## cm3-a-taxonomy-for-communicating-economic-statistics-releases-products-and-product-updates

## Source details

**Canonical URL:** [cm3-a-taxonomy-for-communicating-economic-statistics-releases-products-and-product-updates](https://www.imf.org/-/media/files/data/statistics/bpm6/approved-guidance-notes/cm3-a-taxonomy-for-communicating-economic-statistics-releases-products-and-product-updates.pdf)

## Other formats

- [Markdown version](/-/media/files/data/statistics/bpm6/approved-guidance-notes/cm3-a-taxonomy-for-communicating-economic-statistics-releases-products-and-product-updates.pdf.md)
- [Structured JSON version](/-/media/files/data/statistics/bpm6/approved-guidance-notes/cm3-a-taxonomy-for-communicating-economic-statistics-releases-products-and-product-updates.pdf.json)

---

### Purpose, scope, and rationale
- The Guidance Note (GN) provides recommendations for national statistical organisations (NSOs) to communicate:
  - economic statistics releases / vintages of data,
  - products, and
  - product updates to users.
- Adoption of international recommendations and incorporation into the SNA, BPM and GFSM is expected to assist NSOs in communicating releases, products and product updates to users.
- Implementation depends on tradition, organizational branding, resources, systems, statistical governance and laws.
- The Communication Task Team recommends inclusion of the GN proposals in a chapter of the SNA, BPM, and GFSM titled “Communicating and Disseminating Economic Statistics.”

### Current challenges and need for harmonisation
- Challenges:
  - Diverse vintage descriptors (examples: provisional, first, preliminary, second, and final).
  - Diverse substance descriptors (examples: initial estimates, mature estimates, data revisions, benchmark revisions, rebased estimates, chain-linked volumes, improvements to methods, and corrections).
  - Pandemic-era proliferation of terminologies due to new data sources and real-time indicators, often less timely than official estimates.
- Recommendation:
  - Consistent presentation of terminology across countries to facilitate cross-country and over-time comparisons while balancing domestic user needs.

### Illustrative revision schedule and vintages (first quarter 2021)
- Four new estimates for the first quarter of 2021 are made between 2021 and 2028 with these illustrative release/revision dates:
  - June 30th, 2021: 2021 Q1 may be revised as the quarterly estimates are produced.
  - March 31st, 2022: 2021 Q1 may be revised due to seasonal adjustment for 2021.
  - September 30th, 2023: 2021 Q1 may be revised due to benchmarking through SUTs.
  - May 30th, 2024: 2021 Q1 may be revised due to the results of an economic census.
  - September 30th, 2028: 2021 Q1 may be revised due to the implementation of the new SNA.
- Example vintages timing assumption:
  - First vintage: 30-day after reference period.
  - Second vintage: 60-day after reference period.
  - Third vintage: 90-day after reference period.
- Example communication labels for successive vintages of Quarterly GDP, first quarter 2021:
  - Quarterly National Accounts release, first quarter 2021 - Regular update – 30-day
  - Quarterly National Accounts release, first quarter 2021 - Regular update – 60-day - (2021 Q1 revised)
  - Quarterly National Accounts release, first quarter 2021 - Regular update – 90-day - (2021 Q1 revised)

### Core definitions and vintage terminology (recommended)
- Statistic: a discrete unit of observation derived from measurement and representable numerically.
- Time-series: a series of regular time-ordered statistics taken at successive, in most cases equidistant, periods.
- Vintage of statistics: a statistic or set of statistics for a given reference period released for use at a particular release period; a new vintage when same data for same reference period are released at a different release period.
- Reference period: the time-period represented by the statistic.
- Update period: the time-period over which revisions to a statistic have been applied.
- Release period: the calendar date that data are released to the public.
- Update (revision): the numerical difference between two vintages of the same data point (example: first vintage real GDP 2019 = 0.5 per cent, second vintage = 0.4 per cent, revision = 0.1 percentage points).
- Regular (routine) update: reflects revisions from incorporation of more complete (but not yet final) source data, improved models, and/or methods (occurs for sub-annual and annual estimates; may include seasonal adjustment, correction of compilation errors, minor methodological adjustments).
- Benchmark estimate: the final vintage compiled using the highest quality source data and most advanced methods at that time; generally not revised and often called the “final” estimate (changes in concept or move to a new standard can generate further changes).
- Benchmark update: revisions from incorporation of benchmark estimate(s) (final vintage of all source data) into a set of economic statistics / accounts.
- Comprehensive update: a special case of benchmark update that also integrates new or updated concepts, accounting treatments, classifications or substantially improved methods or base-year updating; generally results in breaks in time series and backcasting needs.

