## 1. Descriptive Statistics for GFS Indicators, 2013

## Source details

**Canonical URL:** [1. Descriptive Statistics for GFS Indicators, 2013](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15188.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15188.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15188.pdf.json)

---

### Overview and motivation
- New database of indicators describing the comprehensiveness of government finance statistics (GFS) drawn from Government Finance Statistics Yearbooks, starting with the 2003 Yearbook.
- Database records, for each country-year (2003–13) and for six subsectors of general government, whether the following six items were reported:
  - (i) liabilities
  - (ii) financial assets
  - (iii) nonfinancial assets
  - (iv) statement of the sources and uses of cash
  - (v) statement of government operations
  - (vi) statement of other economic flows
- Database used to construct indices of comprehensiveness in three dimensions—institutions, stocks, and flows—and an overall GFS comprehensiveness index (converted to a score out of 100).

### Database structure and index construction
- Institutional coverage: three institutional definitions are used (budgetary central government, central government, general government). For each item a country receives:
  - 0 if it does not report the item
  - 1 if it reports the item for budgetary central government only
  - 2 if it reports the item for central government but not general government
  - 3 if it reports the item for general government
- Raw maximum score is 18 (all six items for general government); raw scores are converted to a score out of 100.
- Separate indices:
  - Coverage of institutions: score of 3 if either statement of government operations or statement of sources and uses of cash is reported for general government; 2 if reported for central government; 1 if reported only for budgetary central government.
  - Coverage of stocks: count of balance-sheet subtotals (liabilities, financial assets, nonfinancial assets); scores of 1, 2, 3 reflect reporting one, two, or three subtotals respectively.
  - Coverage of flows: count of flow statements (cash, government operations, other economic flows); scores of 1, 2, 3 reflect reporting one, two, or three statements respectively.

### Key statistics and descriptive results (2013)
- GFS Index Score (Out of 100):
  - Mean 23.5
  - Median 11.1
  - Maximum 100.0
  - Minimum 0.0
- Coverage of Institutions (Out of 3):
  - Mean 1.7
  - Median 1.0
  - Maximum 3.0
  - Minimum 0.0
- Coverage of Stocks (Out of 3):
  - Mean 0.8
  - Median 0.0
  - Maximum 3.0
  - Minimum 0.0
- Coverage of Flows (Out of 3):
  - Mean 0.8
  - Median 0.9
  - Maximum 3.0
  - Minimum 0.0
- Sample coverage:
  - 186 countries included in the database
  - 36 countries reported no government finance statistics at all for the 2013 Yearbook

### Correlations with other indices and governance measures
- Internal correlations among GFS components:
  - Correlation between coverage of stocks and coverage of flows: 0.73
  - Coverage of stocks with coverage of institutions: 0.68
  - Coverage of flows with coverage of institutions: 0.79
  - Overall GFS score correlations:
    - 0.79 with institutions
    - 0.94 with stocks
    - 0.86 with flows
- Correlations with external measures:
  - Correlation between overall GFS index (2003–13 average) and Open Budget Index 2012: 0.60
  - Correlation between overall GFS index (2003–13 average) and Weber’s (2012) index: 0.62
  - GFS-based index is positively correlated with a PEFA-derived indicator (Appendix 2)
- Additional relationships:
  - Countries with no freedom-of-information law (93 countries) have average GFS score 9.3
  - Six countries with freedom-of-information laws older than 30 years have average GFS score 60.3
  - GFS index is loosely negatively correlated with resource-revenue dependence (correlation coefficient 0.36)
  - No clear relationship between GFS score and Wehner’s (2006) measure of legislative power

### Trends 2003–13: aggregate changes
- Overall mean GFS index increased from 15.2 (2003) to 23.5 (2013).
- Improvements occurred across all regions and income groups; Europe has the most comprehensive statistics on average.
- By income group: advanced economies > emerging markets > low-income countries, though all groups show gains (low-income gains are slight).

