## wpiea2023187-print-pdf

## Source details

**Canonical URL:** [wpiea2023187-print-pdf](https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023187-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2023/english/wpiea2023187-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2023/english/wpiea2023187-print-pdf.pdf.json)

---

### 2.1 An overview of the DSSI and Common Framework — Design, timeline, participation, and interpretation
- Design and timeline of the DSSI:
  - Launched by the G20 on 15 April 2020 and entered into force on 1 May.
  - Temporary and “net present value (NPV)-neutral” suspension of debt service on claims owed to all official bilateral creditors; strictly speaking not debt relief.
  - Original scope: suspend debt service due between May and December 2020, to be repaid over three years after a one-year grace period.
  - Extension 1 (November 2020): covered debt service from January to June 2021, to be repaid over five years after a one-year grace period.
  - Final extension (April 2021): covered debt service through December 2021.
  - Formal expiry: end of December 2021.
  - Eligibility: open to all IDA countries and least developed countries (LDCs) with no arrears vis-à-vis the IMF or World Bank — in practice 73 eligible countries (72 IDA countries plus Angola).
  - Conditionality for beneficiaries:
    - Formal request to creditors.
    - Participation in, or at least a request for, an IMF financing arrangement (including IMF emergency facilities).
    - Commit to use freed-up resources to address the COVID-19 crisis subject to IMF–World Bank fiscal monitoring.
    - Disclose public debt composition.
    - Respect IMF and World Bank limits on contracting new non-concessional debt during the suspension period.
  - G20 called on private creditors to participate “on comparable terms”; private creditors ultimately did not provide debt service suspension.
  - MDBs were kept outside the DSSI perimeter and did not provide debt service relief.

- Implementation outcomes and key statistics:
  - 48 out of the 73 eligible countries made requests to take part in the DSSI.
    - Note: this number includes Uganda (signed a Paris Club deal but paid original debt service) and Malawi (appears to have requested but not received relief).
  - Estimated suspended debt service between May 2020 and December 2021: USD 8.9 billion.
    - USD 3.4 billion falling due in 2020.
    - USD 5.5 billion falling due in 2021 (World Bank, 2022).
  - Earlier projected total (for comparison): USD 12.9 billion; Brautigam and Huang (2023) report a higher total of USD 13.1 billion due to alternative inclusions (e.g., USD 4.1 billion savings for Angola from voluntary reprofiling by China Development Bank and ICBC).
  - Private creditor participation: largely absent; very few debtor countries requested private creditor participation (reported requests: Grenada, Chad, Zambia), without success.
  - Example of administrative burden: Pakistan had signed 93 deferral agreements with 21 creditors by June 2022.
  - Rating actions: some countries (e.g., Ethiopia, Pakistan, Cameroon, Senegal, Cote d’Ivoire) were placed under review by Moody’s when they requested DSSI support; these reviews concluded without downgrades once it was clear private creditors would not join.

- Common Framework — rationale, design, and progress:
  - Introduced in November 2020 as the Common Framework for Debt Treatments beyond the DSSI.
  - Purpose: facilitate timely and orderly debt treatments for DSSI-eligible countries, on a case-by-case basis and at the debtor’s request.
  - Debt treatment design and envelope determined based on parameters of a full-fledged IMF-supported programme including conditionality, an accompanying IMF–World Bank DSA, and a collective assessment by participating official bilateral creditors.
  - Creditor coordination via an ad hoc creditor committee.
  - Possible outcomes: range from short-term reprofiling to deep restructuring with large NPV reductions or nominal write-offs; term sheet restricts write-offs to “the most difficult cases” and requires due consideration of creditors’ domestic approval procedures.
  - Comparable treatment: debtors agreeing key parameters with participating official bilateral creditors are formally required to seek at least as favourable treatment from other official bilateral and private creditors; requirement is binding under the Common Framework.
  - MDBs: not expected to participate in Common Framework treatments but asked to explore options to help meet longer-term financing needs while protecting their credit ratings.
  - Common Framework case progress (as of writing):
    - Four requests received by G20 and Paris Club creditors: Chad (January 2021), Ethiopia (February 2021), Zambia (February 2021), Ghana (January 2023).
    - Chad: first participant to conclude a debt treatment (November 2022); limited reprofiling; official bilateral creditors and largest private creditor (Glencore) committed to reconvene if downside risks materialize (IMF, 2023).
    - Zambia: June 2023 agreement with official bilateral creditors expected to result in a substantial NPV reduction; exact interest rates and maturity extensions contingent on a future joint IMF–World Bank assessment.

