advantage of using the technical pilot sandbox is that it would provide a central hub for data cooperation. The main disadvantages associated with its use are that it currently does not support automatic notifications upon arrival of data and that data files need to be manually downloaded. Option 3: Use of EDAMIS: EDAMIS is an application implementing the Single Entry Point policy of Eurostat. A specific dataflow could be created for the pilot and any participating agencies could be set up as sender/receiver.. EDAMIS would provide a central dispatch place for data cooperation. It would allow the data to be transmitted securely and would automatically monitor the reception and re-dispatch of data, including any small delays. Receivers could also choose in which form they want to receive the data, including email. The main drawback of EDAMIS is that not all agencies are familiar with it. However, Eurostat will configure EDAMIS accordingly and provide a short webinar session to explain how it can be used in the context of the pilot. Option 4: Email exchange: All agencies mention email exchange as a possible data cooperation alternative, although none rated it as the preferred option. Under this solution, each participating organization would send the pilot data through encrypted emails to all other agencies. The main advantage of this option is its simplicity. It would however lead to an overall increase in manual work compared to the previous option. Moreover, this option does not contemplate any centralized hub, which would make data cooperation more error-prone (due to e.g. inadvertent omissions of recipients in the mailing lists). 45. The TFIDC concluded that for the GDP pilot the EDAMIS option would be implemented. Should there be problems with longer system delays, the fall back solution considered would be the use of an email exchange. 16 Implications for Pilot Exercises 46. Taking into account the current state of progress of SDMX implementation in Eurostat and the other international agencies, Eurostat has prepared for a transmission of the selected list of GDP and population data building on its public dissemination database (Eurobase). 47. For this purpose, the list of identified indicators, units, frequencies and adjustment methods were mapped to respective series in Eurobase, which are updated twice per day following the validation of data in the internal production database, and a correspondence to the SDMX DSD has been prepared. 48. The files will be pushed via EDAMIS as a preferred option. While an initial transfer would encompass all series (starting in 1995), the extraction of updates could be limited to concerned countries. 49. The OECD is preparing for a transmission of the selected list of GDP and population data built on OECD.Stat. A correspondence between the DSDs representing the native structure of the OECD dissemination data warehouse and the agreed DSD for National accounts is to be prepared. 50. Assuming that the data messages are SDMX-ML 2.0 Compact EDI compliant and Eurostat sends the files via EDAMIS, ECB reception and loading of the data in the ECB test system could be automated. With the OECD, secured e-mail could be used after an exchange of PGP/GPG keys. If EDAMIS or secured e-mail is not used, manual intervention will be needed on the reception of the data to manually load them in our reception database. Manual interventions will also be required if the SDMX-ML data messages are not EDI compliant (see above). Once the data are in the reception database they can be moved in production and disseminated in the ECB test environment. 51. Regarding the sending of the quarterly sector financial accounts, the ECB foresees two steps in both cases using own tools. First, the test transmission planned for January 2014 will require mapping to the new DSD series keys; the ECB test environment will be used and data sending via the agreed and currently established transmission channels to all agencies. The second transmission scheduled in November 2014 will take place after the official go-live of the new 2008 SNA DSD. Therefore the ECB production system will be used, no mapping will be required by then and data will be automatically sent via the agreed and currently established transmission channels to all agencies. The published quarterly sectoral financial accounts data will also be accessible via ECB SDW site at that time as well as web-services. V. CURRENT WORKFLOW PRACTICES AND WORKFLOW FOR THE PILOTS 52. The TFIDC will examine the data cooperation arrangements. This includes arrangements relating to the collection of data from the countries, the transmission of the data to the international agencies, and the work flows among the international agencies with respect to the sharing of consistent data on a timely basis. The TFIDC will also review the workflow practices 17 relating to the flow of information from the national (compiling) agencies to the international agencies. A. Data transmission from country to international agencies 53. The review of workflow practices illustrated a range of methods, timelines, and coverage for the transmission of country data to international agencies (see Annex VI). In addition, the transmission schedules vary among international agencies as the schedules are designed to meet the re-dissemination timetables of the agencies. The pilot exercises will need to determine a suitable schedule for the transmission of data from national agencies to the collecting international agencies that will satisfy the timetables of all receiving international agencies. This implies that, among all the agencies, the strictest schedule currently in place may have to be followed. Further, the validation process should also satisfy the timetables of the receiving agencies. 54. The pilot exercises will also need to determine which national agencies would be responsible for submitting the data. B. Sharing of validated data among international agencies 55. For many years, international agencies have been exchanging national accounts data with each other, to respond to user needs. Following are some examples of current data cooperation: Eurostat submits national accounts data for most EU countries to IMF in a bulk file on a monthly basis. OECD provides quarterly national accounts data to all other six institutions once a month (at the time of the press releases). UNSD receives data from the OECD, the UNECE and CARICOM. The ECE provides data for transition economies. World Bank receives national accounts data for selected high-income economies from the OECD and most of the population data from UNPD. 56. Furthermore, in relation to the G20 Data Gaps Initiative, ECB, Eurostat and the OECD started, from 2012 onwards, providing selected datasets on an ongoing basis for the PGI website hosted by the IMF. 57. More recently, Eurostat and the OECD signed a data exchange protocol for national accounts data. This agreement describes the scope of data exchanges and data validation arrangements4. The protocol includes more detailed validation requirements and maximum delays for processing and validation, which could also serve as a basis for the pilot data cooperation exercise at hand. 4 http://epp.eurostat.ec.europa.eu/portal/page/portal/national_accounts/documents/MoU%20OECD%20NA%202013. pdf 18 Workflow arrangements for the pilots 58. Below the workflows to be followed are given separately for sender and receiver agencies. Workflow for data sender data collection data validation creation SDMX file pushing of data documenting • collection of data from countries • sender IOs procedures in place used • data validation respecting minimum