### Recommended minimum descriptive elements for releases/vintages
- National authorities should include, at a minimum, when describing a release:
  - (1) substance of the release;
  - (2) timeliness;
  - (3) frequency;
  - (4) reference period;
  - (5) update period.
- Adopt standardised definitions and consistent presentation across countries to facilitate cross-country and over-time comparisons.

### Decomposing revisions: categories, presentation guidance, and numeric example
- Recommended categories for decomposing updates (revisions) to show revision source and rationale:
  - Conceptual changes
  - Methodological changes
  - Accounting changes
  - Coverage adjustments
  - Source data changes
  - Quality changes (e.g., improved seasonal adjustment, data validation, balancing adjustments)
  - Presentational changes
- Presentation guidance:
  - In current prices, show revision contributions in monetary terms.
  - In volume terms, show contributions as changes to growth rates.
  - Use combinations of tables and charts; consider showing component contributions for a single period or across all periods revised.
  - Supplement decomposition by showing revisions by industry, expenditure category, or institutional sector.
- Example decomposition of a $10 billion benchmark GDP update:
  - Conceptual Change: $4 — Incorporated estimates of Financial Intermediation Services Indirectly Measured into the accounts for the first time.
  - Methodological change: $1 — Improved method used to calculate imputed rental income.
  - Accounting change: $1 — Recording consumption of fixed capital at replacement cost.
  - Coverage change: $1 — Extended coverage to include a wider geographical boundary.
  - Source data change: $2 — Incorporated new estimates from the latest household budget survey.
  - Quality change: $1 — Improved seasonal adjustment.
  - Presentational change: zero — Presentational changes were made but did not result in any revisions.
  - Total Change/Revision: $10
- Interpretive note from example: nearly half of the revision is due to a change in concept and 20 per cent is due to incorporation of new and more complete data.

### Classifying types of updates and product quality taxonomy
- Classification of updates by extent/substance:
  - Regular updates (reflect vintage of source data and routine incorporation of more complete information).
  - Benchmark updates (incorporation of final vintage source data).
  - Comprehensive updates (addition/changes to concepts, methods, classifications or presentation — substantial).
- Product quality taxonomy (two-tier):
  - Level 1 — Official Statistics:
    - Estimates that incorporate recommended nationally or internationally adopted concepts, methods, accounting rules and classifications and meet all required standards.
    - Includes provisional estimates that use required concepts/methods/accounting/classifications but represent early estimates expected to “graduate” to final official statistics after scrutiny, quality assurance, and user feedback.
  - Level 2 — Experimental Estimates:
    - Estimates where the agency may have departed from recommended concepts/methods/accounting/classifications; often research or indicative in nature, based on untested source data or proxies.
    - Released with “proof of concept” intent and to solicit feedback; may become official statistics if improved to meet standards.

### Product taxonomy and placement within accounting frameworks
- Recommended distinctions:
  - Accounts:
    - Sequence of Economic Accounts — main interconnected sequence with opening and closing items, debits and credits, balancing item(s).
    - Supplementary/Thematic Accounts — operate outside the main sequence or include features not in the main accounts but still have opening/closing items and a debit/credit structure.
  - Tables:
    - Economic Tables (e.g., SUTs / IOTs) — operate within SNA boundaries.
    - Supplementary Tables (e.g., labour, pensions, regional, environmental) — operate within SNA boundaries and provide additional information.
    - Extended Tables (e.g., extended SUTs, unpaid household work) — operate across the SNA boundary and apply similar definitions.
    - Thematic Tables (e.g., health, tourism, sport, creative sector, etc.) — operate beyond the SNA boundary and apply different definitions/valuations.
- Recommendation: prefer terms “supplementary account” or “thematic account” instead of “satellite account”; reserve “table” for outputs without balancing items.

### Practical implementation guidance and phased adoption
- Implementation should be gradual where existing branding, documentation, training, and sources and methods materials make immediate change impractical.
- CMTT suggests phased adoption of recommendations, culminating with full transition to the next update of the SNA.
- Producers are encouraged to:
  - Adopt standard naming conventions and recording formats for vintages.
  - Decompose and clearly present revision sources and magnitudes.
  - Use the two-tier product quality taxonomy to communicate product status and expected evolution.
  - Situate products within the product taxonomy to clarify relationships to the sequence of accounts and supplementary outputs.