### Coverage of institutions: transitions and patterns
- Mean institutional coverage score increased from 1.25 (2003) to 1.71 (2013).
- Transition summary (2003 → 2013):
  - Improved: 83 countries
  - Declined: 22 countries
  - No change: 81 countries
- 2003→2013 matrix (counts):
  - From No Coverage in 2003:
    - 23 remained No Coverage in 2013
    - 39 reported BCG
    - 4 reported CG
    - 15 reported GG
  - From BCG in 2003:
    - 3 → No Coverage
    - 14 → BCG
    - 0 → CG
    - 5 → GG
  - From CG in 2003:
    - 8 → No Coverage
    - 7 → BCG
    - 4 → CG
    - 20 → GG
  - From GG in 2003:
    - 2 → No Coverage
    - 2 → BCG
    - 0 → CG
    - 40 → GG
- Note: Some backward steps occurred (e.g., a few countries moved from GG to lower coverage or dropped out).

### Coverage of stocks: transitions and patterns
- Mean stocks coverage score increased from 0.52 (2003) to 0.82 (2013).
- Countries reporting full balance sheets over much of the decade include: Australia, Hong Kong SAR, El Salvador, Japan, Norway, Russia, Slovak Republic.
- Transition summary (2003 → 2013):
  - Improved: 42 countries
  - Declined: 24 countries
  - No change: 120 countries
- 2003→2013 matrix (counts):
  - No Coverage in 2013:
    - 87 from No Coverage
    - 24 from Liabilities Only
    - 0 from Financial Balance Sheet
    - 0 from Full Balance Sheet
  - Liabilities Only in 2013:
    - 9 from No Coverage
    - 13 from Liabilities Only
    - 0 from Financial Balance Sheet
    - 0 from Full Balance Sheet
  - Financial Balance Sheet in 2013:
    - 7 from No Coverage
    - 7 from Liabilities Only
    - 13 from Financial Balance Sheet
    - 0 from Full Balance Sheet
  - Full Balance Sheet in 2013:
    - 14 from No Coverage
    - 4 from Liabilities Only
    - 0 from Financial Balance Sheet
    - 7 from Full Balance Sheet
- Remark: No country that produced a financial or full balance sheet in 2003 went backwards to a lower stocks category by 2013.

### Coverage of flows: transitions and patterns
- Mean flows coverage score increased from 0.68 (2003) to 0.97 (2013).
- Most countries report only the statement of the sources and uses of cash; many European countries report only a statement of government operations.
- Six economies reported all three statements in the 2013 Yearbook: Hong Kong SAR, Colombia, El Salvador, Lithuania, Russia, Slovak Republic.
- Transition summary (2003 → 2013):
  - Improved: 70 countries
  - Declined: 21 countries
  - No change: 95 countries
- 2003→2013 matrix (counts):
  - No flow statements in 2013:
    - 23 from No flow statements
    - 12 from One
    - 1 from Two
    - 0 from Three
  - One statement in 2013:
    - 54 from No flow statements
    - 65 from One
    - 5 from Two
    - 2 from Three
  - Two statements in 2013:
    - 3 from No flow statements
    - 10 from One
    - 4 from Two
    - 1 from Three
  - Three statements in 2013:
    - 1 from No flow statements
    - 2 from One
    - 0 from Two
    - 3 from Three

### Notable country-level observations
- Only four countries in 2013 report what the authors define as fully comprehensive government statistics (a full balance sheet and a full set of flow statements for general government):
  - El Salvador
  - Hong Kong SAR
  - Russian Federation
  - Slovak Republic
- Most advanced economies submit at least a financial balance sheet and one flow statement for general government; none submitted a full set of flow statements (cash, accrual, and other economic flows) for any definition of government in 2013.

### Conclusions and research uses
- Indicators document gradual average improvements in comprehensiveness of government finance statistics between 2003 and 2013, with some regressions for individual countries.
- Indicators enable cross-country and time-series analysis of the comprehensiveness of aggregate fiscal statistics across 186 countries from 2003–13.
- Potential research applications include studying causes of fiscal transparency (e.g., democracy, reliance on tax revenues, political competition) and whether comprehensive fiscal statistics are associated with better fiscal outcomes.