- Benefits, costs, and participation incentives (summary):
  - Primary DSSI benefit: temporary debt service savings; value increases with the share of scheduled debt service owed to participating official bilateral creditors and where liquidity pressures are acute.
  - Common Framework potential benefits: broader resource availability through debt service relief, potential debt stock relief to address debt overhang, and restoring debt sustainability depending on treatment depth.
  - Perceived costs of participation:
    - Reputational costs, potentially involving credit rating downgrade and higher future borrowing costs.
    - Stigma from requesting IMF assistance (example: Vanuatu withdrew to avoid IMF financing).
    - Administrative costs related to reconciliation and new legal documentation across many creditors (example: Pakistan’s 93 deferral agreements).
    - Extra reporting obligations to IMF and World Bank.
  - Common Framework costs likely higher than DSSI costs due to deeper restructuring, larger rating impacts, and more complex negotiations.

- Real option interpretation of debtor participation choices:
  - Participation modelled as exercising a real (dividend-paying call) option: exercise (participate) to incur uncertain costs (I) for uncertain benefits (V), with option to wait (value C) to gather information.
  - Decision rule with waiting costs: exercise only when V − I ≥ C.
  - Key drivers and directional impacts on option value (C):
    - V (Present value of expected benefits): impact on C: positive.
    - I (Present value of expected costs): impact on C: negative.
    - σ (Volatility): impact on C: positive.
    - t (Time to maturity): impact on C: positive.
    - r (Risk-free rate): impact on C: positive.
    - q (Opportunity costs of holding out): impact on C: negative.
  - Comparative implications: reductions in expected costs or uncertainty lower the waiting value and prompt earlier participation; increased uncertainty or longer time-to-expiry increase the value of waiting and may delay participation.

### 2.3 A qualitative validation — anecdotal support and country cases
- Real-option framing supported by press statements; key value drivers in statements: expected benefits, expected costs (notably reputational costs), and uncertainty.

- DSSI (non-)participation illustrative country cases:
  - Benin (April 2020): concerns about reputational damage and perceived default risk from a moratorium.
  - Rwanda: preference for larger emergency financing over a debt moratorium; emphasis on stimulus regardless of mechanism.
  - Pakistan: early applicant intending to use savings for COVID-related purposes; would not request private-sector payment freeze.
  - Senegal: priority to maintain relationship with private creditors.
  - Kenya: initially ruled out participation due to potential credit rating downgrade; later joined DSSI second round in January 2021 after assurances.
  - Guinea-Bissau: switched from holding out in first round to participating in second round; possibly linked to new IMF–World Bank DSA showing increased risk of debt distress.

- Common Framework (non-)participation illustrative country cases:
  - Chad: first Common Framework requester; heavy external debt burden and oil-price shock; no public external debt or credit rating minimized reputational-cost concerns.
  - Zambia: had defaulted on multiple Eurobonds and deferred payments; committed to transparency and equal treatment of creditors.
  - Ethiopia: signed up to reduce debt vulnerabilities; downplayed private creditor involvement as “very unlikely”.
  - Observed effect: slow progress and uncertain prospects of early Common Framework cases deterred other eligible debtors; Mauritania cited need for clearer expectations; Ghana requested treatment in January 2023 after concerns about long delays.

- Survival analysis — objectives and event definition:
  - Event: request by an eligible debtor to participate in the original DSSI.
  - Objective: explain timing of requests using proxies for real option drivers — perceived benefits, costs, uncertainty, and opportunity costs of waiting.

- Sample construction and key sample statistics:
  - Initial eligible sample: 73 DSSI-eligible countries.
  - First-leg participation: 43 participated in the first leg (May–December 2020).
  - Exclusions: five small economies excluded for lack of debt service data: Kiribati, Marshall Islands, Micronesia, South Sudan and Tuvalu.
  - For 36 out of 43 DSSI participants the exact date of the request to the Paris Club was obtained.
  - Countries dropped for lack of timing information: Afghanistan, Burundi, Fiji, The Gambia, Tonga.
  - Censoring window: launch date 15 April 2020 and censored at 15 December 2020.
  - Highest pre-DSSI credit rating in sample: BB-/Ba3 (Bangladesh).