### Vintages naming conventions, recording examples, and timing options
- General approach:
  - Communicate successive vintages using explicit Type, Timeliness, Frequency, Reference Period, and Update Period labels.
  - Provide human-readable release descriptors (example: “Quarterly National Accounts release, first quarter 2021 - Regular update – 30-day”).
  - Record vintages in real-time tables showing Release Date, Update Type, and series values by reference period.
- Quarterly example (dates and values preserved):
  - Example timeline:
    - On 31st July, 2021 the NSO publishes data for the second quarter of 2021 and updates estimates for the first quarter of 2021.
    - Additional vintages of second quarter estimates are published on 30th August, 2021 and 30th September, 2021.
  - Recording vintages (example values preserved):
    - Series: GDP; Frequency: Quarterly
      - Release Date: June 30th, 2021; Update Type: Regular; Q1 2021: 99; Q4 2020: 95; Q3 2020: 90; Q2 2020: 85; Q1 2020: 80
      - Release Date: May 30th, 2021; Update Type: Regular; Q1 2021: 102; Q4 2020: 95; Q3 2020: 90; Q2 2020: 85; Q1 2020: 80
      - Release Date: April 30th, 2021; Update Type: Regular; Q1 2021: 100; Q4 2020: 95; Q3 2020: 90; Q2 2020: 85; Q1 2020: 80
    - Multiple release-cycle recording (example values preserved):
      - Series: GDP; Frequency: Quarterly
        - Release Date: September 30th, 2021; Regular; Q2 2021: 105; Q1 2021: 100; Q4 2020: 95; Q3 2020: 90; Q2 2020: 85; Q1 2020: 80
        - Release Date: August 30th, 2021; Regular; Q2 2021: 102; Q1 2021: 100; Q4 2020: 95; Q3 2020: 90; Q2 2020: 85; Q1 2020: 80
        - Release Date: July 31st, 2021; Regular; Q2 2021: 104; Q1 2021: 100; Q4 2020: 95; Q3 2020: 90; Q2 2020: 85; Q1 2020: 80
        - Release Date: June 30th, 2021; Regular; Q2 2021: 99; Q1 2021: 95; Q4 2020: 90; Q3 2020: 85; Q2 2020: 80
        - Release Date: May 30th, 2021; Regular; Q2 2021: 102; Q1 2021: 95; Q4 2020: 90; Q3 2020: 85; Q2 2020: 80
        - Release Date: April 30th, 2021; Regular; Q2 2021: 100; Q1 2021: 95; Q4 2020: 90; Q3 2020: 85; Q2 2020: 80
- Alternative timeliness notation and transition:
  - Timeliness may use non-standard numbers of days (examples: 15/45/60) or number of weeks (rounded).
  - For countries familiar with “flash” or “advance”, allow transition period using proposed terms in labels or footnotes.

### Annual, benchmark, and comprehensive revision examples (with preserved values)
- Annual national accounts vintages (example values preserved):
  - If only annual national accounts are produced nine months following the reference period:
    - Communicated as: Regular – 9 month – Annual National Accounts, 2020
    - Recording examples:
      - Series: GDP; Frequency: Annual
        - Release Date: March 31st, 2021; Type: Regular; 2020: 105,000; 2019: 100,000; 2018: 95,000
        - Release Date: September 30th, 2020; Type: Regular; 2020: 102,000; 2019: 100,000; 2018: 95,000
- Benchmark revisions (off-cycle example and recorded values preserved):
  - Example off-cycle benchmark:
    - Benchmark update of 2018 GDP published on December 21st, 2021.
    - Communicated as: Benchmark – Annual GDP release 2018 (2018-2020 updated)
    - Recording example:
      - Series: GDP; Frequency: Annual
        - Release Date: March 31st, 2021; Regular; 2020: 115,000; 2019: 109,000; 2018: 102,000
        - Release Date: December 31st, 2021; Benchmark; 2020: 110,000; 2019: 102,000
        - Release Date: March 31st, 2020; Regular; 2020: 105,000; 2019: 99,000
        - Release Date: March 31st, 2019; Regular; 2019: 100,000
  - Example on-cycle combination (values preserved):
    - Communicated as:
      - Benchmark – Annual National Accounts, 2018 (2018-2019 updated)
      - Regular – Annual National Accounts, 2020
    - Recording example:
      - Series: GDP; Frequency: Annual
        - Release Date: March 31st, 2021; Regular; 2020: 115,000; 2019: 109,000; 2018: 102,000
        - Release Date: March 31st, 2021; Benchmark; 2020: 109,000; 2019: 102,000
        - Release Date: March 31st, 2020; Regular; 2020: 105,000; 2019: 99,000
        - Release Date: March 31st, 2019; Regular; 2019: 100,000
- Comprehensive revisions (values preserved):
  - Example scenario: incorporation of conceptual and methodological changes from release of 2025 SNA in calendar year 2026; benchmark estimates established for 2023 and data revised back to 2020.
  - Naming Convention: Comprehensive – Annual National Accounts release, 2023 (2024-2020 updated)
  - Recording comprehensive revision example:
    - Series: GDP; Frequency: Annual; Release Date: March 31st, 2026; Type: Regular; 2025: 130; 2024: 128; 2023: 124; 2022: 124; 2021: 120; 2020: 119
    - Same release date and series; Type: Comprehensive; 2025: 128; 2024: 124; 2023: 124; 2022: 120; 2021: 119
    - Additional historical examples preserved in the table: Release Date March 31st 2005 Regular 125 121 121 116 115; March 31st, 2024 Regular 122 121 116 115; March 31st, 2023 Regular 119 116 115; March 31st, 2022 Regular 117 114; March 31st, 2021 Benchmark 115

