## ANNEX I–III. FINANCIAL INTERCONNECTEDNESS; SELECTED APPROACHES TO MODIFYING THE MEASURE OF VARIABILITY; FINANCIAL CONTRIBUTIONS

## Source details

**Canonical URL:** [ANNEX I–III. FINANCIAL INTERCONNECTEDNESS; SELECTED APPROACHES TO MODIFYING THE MEASURE OF VARIABILITY; FINANCIAL CONTRIBUTIONS](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/np/pp/eng/2012/_062912.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/np/pp/eng/2012/_062912.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/np/pp/eng/2012/_062912.pdf.json)

---

### ANNEX I. FINANCIAL INTERCONNECTEDNESS — A. Data and Methodology
- Data source and coverage:
  - Utilizes preliminary end-2010 data on equity and debt portfolio assets from the Fund’s Coordinated Portfolio Investment Survey (CPIS).
  - CPIS data treat all countries on a residence basis and follow BPM 6 definitions and classifications.
  - Overall, 65 Fund members (of which 25 are advanced economies), accounting for about 98 percent of the estimated global International Investment Position (IIP), provide cross-border data on their portfolio investment holdings vis-à-vis 178 Fund members.
  - Positions between non-reporting countries and those below USD ½ million are excluded.
  - For China, separately reported data for China’s Mainland, Hong Kong SAR and Macao SAR were aggregated excluding the holdings amongst them.
- Construction of the interconnectedness measure:
  - A global matrix of bilateral portfolio asset positions is constructed; each entry X_ij reports the nominal value (in US Dollars) of country i’s assets located in country j.
  - All diagonal elements X_ii are zero; off-diagonal elements are positive or zero.
  - The matrix is expressed in portfolio shares by scaling each country’s assets in other countries by its total external assets.
  - The interconnectedness indicator for a country is computed as the sum of the interconnectedness indicators of countries investing in it, weighted by their respective portfolio shares (an eigenvector-like measure).
  - Methodological references include Bonacich and Lloyd (2001) and applications in the financial sector such as Von Peter (2007) and Kubelec and Sá (2010).

### ANNEX I. FINANCIAL INTERCONNECTEDNESS — B. Results
- Index properties and distribution:
  - The calculated interconnectedness index ranges from 1 to 15.
  - The United States appears to be the most interconnected with an index of almost 15.
  - Most countries (173 members) have an index value between 1 and 2.
  - There are only five countries with index values between 4 and 8: United Kingdom, Luxembourg, Germany, France, and Netherlands.
- Key portfolio-share statistics explaining dispersion:
  - By far the largest share (20 percent) of total global portfolio assets is invested in the United States.
  - About ten percent is invested in the United Kingdom.
  - Investments located in Luxembourg, Germany, and France each account for around seven percent of total portfolio assets.
- Relationship with IIP and ranking comparison:
  - The interconnectedness index provides country rankings that are broadly similar to rankings according to IIP shares, with the correlation between the two of 0.71.
  - Differences in rankings reflect differing methodologies and underlying data; interconnectedness captures the pattern of bilateral linkages beyond IIP.
- Limitations and considerations for quota formula inclusion:
  - The indicator does not capture all size-related dimensions important for quota variables.
  - The largest country based on this measure (the United States) is only 15 times larger than the smallest member.
  - The index provides almost no differentiation for the vast majority of countries (index between 1 and 2).
  - Treatment of major international financial centers would require additional addressing if the indicator were to be used in the quota formula.
- Supplementary materials referenced (not included here):
  - Figure A1.1: Histogram of Interconnectedness Index (source: Finance Department).
  - Table A1.1: Countries’ Rankings according to Financial Openness and Interconnectedness (top 30 countries ranked by gap-filled IIP shares in 2010 and corresponding interconnectedness rankings); calculated using bilateral portfolio investment data from CPIS; includes China, P.R., Hong Kong SAR, and Macao SAR.

