## ppea2021008

## Source details

**Canonical URL:** [ppea2021008](https://www.imf.org/-/media/files/publications/pp/2021/english/ppea2021008.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/pp/2021/english/ppea2021008.pdf.md)
- [Structured JSON version](/-/media/files/publications/pp/2021/english/ppea2021008.pdf.json)

---

### EXECUTIVE SUMMARY — INTRODUCTION AND CONTEXT
- Paper is part of the workplan on the 15th General Review of Quotas (15th Review) and is a companion to Quota Formula and Realigning Shares.
- No proposals are made in this paper; analysis informs Directors’ judgments and future staff work on possible landing zones for Fund size and quota resources.
- Key near-term institutional facts:
  - The Fund’s current overall lending capacity is about SDR 693 billion (almost $1 trillion).
  - The 2016 Bilateral Borrowing Agreements (BBAs) boost this capacity but will expire at end-2019, or end-2020 at the latest (if a one-year extension is approved).
  - The New Arrangements to Borrow (NAB) Decision was renewed through November 2022.
  - A decision to renew the General Arrangements to Borrow (GAB) must be taken by December 2017 for the current period that ends in December next year.
- Motivation:
  - The Fund’s central role in the Global Financial Safety Net (GFSN) and lessons from the Global Financial Crisis (GFC) underline the need for appropriately sized resources to preserve confidence and enable catalytic assistance.
  - The GFSN has become larger and more multilayered since the GFC.

### FIRST PILLAR: QUANTITATIVE ANALYSIS — FRAMEWORK AND PURPOSE
- Framework: three complementary quantitative methodologies provide indicative ranges for adequacy of Fund resources:
  1. Traditional metrics.
  2. Access-based simulations.
  3. Global scenarios.
- Purpose and limits:
  - These quantitative approaches are not definitive; they are intended to inform judgment and identify possible landing zones.
  - Given projection uncertainties, results should be interpreted as indicative ranges rather than precise targets.
- Fund resource and liquidity definitions (highlights from Box 1 and text):
  - Quotas are the primary financing source for lending.
  - Standing borrowing arrangements supplement quotas: NAB and GAB.
  - Bilateral borrowing (2012 and 2016 BBAs) serves as temporary backstop to quotas and NAB.
  - Lending capacity measures usable resources available for non-concessional lending.
  - Forward Commitment Capacity (FCC) measures resources available to finance new commitments over the next 12 months; currently contains only quota resources due to NAB deactivation in February 2016 and excludes BBAs unless certain conditions are met.
  - Reported Fund financial indicators (as presented in the paper): Lending capacity: about SDR 693 billion.

### A. Traditional Metrics — KEY FINDINGS AND NUMBERS
- Reference levels: average of ratios at the time of the past four quota reviews with an agreed quota increase (8th, 9th, 11th and 14th Reviews: 1983, 1990, 1998, 2010).
- Two reference periods: 2012–16 (current period) and 2015–19 (April 2017 WEO baseline).
- Key findings:
  - Quota-to-GDP ratio historically restored to a similar level (1.2–1.3 percent) in reviews where quotas were increased.
  - Restoring the quota-to-GDP ratio to its reference level would require:
    - a 31 percent increase in quotas for 2012–16, and
    - a 47 percent increase in quotas for 2015–19.
  - Restoring quota ratios to external-flow indicators would require:
    - roughly doubling quotas for 2012–16, and
    - nearly tripling quotas for 2015–19.
  - Restoring the quota-to-EFNs ratio (for past borrowers) would require:
    - increasing quotas by 151 percent for 2012–16, and
    - increasing quotas by 169 percent for 2015–19.
  - Required increase in total Fund resources to restore the quota-to-EFN ratio to its reference level would be 62 to 71 percent.
- Selected entries from Table 1 — additional quotas required to restore quotas relative to economic indicators (SDR billion; percent increase from current quotas):
  - GDP: 124, 149, 222; 26, 31, 47
  - Current Payments: 481, 545, 835; 101, 114, 175
  - Capital Inflows to EMDCs: 850, 944, 1,176; 178, 198, 246
  - EFN: 701, 721, 808; 147, 151, 169
  - Average: 539, 590, 760; 113, 124, 159
- Selected entries for total Fund resources required to restore ratios (SDR billion; percent increase from current resources):
  - GDP: (194), (152), (90); (21), (16), (10)
  - Current Payments: 224, 314, 636; 24, 34, 69
  - Capital Inflows to EMDCs: 642, 765, 1,012; 70, 83, 110
  - EFN: 536, 574, 654; 58, 62, 71
  - Average: 302, 375, 553; 33, 41, 60
- Context:
  - Large increase in Fund borrowing since the GFC raised ratio of overall Fund resources to GDP to slightly above the relevant reference ratio.

### B. Access-Based Simulations — METHODOLOGY AND RESULTS FRAMEWORK
- Two-step approach:
  - Step 1: Identify members that might require Fund assistance using two approaches:
    - Approach A — Large borrower scenarios: Scenario 1 = top 12; Scenario 2 = top 32 (by 2020 GDP as size proxy).
    - Approach B — Panel logit scenarios: binary response model for panel data covering 94 members (1992–2014, 105 GRA arrangements).
  - Step 2: Apply assumptions on arrangement sizes using historical distributions (as share of member GDP).
- Arrangement-size statistics (Table 2 — Size of GRA arrangements, percent of GDP):
  - Last 20 years: Average 3.5; Average +1STD 6.2
  - Last 15 years: Average 4.1; Average +1STD 7.0
  - Last 10 years: Average 5.3; Average +1STD 8.0
  - Since 2008: Average 5.8; Average +1STD 8.1
- Observations:
  - Average size of Fund arrangements (as share of GDP) has tended to increase over time.
  - Normal access limits of 435 percent of quota correspond on average to 6.9–7.4 percent of members’ GDP identified by the panel logit approach depending on threshold; and 3.9 and 4.9 percent of GDP for the top 12 and top 32 past borrowers respectively.
- Results presentation:
  - Table 3 summarizes potential calls on Fund financing across scenarios with arrangement size assumptions from 4 to 8 percent of members’ GDP in 1 percentage point increments; scenarios are color coded in the source to indicate coverage by quotas only, quotas+NAB, quotas+NAB+BBAs, or exceed current overall lending capacity.
- Note:
  - All scenarios use overall lending capacity and do not account for existing loan commitments which reduce available resources for new commitments as reflected in the FCC.

### C. Global Scenarios — APPROACH, SUPPLY ASSUMPTIONS, AND DEMAND RANGES
- Approach:
  - Bottom-up stress test based on members identified as vulnerable by staff’s Winter 2017 Vulnerability Exercise.
  - Calibrations align members’ responses with average historical responses during past crises.
  - Model computes potential demand factoring domestic adjustment and supply of non‑Fund financing (international reserves, RFAs, BSAs).
- Modeling demand — three-step approach:
  - Pervasiveness: members selected using crisis-probability thresholds of more than 1, 3, 5, and 10 percent.
  - Severity: impacts on FDI inflows, rollover rates for short- and medium-term debt, and (for EMs) deposit outflows; crisis assumed to persist for two years (2017–18) with declining intensity; calibrated using empirical distribution (90th, 85th, 75th, 65th percentiles).
  - Domestic adjustment: assumes fiscal adjustment of 1.4 percent of GDP (0.7 percent each year).
- Supply assumptions:
  - Members do not use reserves beyond 25 percent of pre-crisis level; EMs assumed to maintain reserves at 100 percent of the Fund’s ARA metric; AEs and LICs at 100 percent of short-term debt.
  - RFAs included: AMF, CRA, CMIM, EFSD, EU-BoP facility, ESM, FLAR (with specific modality assumptions).
  - Standing unlimited BSAs among major central banks included; RMB swaps largely excluded.
- Demand for financing across considered scenarios: SDR 186 to SDR 2,237 billion (source summary).
- Potential calls on Fund financing (demand minus supply, modeled matrices):
  - Range of potential calls on Fund financing: SDR 143 billion to SDR 1,391 billion across modeled matrices of pervasiveness and intensity (Box 2 summary).
- Historical benchmark:
  - The GFC could have been located in scenario “a” or “b” and the 85th to 90th percentile in absence of buffers and policy stimulus; with buffers, the GFC was toward the lower right corner.

### BOX 2 — TRENDS IN GRA ARRANGEMENT SIZE AND CREDIT OUTSTANDING (SELECTED POINTS)
- Average program size over time:
  - 1960–1976: around 1.5 percent of GDP.
  - Onset of international debt crisis (early 1980s): shifted up to 2.0 percent of GDP.
  - Since 2008: average program size about 5.8 percent of GDP.
- Upper-tail growth:
  - 75th percentile increased from around 2 percent of GDP around 1990 to about 6.5 percent of GDP in 2015.
- Distributional dispersion:
  - 1960–79: 21 percent of programs were larger than the average.
  - 2008–16: 34 percent of programs were larger than the average.
- Arrangement duration and maturity:
  - Average expected program duration in 2016: 32.6 months.
  - Average maturity of GRA credit outstanding increased sharply to over 6½ years (figures in source: 1995–2017 and 1995–2016 series referenced).
- Modeling demand for financing in systemic scenarios:
  - Demand for financing across scenarios ranges from SDR 186 to SDR 2,237 billion.
  - Potential calls on Fund financing range from SDR 143 billion to SDR 1,391 billion after accounting for non‑Fund financing.