### Compilation steps and types of changes creating revisions
- Compilation process stages and example types of changes that create revisions:
  - Source data:
    - Different concepts, basis, coverage, timeliness, etc.; results based on improved survey response rates, etc.
  - Conversion process:
    - Various steps and different types of adjustments required; grouped into categories for the GN.
  - Official economic statistics:
    - Consistent with the SNA, BPM, GFS, etc.
- Types of revisions and examples:
  - Conceptual change: aligning to international guidance, e.g., change to the production boundary, capitalising costs of R&D.
  - Methodological change: replacing modelling algorithm with a survey-based estimate, etc.
  - Accounting rules: change from cash to accruals.
  - Presentational changes: new tables of growth rates, charts, revisions triangles, granular detail, etc.
  - Coverage adjustments: exhaustiveness.
  - Quality: data validation, consistency of source data results, seasonal adjustment, deflation, impact of revised balancing adjustments, etc.

### Committee questions and recommended standardizations (policy recommendations)
- Committee questions for global consultation include approval to consult, feasibility of standardized breakdown of revisions, agreement with two-tier labelling of product quality, and terminology for accounts and tables.
- Recommended standardizations:
  - Standardize breakdown of revisions to explicitly label:
    - Conceptual changes
    - Methodological changes
    - Accounting rules changes
    - Changes to source data
    - Presentational changes
  - Adopt two-tiered labeling of statistical product quality:
    - Official and provisional (distinct from experimental)
    - Define experimental and provisional as described in the GN
  - Adopt consistent terminology across accounts and tables:
    - ‘economic accounts’, ‘supplementary accounts’, ‘thematic accounts’
    - ‘economic tables’, ‘supplementary tables’, ‘extended tables’, ‘thematic tables’

*Prepared by the Communication Task Team (CMTT), 2021; chapter in the BPM, SNA and GFSM titled “Communicating and Disseminating Economic Statistics.”*

### INTRODUCTION

### cm3-a-taxonomy-for-communicating-economic-statistics-releases-products-and-product-updates - INTRODUCTION

### Purpose and scope
- National statistical organisations (NSOs) need to communicate much more than the statistics themselves, including information related to the quality, timeliness, frequency and reference period of the product.
- The Guidance Note (GN) provides a set of recommendations for NSOs to communicate:
  - economic statistics releases / vintages of data,
  - products, and
  - product updates to users.
- The degree of implementation depends on tradition, organizational branding, resources, systems, statistical governance and laws.
- Adoption of international recommendations and incorporation into the SNA, BPM and GFSM is expected to assist NSOs in communicating releases, products and product updates to users.

### Current challenges: terminology and comparability
- A complex landscape exists with a range of different terms and phrases used to describe releases and product updates:
  - Examples of vintage descriptors: provisional, first, preliminary, second, and final.
  - Examples of substance descriptors: initial estimates, mature estimates, data revisions, benchmark revisions, rebased estimates, chain-linked volumes, improvements to methods, and corrections.
- The variety of terminologies makes cross-country comparisons and interpretation of economic statistics a challenge for users.
- The pandemic increased the number of terminologies used to demarcate vintages and metadata descriptions due to:
  - proliferation of new data sources,
  - real-time indicators aimed at giving early insights into economic developments,
  - these new sources are usually provided by less timely official estimates.

### Recommendation for harmonisation and institutional uptake
- A consistent presentation of terminology across countries would facilitate comparisons between countries and over time while balancing domestic user needs.
- The Communication Task Team recommends that the proposals of this GN are included in a chapter of the System of National Accounts (SNA), Balance of Payments Manual (BPM), and Government Finance Statistics Manual (GFSM) titled “Communicating and Disseminating Economic Statistics.”

### Existing guidance and opportunity
- Significant guidance on communicating revisions, products, product updates and releases already exists in the international community.
- Inconsistencies remain across existing guidance and recommendations.
- The update of the current SNA and BPM is an opportune time to:
  - review existing work,
  - identify best practices, and
  - incorporate changes into the existing body of work via recommendations to improve how NSOs communicate data releases, products, and product updates to users.

### Communicating releases / vintages of data: conceptual framing
- Producers must balance users’ needs for:
  - timely, high frequency economic data, and
  - highly accurate economic data.
- Producers must also balance:
  - users’ expectations for long consistent time-series, and
  - the desire for agile economic accounting statistical standards that ensure an exhaustive measure of economic activity.
- An implicit agreement exists between economic accountants and users:
  - economic accountants provide timely, high frequency data reflecting current economic conditions if users accept updates to those estimates in the future.
- The fact that a statistic released today can be updated multiple times creates vintages of economic statistics.
- Vintages should be reflected in a release calendar with dates of forthcoming releases published well in advance.
- Illustration (partial in source): a national authority releasing an estimate of GDP for the first quarter of 2021 on May 30th.