- Variables used as proxies for real-option drivers:
  - Benefits: (potential) debt service savings as a percentage of GDP for 2020 (World Bank).
  - Creditor-share variables from World Bank’s International Debt Statistics, 2021:
    - Shares of total external debt service to official bilateral creditors (excluding China) in 2020.
    - Shares to Chinese (official and non-official) creditor agencies in 2020.
    - Share to international bondholders in 2020.
  - Debt-distress dummy: 0 for low risk; 1 for moderate/high risk or in distress (pre-DSSI DSA).
  - Credit-rating dummy: 0 for unrated or rating below B; 1 for B-level rating.
  - IMF-arrangement dummy: 1 if country had COVID-19-related emergency financing or was involved in a regular IMF arrangement as of 1 May 2020.

- Methodology overview:
  - Kaplan-Meier estimator to plot failure functions F(t) = 1 − S(t); Log-rank tests for equality across groups.
  - Multivariate Cox proportional hazard models and parametric alternatives (Weibull, exponential, Gompertz).

- Descriptive survival-analysis findings:
  - Above-median external debt service savings → more likely and earlier requests.
  - Moderate/high risk of or in debt distress → quicker requests.
  - Pre-DSSI IMF involvement → associated with earlier requests.
  - Higher shares of debt service to official bilateral or Chinese creditors → quicker action.
  - Top-quartile shares to international bondholders or better pre-DSSI credit ratings → slightly more reluctance to request DSSI.
  - Log-rank tests: differences in key panels statistically significant at the 10% level or lower.

### 3.4 Estimation results and discussion — quantitative determinants of DSSI request timing
- Baseline: multivariate Cox proportional hazard model — key qualitative results:
  - Faster DSSI requests associated with:
    - higher debt service savings;
    - larger exposures to willing bilateral creditors (and to a lesser extent Chinese creditors);
    - having to request an IMF arrangement first was less of a hurdle.
  - Slower / less likely requests associated with:
    - larger exposures to bondholders;
    - better pre-DSSI credit ratings.
  - Model diagnostics:
    - Schoenfeld residuals: proportional hazards assumption not rejected.
    - Cox-Snell residuals: overall model fit acceptable.

- Baseline Cox estimation results (hazard ratios; robust standard errors in parentheses):
  - Column (1) — Cox (excludes Bhutan):
    - Debt service savings: 2.2377 (0.5956) ***
    - Risk of external debt distress: 2.1039 (1.1897) †
    - IMF arrangement: 2.3450 (0.8538) **
    - Number of countries: 60
    - Number of DSSI requests: 36
    - Log pseudo-likelihood: -126.6483
    - Akaike information criterion: 259.2965
    - Baysesian information criterion: 265.5796
  - Column (2) — Cox (excludes Bhutan and Haiti):
    - Debt service savings: 2.3002 (1.1222) *
    - Risk of external debt distress: 2.0798 (1.0950) †
    - IMF arrangement: 3.2423 (1.2699) ***
    - Share of bilateral debt service: 1.0301 (0.0162) *
    - Share of Chinese debt service: 1.0098 (0.0159)
    - Share of bonded debt service: 0.9744 (0.0138) *
    - Credit rating: 0.7119 (0.2405)
    - Number of countries: 59
    - Number of DSSI requests: 36
    - Log pseudo-likelihood: -120.2649
    - Akaike information criterion: 252.5297
    - Baysesian information criterion: 264.9949
  - Column (3) — Cox (excludes Bhutan and Haiti; alternative specification):
    - Debt service savings: 1.4224 (0.4988)
    - Risk of external debt distress: 2.2714 (1.1555) †
    - IMF arrangement: 3.0887 (1.2859) ***
    - Number of countries: 59
    - Number of DSSI requests: 36
    - Log pseudo-likelihood: -121.0731
    - Akaike information criterion: 254.1462
    - Baysesian information criterion: 266.6114
  - Significance markers: ***p<0.01, **p<0.05, *p<0.1, †p<0.2.

- Alternative estimators (Exponential, Weibull, Gompertz):
  - Parametric alternatives produce qualitatively similar hazard ratios, often with higher statistical significance.
  - Weibull shape parameter p not significantly different from 1.
  - Gompertz auxiliary parameter γ not significantly different from 0.
  - Selected hazard ratios (representative):
    - Exponential: Debt service savings up to 2.9993 (1.4566) **; IMF arrangement up to 4.0686 (1.7889) ***.
    - Weibull: Debt service savings up to 3.0720 (1.5218) **; IMF arrangement up to 4.1508 (2.0068) ***; shape p: 0.8657; 1.0263; 1.0233 (not significantly different from 1).
    - Gompertz: Debt service savings up to 2.1790 (0.8393) **; IMF arrangement up to 3.0234 (1.1921) ***; auxiliary γ values not significantly different from 0.