### ROBUSTNESS CHECKS — KEY FINDINGS (Annex III highlights)
- Self-insurance (reserves):
  - Relaxing reserve-use assumption from 25 to 35 or 75 percent could reduce potential calls on Fund financing by up to SDR 53 billion (for 75th intensity percentile, depends on crisis pervasiveness).
  - Relaxing ARA assumption to 80 percent of the metric could lower potential calls by up to SDR 125 billion in the most pervasive scenario.
- Fiscal adjustment:
  - Stronger domestic response (2 percent of GDP over two years versus baseline 1.4 percent) can substantially reduce potential calls.
  - Weaker domestic response (1 percent of GDP over two years) increases calls by up to SDR 79 billion.
- Burden sharing between RFAs and the Fund:
  - If RFAs cover 75 percent of a member’s net financing needs, potential calls on Fund financing would be lower by up to SDR 28 billion (75th percentile, illustrative).
  - Equal burden sharing could increase potential calls by up to SDR 44 billion.
  - If RFAs cover only 25 percent, potential calls increase between SDR 5 and SDR 116 billion depending on pervasiveness.
- Global scenario model — financing gap by crisis intensity (Financing gap, billions of SDRs):
  - a. Extremely pervasive global systemic crisis: 65th = 396; 75th = 627; 85th = 1,011; 90th = 1,391
  - b. Very pervasive systemic crisis: 65th = 393; 75th = 533; 85th = 791; 90th = 1,065
  - c. Pervasive systemic crisis: 65th = 310; 75th = 399; 85th = 606; 90th = 830
  - d. Systemic crisis: 65th = 143; 75th = 197; 85th = 298; 90th = 367
- Summary:
  - Quantitative approaches yield a broad range of estimates; scenarios indicate calls on Fund financing from as low as SDR 140 billion to almost ten times this amount.
  - Maintaining Fund lending capacity in the absence of bilateral borrowing agreements would require a quota increase of about 70 percent.

### SECOND PILLAR: QUALITATIVE CONSIDERATIONS — INTERCONNECTEDNESS, TRANSITIONS, AND GFSN IMPLICATIONS
- Overriding qualitative consideration: high interconnectedness of the global economy increases potential calls on Fund resources.
- Drivers increasing interconnectedness and uncertainty:
  - Structural changes: EMDCs’ share in global trade increased from 15 percent in 1990 to 36 percent in 2016.
  - Nature of crises shifting from current account to capital account drivers; capital flows increased more than 25-fold between 1980 and 2007 versus an eight-fold expansion in global trade.
- New and ongoing transitions that could raise systemic risk:
  - Shocks of non-economic origin (refugee flows, global epidemics).
  - Climate change increasing frequency, severity, and cost of natural disasters.
  - Technological adjustment and digital disruptions (cybersecurity risks, virtual currencies, peer-to-peer lending, high-frequency trading).
- Financial system considerations:
  - Global financial cycles have grown in amplitude and duration; nonbank financial institutions are increasingly important yet largely unregulated.
  - Post-crisis regulatory reforms improved resilience but key aspects remain unfinished.
- Narrowed room for policy maneuver:
  - Monetary policy remains accommodative in many AEs.
  - Rising global indebtedness and weaker medium-term growth prospects reduce fiscal space and policy flexibility.
- GFSN multilayering and implications:
  - GFSN has become more multilayered and regional; predictability and durability of RFAs and BSAs are uncertain.
  - Self-insurance (reserve accumulation) is costly and may undermine global demand.
  - Fund’s catalytic role and need for sufficient firepower to maintain credibility emphasized.
- IMF–RFA collaboration:
  - Scope to improve collaboration across lending, surveillance, and capacity development; operational framework could increase effective GFSN firepower but faces challenges (stigma, conditionality, institutional independence).
- Changes in Fund policies since the GFC:
  - Improved bilateral and multilateral surveillance, mainstreaming macro-financial surveillance, expanded risk management, and lending toolkit reforms (including work on a liquidity backstop).
  - Reforms to debt restructuring rules could reduce demand for Fund financing by triggering earlier creditor action; effects are hard to estimate ex ante.
- Role of Fund resources, credibility, and moral hazard:
  - Adequately resourced Fund can foster confidence; larger Fund resources have favorable effects on risk outlook and borrowing costs of emerging markets.
  - Risk of moral hazard from resource size judged low given strict program design, eligibility criteria (FCL/PLL), coordination efforts, and safeguards against superfluous lending.

### THE SIZE OF THE FUND — QUOTAS, BORROWED RESOURCES, AND GOVERNANCE
- Membership view: "The membership has stressed consistently that the Fund is and should remain a quota-based institution."
- Distinct implications of quotas versus borrowed resources:
  - Governance: voting power determined by quotas; borrowed resources do not affect quotas.
  - Permanent versus temporary: quotas are permanent resources; borrowing historically temporary.
  - Participation: quota increases cover entire membership; borrowed resources provided by subset of membership.
  - Resource mobilization: quota increases take considerable time; borrowed resources can be raised more quickly but future mobilization is uncertain.
  - Resource availability and activation conditions:
    - NAB activation requires (i) a threat to the stability of the IMS, and (ii) a majority of 85 percent or more of total credit commitments.
    - BBAs (2016) can be activated only if the “modified FCC” is below SDR 100 billion and activation is approved by creditors representing 85 percent of total commitments under the 2016 BBAs.
  - Burden-sharing: use of quota resources increases the Fund’s burden-sharing capacity; use of borrowed resources reduces burden-sharing capacity relative to credit outstanding.
  - Financial characteristics: both quotas and borrowed resources are liquid reserve assets; remuneration and interest follow SDR interest rate conventions.
- Availability ex-ante versus mobilization ex-post:
  - Having resource commitments available ex-ante can boost market confidence; mobilizing resources ex-post can be costly.
  - Increasing multipolarity and inward-looking policies may make future agreement on additional borrowed resources more difficult.
- Composition history:
  - Before the GFC, the share of quotas in quotas+NAB/GAB was about 84 percent on average (1978–2008).
  - Share of NAB and bilateral borrowing agreements peaked at around 73 percent of total resources in 2013–15.
  - Doubling of quotas under the 14th Review in early 2016 and rollback of the NAB reduced reliance on borrowing; share of borrowed resources remains high at about 50 percent.

### ANNEX HIGHLIGHTS — PANEL LOGIT MODEL (Annex II) AND GLOBAL VOLATILITY SHOCK
- Panel logit model (94 countries, 1992–2014, 105 GRA arrangements):
  - Key marginal effects (dy/dx) and significance:
    - Past program (dummy): 0.306; Robust SE = 0.092; significance = ***
    - External Financing Needs: 2.331; Robust SE = 0.916; significance = **
    - GDP growth: -0.063; Robust SE = 0.026; significance = **
    - GDP per capita: -0.866; Robust SE = 0.218; significance = ***
    - Credit gap: 0.025; Robust SE = 0.010; significance = ***
    - 3M US int. rate variation: 0.370; Robust SE = 0.161; significance = **
    - VIX: 0.121; Robust SE = 0.034; significance = ***
    - Oil price: -0.023; Robust SE = 0.012; significance = *
  - Model fit: Pseudo R2 = 0.504; Observations = 1,597; Countries = 94; GRA Arrangements = 105.
- Global volatility shock scenario:
  - VIX assumed to reach an average level of 30 during the year of the shock (described as below average VIX observed over 2008–09).
  - Thresholds for flagging members based on predicted probabilities:
    - Equal weights (Type I = Type II): threshold = 4.8 percent.
    - Penalizing false positives 2:1: threshold = 6.4 percent (false positives = 16 percent; missed new programs = about 25 percent).
    - Penalizing false positives 3:1: threshold = 16.1 percent.

### ORGANIZATIONAL AND PROCESS POINTS; NEXT STEPS
- The paper builds on the March 2016 staff paper and reflects informal Board discussions, bilateral outreach, and a June 2017 staff briefing.
- Decisions on Fund size will need to be taken as a package with quota formula and share realignment issues; further deliberations and compromise will be required to reach broad membership consensus.
- Next steps:
  - Input from this paper is intended to support Directors in making a judgment on the appropriate size of the Fund and quota increases.
  - Based on Directors’ guidance, staff will prepare a follow-up paper examining possible landing zones for the Fund’s size and quota resources.