standards • creation of the SDMX message agreed • transmission of the SDMX message agreed • date and time of reception of data from countries • person days used for validation • date and time of pushing of the file Outside scope of pilots Inside scope of pilots 19 Workflow for data receiver data reception data validation data dissemination documenting • reception of the SDMX file from data sender • data validation using own procedures in place • note: during the pilots the data will not be made public • date and time of reception of SDMX file (EDAMIS will automatically record) • person days used for validation of the pilot data • date and time of when the pilot data could be loaded into the own public dissemination environment (note: during the pilots the data will not be made public) • date and time of reception of data from countries via the own collection system, if applicable • person days used for validation of the data collected via the own system, if applicable • date and time of dissemination of the data colelcted via the own system in the public environment Procedures and validation rules 59. International agencies will only disseminate the data as soon as the validation process is complete. The validation process will be based on rules and procedures which have been identified as good practice by the ECB-Eurostat-OECD technical working group in 2012 and agreed for the Eurostat-OECD protocol. 60. The agreed minimum checking rules (Annex VII) to validate national accounts data include basic format, structure, encoding and content tests, which are essential for automatic processing, as well as revision checks and consistency checks within and between transmitted national accounts datasets, which are essential for users’ perception of the quality of national accounts data. Further checks (e.g. statistical and economic plausibility checks or cross-checks against related data) are also considered useful for detecting problems in the transmission or comparability of data and should be run occasionally or systematically depending on the dataset in question. While focusing on checks that can be processed and interpreted in a relatively 20 automatic way will help to ensure that data fulfilling minimum quality requirements can be published in a timely way, it is important that data specificities which affect validation checks are clarified with national authorities and documented as metadata. Timetables for Pilot Exercises 61. In terms of the timetables, data received from national sources should preferably be validated by the responsible international organization with the shortest possible delay. One of the objectives of the data cooperation pilot is indeed to monitor these delays in practice. 62. Based on the timeliness agreed in the Eurostat-OECD data exchange protocol, the sharing of validated data is expected to be as follows: Validated data for main aggregates-population will be shared: within three days for quarterly data (one working day for larger economies) within one week for annual data Validated data for sector accounts will be shared: within one to two weeks for quarterly data within one month for annual data 63. These time lags are counted from the moment the data-providing international agencies receive data from the relevant national agency. The pilots will address timeliness issues and will propose feasible timelines for data sharing. Timetable for Pilot Exercise on GDP Aggregates and Population The following timetable is proposed for the first data exchanges. In total the pilot will take 9 months. Eurostat will start pushing data according to the following time table: Test with dummy or old data: 26 September 2013 Initial data dissemination (up to Q2 2013): a. complete national dataset b. aggregates 7 October 2013 Continuous transmission of new and updated data Q3 2013: GDP flash growth rate Q3 2013: aggregates Q3 2013: full transmission Q4 2013: GDP flash growth rate as of 07 October 2013 14 November 04 December 2013 09 January 2014 15 February 2014 (to be confirmed) July 2014 Report on status of pilot exercise 21 OECD and IMF will follow the same timetable. However, OECD will start in late October and IMF in November 2013, in order to allow for some technical adaptations in their own systems to receive and push data according to the agreed DSDs. Timetable for Pilot Exercise on Sectoral Accounts 64. The following tentative timetable is proposed for the first data exchanges: Test of System with dummy or old data February, 2014 Initial data dissemination (up to Q2 2014): end -November 2014 Q2 2014. Sectoral Accounts and Balance Sheets end-November 2014 Annual 2013: Sectoral Accounts and Balance Sheets end-November 2014 Report on the pilot exercise and implementation February 2015 65. The timetables presented above may be adjusted where necessary as the situation warrants and based on the first experiences with the pilot on GDP main aggregates5. A brief status of the pilots could be included in the ISWGNA report to the 2014 meeting of the UN Statistical Commission. A side event may be considered at the time of 2014 UNSC meeting. 5 For instance, whether the dissemination of sector accounts from Eurostat can be organized based on the automatic extractions from the dissemination database (Eurobase), independently of the production database. 22 ANNEX I: MEMBERS OF THE TFIDC Bank for International Settlements Christian Dembiermont European Central Bank Tjeerd Jellema Statistical Office of the European Communities Silke Stapel-Weber (Co-Chair) August Götzfried Christine Gerstberger Daniel Suranyi International Monetary Fund Manik Shrestha (Co-Chair) Thomas Alexander Olga Laveda Gangti Zhu Organization for Economic Co-operation and Development Peter van de Ven Jennifer Ribarsky Rachida Dkhissi Gyorgy Gyomai United Nations Herman Smith World Bank Ibo Levent 23 ANNEX II: COUNTRY COVERAGE FOR THE PILOT EXERCISE (BY GROUP OF COUNTRIES) 6 Pilot 1: GDP and Population BIS ECB ESTAT IMF OECD UN WB UN WB Sender Country data European Union members X EU-candidate countries X EFTA-members X Non-EU and non-EFTA OECD members X OECD key partners and accession countries X Other G20 countries (Argentina, Saudi Arabia) X X Other countries (IMF members) Aggregates EU-28 X EUR-17 X OECD X G20 X World X Pilot 2: Sector accounts BIS ECB7 ESTAT IMF OECD Sender Country data European Union members X X EU-candidate countries X EFTA-members X Non-EU and non-EFTA OECD members X OECD key partners and accession countries X Other G20 countries (Argentina, Saudi Arabia) X Aggregates EU-28 EUR-17 X X X OECD X G20 X World 6 Data are not available for all countries and all indicators; i.e., the tables do not represent actual data availability but rather the allocation of collection/validation responsibilities for the pilot. Transmission will depend on availability from countries. 7 Quarterly sectoral financial positions and flows. 