*Source: Finance Department*

### ANNEX II. SELECTED APPROACHES TO MODIFYING THE MEASURE OF VARIABILITY — Alternative variability measures
- Purpose:
  - Address shortcomings of the current variability measure in the quota formula, notably (i) lack of correlation with an indicator for approval of a Fund arrangement, and (ii) significant year-to-year instability in calculated quota shares.
- Measures examined:
  - 13Y AAD: Average absolute deviation from a three-year centered moving average calculated over a recent 13-year period. Differs from the current measure by using average absolute deviation (reduces impact of extreme observations) instead of root mean squared deviation.
  - 5-year statistical dispersion measures (relative to sample mean/median to limit trend influence):
    - 5Y SD (Standard deviation)
    - 5Y AAD (Average absolute deviation)
    - 5Y MAD (Median absolute deviation)
    - 5Y MAXD (Maximum deviation from the mean)
  - 10Y II (Instability index): variability measured relative to a linear trend. Calculated over a 10-year period; b denotes the slope estimated by OLS.

### ANNEX II. Stability of alternative measures (comparative findings)
- Empirical findings (cross-sectional standard deviation of year-on-year changes in variability shares):
  - 13Y AAD and 10Y II yield the smallest variation in shares in 2009-2010.
  - For 2008-2009, 5Y SD (relative to the sample mean) performs better.
  - Over a longer time span (1990-2010), the 13-year average absolute deviation produces the smallest variation in shares on average.
  - There are periods (e.g., 2004-2007) where the existing variability measure yields more stable results than the alternatives.
- Overall conclusion:
  - None of the alternative measures consistently outperforms the current measure in terms of stability properties.

### ANNEX II. Association with potential financing needs (correlation with Fund arrangements)
- Staff analysis:
  - Alternative variability measures remain uncorrelated with a binary variable for approval of a Fund arrangement; correlations are not significantly different from zero.
  - All variability measures based on current receipts and net capital flows are highly correlated among themselves (even after adjustment for economic size), though outcomes for individual members can differ significantly.

### ANNEX II. Macroeconomic variables signaling vulnerability (determinants of use of Fund resources)
- Candidate variables (three-year averages used for illustrative composite construction):
  - Reserve cover ratio (reserves to imports)
  - Current account (percent of GDP)
  - Per capita GDP
  - Real GDP growth
  - Inflation
  - External debt service (debt service/exports)
  - Government balance (percent of GDP)
  - Stock of external debt (percent of GDP)
- Empirical differences between countries with programs and without programs (year of approval and two preceding years):
  - Current account deficit as percent of GDP: about twice as high for program countries compared to non-program countries (valid for contemporaneous and lagged values).
  - Reserve cover ratio: program countries have lower reserve cover ratios; difference in the year preceding arrangement approval is about 8 percentage points.
  - GDP per capita: countries with lower per capita income are more likely to have Fund arrangements.
  - GDP growth: countries with programs have recorded lower GDP growth on average (robust with lags).
  - Inflation: typically higher in countries with IMF programs.
  - Government budget deficits: higher by 0.9-1.4 percent of GDP on average in countries with programs versus without; peak difference in the year preceding approval.
  - Stock of external debt (percent of GDP): program countries average 77 percent in the year of approval versus 68 percent for countries without programs.
  - External debt service (share of exports): program countries allocate a greater share of exports to debt service payments.

### ANNEX II. Composite indicator construction and properties
- Rationale:
  - No single macro variable robustly predicts use of Fund resources; combining indicators into a composite variable can capture multiple vulnerability types and typically yields higher explanatory/predictive power.
- Weighting approach (illustrative):
  - Weights determined based on the inverse of the standard deviation of each component (standardization prevents domination by a single component).
- Composite variants considered:
  - Comp1: linear combination of all variables except inflation (inflation shows little correlation with likelihood of a Fund arrangement).
  - Comp2: excludes external debt and external debt service.
  - Comp3: excludes external debt, external debt service, and fiscal balance (allows calculation for the largest number of countries; used for illustrative mapping and results).
- Correlation results:
  - Correlations of composite indicators with the binary variable for approval of a Fund arrangement are statistically significant but modest.
  - Example: Comp3 correlation for the full sample is 0.19.
  - Dropping LICs increases Comp3 correlation to 0.24 (for emerging markets and advanced economies).
  - For LICs only, correlations are generally much weaker—between 0.05-0.07.
  - Robustness: correlation coefficients remain highly significant when calculated over 5-year intervals since 1990.
- Transformation to a quota-compatible variable:
  - Composite indicator scores can be negative and are not size-scaled, whereas quota formula variables are expressed as shares of global totals and include significant size components.
  - Proposed mapping: transform x_i (vulnerability score for country i) using a mapping (denoted mapping (1) in the text) that stacks all individual scores in the (0,2) interval.
  - After transformation, multiply transformed vulnerability scores by members’ shares in GDP to impart economic size; rescale and subtract GDP share to adjust for size effect (illustrative approach).
  - Impact on correlations:
    - Comp3 correlation falls from 0.19 to 0.16 after applying the non-linear transformation (1).
    - Correlation falls further to 0.09 after multiplying transformed score by GDP shares, rescaling, and then subtracting the GDP share to adjust for size.