### QUESTIONS FOR DIRECTORS (ISSUES FOR DISCUSSION)
- Do Directors broadly agree with the two-pillar framework (quantitative and qualitative) to assess adequacy of Fund resources?
- Views on the quantitative pillar: merit in using the range of approaches (traditional metrics, access-based, global scenarios)? Are these ranges useful input for assessing size and composition of Fund resources?
- Views on qualitative considerations: relative weights to assign given expectations on how the global economy and IMS will evolve? Do Directors support the view that qualitative considerations can be expected to increase potential calls on Fund resources in future crises?
- Views on composition of Fund resources:
  - Do Directors agree that the Fund should remain a quota-based institution?
  - Do Directors concur that a financing structure of quotas, backstopped by a standing borrowing facility such as the GAB/NAB, has served the Fund well over several decades?

*International Monetary Fund — excerpts from staff chapter and annexes in content unit ppea2021008.*

### EXECUTIVE SUMMARY

### EXECUTIVE SUMMARY

### INTRODUCTION AND CONTEXT
- Paper is part of the workplan on the 15th General Review of Quotas (15th Review) and is a companion to Quota Formula and Realigning Shares.
- No proposals are made in this paper; analysis informs Directors’ judgments and future staff work on possible landing zones for Fund size and quota resources.
- Key near-term institutional facts:
  - The Fund’s current overall lending capacity is about SDR 693 billion (almost $1 trillion).
  - The 2016 Bilateral Borrowing Agreements (BBAs) boost this capacity but will expire at end-2019, or end-2020 at the latest (if a one-year extension is approved).
  - The New Arrangements to Borrow (NAB) Decision was renewed through November 2022.
  - A decision to renew the General Arrangements to Borrow (GAB) must be taken by December 2017 for the current period that ends in December next year.
- Motivation:
  - The Fund’s role at the center of the Global Financial Safety Net (GFSN) and lessons from the Global Financial Crisis (GFC) underline the need for appropriately sized resources to preserve confidence and enable catalytic assistance.
  - The GFSN has become larger and more multilayered since the GFC, with increased participation in official financing arrangements.

### FIRST PILLAR: QUANTITATIVE ANALYSIS
- Framework: three complementary quantitative methodologies provide indicative ranges for adequacy of Fund resources:
  1. Traditional metrics.
  2. Access-based simulations.
  3. Global scenarios.
- Key findings from the three approaches:
  - Traditional metrics:
    - The ratio of current quotas to GDP is below its historical norm.
    - If borrowed resources are factored in, the quotas-to-GDP gap is filled for that specific metric.
    - However, all other traditional metrics indicate a significant resource gap—even when borrowed resources are included.
  - Access-based simulations and global scenarios:
    - Produce a relatively wide range of estimates for required lending capacity.
    - Results depend critically on the pervasiveness and intensity of crises and on access-related considerations.
- Purpose and limits:
  - These quantitative approaches are not definitive; they are intended to inform judgment and identify possible landing zones.
  - Given projection uncertainties, results should be interpreted as indicative ranges rather than precise targets.
- Fund resource definitions and liquidity metrics (Box 1):
  - Quotas are the primary financing source for lending.
  - Standing borrowing arrangements supplement quotas:
    - New Arrangements to Borrow (NAB).
    - General Arrangements to Borrow (GAB) — usable only if NAB activation is not supported.
  - Bilateral borrowing (2012 and 2016 BBAs) serves as temporary backstop to quotas and NAB.
  - Lending capacity measures usable resources available for non-concessional lending.
  - Forward Commitment Capacity (FCC) measures resources available to finance new commitments over the next 12 months; currently contains only quota resources due to NAB deactivation in February 2016 and excludes BBAs unless certain conditions are met.
  - Reported Fund financial indicators (as presented in the paper):
    - Lending capacity: about SDR 693 billion.
    - Total Fund resources and FCC components are detailed in Box 1.

### SECOND PILLAR: QUALITATIVE CONSIDERATIONS
- Overriding qualitative consideration:
  - High interconnectedness of the global economy is the central qualitative factor shaping adequacy assessments.
- Interaction with transitions and vulnerabilities:
  - The interaction of interconnectedness with ongoing and new transitions increases the risk of systemic stress.
  - Vulnerabilities across countries have increased and many countries have less room for policy maneuver.
- Implications for the GFSN and Fund policy:
  - While interconnectedness argues for global solutions, the GFSN has become more multilayered and regional in nature.
  - Other qualitative considerations discussed include:
    - Changes to Fund policies and toolkit reforms.
    - Confidence impacts of resource announcements and perceived adequacy.
    - Moral hazard considerations related to resource size and composition.
- Composition of Fund resources:
  - The membership has underscored that the Fund should remain a quota-based institution.
  - The overall share of borrowed resources, including bilateral borrowing, remains high by historical standards.
  - The paper provides elements relevant to discussions on resource composition:
    - Quantitative assessment of how various adequacy estimates are covered by current quotas and borrowed resources.
    - Discussion of implications of different financing sources across governance, resource mobilization, and burden sharing dimensions.
    - Emphasis on ensuring sufficient resources ex ante before a crisis hits.

### ORGANIZATIONAL AND PROCESS POINTS
- The paper builds on the March 2016 staff paper and reflects informal Board discussions, bilateral outreach, and a June 2017 staff briefing.
- Because decisions on Fund size will need to be taken as a package with quota formula and share realignment issues, further deliberations and compromise will be required to reach broad membership consensus.
- Next steps noted in the paper:
  - Input from this paper is intended to support Directors in making a judgment on the appropriate size of the Fund and quota increases.
  - Based on Directors’ guidance, staff will prepare a follow-up paper examining possible landing zones for the Fund’s size and quota resources.

*International Monetary Fund — EXECUTIVE SUMMARY of ppea2021008*

### 8.      This section details the quantitative pillar of the framework, building on earlier

### 8.      This section details the quantitative pillar of the framework, building on earlier analyses

### Overview
- The quantitative pillar uses three complementary approaches:
  - A metric-based assessment of Fund resources relative to (i) key traditional proxies for demand (global GDP, trade, and capital flows) and (ii) members’ external financing needs (EFNs). This estimates additional resources needed to restore these ratios to their reference levels.
  - An access-based assessment that (i) identifies members that could potentially require Fund financial assistance using two methods and (ii) calculates a range of potential financing needs using past patterns of demand for Fund resources.
  - Global shock scenarios that simulate financing needs using a range of historical systemic crises and the staff’s Vulnerability Exercises.
- The metric-based approach examines both quotas and total Fund resources (quotas, the NAB/GAB, and BBAs), while the access-based and global shock approaches focus on the Fund’s lending capacity.

### A. Traditional Metrics
- Reference definition: reference levels are the average of ratios at the time of the past four quota reviews with an agreed quota increase (the 8th, 9th, 11th and 14th Reviews concluded in 1983, 1990, 1998, and 2010).
- Two reference periods are presented: 2012–16 (current period) and 2015–19 (based on April 2017 WEO baseline).
- Key findings:
  - Quota-to-GDP ratio historically restored to a similar level (1.2–1.3 percent) in reviews where quotas were increased.
  - Since the 14th Review, the quota-to-GDP ratio has declined to levels prevalent at the 10th Review in the mid-1990s.
  - Restoring the quota-to-GDP ratio to its reference level would require:
    - a 31 percent increase in quotas for 2012–16, and
    - a 47 percent increase in quotas for 2015–19.
  - Ratios of quotas to external-flow indicators (current payments, capital inflows) have declined more steeply than for GDP. Restoring quota ratios to these reference levels would require:
    - roughly doubling quotas for 2012–16, and
    - nearly tripling quotas for 2015–19.
  - Restoring the quota-to-EFNs ratio (for past borrowers) to the reference level would require:
    - increasing quotas by 151 percent for 2012–16, and
    - increasing quotas by 169 percent for 2015–19.
  - The required increase in total Fund resources to restore the quota-to-EFN ratio to its reference level would be 62 to 71 percent.
- Table 1 (selected entries) — additional quotas required to restore quotas relative to economic indicators (in SDR billion and percent increase from current quotas):
  - GDP: 124, 149, 222; 26, 31, 47
  - Current Payments: 481, 545, 835; 101, 114, 175
  - Capital Inflows to EMDCs: 850, 944, 1,176; 178, 198, 246
  - EFN: 701, 721, 808; 147, 151, 169
  - Average: 539, 590, 760; 113, 124, 159
- Table 1 (selected entries) — additional resources required to restore total resources relative to economic indicators (in SDR billion and percent increase from current resources):
  - GDP: (194), (152), (90); (21), (16), (10)
  - Current Payments: 224, 314, 636; 24, 34, 69
  - Capital Inflows to EMDCs: 642, 765, 1,012; 70, 83, 110
  - EFN: 536, 574, 654; 58, 62, 71
  - Average: 302, 375, 553; 33, 41, 60
- Additional context:
  - The large increase in Fund borrowing since the GFC raised the ratio of overall Fund resources to GDP to slightly above the relevant reference ratio.
  - Gross external liabilities and broad money (M2) are relevant indicators of external vulnerability but are less central due to data and interpretation issues.