### Appendix 1 — Calculating the Indices (methodological details)
- Country coverage:
  - Only countries "that joined the IMF in 2001 or earlier are included in the database." Timor-Leste, Kosovo, Tuvalu, and South Sudan are excluded.
  - Hong Kong and Macao are counted separately from the People’s Republic of China.
  - Serbia and Montenegro treatment:
    - Serbia and Montenegro became separate countries in 2006.
    - Before 2006 they were one country known as “Serbia and Montenegro.”
    - After the split, the Republic of Serbia retained membership; the Republic of Montenegro joined the IMF.
    - Both the Republic of Serbia and the Republic of Montenegro are counted as separate countries for the whole period. Scores for Serbia and Montenegro before 2006 were counted twice—once for the Republic of Serbia and once for the Republic of Montenegro. After 2006 the countries are scored independently.
  - "Currently, there are 188 IMF member countries." Excluding the four post-2001 countries (excluding Montenegro for the reasons stated) and adding Hong Kong and Macao yields "186 countries and territories in the database."
- Timing conventions:
  - Only values from years t – 3, t – 2, and t – 1 are used to assess reporting in the Yearbook of year t.
  - Example: For data collected from the 2005 Yearbook, only numbers reported for 2002, 2003, and 2004 are counted.
  - Preliminary or provisional numbers (denoted with a “p” in the Yearbooks) and forecasted or projected numbers (denoted with an “f”) are assessed the same as actual numbers.
- Coding conventions in Yearbook extracts:
  - "1" = data were reported for this particular sector and statement.
  - "0" = space provided but no data reported.
  - "x" = combination of sector and statement is not applicable to the country.
  - blank cell = country was not included in the GFS Yearbook.
- Example (Austria, 2012 Yearbook):
  - Austria reports three items for general government, each generating a score of 3:
    - Statement of government operations
    - Two subtotals of the summary balance sheet: liabilities and financial assets
  - Other items receive a score of 0 because they were not reported for any of the main definitions of government.
  - Sum of all six scores = three 3s and three 0s = 9 out of 18 → Scaled up to 50 out of 100 (as reported).
  - Austria result: score of 2 for coverage of fiscal stocks; score of 1 for coverage of fiscal flows; score of 3 for coverage of institutions.
- Coverage of stocks:
  - Subtotals coded:
    - 61 = liabilities
    - 62 = financial assets
    - 63 = nonfinancial assets
  - Balance-sheet completeness definitions:
    - Full balance sheet = all three subtotals.
    - Partial balance sheet = two of the three subtotals.
    - One-subtotal balance sheet = only one subtotal.
  - Scoring for coverage of fiscal stocks:
    - "0" = nothing reported.
    - "1" = one of the balance-sheet subtotals reported.
    - "2" = two balance-sheet subtotals reported.
    - "3" = all three subtotals reported.
  - Institutional coverage rule: a country is counted as reporting liabilities if it reports them for at least one of budgetary central government, central government, and general government. Balance sheets reported only for local government, state government, or extrabudgetary funds are not considered as reporting a balance sheet.
- Coverage of flows:
  - The three statements:
    - Government operations
    - Other economic flows
    - Sources and uses of cash
  - Scoring for coverage of flows:
    - "0" = none reported.
    - "1" = one statement reported.
    - "2" = two statements reported.
    - "3" = all three statements reported.
  - Institutional coverage rule: Statements for budgetary central government, central government, and general government are counted.
- Coverage of institutions:
  - Scoring for coverage of institutions:
    - "0" = no information reported.
    - "1" = information for budgetary central government only.
    - "2" = information for central government but not general government.
    - "3" = information for general government.
  - Permissiveness: reporting either the statement of government operations or the statement of sources and uses of cash for general government counts as reporting for general government.
- Alternatives and methodological choices:
  - Authors prefer a multiplicative approach to overall index construction (gives credit for comprehensive measures of stocks and flows for comprehensive definition of government) rather than simply summing the three indicators.
  - Discussed but not adopted alternatives:
    - Weighting accrual-based statement of government operations higher than cash-flow statements (not adopted).
    - Ordinal scoring giving higher weight to nonfinancial assets over financial assets over liabilities (not adopted).
    - Inclusion of local government, state government, and extrabudgetary institutions (excluded because not all countries have them).
  - Authors note researchers can use the database to construct alternative indices.