24 Overview of division of work for data transmissions for the Pilots 1 and 2 Countries marked in bold will also be covered by sector accounts (Pilot 2) Eurostat or ECB Eurostat OECD European Union (EU28)8 1 Belgium BE 2 Bulgaria* BG Czech 3 CZ Republic 4 Denmark DK 5 Germany DE EU-candidates9 54 55 Egypt Hong Kong EG HK 8 9 29 30 FYROM Iceland MC IS Non-EU and non-EFTA OECD members 37 Australia AU 38 Canada CA 31 Montenegro ME 39 Chile CL 56 Malaysia MY 32 33 Serbia Turkey RS TR 40 41 Israel Japan Korea, Republic of Mexico New Zealand USA IL JP 57 58 Singapore Thailand SG TH KR 59 Taiwan, PC TW MX NZ US 60 Ukraine UA 6 Estonia* EE 7 8 9 10 Ireland Greece Spain France IE GR ES FR 11 Croatia** HR 12 13 14 15 Italy Cyprus Latvia* Lithuania* IT CY LV LT 46 47 48 49 16 Luxembourg* LU 50 17 18 19 20 21 22 23 24 25 26 27 Hungary* Malta* Netherlands Austria Poland Portugal Romania Slovenia Slovakia* Finland Sweden United Kingdom HU MT NL AT PL PT RO SI SK FI SE 51 28 IMF 42 EFTA-Members 34 35 36 Lichtenstein Norway* Switzerland LI NO CH 43 44 45 OECD key partners and accession countries Brazil China India Indonesia Colombia Russian Federation South Africa Other countries (priority) Other countries (available) 61 All other countries BR CN IN ID CO RU ZA Other G20 countries 52 53 Argentina Saudi Arabia AR SA UK Quarterly non-financial accounts:* partial coverage only S13 and S2;**expected . Annual non-financial accounts; Iceland -only limited financial accounts data; Montenegro- only non-financial accounts data from 2013; Serbia-confidential data. 25 ANNEX III: TEMPLATE FOR GDP MAIN AGGREGATES AND POPULATION Proposed selection of indicators for IAG Pilot project 1: If all indicators in Bold are transmitted; indicators in italic could, in principle, be derived; more precise information on preferred concepts, prices or units are provided as footnotes. Q X X X X N x x x x Y or S X x x x V x x x x L X X X X X X X X x x x x x x x X x P3 P3_S13 X X X X x x x x x x X X X X x x x X X X X x X X x x x x x x x x x x X X X X X X X x x x x x x x x x x X X x X X X X X X X x x x x x x x x x x x x x P3_S14&15 P41 P5g P51g P52&P53 P6 P7 B11 Total final consumption expenditure Government final consumption expenditure Household and NPISH final consumption expenditure Actual individual consumption Gross capital formation Gross fixed capital formation Changes in inventories and acquisition less disposals of valuables Exports of goods (fob) and services Imports of goods (fob) and services External balance of goods and services Main GDP aggregates from the income side D1 Compensation of employees D11 Gross wages and salaries B2g&B3g Operating surplus and mixed income, gross D2X3 Taxes less subsidies on production Volumes A X X X X Current prices Code Unit: national currencies (millions) B1gQ Gross domestic product at market prices YA0 Statistical discrepancy (expenditure approach) YA1 Statistical discrepancy (production approach) YA2 Statistical discrepancy (income approach) Main GDP aggregates from the output side P1 Output P2 Intermediate consumption B1g Gross value added at basic prices D21XD31 Taxes less subsidies on products Main GDP aggregates from the expenditure side Indicator yes (2) no (1) Prices (3) Quarterly Frequency Annual I/ Main GDP aggregates Seasonal Adjustment of quarterly series 26 Y or S V Gross national income, saving and net Lending (D1_D4) Net primary income from RoW B5G Gross national income at market prices K1 Consumption of fixed capital B5N Net national income at market prices (D5_D6_D7) Net current transfers from RoW B6N Disposable income, net Adjustments for change in pension D8 entitlements B8N National saving, net D9 Net capital transfers from RoW K2 (+/-) non produced, non--financial assets B9 Net lending or net borrowing of the nation II/ Population and Employment (4) X X X X X X x x x x x x x x x x x x x x x x x x x x x x x x X x x x x X X X X x x x x Frequency x x x x x x x x Q: Adjustment Volumes N Current prices Q yes (2) no (1) Unit: national currencies (millions) Prices (3) A Indicator Code Seasonal Adjustment of quarterly series Quarterly Frequency Annual I/ Main GDP aggregates L x x x x Units (5) Indicator Annual Quarterly no (1) yes (2) Persons Hours worked Code Unit: 1000 A Q N Y or S PERS HR POP EMP_DC SAL_DC SELF_DC Total population Employment, domestic concept Employees, domestic concept Self employed, domestic concept X X X X x x x x x x x x x x x x x x x x x x x (1) Neither seasonally nor working day adjusted (2) Working day and seasonally adjusted (Y) or seasonally adjusted, not working day adjusted (S); if seasonal adjustment is done by the International Organisation, a flag (I) should be added. (3) GDP volume data should preferably be provided as chain-linked volumes (L) or previous year prices (Y); if these are not available constant prices (Q) can also be used. (4) Population and employment are considered auxiliary variables in national accounts, aimed to calculate ratios like value added, output, or labour costs per inhabitant or per employed person. If possible, population data should therefore refer to data compiled for national accounts rather than demographic statistics, i.e. representing averages for the observation period and including adjustments (e.g. for students, immigration, armed forces...). Employment should refer to employment in resident production units irrespective of the place of residence of the employed person (i.e. domestic concept, DC) rather than resident persons in employment (i.e. the socalled national concept of employment, NC) as this concept is more appropriate when examining employment in relation to GDP (the difference being mainly the net number of cross-border workers). (5) If persons (PERS) and hours worked (HR) are not available, Jobs (JOB) or Full-time-equivalents (FTE) could be provided. 27 ANNEX IV: SDMX SURVEY QUESTIONNAIRE June 20, 2013 Task Force on “International Data Cooperation” Draft Questionnaire Infrastructure Data cooperation Pilot Exercise. September 2013 – March 2014. Introdcution The TF-IDS agreed to conduct a data cooperation pilot with the following participating international agencies 1. BIS 2. ECB 3. Eurostat 4. OECD 5. IMF 6. United Nations 7. World Bank The pilot exercise will encompass the transmission of data messages with regards to the agreed template for GDP main aggregates and population statistics; the data transmission will make use of SDMX-ML messages using the now nearly finalized DSD for national accounts. The scope of the data transmission will encompass a list of countries for which it is established that quarterly main aggregates are regularly produced and disseminated. This list, it has been established, is a super set of the data currently collected by the OECD and Eurostat, which implies that for several countries a primary validator/disseminator IO needs to be assigned. This questionnaire provides for a fact finding for the transmission and reception of SDMX type of messages. A concrete follow up could involve a meeting or teleconference of subject matter specialists to establish the precise parameters of the pilot data transmissions under data cooperation. An early opportunity may be at the margin of the Eurostat workshop 10-11 June 2013. 28 Questionnaire 1. SDMX-NA DSD Pilot test participation. Are you/is your organization actively participating to the test technical testing of the SDMX-NA, in May – July 2013 a. Will you obtain the SDMX specifications from the Technical Pilot Sandbox? b. Will you formulate and send test SDMX messages to the technical pilot sandbox? c. Would you like the possibility to download SDMX messages from the technical pilot sandbox? 2. Which SDMX messaging formats are you able to process? a. SDMX-ML V2.0, which format: compact, generic? b. SDMX-ML V2.1, which format: structure specific, generic c. SDMX-EDI / GESMES/TS 3. Which SDMX messages are you able to process in your system? MSD? DSD? 4. Is your current system in terms of SDMX messages handling and data/metadata storage relying on features/concepts used in SDMX-EDI, e.g. use of sibling and/or dataset attachment level attribute for specific attributes across DSDs, mandatory OBS_STATUS, etc? 5. Is your current system able to handle new SDMX artifacts like hierarchical code lists, to handle attributes at any group level, etc? Do you intend to use those features? If yes, by when will this be implemented? Please specify. 