### B. Access-Based Simulations
- Two-step approach:
  - Step 1: Identify members that might require Fund assistance using two approaches.
  - Step 2: Apply plausible assumptions on the size of financial assistance to generate estimates of the impact on the Fund’s lending capacity.
- Step 1 — Identification of members:
  - Approach A — Large borrower scenarios:
    - Scenario 1: the 12 largest members among the sample (12 equals the average number of new arrangements per year since 1990).
    - Scenario 2: the 32 largest members among the sample (32 equals the number of members with a Fund arrangement during the past two crises lending cycles, 1996–98 and 2008–10).
    - Sample: all members that had a Fund GRA arrangement or requested outright disbursements since 1990, ranked by their 2020 GDP as proxy for size.
  - Approach B — Panel logit scenarios:
    - Econometric model: binary response model for panel data covering 94 members to gauge effects of economic and financial variables on GRA lending (panel dataset covers advanced, emerging, and frontier market economies over 1992–2014, total of 105 GRA arrangements).
    - Independent variables include:
      - Member-specific: External financing needs, GDP growth, GDP per capita, credit-to-GDP gap, variation of the bilateral nominal exchange rate versus the U.S. dollar, government stability, and a measure of interconnectedness. A dummy variable controls for access to an RFA.
      - Global: U.S. interest rate variations and the VIX (proxy for risk aversion). The model indicates sustained drops in the oil price increase probabilities of Fund arrangements.
    - Model application: generate probabilities under a global volatility shock assuming the VIX index reaches an average level of 30 during the year of the shock (this is below average VIX observed over 2008–09); all other variables held unchanged.
    - Thresholds for flagging members:
      - Equal weights for Type I and Type II errors sets threshold to 4.8 percent.
      - Penalizing false positives with ratios 2:1 and 3:1 sets thresholds to 6.4 percent and 16.1 percent respectively.
- Step 2 — Arrangement size:
  - Arrangement sizes are modeled using historical distribution of arrangement sizes as a share of members’ GDP.
  - Table 2 — Size of GRA arrangements (in percent of a member’s GDP):
    - Last 20 years: Average 3.5; Average +1STD 6.2
    - Last 15 years: Average 4.1; Average +1STD 7.0
    - Last 10 years: Average 5.3; Average +1STD 8.0
    - Since 2008: Average 5.8; Average +1STD 8.1
  - Observations: average size of Fund arrangements (as share of GDP) has tended to increase over time; normal access limits of 435 percent of quota correspond on average to 6.9–7.4 percent of members’ GDP identified by the panel logit approach depending on threshold; and 3.9 and 4.9 percent of GDP for the top 12 and top 32 past borrowers respectively.
- Results presentation:
  - Table 3 (described) summarizes potential calls on Fund financing across scenarios (rows = five Step 1 scenarios ordered by impact; columns = arrangement size assumptions from 4 to 8 percent of members’ GDP in 1 percentage point increments).
  - Color-coding used in table narrative:
    - Dark green: scenarios potentially covered by current lending capacity using quota resources only.
    - Green: covered using quotas and the NAB.
    - Light green: covered using quotas, NAB, and BBAs.
    - White: scenarios exceed current overall lending capacity.
  - All scenarios are based on overall lending capacity and do not account for existing loan commitments which reduce available resources for new commitments as reflected in the FCC.

### C. Global Scenarios
- Approach:
  - Bottom-up stress test based on members identified as vulnerable by staff’s Winter 2017 Vulnerability Exercise.
  - Calibrations align members’ responses to a global crisis with average historical responses during past crises.
  - The model computes potential demand for financing by vulnerable members factoring in:
    - domestic adjustment in response to the crisis;
    - supply of non-Fund financing including members’ international reserves and, where applicable, contributions from RFAs and bilateral swap arrangements (BSAs).
- Purpose: assess the Fund’s implied lending capacity under a global crisis scenario using granular country-specific estimations of potential calls on Fund resources.

*Source: Excerpt from IMF staff chapter: "This section details the quantitative pillar of the framework, building on earlier analyses."*

### Box 2. Trends in GRA Arrangement Size and Credit Outstanding

### Box 2. Trends in GRA Arrangement Size and Credit Outstanding

### Trends in program size and distribution
- Average program size over time:
  - 1960–1976: hovered around 1.5 percent of GDP.
  - Onset of the international debt crisis (early 1980s): shifted up to 2.0 percent of GDP.
  - Since 2008: average program size about 5.8 percent of GDP.
  - Note: the post-2008 increase also reflects a shift from current to capital account crisis.
- Upper-tail growth:
  - Size at the 75th percentile increased from around 2 percent of GDP around 1990 to about 6.5 percent of GDP in 2015.
- Distributional dispersion:
  - 1960–79: 21 percent of all programs were larger than the average.
  - 2008–16: 34 percent of all programs were larger than the average.
- Graphical references in source: "Average Program Size in Percent of GDP"; "Distribution of Program Size in Percent of GDP" (25th percentile, Median, 75th percentile).

### Duration and maturity of Fund engagements
- Arrangement duration:
  - Prior to mid-1970s: rather constant at 12 months.
  - Began increasing in the mid-1970s.
  - Average expected program duration in 2016 stood at 32.6 months.
- Average maturity of GRA credit outstanding:
  - Followed a declining trend until around 2010 but has since increased sharply to over 6½ years.
- Time windows shown in source figures: "Average Maturity of GRA Credit Outstanding: 1995–2017 (in years)"; "Average GRA program duration, 1995–2016 (in years)".

### Modeling demand for financing in systemic scenarios
- Three-step approach to estimate potential financing needs:
  - First step — Pervasiveness of the crisis:
    - Members selected based on thresholds for crisis probabilities estimated in staff Vulnerability Exercises for AEs, EMs, and LICs.
    - Four systemic crisis scenarios considered using crisis-probability thresholds of more than 1, 3, 5, and 10 percent in a given year.
    - Example: for a member with annual crisis probability of 1 percent, unchanged policies over five years raise the probability to almost 5 percent.
  - Second step — Severity of the crisis:
    - Potential financing needs estimated based on impacts on (i) FDI inflows; (ii) rollover rates for short-term and medium-term debt; and for EMs only (iii) deposit outflows.
    - Crisis assumed to persist for two years (2017–18) with declining intensity.
    - Severity calibrated using empirical distribution of financial crises for EMs; simulations provided for crises broadly in the 90th, 85th, 75th, and 65th percentiles of the distribution.
    - Example (75th percentile for an EM): FDI declines by 20 percent in both crisis years; short-term debt rollover 80 percent first year and 85 percent second year; medium-term rollover 50 percent and 65 percent; deposit outflows of 5 percent in the first year.
  - Third step — Domestic adjustment (new relative to prior paper):
    - Model assumes a member undertakes fiscal adjustment of 1.4 percent of GDP (0.7 percent of GDP each year), consistent with average adjustment in members with Fund-supported programs during 2002–14.
    - The adjustment is assumed to translate into an external adjustment of the same magnitude.

- Crisis intensity and demand:
  - Demand for financing across considered scenarios ranges from SDR 186 to SDR 2,237 billion.
  - The estimates are somewhat lower than those presented in the March paper, mainly because the model now assumes member policy adjustment.

### Supply of non‑Fund financing and assumptions
- Non-Fund financing sources considered:
  - International reserves, Regional Financing Arrangements (RFAs), and active bilateral swap arrangements (BSAs).
  - Multilateral and regional development banks excluded from calculations.
  - Official bilateral borrowing excluded due to ad-hoc nature.
  - Private borrowing during crises implicitly included via debt roll-over assumptions.
- Reserve-use assumptions:
  - A member would not use reserves beyond 25 percent of its pre-crisis level (consistent with share used during the GFC).
  - EMs assumed to maintain reserves at 100 percent of the Fund’s metric for assessing reserve adequacy (ARA).
  - AEs and LICs assumed to maintain reserves at 100 percent of short-term debt.
  - In practice, central banks often reluctant to heavily use reserves and tend to rebuild them after initial losses.
- RFAs and BSAs:
  - RFAs incorporated: AMF, CRA, CMIM, EFSD, EU-BoP facility, ESM, FLAR (specific modality assumptions applied).
  - Uncertainty noted about RFA availability in systemic crises; regional risk pooling may be limited if crises affect entire regions; many RFAs remain untested.
  - Standing unlimited BSAs among major central banks included; RMB swaps largely excluded given uncertainty about activation in a global crisis.
- Joint coverage:
  - On average, use of international reserves, RFAs, and BSAs could jointly cover over one-third of global financial needs.