### Appendix 2 — Deriving an Indicator of Fiscal Transparency from PEFA Reports
- PEFA data coverage:
  - "One hundred and fifty three Public Expenditure and Financial Accountability (PEFA) reports have been done for 91 mainly developing or emerging economies."
  - Report frequency in dataset:
    - 42 countries: single report
    - 37 countries: two reports
    - 11 countries: three reports
    - 1 country: four reports
  - Some reports are self-assessments.
- PEFA framework:
  - PI-10 is the only indicator explicitly measuring public availability of fiscal information; PI-10 has four components (i)–(iv) indicating availability of:
    - (i) the budget
    - (ii) in-year budget-execution reports
    - (iii) year-end financial statements
    - (iv) an audit report
- Complementary PEFA indicators used to assess quality/comprehensiveness:
  - PI-10(i) × average(PI-1, PI-5, PI-6)
    - PI-1: Aggregate expenditure out-turn compared to original approved budget
    - PI-5: Classification of the budget
    - PI-6: Comprehensiveness of information included in budget documentation
  - PI-10(ii) × PI-24
    - PI-24: Quality and timeliness of in-year budget reports
  - PI-10(iii) × PI-25
    - PI-25: Quality and timeliness of annual financial statements
  - PI-10(iv) × PI-26
    - PI-26: Scope, nature, and follow-up of external audit
- Scoring rules and transformations:
  - PI-10 components (i, ii, iii, iv) converted to binary: 1 if publicly available, 0 if not.
  - Letter grades converted to numeric scores:
    - "A" = 4.0
    - "B+" = 3.5
    - "B" = 3.0
    - "C+" = 2.5
    - ... (continuation implied)
    - "D" = 1.0 (lowest non-zero score)
    - All "NA", "NR", or "NU" scores = 0
  - Quality of publicly available fiscal information of each type = PI-10(component) × (relevant PI score or average of relevant PI scores).
  - Total PEFA-based index aggregates these elements (Figure A1 note: "Maximum Score is 16.").
- Empirical highlights and descriptive statistics:
  - Figure A1 top scorers (shown on chart): South Africa, Peru, Serbia, Colombia, Indonesia, Brazil, Trinidad & Tobago, Uganda, Vanuatu, Armenia, Georgia, Norway.
  - Table A6. Descriptive Statistics for PEFA Indicators:
    - Total Score (Out of 100):
      - Mean 34.7
      - Median 34.4
      - Maximum 87.5
      - Minimum 0.0
    - Annual Budget (Out of 4):
      - Mean 1.9
      - Median 2.3
      - Maximum 4.0
      - Minimum 0.0
    - In-Year Execution Reports (Out of 4):
      - Mean 1.4
      - Median 1.5
      - Maximum 4.0
      - Minimum 0.0
    - Year-End Financial Statements (Out of 4):
      - Mean 1.2
      - Median 0.0
      - Maximum 4.0
      - Minimum 0.0
    - External Audit Reports (Out of 4):
      - Mean 1.3
      - Median 1.5
      - Maximum 3.5
      - Minimum 0.0

*Source: _wp15188 - 1. Descriptive Statistics for GFS Indicators, 2013 (PDF chapter/section).*

### 1. Descriptive Statistics for GFS Indicators, 2013 ...................................................................13

### 1. Descriptive Statistics for GFS Indicators, 2013 ...................................................................13

### Major sections listed in this content unit
- 1. Descriptive Statistics for GFS Indicators, 2013 ...................................................................13
- 2. Transitions in Coverage of Institutions, 2003–13 ................................................................20
- 3. Transitions in Coverage of Stocks, 2003–13 .......................................................................22
- 4. Transitions in Coverage of Flows, 2003–13 ........................................................................24

### Figures referenced in this content unit
- 1. General Government and its Subsectors ................................................................................8
- 2. Average Comprehensiveness of GFS, 2011–13: Top Quartile ............................................10
- 3. Relationships among Measures of Comprehensiveness of GFS ..........................................14
- 4. Relationships among Measures of Fiscal Transparency ......................................................15
- 5. Fiscal Transparency and Governance ..................................................................................16
- 6. Fiscal Transparency and Natural Resources ........................................................................17
- 7. Average Comprehensiveness of GFS, 2003–13 ..................................................................18
- 8. Average Comprehensiveness of the Coverage of Institutions, 2003–13 .............................19
- 9. Average Comprehensiveness of the Coverage of Stocks, 2003–13 ....................................21
- 10. Average Comprehensiveness of the Coverage of Flows, 2003–13 ...................................23

### Appendices associated with this content unit
- Appendix 1. Calculating the Indices .........................................................................................................26
- Appendix 2. Deriving an Indicator of Fiscal Transparency from PEFA Reports ....................................33

*Source: _wp15188 - 1. Descriptive Statistics for GFS Indicators, 2013 (PDF chapter/section).*