6. Creation of SDMX messages as of September 2013 a. You have the necessary infrastructure in place to create from your database system SMDX messages , in all SDMX formats you mentioned above (question 2), for any DSDs. Please specify. b. You will need to formulate SDMX type of messages manually (e.g. through extracting data manually and filling a custom prepared Excel workbook). Please specify. c. Any other means of forming SDMX messages, namely. d. No infrastructure present. 29 7. Loading of SDMX messages in your database system a. You have the necessary infrastructure in place to automatically load SDMX messages in your database system, in all SDMX formats you mentioned above (question 2). Please specify. b. You will need to rely on a manual loading of the data in your database. 8. Reception of SDMX messages as of September 2013 a. Do you have the necessary infrastructure in place to receive SDMX messages from all participating agencies in all SDMX formats you mentioned above (question 2). Please specify. b. You will need to receive the SDMX-ML messages by e-mail 9. Monitoring of receipt of SDMX messages and upload of the data on your website (in order to be able to see if the pilots are a success in terms of more consistent data being available earlier to users) a. What would be your system for systematically monitoring the receipt of the SDMX messages? b. What systems do you have in place to monitor the date and time of the upload of the data onto your website? c. What is your suggestion how we share this information? 10. Transmission channels in place on the dissemination side. Can you support any of the following? a. Any type of SDMX messages attached in an E-mail: secured with encryption and signature or not?, requires or not manual intervention in preparing e-mails?, with all data exchange partners?, on both reception and dissemination?) Please specify. b. E-mail: automatically generate and process SDMX files and emails for data exchange c. Any type of SDMX-Messages exchanged via an automated process, i.e. via dedicated data exchange infrastructure between institutions, including all data exchange partners? and on both reception and dissemination? Please specify. 11. Any other relevant observations / remarks / limitations in your system that would affect the success of the data cooperation pilot. 30 ANNEX V: SUMMARY OF SDMX SURVEY RESULTS As of 12 August 2013, only one participating international organization (the United Nations) has not provided input to the survey; all other participants’ replies have been taken into account (BIS, ECB, Eurostat, IMF, OECD and the World Bank). Most of the organizations have provided exhaustive replies to the questionnaire which also helped for better evaluation of the preparedness status of the project. Q1: SDMX-NA DSD Pilot test participation. Are you/is your organization actively participating to the test technical testing of the SDMX-NA, in May – July 2013 All responding international agencies (with the exception of the BIS) confirmed their participation in the SDMX-national accounts DSD Pilot test. a. The ECB, Eurostat, IMF and OECD confirmed that they can obtain the SDMX specifications from the Technical Pilot Sandbox. The World Bank said that it will be monitoring the progress of 2008 SNA through the SDMX Secretariat. b. Eurostat, OECD and IMF will formulate and send test SDMX messages to the technical pilot sandbox. The World Bank will use the technical pilot sandbox when it is ready; an initial analysis of the NA DSD structure and their internal data structure is needed followed by mapping of the two. The ECB will not use the 2008 SNA fusion registry or the sandbox but will use its data exchange infrastructure and statistical environment. c. The OECD, IMF and the World Bank would like to download SDMX messages from the technical pilot sandbox. The ECB and Eurostat will not use this approach. The ECB will set up its system test environment to generate SDMX-ML compact 2.0 and structure specific 2.1 in addition to SDMX-EDI. Eurostat will rely on the “push” approach but still having the technical possibility to download SDMX messages from the technical pilot sandbox. Q2: Which SDMX messaging formats are you able to process? a. SDMX-ML Compact 2.0 format is the preferred format for exchange by all agencies. SDMX-ML Generic 2.0 format is used by Eurostat and IMF; currently developed by the OECD. b. The ECB and the BIS can process SDMX-ML Structure Specific 2.1 format. The World Bank is evaluating its infrastructure to process the SDMX-ML V2.1 messages. No reply from the OECD. 31 c. The ECB and the BIS are able to process SDMX-EDI for this pilot exercise. For Eurostat this format is also possible but would require manual intervention. The OECD, IMF and the World Bank do not support SDMX-EDI (GESMES/TS) technology and the latter two agencies have no plans to provide support to that format. Q3: Which SDMX messages are you able to process in your system? All IOs can process DSD messages. Eurostat can process MSD messages. The IMF and the World Bank would need further investigation to implement it, i.e currently not used by the World Bank. The ECB and the BIS does not currently process MSD messages - reference metadata are defined at the level of the DSD. Q4: Is your current system in terms of SDMX messages handling and data/metadata storage relying on features/concepts used in SDMX-EDI, e.g. use of sibling and/or dataset attachment level attribute for specific attributes across DSDs, mandatory OBS_STATUS, etc? The BIS and the ECB systems rely on features/concepts used in SDMX-EDI (use of sibling and/or dataset attachment level attribute for specific attributes across DSDs, mandatory OBS_STATUS). Eurostat, OECD, IMF and the World Bank do not. Q5: Is your current system able to handle new SDMX artefacts like hierarchical code lists, to handle attributes at any group level, etc? Do you intend to use those features? If yes, by when will this be implemented? Please specify. Only Eurostat indicated that can handle new SDMX artefacts like hierarchical code lists if used with SDMX 2.0. No definite commitment for implementation is being mentioned by the other IOs. Q6: Creation of SDMX messages as of September 2013 a. All agencies in principle have the infrastructure in place to create from database system SMDX messages. The common SDMX format is SDMX-ML v2.0 compact and generic format as well as DSD based on the underlying dataset structure (IMF) or DSD in the native structure of our dissemination data warehouse (OECD), or test environment (the ECB, BIS). Additional mapping will be required for Eurostat and the OECD. b. No agency will need to formulate SDMX type of messages manually (e.g. through extracting data manually and filling a custom prepared Excel workbook) as in all cases these processes are automated. 