## _wp0546

## Source details

**Canonical URL:** [_wp0546](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp0546.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp0546.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp0546.pdf.json)

---

### Introduction: scope and context
- Publication of country staff reports and Public Information Notices has significantly expanded the statistical information published by the Fund since the mid-1990s.
- Prior to this expansion, dissemination was primarily through traditional statistical publications and databases led by International Financial Statistics (IFS) and World Economic Outlook (WEO); other traditional publications referenced include the Balance of Payments Statistics Yearbook (BOPSY), Direction of Trade Statistics (DOTS), and Government Finance Statistics Yearbook (GFSY).
- Before the expansion in publication of country staff reports, Recent Economic Developments (RED) reports and Statistical Appendices were often published.
- Data reported to the Fund and published in country staff reports and the IFS may differ for identical statistical concepts or variables, generating:
  - concerns about the accuracy and reliability of Fund-published data;
  - implications for policy advice based on analyses using these data;
  - reputational risk to the Fund.
- Principal sources of discrepancies include differences in objectives underpinning Fund publications, country-specific adaptations, and more recent data revisions included in staff reports.
- The study emphasizes that data differences should not be interpreted as reflecting deficiencies in data practices of IMF staff or member countries.

### Background and study design
- Purpose: assess consistency of data published in Article IV consultation staff reports and the IFS using only publicly available data and metadata.
- Sample: 66 Article IV consultation staff reports selected from about 150 published between September 2002 and June 2004.
- Variables compared (annual data; period averages for flows, end-of-period for stocks):
  - nominal GDP
  - real GDP growth
  - inflation rate
  - international reserves
  - external current account
  - merchandise exports
  - merchandise imports
  - general government balance
  - total public debt stock
  - bank credit to government
  - broad money stock
- Comparison year: latest year of complete historical data in both staff report and IFS (mostly 2001 or 2002; in some cases shifted to last year of actual data available in IFS).
- Classification outcomes:
  - direct match (M)
  - consistent when put on a comparable basis (m)
  - direct discrepancy (D)
  - inconsistent when put on a comparable basis (d)
  - not comparable (N)
- Matching/consistency threshold: data within 2 percent are considered matching/consistent.

### Key results: sample-wide and by country group
- Overall outcomes for the full sample (11 variables across 66 countries):
  - 64 percent of observations were either direct matches or consistent when put on a comparable basis.
  - 19 percent showed inconsistencies or discrepancies.
  - 17 percent were not comparable.
- By country group:
  - Advanced countries: 61 percent matching/consistent; 20 percent not comparable.
  - Transition countries: 66 percent matching/consistent; 19 percent inconsistencies/discrepancies.
  - Developing countries: 64 percent matching/consistent; 19 percent inconsistencies/discrepancies.
- Variable-specific highlights:
  - Nominal GDP and real GDP growth: matched/consistent for 88 percent of the sample.
    - Advanced countries had 12½ percent noncomparable for nominal GDP.
    - Transition nominal GDP matching 91 percent; transition real GDP noncomparable 27 percent.
    - Developing nominal GDP matching 90 percent.
  - Inflation rate: matched for about 80 percent of the sample. Example: Saudi Arabia 2001 reported as -0.8 percent (staff report) and -1.1 percent (IFS).
  - International reserves: generally comparable for most countries; developing countries had 5 out of 39 noncomparable cases. Example: Costa Rica—IFS 20 percent higher than staff report (staff noted bilateral claims under negotiation excluded). Sweden discrepancy due to different sources and valuation practices.
  - External current account balance: 35 percent incidence of differences in entire sample (25 percent for advanced; 41 percent for developing).
    - Chile and Colombia: current account balances differed by 20–30 percent though exports/imports matched closely.
    - Ecuador and Honduras: significant and persistent differences noted.
  - Government balance: high proportion not comparable—31 percent for advanced countries to 41 percent for transition countries—owing to unclear definitions of government and differing coverage.
    - Brazil example: IFS consolidated central government surplus 0.2 percent of GDP vs. staff report deficit over 4 percent of GDP under staff definition.
  - Public debt: highest incidence of noncomparable reporting—52 percent for the entire sample (31 percent advanced; 55 percent transition; 59 percent developing). Nonreporting to IFS was a major cause.
  - Bank credit to government: 42 percent direct match/consistency for entire sample; 100 percent not comparable for advanced country sample due to absence of monetary survey in staff reports. For IMDP subsample, bank credit to government match rate was 60 percent; broad money match rate 100 percent.
  - Broad money: 73 percent match/consistency for entire sample. For IMDP subsample, match or consistency was 100 percent.