### ANNEX II. Illustrative country and group impacts (selected figures)
- United States:
  - GDP shares: 2009 = 24.5; 2010 = 23.5
  - Existing variability shares: 2009 = 14.1; 2010 = 15.5
  - Composite variability shares (transformed Comp3): 2009 = 27.1; 2010 = 25.4
  - Difference (Composite − Existing): 2009 = 13.0; 2010 = 9.9
  - Note: relative gain driven by size effect and low reserve cover ratio for the US (reserve currency caveat noted).
- China (including P.R., Hong Kong SAR, Macao SAR):
  - Existing variability (2009, 2010): 5.6, 5.2
- Italy and Spain:
  - Composite variability shares are significantly higher than GDP shares; for Greece and Portugal (not shown in the table) composite variability shares based on 2010 data are more than twice their GDP shares.
- Aggregate group comparisons (selected entries):
  - Advanced economies: Existing variability shares (Table A2.1) = 57.9 (Current Var.), 59.0 (13Y AAD), 52.0 (5Y SD), 54.7 (5Y AAD), 58.2 (5Y MAD), 50.2 (5Y MAXD), 57.6 (10Y II).
  - Emerging Market and Developing Countries (Table A2.1): Current Var. = 42.1; 13Y AAD = 41.0; 5Y SD = 48.0; 5Y AAD = 45.3; 5Y MAD = 41.8; 5Y MAXD = 49.8; 10Y II = 42.4.
  - Advanced economies GDP shares (Table A2.2): 2009 GDP = 66.4; 2010 GDP = 64.2; Existing variability = 58.6 (2009), 57.9 (2010); Composite variability = 67.9 (2009), 67.9 (2010); Difference = 11.0 (2009), 10.0 (2010).

### ANNEX II. Overall assessment and challenges
- Composite vulnerability indicators can improve explanatory power relative to single variables, but raise challenges:
  - Choice of variables and weights influences results and may reduce transparency and simplicity.
  - Non-linear transformations required to make composite scores compatible with quota-share scaling can reduce correlation with the arrangement indicator.
  - Data availability constraints (especially for fiscal and external debt series) limit coverage; Comp3 chosen in illustrations because it covers the largest number of countries (175 members).
  - Uncertainty about whether composite indicators that explain past use of Fund resources will be robust predictors of future use.
- Conclusion:
  - Translating macroeconomic vulnerability information into a single quota-formula-compatible variability measure poses major methodological and practical challenges; no alternative examined consistently dominates the current measure across stability and explanatory criteria.

*Source: Finance Department (ANNEX II. SELECTED APPROACHES TO MODIFYING THE MEASURE OF VARIABILITY).*

### ANNEX III. FINANCIAL CONTRIBUTIONS — Pre-NAB Commitments and NAB
- Pre-NAB Commitments:
  - Covers loan commitments by members under the 1974 and 1975 Oil Facilities, the Supplementary Financing Facility (SFF), the Enlarged Access Facility (EAF), and the expanded General Agreements to Borrow (GAB).
  - Includes bilateral borrowing agreements with Italy (1966), Switzerland (1977), and Japan (1986).
- NAB:
  - Data cover the commitments by members under NAB credit arrangements and reflecting the rollback agreed by the Executive Board.
  - Commitments by those members that have not yet adhered to the NAB decision are not included.
  - For members with bilateral borrowing arrangements being folded into the NAB, only the NAB commitment is included.
  - Bilateral loan commitments from members that are not participants in the NAB (Czech Republic, Malta, Slovak Republic, and Slovenia) have been included.