*Prepared by Claire Evans, Robert Kent-Smith, Kwangwon Lee, David Wasshausen, Orestis Tsigkas, Francien Berry, James Tebrake and Richard Kimuli under the supervision of Mr. Sanjiv Mahajan (Communication Task Team).*

### 2021. Between 2021 and 2028 four new estimates are made

### cm3-a-taxonomy-for-communicating-economic-statistics-releases-products-and-product-updates - 2021. Between 2021 and 2028 four new estimates are made

### Illustrative revision schedule and vintages (first quarter 2021)
- Four new estimates for the first quarter of 2021 are made between 2021 and 2028 with these illustrative release/revision dates:
  - June 30th, 2021: 2021 Q1 may be revised as the quarterly estimates are produced.
  - March 31st, 2022: 2021 Q1 may be revised due to seasonal adjustment for 2021.
  - September 30th, 2023: 2021 Q1 may be revised due to benchmarking through SUTs.
  - May 30th, 2024: 2021 Q1 may be revised due to the results of an economic census.
  - September 30th, 2028: 2021 Q1 may be revised due to the implementation of the new SNA.
- Example vintages timing assumption:
  - First vintage: 30-day after reference period.
  - Second vintage: 60-day after reference period.
  - Third vintage: 90-day after reference period.
- Example communication labels for successive vintages of Quarterly GDP, first quarter 2021:
  - Quarterly National Accounts release, first quarter 2021 - Regular update – 30-day
  - Quarterly National Accounts release, first quarter 2021 - Regular update – 60-day - (2021 Q1 revised)
  - Quarterly National Accounts release, first quarter 2021 - Regular update – 90-day - (2021 Q1 revised)

### Core definitions and vintage terminology (Box I)
- Key terms and their definitions as recommended:
  - Statistic: a discrete unit of observation derived from measurement and representable numerically.
  - Time-series: a series of regular time-ordered statistics taken at successive, in most cases equidistant, periods.
  - Vintage of statistics: a statistic or set of statistics for a given reference period released for use at a particular release period; a new vintage when same data for same reference period are released at a different release period.
  - Reference period: the time-period represented by the statistic.
  - Update period: the time-period over which revisions to a statistic have been applied.
  - Release period: the calendar date that data are released to the public.
  - Update (revision): the numerical difference between two vintages of the same data point (example: first vintage real GDP 2019 = 0.5 per cent, second vintage = 0.4 per cent, revision = 0.1 percentage points).
  - Regular (routine) update: reflects revisions from incorporation of more complete (but not yet final) source data, improved models, and/or methods (occurs for sub-annual and annual estimates; may include seasonal adjustment, correction of compilation errors, minor methodological adjustments).
  - Benchmark estimate: the final vintage compiled using the highest quality source data and most advanced methods at that time; generally not revised and often called the “final” estimate (changes in concept or move to a new standard can generate further changes).
  - Benchmark update: revisions from incorporation of benchmark estimate(s) (final vintage of all source data) into a set of economic statistics / accounts.
  - Comprehensive update: a special case of benchmark update that also integrates new or updated concepts, accounting treatments, classifications or substantially improved methods or base-year updating; generally results in breaks in time series and backcasting needs.

### Recommended minimum descriptive elements for releases/vintages
- The CMTT recommends that national authorities include, at a minimum, the following when describing a release:
  - (1) substance of the release;
  - (2) timeliness;
  - (3) frequency;
  - (4) reference period;
  - (5) update period.
- The CMTT recommends adopting standardised definitions and consistent presentation across countries to facilitate cross-country and over-time comparisons.

### Decomposing revisions: categories and presentation
- Recommended categories for decomposing updates (revisions) to show revision source and rationale:
  - Conceptual changes
  - Methodological changes
  - Accounting changes
  - Coverage adjustments
  - Source data changes
  - Quality changes (e.g., improved seasonal adjustment, data validation, balancing adjustments)
  - Presentational changes
- Presentation guidance:
  - In current prices, show revision contributions in monetary terms.
  - In volume terms, show contributions as changes to growth rates.
  - Use combinations of tables and charts; consider showing component contributions for a single period or across all periods revised.
  - Supplement decomposition by showing revisions by industry, expenditure category, or institutional sector.
- Example decomposition of a $10 billion benchmark GDP update:
  - Conceptual Change: $4 — Incorporated estimates of Financial Intermediation Services Indirectly Measured into the accounts for the first time.
  - Methodological change: $1 — Improved method used to calculate imputed rental income.
  - Accounting change: $1 — Recording consumption of fixed capital at replacement cost.
  - Coverage change: $1 — Extended coverage to include a wider geographical boundary.
  - Source data change: $2 — Incorporated new estimates from the latest household budget survey.
  - Quality change: $1 — Improved seasonal adjustment.
  - Presentational change: zero — Presentational changes were made but did not result in any revisions.
  - Total Change/Revision: $10
- Interpretive note from example: nearly half of the revision is due to a change in concept and 20 per cent is due to incorporation of new and more complete data.