### APPENDIX TABLES

### _wp15188 - APPENDIX TABLES

### Overview and motivation
- The paper presents a new database of indicators describing the comprehensiveness of government finance statistics (GFS) drawn from Government Finance Statistics Yearbooks, starting with the 2003 Yearbook.
- The database records, for each country-year (2003–13) and for six subsectors of general government, whether the following six items were reported: (i) liabilities, (ii) financial assets, (iii) nonfinancial assets, (iv) statement of the sources and uses of cash, (v) statement of government operations, and (vi) statement of other economic flows.
- The database is used to construct indices of comprehensiveness in three dimensions—institutions, stocks, and flows—and an overall GFS comprehensiveness index (converted to a score out of 100).

### Database structure and index construction
- Institutional coverage: three institutional definitions are used (budgetary central government, central government, general government). For each item a country receives:
  - 0 if it does not report the item
  - 1 if it reports the item for budgetary central government only
  - 2 if it reports the item for central government but not general government
  - 3 if it reports the item for general government
- Raw maximum score is 18 (all six items for general government); raw scores are converted to a score out of 100.
- Separate indices:
  - Coverage of institutions: score of 3 if either statement of government operations or statement of sources and uses of cash is reported for general government; 2 if reported for central government; 1 if reported only for budgetary central government.
  - Coverage of stocks: count of balance-sheet subtotals (liabilities, financial assets, nonfinancial assets); scores of 1, 2, 3 reflect reporting one, two, or three subtotals respectively.
  - Coverage of flows: count of flow statements (cash, government operations, other economic flows); scores of 1, 2, 3 reflect reporting one, two, or three statements respectively.

### Key statistics and descriptive results (2013)
- GFS Index Score (Out of 100): Mean 23.5; Median 11.1; Maximum 100.0; Minimum 0.0
- Coverage of Institutions (Out of 3): Mean 1.7; Median 1.0; Maximum 3.0; Minimum 0.0
- Coverage of Stocks (Out of 3): Mean 0.8; Median 0.0; Maximum 3.0; Minimum 0.0
- Coverage of Flows (Out of 3): Mean 0.8; Median 0.9; Maximum 3.0; Minimum 0.0
- Sample coverage: 186 countries included in the database; 36 countries reported no government finance statistics at all for the 2013 Yearbook.

### Correlations with other indices and governance measures
- Internal correlations among GFS components:
  - Correlation between coverage of stocks and coverage of flows: 0.73
  - Coverage of stocks with coverage of institutions: 0.68
  - Coverage of flows with coverage of institutions: 0.79
  - Overall GFS score correlations: 0.79 (institutions), 0.94 (stocks), 0.86 (flows)
- Correlations with external measures:
  - Correlation between overall GFS index (2003–13 average) and Open Budget Index 2012: 0.60
  - Correlation between overall GFS index (2003–13 average) and Weber’s (2012) index: 0.62
  - GFS-based index is positively correlated with a PEFA-derived indicator (details in Appendix 2).
- Additional relationships:
  - Countries with no freedom-of-information law (93 countries) have average GFS score 9.3; six countries with freedom-of-information laws older than 30 years have average GFS score 60.3.
  - GFS index is loosely negatively correlated with resource-revenue dependence (correlation coefficient 0.36).
  - No clear relationship between GFS score and Wehner’s (2006) measure of legislative power.

### Trends 2003–13: aggregate changes
- Overall mean GFS index increased from 15.2 (2003) to 23.5 (2013).
- Improvements occurred across all regions and income groups; Europe has the most comprehensive statistics on average.
- By income group: advanced economies > emerging markets > low-income countries, though all groups show gains (low-income gains are slight).

### Coverage of institutions: transitions and patterns
- Mean institutional coverage score increased from 1.25 (2003) to 1.71 (2013).
- Transition summary (2003 → 2013):
  - Improved: 83 countries
  - Declined: 22 countries
  - No change: 81 countries
- 2003→2013 matrix (counts):
  - From No Coverage in 2003: 23 remained No Coverage in 2013; 39 reported BCG; 4 reported CG; 15 reported GG
  - From BCG in 2003: 3 → No Coverage; 14 → BCG; 0 → CG; 5 → GG
  - From CG in 2003: 8 → No Coverage; 7 → BCG; 4 → CG; 20 → GG
  - From GG in 2003: 2 → No Coverage; 2 → BCG; 0 → CG; 40 → GG
- Note: Some backward steps occurred (e.g., a few countries moved from GG to lower coverage or dropped out).