### ANNEX III. NAB + New Pledges; PRGT Loans and Subsidies
- NAB + New Pledges:
  - In addition to the NAB amounts, data reflect bilateral pledges under the current fund raising exercise, including the announcements made at the G-20 Leaders’ Summit in Los Cabos.
- PRGT Loans:
  - Reflects all loan commitments to the PRGT Trust as of December 31, 2011.
  - Includes Saudi Arabia’s pledge of SDR 11 million as communicated to the Fund in a letter received by the Fund on June 5, 2012.
- PRGT Subsidies:
  - Includes total bilateral resources provided since 1987 for subsidizing concessional lending, HIPC and MDRI debt relief as of December 31, 2011 plus all pledges made under current fundraising as of December 31, 2011.

### ANNEX III. Technical Assistance and FTP Participation metrics
- Technical Assistance:
  - Reflects actual cash disbursements to the Fund for technical assistance and training excluding in kind contributions over the period FY1999-FY2012.
  - These data do not include pledges.
- FTP Participation:
  - Two metrics developed to account for FTP participation:
    - Duration-based metric:
      - Derived from the number of quarters out of the 80 quarters covering 1992–2011 that a member was included in the Financial Transactions Plan (FTP).
      - A member’s share = total number of quarters a member participated in the FTP over the 20-year period January 1992–December 2011 (i.e., 80 quarters) as a share of total participant quarters aggregated over the 80 quarters (these sum to 3,197).
      - The maximum share for a member that participated in all 80 quarters is 2.5 = (80/3197)*100.
      - Figure A3.1 shows participation in the FTP during the period in question.
    - Resources-based metric:
      - Obtained by weighting each member’s participation in the FTP with its nominal quota.
      - A member’s share in the resources-based FTP participation metric is a proxy for the share of resources a member has made available over time in the FTP.
  - Notes:
    - The period 1992-2011 includes the move from the Operational Budget to the FTP framework in mid-2000 as well as the period of expanding membership in the early 1990s which was reflected in FTP participation.
    - For quarters prior to June-August, 2000, the Operational Budget participation is shown.

*Source: ANNEX III. FINANCIAL CONTRIBUTIONS (excerpt).*

### ANNEX I.  FINANCIAL INTERCONNECTEDNESS

### ANNEX I.  FINANCIAL INTERCONNECTEDNESS

### A. Data and Methodology
- Data source and coverage:
  - Utilizes preliminary end-2010 data on equity and debt portfolio assets from the Fund’s Coordinated Portfolio Investment Survey (CPIS).
  - CPIS data treat all countries on a residence basis and follow BPM 6 definitions and classifications.
  - Overall, 65 Fund members (of which 25 are advanced economies), accounting for about 98 percent of the estimated global International Investment Position (IIP), provide cross-border data on their portfolio investment holdings vis-à-vis 178 Fund members.
  - Positions between non-reporting countries and those below USD ½ million are excluded.
  - For China, separately reported data for China’s Mainland, Hong Kong SAR and Macao SAR were aggregated excluding the holdings amongst them.
- Construction of the interconnectedness measure:
  - A global matrix of bilateral portfolio asset positions is constructed; each entry X_ij reports the nominal value (in US Dollars) of country i’s assets located in country j.
  - All diagonal elements X_ii are zero; off-diagonal elements are positive or zero.
  - The matrix is expressed in portfolio shares by scaling each country’s assets in other countries by its total external assets.
  - The interconnectedness indicator for a country is computed as the sum of the interconnectedness indicators of countries investing in it, weighted by their respective portfolio shares (an eigenvector-like measure).
  - Methodological references include Bonacich and Lloyd (2001) and applications in the financial sector such as Von Peter (2007) and Kubelec and Sá (2010).