### Primary reasons for differences and noncomparability
- Revisions captured at different times:
  - Staff reports may include updated information not yet transmitted to STA; IFS may incorporate revisions not reflected in staff reports (examples: Uruguay current account revision pattern; Swedish HICP rebasings).
- Differences in coverage or classification:
  - Use of fiscal-year reporting in staff reports vs. calendar-year in IFS; differing definitions of government and public sector coverage (examples: Ecuador, Brazil, Iran, Romania, Thailand).
  - Balance of payments classification differences can shift components (example: Thailand merchandise imports vs. services).
- Different methodologies or staff adjustments:
  - IFS follows international standards (BPM5, 1993 SNA); staff reports may use country-specific definitions or staff adjustments/estimates for analytic purposes (example: monetary aggregates per MFSM allowing national definitions; Algeria bank restructuring impact on bank credit to government).
  - Staff reports limited by space—may lack documentation of definitions and data quality.
- Different data sources within Fund and national agencies:
  - Area departments and STA may rely on different government contact points or databases (examples: Sweden reserves from Riksbank website vs. STA submissions; Korea GDP from central bank vs. economic planning board).
  - Fund Finance Department data adjustments (e.g., reserve position in the Fund, SDR holdings) can differ from publicly posted central bank datasets.

### Conclusions and recommended measures
- Study conclusions:
  - Noncomparable data and data differences between staff reports and IFS are significant for core indicators.
  - Better documentation of staff reports and STA/area department databases, and improved internal coordination, would likely have reduced incidence and persistence of differences.
  - Area departments have not fully utilized STA’s databases and methodological expertise; STA’s databases have not fully benefited from area department country knowledge.
- Recommended measures:
  - Integrated, country-by-country coordinated data collection arrangements jointly by area departments and STA to improve consistency.
  - Promote use of standard templates for data reporting rooted in accepted statistical methodologies; staff reports should document coverage and definitions of key variables (e.g., in statistical issues appendix) when they diverge from STA publications.
  - In program countries, consider supporting key data-improvement recommendations with performance criteria and benchmarks.
  - Standard templates should cover core Article IV surveillance indicators: nominal GDP, international reserves, essential balance of payments statistics, key monetary aggregates from the monetary survey, and financial soundness and external vulnerability indicators.
  - Encourage reporting of international reserves using the Data Template on International Reserves and Foreign Currency Liquidity to minimize discrepancies related to Fund financial variables (using Finance Department values).
  - Expand Integrated Monetary Data Project (IMDP) where feasible; IMDP benefits observed include:
    - reduced reporting burden for country authorities;
    - reduced and better-harmonized databases across area departments and STA;
    - easier identification and reduction of differences in monetary aggregates;
    - improved quality and transparency of monetary statistics;
    - significant staff-time economies for monetary surveillance.
  - Apply common sourcing selectively to other sectors where resource and operational constraints allow, acknowledging challenges where area departments require the most up-to-date data.
  - Where common sourcing is not feasible, ensure better metadata and clear explanations in staff reports when differences with IFS exist.

*Source: _wp0546*

### References..............................................................................................................

### References

### Tables and Figures (listed)
- Tables:
  - 1. Sample of Economies: Selected Characteristics and Fund Document Reference
  - 2. Advanced Economies: Comparability and Consistency of IMF IFS and Country Staff Reports, by Selected Indicators
  - 3. Transition Countries: Comparability of IMF IFS and Country Staff Reports, by Selected Indicators
  - 4. Developing Countries: Comparability and Consistency of IMF IFS and Country Staff Reports, by Selected Indicators
- Figures:
  - 1. Comparison of IMF Country Staff Report and IFS Data

### Introduction: scope and context
- The publication of country staff reports and Public Information Notices has significantly expanded the statistical information published by the Fund since the mid-1990s.
- Prior to this expansion, dissemination of economic and financial statistics by the Fund was primarily through its traditional statistical publications and databases, led by International Financial Statistics (IFS) and World Economic Outlook (WEO).
- Other traditional IMF statistical publications referenced include the Balance of Payments Statistics Yearbook (BOPSY), Direction of Trade Statistics (DOTS), and Government Finance Statistics Yearbook (GFSY).
- Before the expansion in publication of country staff reports, the Recent Economic Developments (RED) reports and Statistical Appendices that accompanied them were often published.