### Classifying types of updates and changes
- Recommended terminology to communicate the “extent” or “substance” of revisions:
  - Regular updates (reflect vintage of source data and routine incorporation of more complete information).
  - Benchmark updates (incorporation of final vintage source data).
  - Comprehensive updates (addition/changes to concepts, methods, classifications or presentation — substantial).
- Encouraged practice: categorise and decompose updates by source (conceptual, methodological, accounting, coverage, source data, quality, presentational) and, where relevant, further break down by affected areas of the accounts.

### Product quality taxonomy (two-tier)
- Level 1 — Official Statistics:
  - Estimates that incorporate recommended nationally or internationally adopted concepts, methods, accounting rules and classifications and meet all required standards.
  - Includes provisional estimates that use required concepts/methods/accounting/classifications but represent early estimates expected to “graduate” to final official statistics after scrutiny, quality assurance, and user feedback.
- Level 2 — Experimental Estimates:
  - Estimates where the agency may have departed from recommended concepts/methods/accounting/classifications; often research or indicative in nature, based on untested source data or proxies.
  - Released with “proof of concept” intent and to solicit feedback; may become official statistics if improved to meet standards.
- Communication aims of labels: primarily to convey quality, methodological compliance, and whether the product is in exploration/development and open to user feedback.

### Product taxonomy and placement within accounting frameworks
- Recommended product distinctions:
  - Accounts:
    - Sequence of Economic Accounts — main interconnected sequence with opening and closing items, debits and credits, balancing item(s).
    - Supplementary/Thematic Accounts — operate outside the main sequence or include features not in the main accounts but still have opening/closing items and a debit/credit structure.
  - Tables:
    - Economic Tables (e.g., SUTs / IOTs) — operate within SNA boundaries.
    - Supplementary Tables (e.g., labour, pensions, regional, environmental) — operate within SNA boundaries and provide additional information.
    - Extended Tables (e.g., extended SUTs, unpaid household work) — operate across the SNA boundary and apply similar definitions.
    - Thematic Tables (e.g., health, tourism, sport, creative sector, etc.) — operate beyond the SNA boundary and apply different definitions/valuations.
- Recommendation: prefer terms “supplementary account” or “thematic account” instead of “satellite account” to reduce misunderstanding; reserve “table” for outputs without balancing items.

### Practical implementation guidance
- Implementation should be gradual where existing branding, documentation, training, and sources and methods materials make immediate change impractical.
- CMTT suggests phased adoption of recommendations, culminating with full transition to the next update of the SNA.
- Producers are encouraged to:
  - Adopt standard naming conventions and recording formats for vintages (examples provided).
  - Decompose and clearly present revision sources and magnitudes.
  - Use the two-tier product quality taxonomy to communicate product status and expected evolution.
  - Situate products within the product taxonomy to clarify relationships to the sequence of accounts and supplementary outputs.

*Source: CMTT, 2021.*

### chapter in the BPM, SNA and GFSM    titled “Communicating and Disseminating Economic

### Communicating and Disseminating Economic Statistics

### Committee questions and recommended standardizations
- Questions for the Committee:
  - 1. Do you approve this note for global consultation?
  - 2. Do you agree with the recommendation for countries to standardize the breakdown of revisions to highlight conceptual changes, methodological changes, accounting rules changes, changes to source data, and presentational changes?
    - a. Is this breakdown feasible for your economy without significant change to the statistical infrastructure?
  - 3. Do you agree with the two-tiered standardized labelling of the ‘quality’ of statistical products as official and provisional separate from experimental estimates?
    - a. Do you agree with the description of experimental estimates as: “estimates released by a NSO in which the NSO may have gone outside of nationally or internationally recommended concepts, methods, accounting rules or classifications in the production of the estimates”?
    - b. Do you agree with the description of provisional estimates as: “estimates that incorporate nationally or internationally adopted concepts, methods, accounting rules and classifications but for which the scope of the output has not been finalized, for example, they reflect early vintages of source data”?
  - 4. Do you agree with the recommendation to adopt the terms, ‘economic accounts’, ‘supplementary accounts’, and ‘thematic’ accounts?
  - 5. Do you agree with the recommendation to adopt the terms, ‘economic tables’, ‘supplementary tables’, ‘extended tables’ and ‘thematic tables’ across all domains when referring to those tables outside the main sequence of accounts?
  - 6. Do you have other comments on this guidance note?