### Coverage of stocks: transitions and patterns
- Mean stocks coverage score increased from 0.52 (2003) to 0.82 (2013).
- Countries reporting full balance sheets over much of the decade include: Australia, Hong Kong SAR, El Salvador, Japan, Norway, Russia, Slovak Republic.
- Transition summary (2003 → 2013):
  - Improved: 42 countries
  - Declined: 24 countries
  - No change: 120 countries
- 2003→2013 matrix (counts):
  - No Coverage in 2013: 87 from No Coverage; 24 from Liabilities Only; 0 from Financial Balance Sheet; 0 from Full Balance Sheet
  - Liabilities Only in 2013: 9 from No Coverage; 13 from Liabilities Only; 0 from Financial Balance Sheet; 0 from Full Balance Sheet
  - Financial Balance Sheet in 2013: 7 from No Coverage; 7 from Liabilities Only; 13 from Financial Balance Sheet; 0 from Full Balance Sheet
  - Full Balance Sheet in 2013: 14 from No Coverage; 4 from Liabilities Only; 0 from Financial Balance Sheet; 7 from Full Balance Sheet
- Remark: No country that produced a financial or full balance sheet in 2003 went backwards to a lower stocks category by 2013.

### Coverage of flows: transitions and patterns
- Mean flows coverage score increased from 0.68 (2003) to 0.97 (2013).
- Most countries report only the statement of the sources and uses of cash; many European countries report only a statement of government operations.
- Six economies reported all three statements in the 2013 Yearbook: Hong Kong SAR, Colombia, El Salvador, Lithuania, Russia, Slovak Republic.
- Transition summary (2003 → 2013):
  - Improved: 70 countries
  - Declined: 21 countries
  - No change: 95 countries
- 2003→2013 matrix (counts):
  - No flow statements in 2013: 23 from No flow statements; 12 from One; 1 from Two; 0 from Three
  - One statement in 2013: 54 from No flow statements; 65 from One; 5 from Two; 2 from Three
  - Two statements in 2013: 3 from No flow statements; 10 from One; 4 from Two; 1 from Three
  - Three statements in 2013: 1 from No flow statements; 2 from One; 0 from Two; 3 from Three

### Notable country-level observations
- Only four countries in 2013 report what the authors define as fully comprehensive government statistics (a full balance sheet and a full set of flow statements for general government): El Salvador, Hong Kong SAR, Russian Federation, Slovak Republic.
- Most advanced economies submit at least a financial balance sheet and one flow statement for general government; none submitted a full set of flow statements (cash, accrual, and other economic flows) for any definition of government in 2013.

### Conclusions and research uses
- The new GFS-based indicators document gradual average improvements in comprehensiveness of government finance statistics between 2003 and 2013, with some regressions for individual countries.
- The indicators enable cross-country and time-series analysis of the comprehensiveness of aggregate fiscal statistics across 186 countries from 2003–13.
- Potential research applications include studying causes of fiscal transparency (e.g., democracy, reliance on tax revenues, political competition) and whether comprehensive fiscal statistics are associated with better fiscal outcomes.

*Source: APPENDIX TABLES, _wp15188 - APPENDIX TABLES (Government Finance Statistics Yearbooks 2003–13 as described in the content unit).*

### Appendix 1. Calculating the Indices

### Appendix 1. Calculating the Indices

### Country coverage
- Only countries "that joined the IMF in 2001 or earlier are included in the database." Timor-Leste, Kosovo, Tuvalu, and South Sudan are excluded.
- Practice for China SARs: Hong Kong and Macao are counted separately from the People’s Republic of China (follows GFS Yearbooks).
- Serbia and Montenegro treatment:
  - Serbia and Montenegro became separate countries in 2006.
  - Before 2006 they were one country known as “Serbia and Montenegro.”
  - After the split, the Republic of Serbia retained membership; the Republic of Montenegro joined the IMF.
  - Both the Republic of Serbia and the Republic of Montenegro are counted as separate countries for the whole period. Scores for Serbia and Montenegro before 2006 were counted twice—once for the Republic of Serbia and once for the Republic of Montenegro. After 2006 the countries are scored independently.
- Membership counts:
  - "Currently, there are 188 IMF member countries."
  - Excluding the four post-2001 countries (excluding Montenegro for the reasons stated) and adding Hong Kong and Macao yields "186 countries and territories in the database."

