SDMX AN OECD PERSPECTIVE Paul Schreyer OECD CCSA

SDMX – AN OECD PERSPECTIVE Paul Schreyer OECD CCSA Special Session, September 2014 Rome

Outline • Why we think SDMX is important • Key OECD Activities • What Have We learned: Lessons and Challenges • Looking Ahead

Why we think SDMX is important

Why we think SDMX is important • Standardisation of data transmission • Speed • Quality: accuracy, readability • Consistency between international sources

Au st r Po ia la n Fr d an Ir ce el Sl ov Ge and r U ak ma ni R te ep ny d u K bl N ing ic et do he m rla Po nds rt ug Lu S al xe pa m i Sw bo n itz ur er g la Cz n ec No d h rw Re a pu y b G lic re Be ece lg H ium un ga ry It Es aly t D on en ia m ar Is k r Fi ael nl a Sw nd ed en Ch Tu ili rk e Ko y Au re st a r Sl alia ov en Ca ia na d U ni Ja a te pa d St n at M es ex i N Ice co ew la Ze nd al an d Example: data differences between international sources Example: Government Deficit 2010 Differences between highest and lowest result, in %-points of GDP 4, 5 4, 0 3, 5 3, 0 2, 5 2, 0 1, 5 1, 0 0, 5 0, 0 5

Why we think SDMX is important (2) • The Process itself is useful: • Detailed discussions with other IOs and with countries concerning: – Data requirements – Templates – Data sharing

Why we think SDMX is important (3) • Example Mexico • SDMX is enabler of co-ordination of data production and dissemination at the national level

Key OECD Activities

1. Definition and maintenance of Global Data Structure Definitions (DSDs) • Already established: – National Accounts (SNA 2008) – Balance of Payments – FDI (OECD is maintenance agency) • Forthcoming: – Education, R&D, Merchandise Trade Statistics • Ongoing: Management of Global Registry in 2013 -2014 with Eurostat – Administrative duties for users – Maintaining the registry content – Coordination with SDMX working groups

2 a. Implementation: OECD Short Term Economic Statistics collection • DSD for OECD-specific data transmission of short-term indicators (prices, real indicators, etc. ) • 8 countries providing STES SDMX data, 9 countries ongoing implementation work • Goals: – For disseminators: • An open format specification with which to transmit data • Avoids having to package and push the data to OECD – For OECD: • Timeliness • One format and structure enables automated processing and checking of collected data • Ease of validation of data structure (correct coding, file format)

2 b. Implementation: Task Force on International Data Co-operation • Pilot exercise under the IAG (see ECB presentation) – testing SDMX data exchanges (push mode) – decreasing respondent burden of national data providers – Minimising data differences between IOs • First phase, since 2013: Key National Accounts and population data (annual and quarterly) • Second phase, 2015: institutional sector accounts • The first exchanges: – proved technical feasibility – revealed problems in coding data available according to pre 2008 SNA – helped clarifying formulation of SDMX messages – were extremely useful in understanding other IOs’ data collection

2 c. Implementation: OECD Preparations for regular data collection and dissemination • Implement SDMX IT Infrastructure for Global DSDs, both collection and dissemination – Focus on existing tools, reusability, generic tools • Build SDMX capacity for IT and non-IT staff • Align existing questionnaires with SDMX coding

Announcement: SDMX Expert Group meeting in Korea • Co-organised with the Ko. Stat, and SDMX Sponsor organisations • Seoul, 27 -30 October 2014 • Focus on – Experience with implementing Global DSDs – SDMX Working Groups new guidelines and improvements to standards – SDMX Technical solutions • 2 days training – Using SDMX Reference Infrastructure – Implementing Global DSDs • Enquiries: Gyorgy. Gyomai@OECD. org

Lessons learned

What has OECD learned after over 10 years of SDMX? • SDMX adoption is not simply a technological challenge • Methodological and subject-matter knowledge and resources are key • ‘Business process’ led by statisticians • SDMX knowledge good in Ios • Knowledge base in NSOs and central banks growing but still unevenly between countries • The early focus was on developing the SDMX technical standard. Now the main focus is how to ease adoption and provide guidelines for common issues and use cases

Key Challenges • Motivating statistical agencies to adopt SDMX – Make business case – IO support to implement SDMX • Motivating broader group of IOs to broaden subject matters (CCSA) • Dealing with cost of adoption – Legacy systems and standards must be maintained until all providers have phased out non-SDMX dissemination – Building SDMX knowledge in NSOs – Provide shared tools to ease the move from legacy systems (such as the SDMX Convertor) • Governance structure may need to evolve with take-up of SDMX by more countries and IOs

Way Forward

The way forward – general considerations • Near future should be consolidation phase • Demonstrate real-life workability • Involving more non-European countries and other CCSA members • Addresses practical implementation problems • Provide practical guidance; revisit and consolidate existing guidelines where required • Clearer prioritisation of tasks developments • Seek closer co-ordination with related initiatives, in particular High Level Group on Modernisation of Statistics

The way forward – specific issues • Complete global DSDs for major statistical domains – e. g. currently the SNA DSD can only code properly SNA 2008 data, and activity breakdowns with ISIC rev 4; • Further integration with Statistical Information Collaboration Community (SIS-CC) • OECD. Stat fully SDMX compatible • Use SDMX for PGI

Thank you!