### Potential calls on Fund financing (demand minus supply)
- Calculation:
  - Potential calls on Fund financing are modeled mechanically as the difference between demand and supply; only a subset of members identified will have a financing gap requiring Fund assistance.
- Range of potential calls:
  - The potential call on Fund financing ranges from SDR 143 billion to SDR 1,391 billion across modeled matrices of pervasiveness and intensity.
- Comparison with prior work:
  - Estimates are lower than in the March 2016 paper, largely because the model now includes members’ policy response (assumed fiscal contraction of 1.4 percent of GDP).
  - The assumed fiscal contraction may not be plausible during a global crisis when global demand is weak; it is used for benchmarking and to capture domestic adjustment effects.
- Historical benchmark:
  - The Global Financial Crisis (GFC) could have been located in scenario “a” or “b” and the 85th to 90th percentile in the absence of large buffers and policy stimulus; with buffers then available, the GFC was located toward the lower right corner. With buffers more depleted now, a new crisis similar to the GFC could involve more and larger countries.

*Source: IMF staff (Box 2. Trends in GRA Arrangement Size and Credit Outstanding).*

### 29.      The model results are relatively robust to alternative assumptions (see Annex III):

### 29.      The model results are relatively robust to alternative assumptions (see Annex III):

### Robustness checks — key findings
- Self-insurance:
  - Assuming a crisis in the 75th intensity percentile, relaxing the assumptions on the maximum level of reserve use from 25 to 35 or 75 percent of total reserves could reduce potential calls on Fund financing by up to SDR 53 billion, depending on the pervasiveness of the crisis (Table AIII.4 panel 1).
  - Scenarios with a less pervasive crisis are not affected because members with financing needs hold already very low reserves.
  - Relaxing the assumption on ARA to 80 percent of the metric would lower potential calls on Fund financing by SDR 4 billion in the less pervasive crisis scenario and SDR 125 billion in the most pervasive adverse crisis scenario (Table AIII.4 panel 2).

- Fiscal adjustment:
  - Assuming a stronger domestic policy response—an adjustment of 2 percent of GDP over the two crisis years compared with 1.4 percent of GDP during recent adjustment programs—could substantially reduce potential calls on Fund financing, depending on the pervasiveness of the crisis (Table AIII.5 panel 2 for a crisis with an intensity at the 75th percentile).
  - The assumed adjustment of 1.4 percent of GDP would probably not be plausible (and advisable) during a pervasive global crisis scenario.
  - Assuming a weaker domestic policy response—an adjustment of 1 percent of GDP over the two crisis years—results in additional calls on Fund financing of up to SDR 79 billon (Table AIII.5 panel 1).

- Burden sharing between RFAs and the Fund:
  - A robustness check assuming equal burden sharing for illustrative purposes (Table AIII.6).
  - If RFAs cover 75 percent of a member’s net financing needs, potential calls on Fund financing would be somewhat lower (up to SDR 28 billion depending on crisis pervasiveness for a crisis with an intensity at the 75th percentile).
  - Assuming equal burden sharing between the Fund and RFAs, potential calls on Fund financing would increase by up to SDR 44 billion.
  - If RFAs can cover only 25 percent of the financing needs, potential calls on Fund financing would increase between SDR 5 and SDR 116 billion, depending on crisis pervasiveness.

### Global scenario model — financing gap by crisis intensity
- Crisis Intensity (percentile) — Financing gap (billions of SDRs)
  - a. Extremely pervasive global systemic crisis: 65th = 396; 75th = 627; 85th = 1,011; 90th = 1,391
  - b. Very pervasive systemic crisis: 65th = 393; 75th = 533; 85th = 791; 90th = 1,065
  - c. Pervasive systemic crisis: 65th = 310; 75th = 399; 85th = 606; 90th = 830
  - d. Systemic crisis: 65th = 143; 75th = 197; 85th = 298; 90th = 367

### Summary of quantitative approaches and range of estimates
- The quantitative approaches point to a broad range of estimates for the size of the Fund.
- For traditional metrics (Table 1): most indicators suggest Fund resources would have to rise from current levels to align ratios to historical reference points; for GDP, an increase in quotas (but not in total resources) would be needed to restore its ratio to the reference point.
- For the access-based approach and the global scenarios, the Fund’s current overall lending capacity could cover those scenarios presented in Tables 3 and 6 (indicated by the shading).
- The scenarios indicate calls on Fund financing from as low as SDR 140 billion to almost ten times this amount.
- Maintaining the Fund’s lending capacity in the absence of bilateral borrowing agreements would require a quota increase of about 70 percent.

### Pros and cons of the methodologies (high-level)
- Metric-based methodology:
  - Pros: refined over years; straightforward and easy to understand and replicate; relevant starting point.
  - Cons: requires judgment on reference levels and how structural changes affect benchmark levels.
- Access-based approach:
  - Pros: direct estimation of potential calls on Fund financing; expanded with econometric modeling; accounts for interconnectedness and foreign exchange risk.
  - Cons: results sensitive to choice of probability threshold.
- Global scenario analysis:
  - Pros: granular—size of the Fund determined by financing needs of each member under varying crisis pervasiveness and intensity.
  - Cons: model is a simple stress test with some stylized assumptions; crisis intensity based on historical data though crisis nature has evolved; model does not explicitly take into account contagion (implicitly assumes simultaneous access); maps individual member probabilities into a global crisis incidence with limitations under independence assumptions; implicit assumption that Fund financing is a residual source has shortcomings.

### Qualitative considerations affecting future demand for Fund resources
- Overarching point:
  - High interconnectedness is the overriding qualitative consideration likely to increase potential calls on Fund resources in future crises.
- Drivers and transitions increasing interconnectedness and uncertainty:
  - Structural changes: rise in cross-border trade and financial flows increases spillover and contagion risk.
  - Shifting center of global economic gravity: EMDCs’ share in global trade increased from 15 percent in 1990 to 36 percent in 2016; EMDCs are less financially integrated but the gap is closing.
  - Nature of crises evolving: shift from current account to capital account drivers; capital flows increased more than 25-fold between 1980 and 2007 versus an eight-fold expansion in global trade.
- New and ongoing transitions that could raise systemic risk:
  - Shocks of non-economic origin (refugee flows, global epidemics, demographic challenges).
  - Climate change increasing frequency, severity, and cost of natural disasters.
  - Adjustment to technological progress and global integration (labor market dislocations, need for policy support).
  - Digital disruptions (cybersecurity risks, rapid technological change, virtual currencies, peer-to-peer lending, high-frequency trading).
- Financial system considerations:
  - Global financial cycles are growing in amplitude and duration; increased interconnectedness can produce rapid propagation of shocks and correlated crises.
  - Post-crisis regulatory reforms have improved resilience but key aspects remain unfinished, and nonbank financial institutions are increasingly important yet largely unregulated.
- Increased vulnerabilities and reduced room for policy maneuver:
  - Medium-term growth prospects clouded by weak productivity growth and crisis legacies; rising global indebtedness limits policy adjustment scope.
  - Financial vulnerabilities remain elevated, with emerging market risks particularly through higher external financing risks and rising financial vulnerabilities in some members.

*Source: IMF staff*

### 41.      At the same time, room for policy maneuver has narrowed. Monetary policy continues

### 41. At the same time, room for policy maneuver has narrowed

### Narrowed policy maneuver and macroeconomic context
- Monetary policy continues to chart an accommodative course in many AEs, often relying on unconventional strategies to help raise inflation expectations and lower the real costs of borrowing for households and firms.
- With an increase in debt levels, fiscal space is also not available in all members to do more to close the output gap and to share the burden with monetary policy.
- EMs face challenges, including commodity exporters where the outlook for export prices remains subdued compared with the past.

### Interconnectedness and weakening external adjustment mechanisms
- In the recovery from the GFC, real exchange rate adjustments played a limited role in reducing external current account imbalances; imbalances and subsequent exchange rate movements were only loosely related, and trade flows showed a weak response to exchange rate movements.
- Factors contributing to a longer-term weakening of adjustment mechanisms include:
  - Capital flow volatility, resulting at times in sharp and sudden adjustments in exchange rates and domestic demand.
  - Larger cross-border liability positions, which can mute the impact of exchange rate movements on economic activity as contractionary effects via financial balance sheets can offset expansionary trade effects.
  - Changes in the structure of trade: the concentration of trade pricing in a small number of dominant currencies can inhibit external adjustment due to an asymmetric short-run response of trade flows to exchange rate movements. Specifically, a depreciation against the dominant currency may reduce imports on impact but not increase exports, as exporters maintain their prices in the dominant currency.
- The rising share globally in trade of EMDCs, where the dollar is more widely used in pricing, may exacerbate the asymmetric pricing effect.
- There is an open debate as to whether the relationship between real exchange rate adjustment and trade flows has been weakened by the build-up of global value chains.

### Global Financial Safety Net (GFSN): multilayering and implications
- The GFSN has expanded and become more multilayered since the GFC; the growing relative importance of BSAs and RFAs—some of which are untested—has led to a more decentralized and uncertain safety net.
- Inadequate predictability of some GFSN resources (in particular BSAs and RFAs) and the lack of reliable coverage for the full duration of shocks (most elements provide time-bound support) incentivize an over-accumulation of reserves.
  - This self-insurance is costly for members accumulating reserves and entails potential systemic costs and coordination problems that can undermine the resilience of the IMS and reduce global demand.
- The future development of the GFSN, including the size of the Fund, will need to take a holistic view of gaps in the system as well as the role of individual elements.
  - The Fund has a catalytic role given its experience in macroeconomic adjustment, lower susceptibility to political interference and universal risk pooling; successful mobilization of GFSN elements in a crisis requires effective collaboration across elements.
  - For the Fund to continue to play its catalytic role, it will need sufficient firepower of its own to maintain credibility.