### B. Results
- Index properties and distribution:
  - The calculated interconnectedness index ranges from 1 to 15.
  - The United States appears to be the most interconnected with an index of almost 15.
  - Most countries (173 members) have an index value between 1 and 2.
  - There are only five countries with index values between 4 and 8: United Kingdom, Luxembourg, Germany, France, and Netherlands.
- Key portfolio-share statistics explaining dispersion:
  - By far the largest share (20 percent) of total global portfolio assets is invested in the United States.
  - About ten percent is invested in the United Kingdom.
  - Investments located in Luxembourg, Germany, and France each account for around seven percent of total portfolio assets.
- Relationship with IIP and ranking comparison:
  - The interconnectedness index provides country rankings that are broadly similar to rankings according to IIP shares, with the correlation between the two of 0.71.
  - Differences in rankings reflect differing methodologies and underlying data; interconnectedness captures the pattern of bilateral linkages beyond IIP.
- Limitations and considerations for quota formula inclusion:
  - The indicator does not capture all size-related dimensions important for quota variables.
  - The largest country based on this measure (the United States) is only 15 times larger than the smallest member.
  - The index provides almost no differentiation for the vast majority of countries (index between 1 and 2).
  - Treatment of major international financial centers would require additional addressing if the indicator were to be used in the quota formula.
- Supplementary materials referenced:
  - Figure A1.1: Histogram of Interconnectedness Index (source: Finance Department).
  - Table A1.1: Countries’ Rankings according to Financial Openness and Interconnectedness (top 30 countries ranked by gap-filled IIP shares in 2010 and corresponding interconnectedness rankings); calculated using bilateral portfolio investment data from CPIS; includes China, P.R., Hong Kong SAR, and Macao SAR.

*Source: Finance Department*

### ANNEX II. SELECTED APPROACHES TO MODIFYING THE MEASURE OF VARIABILITY

### ANNEX II. SELECTED APPROACHES TO MODIFYING THE MEASURE OF VARIABILITY

### Alternative variability measures based on current receipts and net capital flows
- Purpose: address shortcomings of the current variability measure in the quota formula, notably (i) lack of correlation with an indicator for approval of a Fund arrangement, and (ii) significant year-to-year instability in calculated quota shares.
- Measures examined:
  - 13Y AAD: Average absolute deviation from a three-year centered moving average calculated over a recent 13-year period. Differs from the current measure by using average absolute deviation (reduces impact of extreme observations) instead of root mean squared deviation.
  - 5-year statistical dispersion measures (relative to sample mean/median to limit trend influence):
    - 5Y SD (Standard deviation)
    - 5Y AAD (Average absolute deviation)
    - 5Y MAD (Median absolute deviation)
    - 5Y MAXD (Maximum deviation from the mean)
  - 10Y II (Instability index): variability measured relative to a linear trend. Calculated over a 10-year period; b denotes the slope estimated by OLS.

### Stability of alternative measures (comparative findings)
- Empirical findings (cross-sectional standard deviation of year-on-year changes in variability shares):
  - 13Y AAD and 10Y II yield the smallest variation in shares in 2009-2010.
  - For 2008-2009, 5Y SD (relative to the sample mean) performs better.
  - Over a longer time span (1990-2010), the 13-year average absolute deviation produces the smallest variation in shares on average.
  - There are periods (e.g., 2004-2007) where the existing variability measure yields more stable results than the alternatives.
- Overall conclusion: none of the alternative measures consistently outperforms the current measure in terms of stability properties.

### Association with potential financing needs (correlation with Fund arrangements)
- Staff analysis: alternative variability measures remain uncorrelated with a binary variable for approval of a Fund arrangement; correlations are not significantly different from zero.
- All variability measures based on current receipts and net capital flows are highly correlated among themselves (even after adjustment for economic size), though outcomes for individual members can differ significantly.