- Robustness: extended Cox specifications and additional variables:
  - IMF arrangement split:
    - Both emergency and regular arrangements significantly drive earlier DSSI requests.
    - Example (Column (2), Table 4): Emergency IMF arrangement: 2.8542 (0.9473) ***; Regular IMF arrangement: 2.6898 (0.9267) ***.
  - Macro-economic controls:
    - 2019 log GDP per capita: 0.6789 (0.1401) * — poorer countries quicker to request DSSI.
    - Forecasted real GDP growth: 1.1984 (0.0974) ** — higher expected growth associated with earlier requests.
    - Forecasted fiscal balance: 1.1952 (0.0600) *** — better expected fiscal balance associated with earlier requests.
    - Other macro variables (expected current account balance, inflation) not significant.
  - COVID-19 health indicators:
    - Government health expenditures (2019 % GDP): 0.7043 (0.0961) ** — countries that spent less on health were faster to sign up.
    - Cumulative confirmed COVID cases per million as of 30 April 2020 and the Oxford stringency index do not change baseline results (sample size issues).
  - Political economy variable:
    - Chief executive up for re-election between May and December 2020: no significant relation with DSSI requests.

- Robustness Cox results (selected columns from Table 4):
  - Column (1) baseline:
    - Debt service savings: 2.3002 (1.1222) *
    - Risk of external debt distress: 2.0798 (1.0950) †
    - IMF arrangement: 3.2423 (1.2699) ***
    - Share of bilateral debt service: 1.0301 (0.0162) *
    - Share of bonded debt service: 0.9744 (0.0138) *
    - Number of countries: 59; DSSI requests: 36; Log pseudo-likelihood: -120.2649
  - Column (3) adding GDP per capita:
    - Debt service savings: 2.5780 (1.2299) **
    - GDP per capita (log): 0.6789 (0.1401) *
    - Number of countries: 59; DSSI requests: 36; Log pseudo-likelihood: -118.5922
  - Column (6) adding government health expenditures:
    - Debt service savings: 2.7620 (1.4597) *
    - Government health expenditures: 0.7043 (0.0961) **
    - Share of bonded debt service: 0.9606 (0.0155) **
    - Number of countries: 56; DSSI requests: 35; Log pseudo-likelihood: -112.1060

- Notes on sample exclusions and outliers:
  - Bhutan excluded as outlier on debt service savings; Haiti excluded in many regressions as outlier on share of bilateral debt service; other exclusions noted per column (Guyana, Somalia, Kosovo, Yemen).

- Implications linked to policy levers:
  - Empirical evidence highlights drivers affecting timing and likelihood of DSSI requests:
    - Magnitude of expected debt service savings.
    - Exposure composition (bilateral, Chinese, bonded).
    - Presence and type of IMF arrangement.
    - Pre-DSSI economic conditions and forecasts.
    - Pre-existing government health spending.
  - These map to real option levers:
    - Lever 1 (V): Increase present value of expected benefits.
    - Lever 2 (I): Reduce present value of expected costs.
    - Lever 3 (σ): Reduce uncertainty.
    - Lever 4 (q): Increase opportunity costs of holding out.
    - Lever 6 (t): Narrow window of opportunity.
    - Lever 5 (r): risk-free rate treated as exogenous.