### Timing conventions
- Only values from years t – 3, t – 2, and t – 1 are used to assess reporting in the Yearbook of year t (to ensure relatively recent data).
  - Example: For data collected from the 2005 Yearbook, only numbers reported for 2002, 2003, and 2004 are counted.
  - Example consequence: Belize is marked as “no assets or liabilities reported” for the 2004 Yearbook because the most recent data were for the year 1996.
- Preliminary or provisional numbers (denoted with a “p” in the Yearbooks) and forecasted or projected numbers (denoted with an “f”) are assessed the same as actual numbers.

### Calculating Index Scores: An Example (Austria, 2012 Yearbook)
- Scope:
  - Only the four summary tables reported by Austria in the 2012 Yearbook are examined: statement of government operations, statement of other economic flows, balance sheet, and statement of sources and uses of cash.
  - Yearbook allows only four sets of institutions in this example: central government, state government, local government, and general government (no room for budgetary central government or extrabudgetary agencies in this Yearbook extract).
- Database coding for each sector × statement:
  - "1" = data were reported for this particular sector and statement.
  - "0" = space provided but no data reported.
  - "x" = combination of sector and statement is not applicable to the country.
  - blank cell = country was not included in the GFS Yearbook (one possible reason: country did not report any data to the GFS system in that year).

### Overall GFS-based index (Austria example)
- Austria reports three items for general government, each generating a score of 3:
  - Statement of government operations
  - Two subtotals of the summary balance sheet: liabilities and financial assets
- Other items receive a score of 0 because they were not reported for any of the main definitions of government (general government, central government, budgetary central government).
- Final overall score calculation:
  - Sum of all six scores = three 3s and three 0s = 9 out of 18
  - Scaled up to 50 out of 100 (as reported).

### Coverage of stocks
- Subtotals in summary balance sheet coded as numbers:
  - 61 = liabilities
  - 62 = financial assets
  - 63 = nonfinancial assets
- Balance-sheet completeness definitions:
  - Full balance sheet = all three subtotals.
  - Partial balance sheet = two of the three subtotals (most often liabilities and financial assets).
  - One-subtotal balance sheet = only one subtotal (almost always liabilities).
- Notable exceptions:
  - Kazakhstan and Mongolia reported nonfinancial assets and liabilities in the 2004 and 2011 Yearbooks, respectively.
  - France reported nonfinancial assets only in the 2003 and 2004 Yearbooks.
  - Jordan reported nonfinancial assets only in the 2005 Yearbook.
- Scoring for coverage of fiscal stocks (overall rating from number of subtotals reported):
  - "0" = nothing reported.
  - "1" = one of the balance-sheet subtotals reported.
  - "2" = two balance-sheet subtotals reported.
  - "3" = all three subtotals reported.
- Institutional coverage rule:
  - A country is counted as reporting liabilities if it reports them for at least one of budgetary central government, central government, and general government.
  - Balance sheets reported only for local government, state government, or extrabudgetary funds are not considered as reporting a balance sheet.
- Austria result: score of 2 for coverage of fiscal stocks.

### Coverage of flows
- The three statements indicating comprehensiveness of fiscal flows:
  - Government operations
  - Other economic flows
  - Sources and uses of cash
- Scoring for coverage of flows (overall rating from number of statements reported):
  - "0" = none reported.
  - "1" = one statement reported.
  - "2" = two statements reported.
  - "3" = all three statements reported.
- Institutional coverage rule:
  - Statements for budgetary central government, central government, and general government are counted.
- Austria result: score of 1 for coverage of fiscal flows (based on Table A4 data).

### Coverage of institutions
- Comprehensiveness indicated by the broadest sector reported among budgetary central government, central government, and general government.
- Scoring for coverage of institutions (overall rating from government sectors reported):
  - "0" = no information reported.
  - "1" = information for budgetary central government only (not central government or general government).
  - "2" = information for central government but not general government.
  - "3" = information for general government.
- Note on permissiveness:
  - For institutions, the methodology is not strict about which financial statements are reported; e.g., reporting either the statement of government operations or the statement of sources and uses of cash for general government counts as reporting for general government.
- Austria result: score of 3 for coverage of institutions.