32 c. Eurostat replied that if needed SDMX Reference Infrastructure or SDMX Converter may be used as any other means of forming SDMX messages. d. The necessary infrastructure is present in all agencies. Q7: Loading of SDMX messages in your database system a. All agencies have the necessary infrastructure in place to automatically load SDMX messages in their database systems, in all SDMX formats they support. The most common format is SDMX-ML V2.0. BIS and the ECB support also SDMX-EDI format (EDI or EDI compatible ML). OECD supported data sources are Web Service (pull method) and file (push method). b. Agencies do not need to rely on a manual loading of the data in their databases. Eurostat may need partially manual intervention, depending on transmission format and transmission mean. The World Bank have manual loading of data from SDMX-ML to the system for some datasets. Q8: Reception of SDMX messages as of September 2013 a. All agencies have the necessary infrastructure in place to receive SDMX messages from all participating agencies in all SDMX formats they support. BIS uses file exchange based on Web or Java application. The ECB uses direct data exchanges with BIS, Eurostat and IMF; data exchanges with the OECD are generally done via Eurostat EDAMIS; no exchange with the UN and World Bank thus it is foreseen to set up infrastructure for the exchange of secured e-mails that would contain a SDMX data message (PGP-GPG encrypted and signed message). The most common format is SDMX-ML V2.0. BIS and the ECB support also SDMX-EDI format (extension for SDMX-ML data files should be available in a near future for the ECB). OECD supported data sources are Web Service (pull method) and file (push method). The World Bank can receive SDMX messages but will need to map to their our data structure. b. The need to receive the SDMX-ML messages by e-mail would, for the agencies, depend on the data exchange partner. This is not seen as a preferred method but more as last resort because manual treatment would be required in such cases. Q9: Monitoring of receipt of SDMX messages and upload of the data on your website (in order to be able to see if the pilots are a success in terms of more consistent data being available earlier to users) a. Systems for systematically monitoring the receipt of the SDMX messages differ between agencies but are basically using their current monitoring systems of data reception. The 33 ECB will rely on its current data exchange infrastructure, including a Web UI, to monitor the data exchange in the SNA context like for any other data flows. Eurostat will use manual monitoring it the reception is by email and automatic monitoring for reception by EDAMIS. The OECD may use mailbox, web service or file server depending on the data transmission. The IMF will use the NA Registry Service notification, NA sandbox service and SDMX reader service forming a system that currently being developed. The World Bank is evaluating tools for this capability. b. Systems in place to monitor the date and time of the upload of the data onto agencies website are generally within the scope of the agencies systems for systematically monitoring the receipt of the SDMX messages. In addition the ECB has developed an interface for its data producers to monitor their data flows from reception to dissemination, including scheduling actions for compilation of derived statistics and data validation; during the pilot phase data will not be uploaded on website until the go-live of the SNA transmission. BIS do not consider that it is necessary to post the data on the web during the pilot phase. The OECD has the .Stat entry gate system, which can be configured for this pilot (for the time being, QNA data is loaded on OECD.Stat twice a day at fixed times at 6am and 12pm). c. The information is proposed to be shared in various ways: web services (ECB); EDAMIS (Eurostat) can also be setup as a message dispatcher and forward incoming messages by email to agencies without EDAMIS client or not using the EDAMIS portal; E-mail and RSS (OECD); web service and data API (IMF) as well as Sandbox support of web service call for extracting the data; Registry service (IMF and World Bank). Q10: Transmission channels in place on the dissemination side. Can you support any of the following? a. The ECB (for reception and dissemination) and Eurostat (for reception via EDAMIS) can support SDMX messages attached in an E-mail: secured with encryption and signature given that there are no manual interventions, i.e. data exchange is automated assuming the exchange of the public keys took place and installation of the relevant keys in their data exchange infrastructure. OECD is in principle able also to support this option. For all e-mail data exchanges agencies will require manual work as such automated infrastructure is not in place. b. E-mail: automatically generating and processing of SDMX files and emails for data exchange is possible for the ECB given public keys exchange and installation on both sender and receiver - exchange systems exists. For Eurostat it would be also possible but with manual intervention. Other agencies do not support this data exchange type. 34 c. SDMX-Messages exchanged via an automated process, i.e. via dedicated data exchange infrastructure between institutions, including all data exchange partners is the supported or preferred option by the participating agencies using their current systems. The World Bank currently do not have any automated exchange of data. Q11: Any other relevant observations / remarks / limitations in your system that would affect the success of the data sharing pilot. The ECB will set up the new SNA DSDs in its test environment until the SNA data exchanges go live in 2014Q3. As result the data will not be available through the ECB SDW web services until then but data files will be exchanged with agencies using the ECB data exchange infrastructure. After the third quarter of 204 (when in production) ECB SDW web services 2.0 or 2.1 will be available. No use of the Sandbox is foreseen. Eurostat would not exclude that, depending on the solution chosen, if manual intervention is needed then the throughput time might be delayed compared to the "real" production process in the future. The OECD foresees the following challenges: the mapping of NA DSDs; handling different message formats, especially the legacy SDMX-EDI if they have to; handling custom and grouplevel attributes such as sibling; validating the data messages because SDMX does not currently provide integrity rule checks. The IMF would like to develop the project in the direction of SDMX registry, Sandbox, notification, monitoring system, web service and open data API platform. No further comments were provided by BIS and the World Bank. 35 ANNEX VI: DATA WORKFLOW PRACTICES Data Workflow Practices in International Agencies GDP Main Aggregates Questions 1. Which countries do you cover in your data collection? ESTAT BIS EU27, HR and EFTA (CH, IS, NO) coutries based on ESA 95 TP and Collects data from 56 central banks, published following national methodology (legal basis). EU candidate countries (ME, MK, SE, methodologies. RS) also provide some of these data. OECD The 34 OECD MCs are covered by data collection plus other countries depending on the subject area. 2. Which indicators do you cover? A and Q data via Table 1 on the ESA TP, include subtables on GDP form the output side (T0101), expenditure side (T0102) and income side (T0103) and savings and net lending (T0107). EU/EA aggregates are compiled by ESTAT. For QNA (GDP and main expenditure components) volume and price indices as well as growth rates. Some zones' aggregates are compiled for the OECD-Total, OECD-Europe, G20 (only GDP), G7 and NAFTA. 3. Which templates do you use (please provide any templates or webforms)? The requirements of data transmissions are specified in the ESA TP. No template. National methodologies are used. Eurostat provides specific EXCEL questionnaires for each subtable, whith a converter to EDAMIS files for transmission to Eurostat. For the ESA 2010 TP, a standardised SDMX framework is being prepared. For EU countries, ESA 95 questionnaires transmitted automatically by Eurostat to the OECD. For financial subjects, ECB transmits to the OECD one file inc.only publishable data. For non-European countries, these templates are slightly adapted to refer to the SNA 93. 