### Key findings on data differences and implications
- Data reported to the Fund by member countries and published in country staff reports and the IFS may differ for identical statistical concepts or variables.
- Potential consequences of such differences:
  - Concerns about the accuracy and reliability of data published by the Fund.
  - Implications for policy advice based on analysis using these data.
  - Reputational risk to the Fund as its published data come under increased external scrutiny.
- Principal sources of discrepancies:
  - Differences in the objectives underpinning the various Fund publications.
  - Adaptations to suit country-specific analysis.
  - More recent data revisions in staff reports.
- External users may lack time or access to Fund staff to determine reasons for differences.
- The study emphasizes that data differences should not be interpreted as reflecting deficiencies in data practices of IMF staff, or of member countries in their provision of data to the Fund.

### Historical management concern and past review
- The reputational risk to the Fund has been a long-standing concern of Fund management.
- In 1989, at management’s request, the then Bureau of Statistics undertook a study of 20 country desk databases and found numerous divergences between the Bureau’s and area departments’ databases.
- Many of the divergences from the 1989 review stemmed from poor coordination of statistical activities.
- The issue has practical significance for the financial operations and size of the Fund, as well as for the representation in it of member countries, because of the use of country data for calculating members’ quotas.

*Source: _wp0546 - References*

### Appendix II provides a detailed description of how the Finance Department handles such

### _wp0546 - Appendix II provides a detailed description of how the Finance Department handles such

### Background and study design
- Purpose: assess consistency of data published in Article IV consultation staff reports and the IFS using only publicly available data and metadata.
- Sample: 66 Article IV consultation staff reports selected from about 150 published between September 2002 and June 2004.
- Variables compared (annual data; period averages for flows, end-of-period for stocks):
  - nominal GDP
  - real GDP growth
  - inflation rate
  - international reserves
  - external current account
  - merchandise exports
  - merchandise imports
  - general government balance
  - total public debt stock
  - bank credit to government
  - broad money stock
- Comparison year: latest year of complete historical data in both staff report and IFS (mostly 2001 or 2002; in some cases shifted to last year of actual data available in IFS).
- Classification outcomes:
  - direct match (M)
  - consistent when put on a comparable basis (m)
  - direct discrepancy (D)
  - inconsistent when put on a comparable basis (d)
  - not comparable (N)
- Matching/consistency threshold: data within 2 percent are considered matching/consistent.

### Key results (sample-wide and by country group)
- Overall outcomes for the full sample (11 variables across 66 countries):
  - 64 percent of observations were either direct matches or consistent when put on a comparable basis.
  - 19 percent showed inconsistencies or discrepancies.
  - 17 percent were not comparable.
- By country group:
  - Advanced countries: 61 percent matching/consistent; 20 percent not comparable (highest); inconsistencies/discrepancies share noted.
  - Transition countries: 66 percent matching/consistent; 19 percent inconsistencies/discrepancies.
  - Developing countries: 64 percent matching/consistent; 19 percent inconsistencies/discrepancies.
- Variable-specific highlights (sample-wide unless noted):
  - Nominal GDP and real GDP growth: matched/consistent for 88 percent of the sample. Advanced countries had 12½ percent noncomparable for nominal GDP. Transition nominal GDP matching 91 percent; transition real GDP noncomparable 27 percent. Developing nominal GDP matching 90 percent.
  - Inflation rate: matched for about 80 percent of the sample. Example discrepancy: Saudi Arabia 2001 reported as -0.8 percent (staff report) and -1.1 percent (IFS).
  - International reserves: generally comparable for most countries; developing countries had 5 out of 39 noncomparable cases. Example: Costa Rica—IFS 20 percent higher than staff report (staff noted bilateral claims under negotiation excluded). Sweden discrepancy due to different sources and valuation practices.
  - External current account balance: 35 percent incidence of differences in entire sample (25 percent for advanced; 41 percent for developing).
    - Chile and Colombia: current account balances differed by 20–30 percent though exports/imports matched closely.
    - Ecuador and Honduras: significant and persistent differences noted.
  - Government balance: high proportion not comparable—31 percent for advanced countries to 41 percent for transition countries—due to unclear definitions of government and differing coverage.
    - Brazil example: IFS consolidated central government surplus 0.2 percent of GDP vs. staff report deficit over 4 percent of GDP under staff definition.
  - Public debt: highest incidence of noncomparable reporting—52 percent for the entire sample (31 percent advanced; 55 percent transition; 59 percent developing). Nonreporting to IFS was a major cause.
  - Bank credit to government: 42 percent direct match/consistency for entire sample; 100 percent not comparable for advanced country sample due to absence of monetary survey in staff reports. For IMDP subsample, bank credit to government match rate was 60 percent; broad money match rate 100 percent.
  - Broad money: 73 percent match/consistency for entire sample. For IMDP subsample, match or consistency was 100 percent.