### Alternatives and methodological choices
- Overall index construction:
  - Authors prefer a multiplicative approach (gives credit for comprehensive measures of stocks and flows for comprehensive definition of government) rather than simply summing the three indicators for institutions, stocks, and flows.
  - Rationale: additive approach would not reward countries that combine broad institutional coverage with comprehensive stock and flow reporting more than countries with narrow definitions.
- Possible alternative scorings discussed (not adopted):
  - Flows: weighting accrual-based statement of government operations higher than cash-flow statements (authors did not adopt to avoid controversial judgments).
  - Stocks: ordinal scoring giving higher weight to nonfinancial assets over financial assets over liabilities (authors did not adopt; would have made little difference in practice).
  - Institutions: authors choose an ordinal ranking (budgetary central government < central government < general government) and exclude extrabudgetary institutions, local government, and state government because not all countries have them.
- Authors note researchers can use the database to construct alternative indices.

---

### Appendix 2. Deriving an Indicator of Fiscal Transparency from PEFA Reports

### PEFA data coverage and reports
- "One hundred and fifty three Public Expenditure and Financial Accountability (PEFA) reports have been done for 91 mainly developing or emerging economies."
- Report frequency in dataset:
  - 42 countries: single report
  - 37 countries: two reports
  - 11 countries: three reports
  - 1 country: four reports
- Some reports are self-assessments.
- PEFA framework: PI-10 is the only indicator explicitly measuring public availability of fiscal information; PI-10 has four components (i)–(iv) indicating availability of:
  - (i) the budget,
  - (ii) in-year budget-execution reports,
  - (iii) year-end financial statements,
  - (iv) an audit report.

### Complementary PEFA indicators used
- To assess quality/comprehensiveness of publicly available documentation, PI-10 components are paired with other PI indicators:
  - PI-10(i) × average(PI-1, PI-5, PI-6)
    - PI-1: Aggregate expenditure out-turn compared to original approved budget
    - PI-5: Classification of the budget
    - PI-6: Comprehensiveness of information included in budget documentation
  - PI-10(ii) × PI-24
    - PI-24: Quality and timeliness of in-year budget reports
  - PI-10(iii) × PI-25
    - PI-25: Quality and timeliness of annual financial statements
  - PI-10(iv) × PI-26
    - PI-26: Scope, nature, and follow-up of external audit

### Scoring rules and transformations
- PI-10 components (i, ii, iii, iv) are converted to binary: 1 if the documentation is publicly available, 0 if not.
- Letter grades in PEFA reports are converted to numeric scores:
  - "A" = 4.0
  - "B+" = 3.5
  - "B" = 3.0
  - "C+" = 2.5
  - ... (continuation implied by "and so on")
  - Lowest non-zero score: "D" = 1.0
  - All "NA", "NR", or "NU" scores = 0
- Quality of publicly available fiscal information of each type = PI-10(component) × (relevant PI score or average of relevant PI scores).
- The total PEFA-based index aggregates these elements (note: Figure A1 "Note: Maximum Score is 16.").

### Empirical highlights and descriptive statistics
- Figure A1 top scorers (scores shown on chart):
  - "South Africa" (highest)
  - "Peru"
  - "Serbia"
  - "Colombia"
  - "Indonesia"
  - "Brazil"
  - "Trinidad & Tobago"
  - "Uganda"
  - "Vanuatu"
  - "Armenia"
  - "Georgia"
  - "Norway"
  - Note on figure: "Maximum Score is 16."
- Table A6. Descriptive Statistics for PEFA Indicators (preserve exact reported numbers):
  - Total Score (Out of 100): Mean 34.7, Median 34.4, Maximum 87.5, Minimum 0.0
  - Annual Budget (Out of 4): Mean 1.9, Median 2.3, Maximum 4.0, Minimum 0.0
  - In-Year Execution Reports (Out of 4): Mean 1.4, Median 1.5, Maximum 4.0, Minimum 0.0
  - Year-End Financial Statements (Out of 4): Mean 1.2, Median 0.0, Maximum 4.0, Minimum 0.0
  - External Audit Reports (Out of 4): Mean 1.3, Median 1.5, Maximum 3.5, Minimum 0.0

*Source: Appendix 1 and Appendix 2, "_wp15188 - Appendix 1. Calculating the Indices" (IMF content provided).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15188.pdf_