### IMF–RFA collaboration (Box summary)
- Scope exists to improve IMF–RFA collaboration and the use of RFA resources; collaboration has been based on broad non-binding principles endorsed by G-20 leaders.
- Close collaboration could enhance the effectiveness of individual institutions and the GFSN, depending on mutual respect for institutional independence, recognition of mandates, technical expertise, and information sharing.
- An operational framework could: (i) increase effective firepower of the GFSN for large-scale crises; (ii) combine deeper regional knowledge of RFAs and global expertise of the Fund; and (iii) enhance country ownership.
- Possible forms of collaboration:
  - Collaboration in lending to strengthen timeliness, effectiveness and credibility; potential challenges include stigma associated with Fund programs, loss of RFA independence in joint lending, substitute nature of many RFAs, and tensions over conditionality.
  - Collaboration in surveillance to design sound policies and strengthen crisis prevention capacity; challenges include informational asymmetries and confidentiality assurances.
  - Collaboration in capacity development to improve RFA capabilities for program design, monitoring, and surveillance; currently no widespread engagement in this area.

### Changes in Fund policies since the GFC
- Surveillance framework: Bilateral and multilateral surveillance is now better integrated; analysis of spillovers has been deepened, macro-financial surveillance has been mainstreamed, and the institutional view on the liberalization and management of capital flows has been adopted. Greater attention has been put on the role of macroprudential policies.
- Risk management: The Fund has expanded its approach to risk management, including general surveillance products covering global and member-specific risks and specific risk-focused products (for instance, the VE).
- Lending toolkit: Since 2009, the Fund has substantially reformed its lending toolkit, including expanding the range of crisis prevention instruments. Staff continues to work on a proposal for a new liquidity backstop to provide renewable and reliable liquidity support against potential short-term moderate volatility of capital flows.
- Changes in debt restructuring rules: Reform of the Fund’s lending framework with regards to treatment of sovereign debt—including the removal of the systemic exemption—could reduce demand for Fund financing by triggering earlier action and aiming at bailing in more creditors; the extent of such effects is difficult to estimate ex ante.

### Role of Fund resources, credibility, and moral hazard
- An adequately resourced Fund can help foster confidence; IMFC and G20 communiqués reiterate commitment to maintaining a strong, quota-based, and adequately resourced IMF to preserve its role at the center of the GFSN.
- Anecdotal evidence suggests announcements to expand Fund resources during periods of stress have been associated with drops in uncertainty and easing of credit spreads.
- Empirical work shows that larger Fund resources have a favorable effect on the risk outlook and borrowing costs of emerging markets.
- The risk of moral hazard related to the size of Fund resources appears low; a strong framework limits risks through:
  - Fund program design with conditionality and financing to smooth adjustment while ensuring capacity to repay.
  - FCL and PLL addressing moral hazard through strict eligibility criteria related to fundamentals, economic policy frameworks, policies, and track record of policy implementation.
  - Coordination efforts to address possible creditor moral hazard (for instance, the ‘Vienna Initiative’).
  - Safeguards against superfluous lending, including requirement of a balance of payments need and Executive Board approval.

### Quantitative approaches, scenarios and judgments on Fund size
- Quantitative approaches provide a broad range of estimates for the size of the Fund, requiring judgment to weigh different approaches and results.
- Discussions on adequacy of Fund resources need to consider the state of the world by the mid-2020s because quota increases take time to become effective and will determine the Fund’s quota-based resources through the middle of the next decade.
- Illustration from the text: taking the “global scenario c” at the 75th percentile and applying the GDP deflator through 2021 as projected in the WEO (rather than 2018, the basis of the calculation in the scenarios) would result in an increase by 20 percent of potential calls on Fund financing.
- In a severe crisis scenario, the Fund’s lending capacity would decline as members under stress drop out of the Financial Transactions Plan (FTP), reducing usable quota resources and the FCC.
  - Access to borrowed resources would also decline if the affected member is a NAB participant or has a bilateral borrowing agreement with the Fund.
  - Example: under the “access-based scenario c,” the net reduction in the supply of usable quota resources would be about SDR 65 billion.
  - Furthermore, access to borrowed resources would be reduced by an additional SDR 60 billion. A further decline in the FCC could potentially be possible if the affected member(s) also purchases its reserve tranche.
- Models are based on historical data; increasingly longer program periods and duration are incorporated only to some extent.

### Considerations on the composition of Fund resources
- The Fund has a long history of supplementing primary resources—quotas—with standing borrowing facilities (for example, the GAB and the NAB), recognizing that it takes time to mobilize new financing when a crisis takes hold.
- Historically, quotas provided the bulk of resources but NAB/GAB provided an important supplement.
  - Before the GFC, the share of quotas in the total of quotas plus NAB/GAB was about 84 percent on average over the period 1978–2008.
- The share of borrowed resources increased markedly in the wake of the GFC and remains high by historical standards.
  - The share of the NAB and bilateral borrowing agreements peaked at around 73 percent of total resources in 2013–15, prior to the effectiveness of the 14th Review.
  - The doubling of quotas under the 14th Review in early 2016 and the corresponding rollback of the NAB reduced reliance on borrowing, although the share of borrowed resources in the Fund’s total resource envelope remains high at about 50 percent.

*Source: ppea2021008 (excerpts as provided).*

### 50.      The membership has stressed consistently that the Fund is and should remain a quota-

### The Size of the Fund

### Quotas, Borrowed Resources, and Governance
- "The membership has stressed consistently that the Fund is and should remain a quota-based institution."
- Quotas underpin the Fund’s finances, governance, and risk management framework and determine:
  - normal access to Fund resources and thresholds for exceptional access procedures,
  - application of surcharges and commitment fees,
  - proportional allocation of SDRs,
  - activation of post-program monitoring.
- Distinct implications of quotas versus borrowed resources:
  - Governance:
    - Voting power in Fund decisions is determined by quotas; borrowed resources do not affect quotas.
    - The NAB and the 2016 BBAs have their own governance frameworks; the governance framework for the BBAs was strengthened in 2016 to give more voice to participating creditors.
  - Permanent versus temporary nature:
    - Quotas provide permanent resources.
    - Borrowing has historically been temporary, often as a bridge to the next quota increase.
  - Participation:
    - General quota increases cover the entire membership.
    - Borrowed resources have been provided by a subset of the membership; overlap exists between FTP members and NAB participants.
  - Resource mobilization:
    - Quota increases generally take considerable time.
    - Borrowed resources have been raised more quickly at times (it took about one year to agree on the 2012 BBAs), but future mobilization is uncertain.
  - Resource availability:
    - Quota resources are accessible solely by Executive Board decisions.
    - NAB activation requires: (i) a threat to the stability of the IMS, and (ii) a majority of 85 percent or more of total credit commitments.
    - BBAs (2016) can be activated only if the “modified FCC” (including uncommitted NAB resources) is below SDR 100 billion and activation is approved by creditors representing 85 percent of total commitments under the 2016 BBAs.
  - Burden-sharing:
    - Use of quota resources increases the Fund’s burden-sharing capacity that protects cash flow against unpaid charges by members in arrears.
    - There is no burden-sharing adjustment for interest paid to creditors on borrowed resources; use of borrowed resources reduces burden-sharing capacity relative to credit outstanding.
  - Financial characteristics:
    - Both quota and borrowed resources are liquid reserve assets; members can encash to obtain freely usable currencies or SDRs on representation of a balance of payments need.
    - Members receive remuneration and interest payments calculated based on the SDR interest rate.

### Availability Ex‑Ante versus Mobilization Ex‑Post
- Having resource commitments available ex-ante can boost market confidence and prevent crises; contingent commitments entail limited financial costs if not called.
- Mobilizing resources ex-post can be costly; the Fund entered the GFC clearly under-resourced (see paragraph 3 cited in source).
- Increasing multipolarity and inward-looking policies may make future agreement on additional borrowed resources more difficult and time consuming.

### Quantitative and Qualitative Framework for Resource Adequacy
- The paper provides a two-pillar framework to assess adequacy of Fund resources:
  - Pillar 1: Quantitative analysis using three approaches to provide indicative ranges (traditional metrics, access-based approaches, global scenarios).
  - Pillar 2: Qualitative considerations on how the global economy and the IMS will evolve over the next decade.
- Updated analysis conclusions:
  - The updated analysis does not point to a case for revisiting staff’s earlier conclusion that the current lending capacity should be seen as a minimum.
  - Most quantitative approaches yield similar results to the March paper; exception: the global scenario approach now includes a domestic policy response that would not be plausible nor advisable to the full extent in a pervasive global crisis.
  - Qualitative considerations argue for a longer-term perspective as quota increases agreed as part of the 15th Review will likely determine quota-based resources through at least the middle of the next decade.
  - Greater interconnectedness increases risk of rapid, correlated crises; the GFSN would need to grow commensurately in size.
  - Staff highlights the advantages of having resources available ex-ante for boosting market confidence and crisis prevention.
- Follow-up work: staff will examine possible landing zones for the Fund’s size and quota resources based on Directors’ guidance.