### Macroeconomic variables signaling vulnerability (determinants of use of Fund resources)
- Candidate variables identified from literature and staff tests (eight variables; three-year averages used for illustrative composite construction):
  - Reserve cover ratio (reserves to imports)
  - Current account (percent of GDP)
  - Per capita GDP
  - Real GDP growth
  - Inflation
  - External debt service (debt service/exports)
  - Government balance (percent of GDP)
  - Stock of external debt (percent of GDP)
- Empirical differences between countries with programs and without programs (year of approval and two preceding years):
  - Current account deficit as percent of GDP: about twice as high for program countries compared to non-program countries (valid for contemporaneous and lagged values).
  - Reserve cover ratio: program countries have lower reserve cover ratios; difference in the year preceding arrangement approval is about 8 percentage points.
  - GDP per capita: countries with lower per capita income are more likely to have Fund arrangements.
  - GDP growth: countries with programs have recorded lower GDP growth on average (robust with lags).
  - Inflation: typically higher in countries with IMF programs.
  - Government budget deficits: higher by 0.9-1.4 percent of GDP on average in countries with programs versus without; peak difference in the year preceding approval.
  - Stock of external debt (percent of GDP): program countries average 77 percent in the year of approval versus 68 percent for countries without programs.
  - External debt service (share of exports): program countries allocate a greater share of exports to debt service payments.

### Composite indicator construction and properties
- Rationale: no single macro variable robustly predicts use of Fund resources; combining indicators into a composite variable can capture multiple vulnerability types and typically yields higher explanatory/predictive power.
- Weighting approach (illustrative): weights determined based on the inverse of the standard deviation of each component (standardization prevents domination by a single component).
- Composite variants considered:
  - Comp1: linear combination of all variables except inflation (inflation shows little correlation with likelihood of a Fund arrangement).
  - Comp2: excludes external debt and external debt service.
  - Comp3: excludes external debt, external debt service, and fiscal balance (allows calculation for the largest number of countries; used for illustrative mapping and results).
- Correlation results:
  - Correlations of composite indicators with the binary variable for approval of a Fund arrangement are statistically significant but modest.
  - Example: Comp3 correlation for the full sample is 0.19.
  - Dropping LICs increases Comp3 correlation to 0.24 (for emerging markets and advanced economies).
  - For LICs only, correlations are generally much weaker—between 0.05-0.07.
  - Robustness: correlation coefficients remain highly significant when calculated over 5-year intervals since 1990.
- Transformation to a quota-compatible variable:
  - Issue: composite indicator scores can be negative and are not size-scaled, whereas quota formula variables are expressed as shares of global totals and include significant size components.
  - Proposed mapping: transform x_i (vulnerability score for country i) using a mapping (denoted mapping (1) in the text) that stacks all individual scores in the (0,2) interval.
  - After transformation, multiply transformed vulnerability scores by members’ shares in GDP to impart economic size; rescale and subtract GDP share to adjust for size effect (illustrative approach).
  - Impact on correlations:
    - Comp3 correlation falls from 0.19 to 0.16 after applying the non-linear transformation (1).
    - Correlation falls further to 0.09 after multiplying transformed score by GDP shares, rescaling, and then subtracting the GDP share to adjust for size.

### Illustrative country and group impacts (selected figures from Table A2.2)
- United States:
  - GDP shares: 2009 = 24.5; 2010 = 23.5
  - Existing variability shares: 2009 = 14.1; 2010 = 15.5
  - Composite variability shares (transformed Comp3): 2009 = 27.1; 2010 = 25.4
  - Difference (Composite − Existing): 2009 = 13.0; 2010 = 9.9
  - Note: relative gain driven by size effect and low reserve cover ratio for the US (reserve currency caveat noted).
- China (including P.R., Hong Kong SAR, Macao SAR):
  - Existing variability (2009, 2010): 5.6, 5.2 (Table A2.1)
  - Composite variability outcomes in Table A2.2 reflect lower variability share relative to GDP share (examples in text).
- Italy and Spain:
  - Composite variability shares are significantly higher than GDP shares; for Greece and Portugal (not shown in the table) composite variability shares based on 2010 data are more than twice their GDP shares.
- Aggregate group comparisons (selected entries from Table A2.1 and Table A2.2):
  - Advanced economies: Existing variability shares (Table A2.1) = 57.9 (Current Var.), 59.0 (13Y AAD), 52.0 (5Y SD), 54.7 (5Y AAD), 58.2 (5Y MAD), 50.2 (5Y MAXD), 57.6 (10Y II).
  - Emerging Market and Developing Countries (Table A2.1): Current Var. = 42.1; 13Y AAD = 41.0; 5Y SD = 48.0; 5Y AAD = 45.3; 5Y MAD = 41.8; 5Y MAXD = 49.8; 10Y II = 42.4.
  - Advanced economies GDP shares (Table A2.2): 2009 GDP = 66.4; 2010 GDP = 64.2; Existing variability = 58.6 (2009), 57.9 (2010); Composite variability = 67.9 (2009), 67.9 (2010); Difference = 11.0 (2009), 10.0 (2010).