4. What is the time schedule of data collection and dissemination? Q: t+70 days (t+90 days); A: usually in March (t+70) and at t+9 Data are received within 24 hours after publication. Data are months, but MSs follow national release calendars, i.e. data are disseminated to external users within 30 minutes after reception. received between t+30 and t+70/90, and updated on the ESTAT website. EU/EA estimates are currently released at t+45 (flash), t+65 (including GVA and expenditure breakdowns) t+75 (income) and t+100 (third estimate for all variables). for EU countries:QNA and QPOP: T+70 days in general and following any national release; for non-EU countries: QNA, QPOP, QNFSA and ANFA: following national releases; QNA, QPOP and QNFSA data are disseminated right after the validation. 5. How is data validation organised (timing, responsibilities, methods)? Standard checks upon reception and loading to the production database include basic format, structure, encoding and content tests, which are essential for automatic processing, as well as revision checks and consistency checks. A preer review of valdation checks recently performed by a ECB-ESTAT-OECD WG. Data are validated within 24 hours after reception. Data are validated within 24 hours . Automatic checks with thresholds are activated .Validation covers consistency checks gaps and outlier detections, vintage consistency timeliness monitoring. Data validation of a dataset takes on average 2 to 4 hours, depending on the dataset. The statisticians are responsible for data validation. The type of checks performed to validate data is summarized in the separate table. 6. Do internal stakeholders have access to nonvalidated data? The access to non-validated data is limited to the production unit. Non-validated data for OECD members are currently forwarded to the OECD. Non-validated data are available to internal stakeholders as well as No to external users. 7. Which data sharing methods with other international organisations are already in place? 8. What is your revision policy? Eurostat automatically forwards non-validated data to OECD, but None following a forthcoming SLA, ESTAT will regulrarly transmit validated data to OECD, as is already the case for ECB. In relation to the G20 IAG initiative, Eurostat provides data for the PGI website. To IMF: QNA (GDP and main exp.comp; to UN: twice a year ANFA series are sent; To all institutions (BIS, ECB, Eurostat, IMF, UN, WB): 1 a month QNA CSV file is sent; from Eurostat: non validated EU data for QNA, QPOP and ANFA are automatically forwarded; validated will be sent from Eurostat following the new procedures . The revision policy for country data varies across MSs. Eurostat As soon as revised data are reported, they are re-disseminated. processes any update of MS data received within a few days. There is no specific BIS policy. Estimations for EU/EA aggregates are updated at the t+65 and t+100 releases. At the t+45 EU/EA flash estimates only the volume series for the last quarter are updated. Revisions erase previous data ,methodological revisions are updated in a new set of series. QNA zones’ aggreg. (OECD-Total, OECD-Europe, G7 and NAFTA) are revised on an ongoing basis incl. data up to the second last quarter. First OECD GDP growth at T+50 days; revisions t T+70 days and T+90 days. For the G20 GDP aggregate, the first estimate at T+70 days and revisions to the at T+90 days and the following quarters. GDP main aggregates, main expenditures in nominal and volume for all countries. Main income series for important countries. 36 Data Workflow Practices in International Agencies GDP Main Aggregates IMF UNSD WB 1. Which countries do you cover in your data collection? Most of the 188 countries of the IMF as well as some nonsovereign entities. The UNSD national accounts data covers 235 countries, areas, and GDP data mostly collected from NSIs/NCBs by visiting and resident territories (including all UN Member States). The current database WB missions. contains detailed annual national accounts estimates for 204 of these entities. 2. Which indicators do you cover? Q and A, at current prices and in volume terms, reported or calculated main GDP aggregates, including expenditure, stat.discrepancy with GDP by type of activity, net primary and secondary income from abroad, national disposable income,GNI and gross saving. Main aggregates (Table 1) GDP by expenditures at current and constant prices ; relations among product, income, saving and net lending aggregates at current prices; Domestic production by industries (value added by industries and fixed assets at current and constant prices) 3. Which templates do you use (please provide any templates or webforms)? Countries submit the data on the prescribed forms . The forms may contain country-specific elements. Countries submit the data in the NC. The data are transmitted using special system or by email. The ICS is a web-based data reporting system for use by country data reporters. The National Accounts Questionnaire (NAQ) template for all but EU and OECD countries, for which data are obtained from the OECD. In addition, UNECE and the CARICOM secretariat collect data from their constituents. Countries are requested to update the data tables and complete them as comprehensively as possible. 4. What is the time schedule of data collection and dissemination? A calendar indicating the monthly cut-off dates for data reporting is posted on the ICS website. The cut-off date is usually one week before the end of the cycle. Quarterly and annual reporters are invited to report every month even if no updates/revisions are available. Data are disseminated to an international dissemination space on a daily basis,they are disseminated to the external users once a month. Section is in charge of collecting, validating and The Database NAQ is sent out to all countries in February/March; countires return completed questionnaires in April; but submissions continue to be received and validated between April-August; End of validation cycle: end August, no additional submissions after this point; Data published online in Sept/Oct (data.un.org) 5. How is data validation organised (timing, responsibilities, methods)? disseminating these indicators every month. Staff members of perform preliminary checks . Color codes signal if data are first transmissions (cells are turned blue) or have been revised (a light-todark coloring reflects the size of the revisions); every quarter, data are compared (with IMF databases and international organizations). Once Excel questionnaires are uploaded, a validation worksheet is created from the internal database with a, extensive set of validation rules (over 800). Inconsistencies may be corrected for minor aggregates if possible; otherwise the country is contacted or a footnote added to indicate the discrepancy. 6. Do internal stakeholders have Internal stakeholders can only access the data once they have been access to non-validated data? validated and disseminated. A working group is currently examining the possibility of authorizing internal stakeholders to access nonvalidated data. Internal stakeholders do not have access to non-validated data. 7. Which data sharing methods with other international organisations are already in place? In most cases, the IMF receives the NA data directly from the countries. Eurostat submits national accounts data for most EU countries in a bulk file (using GES.) on a monthly basis. Currently working on a data exchange process where would directly query Eurostat's databases using their "Bulk download facility". UNSD receives data from the OECD, the UNECE and CARICOM. Data for selected high-income economies are from the OECD. The ECE provides data for transition economies. CARICOM uses the NAQ to collect data, i.e. United Nations questionnaire based on the 1993 SNA is sent to about 160 countries, areas, and territories out of the 235 (204 have provided sufficient data to be published) . 