### Questions for Directors (Issues for Discussion)
- Do Directors broadly agree with the two-pillar framework (quantitative and qualitative) to assess adequacy of Fund resources?
- Views on the quantitative pillar: merit in using the range of approaches (traditional metrics, access-based, global scenarios)? Are these ranges useful input for assessing size and composition of Fund resources?
- Views on the qualitative considerations: relative weights to assign given expectations on how the global economy and IMS will evolve? Do Directors support the view that qualitative considerations can be expected to increase potential calls on Fund resources in future crises?
- Views on composition of Fund resources:
  - Do Directors agree that the Fund should remain a quota-based institution?
  - Do Directors concur that a financing structure of quotas, backstopped by a standing borrowing facility such as the GAB/NAB, has served the Fund well over several decades?

### Key Historical and Numerical Indicators (selected table values preserved exactly as in source)
- Figure and table annotations:
  - NAB activation condition: majority of 85 percent or more of total credit commitments.
  - Activation condition for BBAs (2016): modified FCC below SDR 100 billion and approval by creditors representing 85 percent of total commitments.
- Table AI.1 — Selected entries (All values in billions of SDRs unless otherwise indicated; ratios in percent preserved exactly)
  - Agreed Quotas (selected columns): 61.1; 90.0; 135.2; 146.1; 212.0; 213.7; 217.6; 477.0; 477.0; 477.0; 477.0
  - Economic indicators (selected rows):
    - a. GDP (selected columns): 4,253; 7,588; 11,083; 15,744; 17,884; 22,442; 29,912; 35,906; 48,889; 50,900; 56,829
    - b. Current payments 5/ (selected columns): 718; 1,341; 2,168; 2,852; 3,700; 5,785; 8,026; 12,112; 16,941; 18,074; 23,210
    - c. Capital inflows to EMDCs 6/ (selected columns): 31; 50; 46; 49; 173; 233; 373; 689; 795; 851; 990
    - d. EFN (selected columns): 200; 269; 450; 552; 963; 1,834; 1,865; 2,000
  - Ratio of Quota resources to economic indicators (in percent, selected entries):
    - a. GDP: 1.4; 1.2; 1.2; 0.9; 1.2; 1.0; 0.7; 1.3; 1.0; 0.9; 0.8
    - b. Current payments 5/: 8.5; 6.7; 6.2; 5.1; 5.7; 3.7; 2.7; 3.9; 2.8; 2.6; 2.1
    - c. Capital inflows to EMDCs 6/: 195.1; 181.4; 293.9; 299.0; 122.9; 91.8; 58.4; 69.3; 60.0; 56.0; 48.2
    - d. EFN: n.a.; n.a.; n.a.; 73.2; 78.9; 47.4; 39.4; 49.6; 26.0; 25.6; 23.8
  - Additional Quota resources needed to restore quotas in 2016 relative to ratio calculated in column (in billions of SDR) — based on data through 2016 (selected entries):
    - a. GDP: 254; 127; 144; -126; 8; -199; 20
    - b. Current payments 5/: 1,061; 736; 650; 449; 559; 191; 13; 235; 32
    - c. Capital inflows to EMDCs 6/: 1,184; 1,067; 2,025; 2,069; 569; 305; 20; 113; 34
    - d. EFN: n.a.; n.a.; n.a.; 888; 994; 408; 258; 447; 8
- Table AI.2 — Selected entries (Total Fund resources = Quotas+NAB+BBA)
  - Total Fund resources (selected columns): 79.6; 108.5; 153.7; 164.6; 246.0; 247.7; 251.6; 659.4; 931.3; 918.7; 947.0
  - Ratio of Fund resources to economic indicators (in percent, selected entries):
    - a. GDP: 1.9; 1.4; 1.4; 1.0; 1.4; 1.1; 0.8; 1.8; 1.9; 1.8; 1.7
    - b. Current payments 6/: 11.1; 8.1; 7.1; 5.8; 6.6; 4.3; 3.1; 5.4; 5.5; 5.1; 4.1
    - c. Capital inflows to EMDCs 7/: 254.1; 218.7; 334.1; 336.9; 142.6; 106.4; 67.5; 95.7; 117.1; 107.9; 95.6
    - d. EFN: n.a.; n.a.; n.a.; 82.5; 91.5; 55.0; 45.5; 68.5; 50.8; 49.3; 47.3
  - Additional Fund resources needed to restore Fund resources in 2016 relative to ratio calculated in column (in billions of SDR) — based on data through 2016 (selected entries):
    - a. GDP: 34; 16; 51
    - b. Current payments 6/: 1,085; 544; 363; 124; 283; 65; 75
    - c. Capital inflows to EMDCs 7/: 1,245; 943; 1,926; 1,949; 295; 78
    - d. EFN: n.a.; n.a.; n.a.; 619; 789; 107; 359; 28

### Annex II — Modeling Fund Financing: Panel Logit Approach
- Purpose: estimate likelihood that a member may require Fund assistance using a binary response panel logit model covering 94 advanced, emerging, and frontier market economies over 1992–2014 (total of 105 GRA arrangements).
- Model interprets joint decision by a member to request a GRA arrangement and by the Fund to agree.
- Country-specific independent variables (lagged):
  - External financing needs (EFNs, in percent of GDP) as proxy for balance of payments need.
  - GDP growth.
  - GDP per capita.
  - Credit-to-GDP gap (deviation from trend).
  - 12-month variation of the bilateral nominal exchange rate versus the U.S. dollar.
  - Government stability (ICRG indicator).
  - Interconnectedness variable (weighted GDP growth of trading partners): px_i,t = Σ x_j,t φ_j,i,t (j≠i).
  - Dummy for access to an RFA; controls for size and past Fund engagement.
- Global variables:
  - U.S. interest rate variations.
  - VIX (proxy for risk aversion).
  - Oil price drops (sustained declines increase probabilities of Fund arrangements).
- Empirical outcomes:
  - All variables are statistically significant and have the expected sign.
  - Probability of a GRA arrangement increases with higher EFNs and larger credit-to-GDP gaps, and with drops in member or partner GDP growth.
- Robustness note:
  - Logit specification used; probit yields very similar results.
  - See IMF Lending in an Interconnected World, WP/17/155, Poulain and Reynaud for further details and robustness checks.

*Source: International Monetary Fund (Finance Department and staff analysis) — content unit ppea2021008.*

### 3.       The model is then used to generate probabilities for each member under a global

### 3.       The model is then used to generate probabilities for each member under a global

### Global volatility shock scenario
- The shock assumes that the VIX index would reach an average level of 30 during the year of the shock.
- This level is described as significantly below the average VIX level observed over the period 2008–09, but slightly above the level reached during other Fund lending cycles.
- It is assumed that all other variables remain unchanged; in practice, a global volatility shock would likely be accompanied by a drop in GDP growth and an increase in external financing needs, both increasing the probability that a member may require Fund financing, while required policy adjustment and stigma of approaching the Fund would work in the opposite direction.

### Estimation results (Dependent variable: Start of a GRA Arrangement (dummy))
- Estimation method: panel logit using random effects; marginal effects (dy/dx) and Robust SE reported; a constant is estimated but not reported.
- Key coefficients and significance:
  - Past program (dummy): dy/dx = 0.306; Robust SE = 0.092; significance = *** (1 percent)
  - External Financing Needs: dy/dx = 2.331; Robust SE = 0.916; significance = ** (5 percent)
  - GDP growth: dy/dx = -0.063; Robust SE = 0.026; significance = ** (5 percent)
  - GDP per capita: dy/dx = -0.866; Robust SE = 0.218; significance = *** (1 percent)
  - GDP: dy/dx = 0.001; Robust SE = 0.133; (not significant)
  - Credit gap: dy/dx = 0.025; Robust SE = 0.010; significance = *** (1 percent)
  - Exchange rate variation: dy/dx = -1.496; Robust SE = 0.813; significance = * (10 percent)
  - Government stability: dy/dx = -0.407; Robust SE = 0.091; significance = *** (1 percent)
  - Potential contagion: dy/dx = -0.793; Robust SE = 0.366; significance = ** (5 percent)
  - 3M US int. rate variation: dy/dx = 0.370; Robust SE = 0.161; significance = ** (5 percent)
  - VIX: dy/dx = 0.121; Robust SE = 0.034; significance = *** (1 percent)
  - Oil price: dy/dx = -0.023; Robust SE = 0.012; significance = * (10 percent)
  - Access to RFA (dummy): dy/dx = 0.189; Robust SE = 0.400; (not significant)
- Model fit and sample:
  - Pseudo R2 = 0.504
  - Observations = 1,597
  - Countries = 94
  - GRA Arrangements = 105
  - Likelihood ratio (p-value) = 0.000
- Notes: ***, **, and * denote significance at the 1, 5, and 10 percent levels, respectively.