### Primary reasons for differences and noncomparability
- Revisions captured at different times:
  - Staff reports may include updated information not yet transmitted to STA; IFS may incorporate revisions not reflected in staff reports (example: Uruguay current account revision pattern; Swedish HICP rebasings).
- Differences in coverage or classification:
  - Use of fiscal-year reporting in staff reports vs. calendar-year in IFS; differing definitions of government and public sector coverage (examples: Ecuador, Brazil, Iran, Romania, Thailand).
  - Balance of payments classification differences can shift components (example: Thailand merchandise imports vs. services).
- Different methodologies or staff adjustments:
  - IFS follows international standards (BPM5, 1993 SNA); staff reports may use country-specific definitions or staff adjustments/estimates for analytic purposes (e.g., monetary aggregates per MFSM allowing national definitions; Algeria bank restructuring impact on bank credit to government).
  - Staff reports limited by space—may lack documentation of definitions and data quality.
- Different data sources within Fund and national agencies:
  - Area departments and STA may rely on different government contact points or databases (example: Sweden reserves from Riksbank website vs. STA submissions; Korea GDP from central bank vs. economic planning board).
  - Fund Finance Department data adjustments (e.g., reserve position in the Fund, SDR holdings) can differ from publicly posted central bank datasets.

### Conclusions and recommended measures
- Study conclusions (mirroring a prior STA predecessor study):
  - Noncomparable data and data differences between staff reports and IFS are significant for core indicators.
  - Better documentation of staff reports and STA/area department databases, and improved internal coordination, would likely have reduced incidence and persistence of differences.
  - Area departments have not fully utilized STA’s databases and methodological expertise; STA’s databases have not fully benefited from area department country knowledge.
- Recommended measures:
  - Integrated, country-by-country coordinated data collection arrangements jointly by area departments and STA to improve consistency.
  - Promote use of standard templates for data reporting rooted in accepted statistical methodologies; staff reports should document coverage and definitions of key variables (e.g., in statistical issues appendix) when they diverge from STA publications.
  - In program countries, consider supporting key data-improvement recommendations with performance criteria and benchmarks.
  - Standard templates should cover core Article IV surveillance indicators: nominal GDP, international reserves, essential balance of payments statistics, key monetary aggregates from the monetary survey, and financial soundness and external vulnerability indicators.
  - Encourage reporting of international reserves using the Data Template on International Reserves and Foreign Currency Liquidity to minimize discrepancies related to Fund financial variables (using Finance Department values).
  - Expand Integrated Monetary Data Project (IMDP) where feasible—IMDP benefits observed:
    - reduced reporting burden for country authorities;
    - reduced and better-harmonized databases across area departments and STA;
    - easier identification and reduction of differences in monetary aggregates;
    - improved quality and transparency of monetary statistics;
    - significant staff-time economies for monetary surveillance.
  - Apply common sourcing selectively to other sectors where resource and operational constraints allow, acknowledging challenges where area departments require the most up-to-date data.
  - Where common sourcing is not feasible, ensure better metadata and clear explanations in staff reports when differences with IFS exist.

*Source: Country staff reports, IFS, and related analysis presented in the study summarized in this document.*

### REFERENCES

### _wp0546 - REFERENCES

### References cited

- Berg, Andrew, Eduardo Borensztein, and Catherine Patillo, 2004, “Assessing Early Warning Systems: How Have They Worked in Practice?” IMF Working Paper 04/52 (Washington: International Monetary Fund).

- Carson, Carol S., Sarmad Khawaja, and Thomas K. Morrison, 2004, “Revisions Policy for Official Statistics: A Matter of Governance” IMF Working Paper 04/87 (Washington: International Monetary Fund).

- International Monetary Fund, 2004, International Financial Statistics, Vol. LVII, No. 6, June 2004 (Washington: International Monetary Fund).

*Source: _wp0546 - REFERENCES*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp0546.pdf_