8. What is your revision policy? Countries are required to submit all updates and revisions. The data provided by countries each year replace previously submitted data. In general, figures for the most recent year are regarded as provisional. Where large changes have been made to country data due to changes in currency, adoption of new statistical standards/methods, etc, new series are created, thereby preserving the older data in overlapping years for analysis. 37 Data Workflow Practices in International Agencies Population ESTAT BIS 1. Which countries do you cover in your data collection? 2. Which indicators do you cover? Population data (T0110) and employment data (T0110/1). 3. Which templates do you use (please provide any templates or webforms)? 4. What is the time schedule of data collection and dissemination? OECD 26 countries are covered by data collection. One series: total population Collect data on QPOP. Standard questionnaire, national files or data is extracted from national websites. EA/EU data released at t+75 and t+100. QPOP: following national releases for dissemination: QPOP data are disseminated right after the validation. 5. How is data validation organised (timing, responsibilities, methods)? 6. Do internal stakeholders have access to non-validated data? 7. Which data sharing methods with other international organisations are already in place? 8. What is your revision policy? Eurostat will transmit validated European data for QPOP following the new data exchange procedures . EA/EU data released at t+75 and t+100 38 Data Workflow Practices in International Agencies Population IMF 1. Which countries do you cover in your data collection? 2. Which indicators do you cover? STA collects data only on "total population". 3. Which templates do you use (please provide any templates or webforms)? No specific template is used. The UNSD provides the data in an Excel template . 4. What is the time schedule of data collection and dissemination? Annual data usually in midJune, internal dissemination after one day, external with IFS Yearbook. 5. How is data validation organised (timing, responsibilities, methods)? No validation is performed on this dataset as all validation is the responsibility of the UNSD. UNSD WB The World Bank Development Data Group (DECDG) covers most of the World Bank member economies and all other DECDG collects/estimates many population/demographic data including total population, population by age/sex/place of residence (urban/rural), crude birth No specific template is used. The total population data are collected variety of sources (NSI, Eurostat, UNDP). The World Bank produces its own population estimates for a Data are collected twice year. Data are updated in April (WDI book, WDI database and Health Nutrition Population database) and in July in (WDI database Population dataand are Health reviewed through WDI review process around December-January, and Operational Guidelines exercise around April-May. 6. Do internal stakeholders have Not applicable. access to non-validated data? Not applicable. 7. Which data sharing methods with other international organisations are already in place? All population data published by STA are sourced from the UNSD as part of a data sharing agreement. 8. What is your revision policy? Countries are required to submit all available updates and revisions to the UNSD. About 35% of the total population data are collected from national statistical offices or Eurostat, and about 65% of the total population data received from the Unitedfor The group of countries which the World Bank considers using country estimates are the developed countries which produce high quality estimates every year or even more often 39 Data Workflow Practices in International Agencies Sector Accounts ECB 1. Which countries do you cover in 2. Which indicators do you cover? We cover all 27 EU countries 3. Which templates do you use (please provide any templates or webforms)? Data is always received through SDMX-EDI data files (from 2014 SDMXML 2.0/2.1). National compilers are provided only with the amended TP tables that include the expected codes for the series to be transmitted. These tables are included in Appendix I. No templates or webforms that convert actual data in SDMX-EDI data files are provided to national compilers. BIS Same as for GDP main aggregates. The Euro Area Accounts (EAA) present a complete and consistent set of Main items of the 5 sectors quarterly data for all resident institutional sectors and the rest of the for the main countries. world. In addition, the EAA integrates financial and non-financial statistics , thereby allowing for an integrated analysis of non-financial economic activities and financial transactions. The euro area accounts also contain consistent financial balance sheets. 4. What is the time schedule of data collection and dissemination? Same as for GDP main aggregates. For the regular quarterly financial accounts data production at the EAA the following time schedule exists: T+80 Transmission of MUFA Early Estimates (to ECB from national compilers); T+110 Transmission regular MUFA (to ECB from national compilers); T+120 Publication and dissemination of euro area and national accounts (from ECB to national compilers, international organizations, external/internal users). For the QSA dataflow received from Eurostat the following schedule exists: around T+94 Incomplete MS QSA transmissions (from Eurostat to ECB); around T+98 Validated MS QSA transmission (from Eurostat to ECB); around T+108 EA QSA transmission (from Eurostat to ECB). For the annual financial and non-financial accounts data received from Eurostat there is no time line since there is no revision policy. Annual data are disseminated automatically upon reception. 5. How is data In order to ensure the efficient exchange of high quality national MUFA validation data, a set of data validation and consistency procedures are organised implemented at the ECB, in the statistical production environment. These (timing, data checks refer to: completeness checks; horizontal consistency; responsibilities, balancing items consistency; aggregation consistency; who-to-whom methods)? consistency; Sizable revisions; Overall plausibility checks on other changes; Negative stocks. The data are said to be consistent if inconsistencies in the above mentioned cases do not exceed a threshold of 10 million for both stocks and transactions. The validation of country data is allocated between the EA national accounts team members. Each member validates its own set of countries and communicates any inconsistency issues directly with the country. 6. Do internal No. stakeholders have access to 7. Which data The ECB is responsible for the quarterly national financial accounts data sharing which are disseminated to the users through the Statistical Data methods with Warehouse and, at the same time, the data are also transmitted to the other NCBs/NSIs, Eurostat, BIS, OECD and IMF via SDMX-EDI data files. In international addition ECB receives validated non-financial accounts and annual organisations financial and non-financial data from Eurostat by means of SDMX-EDI are already in data files. The national accounts data flow at the ECB is visualized in the place? graphic blow Same as for GDP main aggregates. 