### Threshold determination and goodness-of-fit
- Threshold selection: countries are flagged as having a new arrangement in a year if predicted probability > chosen threshold. Main approach minimizes loss function of Type I and Type II errors.
- Equal weighting (1:1) of Type I and Type II errors sets the threshold to 4.8 percent.
- Alternative weightings penalize false positives (Type II) more than missed programs (Type I):
  - A 2:1 ratio (Type II weighted twice Type I) yields threshold = 6.4 percent.
    - At 6.4 percent threshold: rate of false positives = 16 percent; rate of missed new programs = about 25 percent.
  - A 3:1 ratio yields threshold = 16.1 percent.
- Goodness-of-fit statistics (Table AII.2):
  - Weight Ratio 1:1 — Threshold 4.8% — Type I errors (missed new) 18.1% — Type II errors (false positives) 19.4%
  - Weight Ratio 2:1 — Threshold 6.4% — Type I errors 24.8% — Type II errors 16.0%
  - Weight Ratio 3:1 — Threshold 16.1% — Type I errors 49.5% — Type II errors 6.1%
- Definitions:
  - Rate of missed new programs (Type I errors) = number of actual new programs not predicted / number of new program observations in sample.
  - Rate of false positives (Type II errors) = number of programs predicted but not realized / number of non-program observations in sample.
- Comment: Penalizing false positives results in smaller estimates for the need for Fund resources compared with equal weights; characterized as a relatively conservative approach.

### Annex III — Global Scenarios: Assumptions and Methodology

A. Demand for Financing
- Framework: granular, bottom-up approach identifying countries vulnerable to crises, shocking individual countries based on past crises, accounting for policy responses, and estimating demand for financing.
- Step 1 — Pervasiveness: identify members facing a BOP shock using each country’s crisis probability threshold from the Fund’s vulnerability exercises. Four systemic shock scenarios defined by crisis-probability thresholds:
  - Scenario a (extreme): crisis probability threshold ≥ 1 percent — includes members with at least a one percent probability of crisis; described as the largest country group and a highly pervasive systemic crisis.
  - Scenario b: crisis probability threshold ≥ 3 percent.
  - Scenario c: crisis probability threshold ≥ 5 percent.
  - Scenario d (moderate): crisis probability threshold ≥ 10 percent — with no advanced economies facing such probability; affects primarily emerging and frontier economies.
- Note: the probability threshold for country selection is different from the likelihood of the scenario; the likelihood of the scenario depends on joint probabilities and cannot be inferred directly from univariate country probabilities.
- Table AIII.1 — Number of Members Included:
  - Scenario a (≥ 1 percent): AE = 27; EMDCs = 64
  - Scenario b (≥ 3 percent): AE = 27; EMDCs = 45
  - Scenario c (≥ 5 percent): AE = 26; EMDCs = 24
  - Scenario d (≥ 10 percent): AE = 0; EMDCs = 15

- Step 2 — Severity: simulate crisis impacts during 2017–18 calibrated on empirical distribution of historical crises in emerging market economies over past 30 years using kernel density estimators.
  - Crisis periods used: 1990, 2001, 2009, and 2011 (selected based on global financial stress index and domestic demand breaches).
  - Shock ranges aligned with the 90th, 85th, 75th, and 65th percentiles of the distribution; for advanced economies, somewhat less severe impacts assumed.
  - Shock smoothing: due to limited tail observations, shock values are smoothed and adjusted downward in absolute terms relative to raw data, producing more conservative resource needs than raw estimates.
- Step 3 — Policy response: assumes domestic adjustment of 0.7 percent of GDP each year (1.4 percent of GDP for entire period), consistent with average fiscal adjustment in Fund-supported programs during 2002–14.
- Financing need estimation per country combines:
  - Net external financing requirements = external financing needs (current account deficit and debt amortization) minus sources (FDI inflows and total new borrowing);
  - Plus additional financing needs to cover potential deposit outflows for emerging markets;
  - Minus domestic adjustment.

B. Financing Sources other than the IMF
- Self-insurance (use of reserves): two constraints:
  - Reserves should remain above 100 percent of the level suggested by the Fund’s ARA metric for emerging markets and above 100 percent of short-term debt for advanced economies and low-income countries.
  - Reserves are assumed not to fall by more than 25 percent relative to their initial level.
  - Countries can use reserves only until they drop by 25 percent or reach the relevant metric threshold.
- Role of Regional Financing Arrangements (RFAs):
  - Burden sharing assumptions based on 2000–16 historical data:
    - For ESM and EU-BoP, Fund financing expected to cover 28 percent and 61 percent of the financing gap after use of reserves, respectively.
    - For EFSD and ACF, Fund financing expected to cover 85 percent and 25 percent, respectively.
  - For BRICS CRA and CMIM (swap arrangements): members assumed to first tap unconditional part (30 percent) of access limit; remaining access assumed co-financed with Fund covering 1/3 and RFAs covering the rest; RFA coverage constrained by country-specific access limits and ability of members to provide financing.
  - Members assumed to use full access limit for the FLAR.
- Bilateral swap arrangements:
  - Standing unlimited BSAs among major central banks (Bank of Canada, Bank of England, Bank of Japan, ECB, Federal Reserve, Swiss National Bank) included but beneficiaries generally not expected to demand financing in simulations.
  - Most crisis-era swap lines between major central banks and EMDCs expired and are not considered.
  - Swap lines between China and many AEs and EMDCs remain active but are not counted due to uncertainties about scope of use in protracted needs.
  - Several small bilateral swap lines (< USD 20 billion each) are excluded due to small size and uncertainty about long-term availability.

- Table AIII.3 — Regional Financing Arrangements (selected entries):
  - AMF (Middle East & North Africa): No. of countries = 22; Size (SDR billion) = 2.7; Max access = 75-200% of paid subscription
  - BRICS: No. of countries = 5; Size (SDR billion) = 74.4; Max access = 50-100% of committed resources
  - CMIM (Southeast Asia): No. of countries = 15; Size (SDR billion) = 174; Max access = 50-500% of financial contribution
  - EFSD (Eurasian Union): No. of countries = 6; Size (SDR billion) = 6.3; Max access = In proportion to GNI per capita
  - EU-BoP (Non-euro zone): No. of countries = 10; Size (SDR billion) = - ; Max access = SDR 35.5 billion
  - ESM (Eurozone): No. of countries = 19; Size (SDR billion) = - ; Max access = SDR 296.3 billion
  - FLAR (Latin America): No. of countries = 8; Size (SDR billion) = 1.9; Max access = 100-250% of paid-in capital

C. Robustness Checks
- Use of reserves and ARA thresholds (simulations at 75th percentile of crisis intensity):
  - Options examined: use of reserves = 25, 35, and 75 percent of countries’ reserves; ARA metric thresholds = 80 and 100 percent.
  - Selected results (differences from baseline, in SDR billion):
    - Use of reserves (percent of total reserves) with ARA = 100 percent of metric and baseline = 75th percentile:
      - For a. Extremely pervasive global systemic crisis: 0 (25%), 45 (35%), 53 (75%)
      - For b. Very pervasive systemic crisis: 0, 9, 15
      - For c. Pervasive systemic crisis: 0, 0, 0
      - For d. Systemic crisis: 0, 0, 0
    - For ARA = 80 percent of metric:
      - For a. Extremely pervasive global systemic crisis: 50 (25%), 107 (35%), 125 (75%)
      - For b. Very pervasive systemic crisis: 15, 29, 38
      - For c. Pervasive systemic crisis: 4, 6, 6
      - For d. Systemic crisis: 4, 6, 6
- Fiscal adjustment robustness:
  - Baseline assumed domestic adjustment = 0.7 percent of GDP each year (1.4 percent total).
  - Additional checks: larger fiscal adjustment = 2 percent of GDP in total (1 percent per year); smaller fiscal adjustment = 1 percent of GDP in total (0.5 percent per year).
  - Table AIII.5 presents differences from baseline (in SDR billion) — positive numbers suggest less potential calls on Fund financing; negative numbers suggest more.
- Burden sharing robustness:
  - Table AIII.6 shows robustness checks for IMF share in co-financing with RFAs ranging from 25 to 75 percent of potential financing needs; simulations done for 75th percentile of crisis intensity.
  - Example entries (In SDR billion, difference from baseline):
    - IMF share in co-financing with the RFAs = 25/50/75:
      - a. Extremely pervasive global systemic crisis: 28, -40, -116
      - b. Very pervasive systemic crisis: 10, -44, -98
      - c. Pervasive systemic crisis: 9, -43, -95
      - d. Systemic crisis: 1, -2, -5

*Source: IMF staff estimates (content unit: ppea2021008).*

---


_Source: https://www.imf.org/-/media/files/publications/pp/2021/english/ppea2021008.pdf_