- Recommended standardizations (policy recommendations):
  - Standardize breakdown of revisions to explicitly label:
    - Conceptual changes
    - Methodological changes
    - Accounting rules changes
    - Changes to source data
    - Presentational changes
  - Adopt two-tiered labeling of statistical product quality:
    - Official and provisional (distinct from experimental)
    - Define experimental and provisional as per the descriptions above
  - Adopt consistent terminology across accounts and tables:
    - ‘economic accounts’, ‘supplementary accounts’, ‘thematic accounts’
    - ‘economic tables’, ‘supplementary tables’, ‘extended tables’, ‘thematic tables’

### Vintages of economic statistics — naming conventions, examples, and recording
- General approach:
  - Communicate successive vintages for the same reference period using explicit Type, Timeliness, Frequency, Reference Period, and Update Period labels.
  - Provide human-readable release descriptors (e.g., “Quarterly National Accounts release, first quarter 2021 - Regular update – 30-day”).
  - Record vintages in real-time tables showing Release Date, Update Type, and series values by reference period.

- Quarterly example (dates and timeliness preserved):
  - Example timeline:
    - On 31st July, 2021 the NSO publishes data for the second quarter of 2021 and updates estimates for the first quarter of 2021.
    - Additional vintages of second quarter estimates are published on 30th August, 2021 and 30th September, 2021.
  - Naming Convention: Quarterly GDP (successive releases)
    - Communicated vintages:
      - Quarterly National Accounts release, first quarter 2021 - Regular update – 30-day
      - Quarterly National Accounts release, first quarter 2021 - Regular update – 60-day -  (2021Q1 revised)
      - Quarterly National Accounts release, first quarter 2021 - Regular update – 90-day -  (2021Q1 revised)
  - Recording vintages (example values preserved):
    - Table: Recording Vintages of Data in Real-time Tables: Quarterly GDP
      - Series: GDP; Frequency: Quarterly
        - Release Date: June 30th, 2021; Update Type: Regular; Q1 2021: 99; Q4 2020: 95; Q3 2020: 90; Q2 2020: 85; Q1 2020: 80
        - Release Date: May 30th, 2021; Update Type: Regular; Q1 2021: 102; Q4 2020: 95; Q3 2020: 90; Q2 2020: 85; Q1 2020: 80
        - Release Date: April 30th, 2021; Update Type: Regular; Q1 2021: 100; Q4 2020: 95; Q3 2020: 90; Q2 2020: 85; Q1 2020: 80
    - Multiple release-cycle recording (example values preserved):
      - Series: GDP; Frequency: Quarterly
        - Release Date: September 30th, 2021; Regular; Q2 2021: 105; Q1 2021: 100; Q4 2020: 95; Q3 2020: 90; Q2 2020: 85; Q1 2020: 80
        - Release Date: August 30th, 2021; Regular; Q2 2021: 102; Q1 2021: 100; Q4 2020: 95; Q3 2020: 90; Q2 2020: 85; Q1 2020: 80
        - Release Date: July 31st, 2021; Regular; Q2 2021: 104; Q1 2021: 100; Q4 2020: 95; Q3 2020: 90; Q2 2020: 85; Q1 2020: 80
        - Release Date: June 30th, 2021; Regular; Q2 2021: 99; Q1 2021: 95; Q4 2020: 90; Q3 2020: 85; Q2 2020: 80
        - Release Date: May 30th, 2021; Regular; Q2 2021: 102; Q1 2021: 95; Q4 2020: 90; Q3 2020: 85; Q2 2020: 80
        - Release Date: April 30th, 2021; Regular; Q2 2021: 100; Q1 2021: 95; Q4 2020: 90; Q3 2020: 85; Q2 2020: 80

- Alternative timeliness notation and transition:
  - Timeliness may not be a standard number of days (countries may use 15/45/60).
  - Alternative: indicate number of weeks (rounded).
  - Transition guidance: for countries familiar with terms like "flash" or "advance", consider a transition period using proposed terms in labels or footnotes before switchover.