8. What is your There is no revision policy national and international data providers are revision policy? allowed to revise and resubmit their data at any time. Same as for GDP main aggregates. Same as for GDP main aggregates. Same as for GDP main aggregates. Same as for GDP main aggregates. 40 Data Workflow Practices in International Agencies Sector Accounts OECD UNSD 1. Which countries do you Very few countries are covered by data collection. cover in your data collection? 2. Which indicators do you We collect data for non-financial and financial national accounts, of which cover? the following subject areas covered by this TF: ANFA, QNFSA, QFSA. Same as for GDP main aggregates. 3. Which templates do you use (please provide any templates or webforms)? Same as for GDP main aggregates. Part IV: Integrated economic accounts (from production to financial accounts); Total economy (S.1); Rest of the world (S.2); Non-financial corporations (S.11);Financial corporations (S.12); General government (S.13); Households (S.14); Non-profit institutions serving households (S.15); Combined Sectors: Non-Financial and Financial Corporations (S.11 + S.12); Households and NPISH (S.14 + S.15) 4. What is the time EU countries - QNFSA: These data are transmitted to the OECD by Eurostat at Same as for GDP main aggregates. schedule of data collection T+105 days for full QSA and free QSA received before t+97, around T+113 and dissemination? (received between t+98 and t+105) and around T+120 (received by Eurostat between t+105 and t+111). For free QSA datasets received later, individual data deliveries are provided.ANFA: T+24 months; QFSA: T+120 days. non-EU countries - QNFSA and ANFA: following national releases; QFSA: T+105 days 5. How is data validation for EU countries, for QNFSA and QFSA data are first validated by Eurostat and Same as for GDP main aggregates. organised (timing, ECB respectively and then transmitted by these institutions to the OECD responsibilities, methods)? which performs its own checks. 6. Do internal stakeholders have access to nonvalidated data? 7. Which data sharing methods with other international organisations are already in place? No. 8. What is your revision policy? Countries’ data are revised according to the national revisions policy. Same as for GDP main aggregates. To IMF: QNFSA, QFSA data released in OECD.stat are used to feed the Same as for GDP main aggregates. Principal Global Indicators website (in the context of Recommendation 15 of the DGI). To UN: twice a year ANFA series are sent by OECD to UNSD. From Eurostat: non validated EU data ANFA are automatically forwarded to the OECD through the e-Damis system on the OECD generic account SNA.contact@oecd.org. Validated EU data for QNFSA (Full and Free datasets) are transmitted to the OECD generic account SNA.contact@oecd.org. From ECB: EU validated QFSA data are transmitted to the OECD on the OECD generic account SNA.contact@oecd.org. Same as for GDP main aggregates. 41 Data Workflow Practices in International Agencies Sector Accounts Questions ESTAT 1. Which countries do you cover in your data collection? Sector Accounts EU27, HR and EFTA (CH, IS, NO) countries based on ESA 95 TP and methodology (legal basis).Some EU candidate countries (ME, RS) also provide some annual data. 2. Which indicators do you cover? Annual Sector Accounts data are transmitted via table 8 of the ESA TP, whereas Quarterly Sector Accounts data are transmitted via table 801. 3. Which templates do you use (please provide any templates or webforms)? The requirements of data transmissions are specified in the ESA TP. Eurostat provides specific EXCEL questionnaires for each sub-table. Sector Accounts templates are available at: https://circabc.europa.eu/w/browse/ab340f5e-fd71-47b6-b8b6-486c437bfb2d 4. What is the time schedule of data collection and dissemination? At present (ESA95 TP), Annual Sector Accounts have to be transmitted by t+9 months; countries data and aggregates are disseminated on the website and on the database of Eurostat at around t+10 months; Full Quarterly Sector Accounts (full QSA) transmitted by the countries at t+90 days; key Indicators by country are published at around t+105; EU and EA aggregates around t+120 days; Publishable Quarterly Sector Accounts (as agreed with countries for free publication) are transmitted to OECD (for international data sharing): at around t+105, t+113 and t+120 days. Standard checks upon reception and loading to the production database include basic format, structure, encoding, content tests, outlier detection, which are essential for automatic processing, as well as revision checks, and consistency checks 5. How is data validation organised (timing, responsibilities, methods)? 6. Do internal stakeholders have access to non-validated data? Non-validated Sector Accounts data are not shared, except for Annual Sector Accounts as given in section 4 above. 7. Which data sharing methods with other international organisations are already in place? The data sharing of Sector Accounts data with the other international organisations is explained in section 4 above. 8. What is your revision policy? The Sector Accounts data EU/EA aggregates and Publishable QSA data by country published quarterly at round t+120 days (or before: see section 4) are not subject to revisions outside the quarterly disseminations. However, the Annual Sector Accounts countries’ data may be revised during the year following the revisions policies of the countries. 42 ANNEX VII: EUROSTAT-OECD PROTOCOL: AGREED MINIMUM CHECKING RULES Implications for validation process Type of check Target Requirement Minimum check Preliminary checks on data reception, conversion, loading Sender, formats, codes, etc. Full respect of transmission and encoding conventions Essential Identification of key data characteristics ESA table number, unit, frequency, time span, series Consistency of data specifications Essential Checks on dubious values Empty, zero or negative values, etc. Respect of coding conventions and expected range of values Revision checks Comparison with previous transmissions Significant revisions not due to regular updates should be explained (metadata) Intra-file/table checks on the consistency of totals and breakdowns Various types of possible breakdowns (indicator, industry, sector/sub-sectors, etc.) Sum of breakdowns should be equal to respective total (for additive series) Other types of intrafile/table consistency checks Current prices/volumes, raw/adjusted data, annual/quarterly data, assets/liabilities, uses/resources, etc. Specific relations between some series are expected Extra-file/table consistency checks Variables transmitted via different files/tables Series should be coherent (except for vintages) Essential Unexplained major discrepancies can lead to refusal of dataset Statistic plausibility checks Standard deviation, etc. Significant deviations may indicate errors Useful Possible follow-up questions Economic plausibility checks Economic ratios, growth rates, etc. Significant deviations may indicate errors Useful Possible follow-up questions Consistency checks against other statistics Consistency checks against data published by other institution Related statistics (possible conceptual differences) Significant deviations may indicate errors Essential Essential Essential (Automatic) correction or back to sender (Automatic) correction or back to sender (Automatic) correction or back to sender Unexplained major revisions can lead to refusal of the dataset Misalignments can lead to refusal of dataset Misalignments can lead to refusal of dataset Essential Useful Possible follow-up questions Possible follow-up questions Same series published by other institutions Significant deviations may indicate errors Useful