### Overall assessment and challenges
- Composite vulnerability indicators can improve explanatory power relative to single variables, but they raise several challenges:
  - Choice of variables and weights influences results and may reduce transparency and simplicity.
  - Non-linear transformations required to make composite scores compatible with quota-share scaling can reduce correlation with the arrangement indicator.
  - Data availability constraints (especially for fiscal and external debt series) limit coverage; Comp3 chosen in illustrations because it covers the largest number of countries (175 members).
  - Uncertainty about whether composite indicators that explain past use of Fund resources will be robust predictors of future use.
- Conclusion: translating macroeconomic vulnerability information into a single quota-formula-compatible variability measure poses major methodological and practical challenges; no alternative examined consistently dominates the current measure across stability and explanatory criteria.

*Source: Finance Department (ANNEX II. SELECTED APPROACHES TO MODIFYING THE MEASURE OF VARIABILITY).*

### ANNEX III. FINANCIAL CONTRIBUTIONS

### ANNEX III. FINANCIAL CONTRIBUTIONS

### Pre-NAB Commitments
- Covers loan commitments by members under the 1974 and 1975 Oil Facilities, the Supplementary Financing Facility (SFF), the Enlarged Access Facility (EAF), and the expanded General Agreements to Borrow (GAB).
- Includes bilateral borrowing agreements with Italy (1966), Switzerland (1977), and Japan (1986).

### NAB
- Data cover the commitments by members under NAB credit arrangements and reflecting the rollback agreed by the Executive Board.
- Commitments by those members that have not yet adhered to the NAB decision are not included.
- For members with bilateral borrowing arrangements being folded into the NAB, only the NAB commitment is included.
- Bilateral loan commitments from members that are not participants in the NAB (Czech Republic, Malta, Slovak Republic, and Slovenia) have been included.

### NAB + New Pledges
- In addition to the NAB amounts, data reflect bilateral pledges under the current fund raising exercise, including the announcements made at the G-20 Leaders’ Summit in Los Cabos.

### PRGT Loans
- Reflects all loan commitments to the PRGT Trust as of December 31, 2011.
- Includes Saudi Arabia’s pledge of SDR 11 million as communicated to the Fund in a letter received by the Fund on June 5, 2012.

### PRGT Subsidies
- Includes total bilateral resources provided since 1987 for subsidizing concessional lending, HIPC and MDRI debt relief as of December 31, 2011 plus all pledges made under current fundraising as of December 31, 2011.

### Technical Assistance
- Reflects actual cash disbursements to the Fund for technical assistance and training excluding in kind contributions over the period FY1999-FY2012.
- These data do not include pledges.

### FTP Participation
- Two metrics developed to account for FTP participation:
  - Duration-based metric:
    - Derived from the number of quarters out of the 80 quarters covering 1992–2011 that a member was included in the Financial Transactions Plan (FTP).
    - A member’s share = total number of quarters a member participated in the FTP over the 20-year period January 1992–December 2011 (i.e., 80 quarters) as a share of total participant quarters aggregated over the 80 quarters (these sum to 3,197).
    - The maximum share for a member that participated in all 80 quarters is 2.5 = (80/3197)*100.
    - Figure A3.1 shows participation in the FTP during the period in question.
  - Resources-based metric:
    - Obtained by weighting each member’s participation in the FTP with its nominal quota.
    - A member’s share in the resources-based FTP participation metric is a proxy for the share of resources a member has made available over time in the FTP.
- Notes:
  - The period 1992-2011 includes the move from the Operational Budget to the FTP framework in mid-2000 as well as the period of expanding membership in the early 1990s which was reflected in FTP participation.
  - For quarters prior to June-August, 2000, the Operational Budget participation is shown.

*Source: ANNEX III. FINANCIAL CONTRIBUTIONS (excerpt).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/np/pp/eng/2012/_062912.pdf_