- Annual national accounts vintages:
  - If only annual national accounts are produced nine months following the reference period:
    - Naming Convention: Annual GDP (New Vintage)
      - Type: Regular; Timeliness: 9 month; Frequency: Annual; Reference Period: 2020
      - Communicated as: Regular – 9 month – Annual National Accounts, 2020
  - If another vintage of 2020 annual GDP estimates is published before releasing 2021 estimates:
    - Naming Convention: Annual GDP (New Vintage)
      - Type: Regular; Timeliness: 15 month; Frequency: Annual; Reference Period: 2020; Update Period: 2020
      - Communicated as: Regular – 15 month – Annual National Accounts, 2020 (2020 updated)
  - Recording examples (values preserved):
    - Series: GDP; Frequency: Annual
      - Release Date: March 31st, 2021; Type: Regular; 2020: 105,000; 2019: 100,000; 2018: 95,000
      - Release Date: September 30th, 2020; Type: Regular; 2020: 102,000; 2019: 100,000; 2018: 95,000

- Benchmark revisions (off-cycle and on-cycle):
  - NSOs may adopt benchmark estimates reflecting final source data vintages either off-cycle or during regular release cycles.
  - Example off-cycle benchmark:
    - Benchmark update of 2018 GDP published on December 21st, 2021.
    - Naming Convention: Annual GDP (Release of Benchmark Revisions)
      - Type: Benchmark; Timeliness: Annual; Frequency: Annual; Reference Period: 2018; Update Period: 2020-2018
      - Communicated as: Benchmark – Annual GDP release 2018 (2018-2020 updated)
    - Recording example (values preserved):
      - Series: GDP; Frequency: Annual
        - Release Date: March 31st, 2021; Regular; 2020: 115,000; 2019: 109,000; 2018: 102,000
        - Release Date: December 31st, 2021; Benchmark; 2020: 110,000; 2019: 102,000
        - Release Date: March 31st, 2020; Regular; 2020: 105,000; 2019: 99,000
        - Release Date: March 31st, 2019; Regular; 2019: 100,000
  - Example on-cycle combination:
    - Naming Convention - Annual GDP (Successive Releases of Different Vintage)
      - Revision Types combined (example): Benchmark (2018; 2020-2018) and Regular (2020; 15 month)
      - Communicated as:
        - Benchmark – Annual National Accounts, 2018 (2018-2019 updated)
        - Regular – Annual National Accounts, 2020
      - Recording example (values preserved):
        - Series: GDP; Frequency: Annual
          - Release Date: March 31st, 2021; Regular; 2020: 115,000; 2019: 109,000; 2018: 102,000
          - Release Date: March 31st, 2021; Benchmark; 2020: 109,000; 2019: 102,000
          - Release Date: March 31st, 2020; Regular; 2020: 105,000; 2019: 99,000
          - Release Date: March 31st, 2019; Regular; 2019: 100,000

- Comprehensive revisions:
  - NSOs should distinguish and communicate conceptual and methodological changes (e.g., adoption of new SNA guidance) separately from regular updates.
  - Example scenario:
    - Compiling organisation incorporates conceptual and methodological changes resulting from the release of 2025 SNA in calendar year 2026; benchmark estimates established for 2023 and data revised back to 2020.
    - Naming Convention: Comprehensive Revision
      - Revision Type: Comprehensive; Timeliness: Annual; Frequency: Annual; Reference Period: 2023; Update Period: 2024-2020
      - Regular: Annual 2025
      - Communicated as:
        - Comprehensive – Annual National Accounts release, 2023 (2024-2020 updated)
        - Regular – Annual National Accounts release, 2025
    - Recording comprehensive revision example (values preserved):
      - Series: GDP; Frequency: Annual; Release Date: March 31st, 2026; Type: Regular; 2025: 130; 2024: 128; 2023: 124; 2022: 124; 2021: 120; 2020: 119
      - Same release date and series; Type: Comprehensive; 2025: 128; 2024: 124; 2023: 124; 2022: 120; 2021: 119
      - Additional historical examples preserved in the table: Release Date March 31st 2005 Regular 125 121 121 116 115; March 31st, 2024 Regular 122 121 116 115; March 31st, 2023 Regular 119 116 115; March 31st, 2022 Regular 117 114; March 31st, 2021 Benchmark 115

### Summary of compilation steps and types of changes creating revisions (Annex III)
- Compilation process stages and example types of changes that create revisions:
  - Source data
    - Different concepts, basis, coverage, timeliness, etc.
    - Results based on improved survey response rates, etc.
  - Conversion process
    - Various steps and different types of adjustments required; grouped into categories for the GN
  - Official economic statistics
    - Consistent with the SNA, BPM, GFS, etc.
- Types of revisions and examples:
  - Conceptual change
    - Aligning to international statistical guidance, e.g., change to the production boundary, capitalising costs of R&D, etc.
  - Methodological change
    - Change to the accounting rules, replacing modelling algorithm with a survey based estimate, etc.
  - Accounting rules
    - Change from cash to accruals
  - Presentational changes
    - New tables of growth rates, charts, revisions triangles, granular detail, etc.
  - Coverage adjustments
    - Exhaustiveness
  - Quality
    - Data validation, consistency of source data results, seasonal adjustment, deflation, impact of revised balancing adjustments, etc.

---


_Source: https://www.imf.org/-/media/files/data/statistics/bpm6/approved-guidance-notes/cm3-a-taxonomy-for-communicating-economic-statistics-releases-products-and-product-updates.pdf_