### 4.2 Policy interventions using non-NPV levers — reducing uncertainty and increasing opportunity costs
- Uncertainty:
  - Reducing uncertainty about expected benefits and costs is a key non-NPV lever to incentivise earlier participation.
  - Further clarification needed on debt treatment procedures and indicative timelines.
  - July 2022 Paris Club Secretariat compendium describes:
    - i) preconditions for informal and formal discussions with the Paris Club,
    - ii) parties expected to participate in Paris Club meetings,
    - iii) information-sharing requirements,
    - iv) indicative timelines for financing assurances to the IMF by Paris Club creditors, for signing bilateral agreements between debtor and Paris Club creditors, and for concluding negotiations with non-Paris Club official and private creditors.
  - The G20 has not agreed a similar set of guidelines for the Common Framework.
  - Policy options to reduce uncertainty:
    - Tighter (but not overly strict) deadlines for key steps.
    - Greater transparency on creditors’ domestic processes and approval procedures.
    - Improved debt transparency by debtors (with technical assistance if necessary) and creditors.
    - Early engagement by the IMF and World Bank on macroeconomic projections and DSAs.
    - Clarification of the perimeter of debt eligible for treatment (including treatment of non-government debt with government guarantees, arrears, syndicated loans, collateralisation, and domestic debt implications).
    - Continued dialogue with rating agencies and private creditor organisations; IMF positioned to act as intermediary or facilitator.
  - Comparability of treatment:
    - Assessment and enforcement of comparability is contentious and a key source of uncertainty.
    - Paris Club’s “broad-based approach” compares treatments across creditors based on changes in nominal debt service, NPV reductions, and extension of duration; some details (e.g., discount rate) remain unspecified.
    - Quantitative evidence shows Paris Club creditors on average faced larger NPV reductions than private creditors in restructurings (Schlegl et al., 2019).
    - Common Framework references same three comparability parameters but provides less detail than Paris Club guidelines.
  - Proposed approaches to comparability and enforcement:
    - World Bank proposal: standardised and transparent NPV formula using a single market-based discount rate; involve private creditors sooner.
    - Lazard (2022): single assessment metric — NPV measure using fixed 5% discount rate when debt is unsustainable; proportional contribution to financing envelope for temporary liquidity constraints; IMF to assess comparability.
    - Reichert-Facilides (2023): confirm NPV discount rate by majority vote across affected creditors.
    - Targeted legislative actions to incentivise private creditor participation: codify ‘good faith’ negotiations; immunise sovereign assets from attachment; limit recovery value for holdouts; retrofit collective action clauses.
    - Buchheit and Gulati (2023) and Lazard (2023): propose cross-creditor group ‘most favoured creditor’ clauses to help enforce comparability.
  - Forums and demonstration effects:
    - Global Sovereign Debt Roundtable (GSDR) launched February 2023 (co-chaired by IMF, World Bank and the G20 Presidency) to build common understanding and address shortcomings.
    - Demonstration effects: efficient handling and progress on ongoing cases reduce uncertainty and boost confidence for other debtors.

- Opportunity costs and time window:
  - Increasing benefits of an early request (e.g., a non-retroactive debt service standstill triggered by the request) would increase opportunity costs of holding out.
  - Narrowing the real option time window may be less relevant for the Common Framework (no sunset clause).
  - DSSI extension in April 2021 served as a time-window lever; G20 communicated final extension and expiry end-2021 to incentivise movement toward permanent solutions (IMF and World Bank, 2021).

- Key empirical and analytical findings (summary):
  - Decisions analysed through a real options framework: compare net benefits of participating now versus waiting for more information.
  - Press statements corroborate importance of expected benefits, expected costs, and uncertainty.
  - Survival analysis finds:
    - Countries with higher expected debt service savings, larger exposures to willing bilateral creditors (and to a lesser extent Chinese creditors), or pre-existing IMF arrangements were quicker to request DSSI support.
    - Debtors with larger exposures to bondholders or better pre-DSSI credit ratings were less likely to make early requests.

- Policy implications and recommendations:
  - NPV-related levers (increase expected benefits or reduce expected costs):
    - Temporary debt service standstills during restructuring negotiations.
    - Increased focus on post-treatment growth.
    - Protection of new financing.
    - Capacity building support.
  - Non-NPV levers (reduce option value of waiting and encourage earlier participation):
    - Clarify debt treatment procedures and timelines.
    - Clarify debt perimeter eligible for treatment.
    - Clarify methodology for assessing and enforcing comparability of treatment.
    - Clarify likely credit rating reactions and maintain dialogue with rating agencies.
    - Demonstrable progress on ongoing debt treatment cases to create positive demonstration effects.

_Italian line: Source: IMF working paper — sections 2.1, 2.3, 3.4 and 4.2 from "wpiea2023187-print-pdf"._

### 2.1    An overview of the DSSI and Common Framework

### 2.1    An overview of the DSSI and Common Framework

### Design and timeline of the DSSI
- Launched by the G20 on 15 April 2020 and entered into force on 1 May.
- Provided a temporary and “net present value (NPV)-neutral” suspension of debt service payments on claims owed to all official bilateral creditors (lenders would be fully repaid and receive interest on deferred sums); strictly speaking not debt relief.
- Original scope: suspend debt service due between May and December 2020, to be repaid over three years after a one-year grace period.
- Extension 1 (November 2020): covered debt service from January to June 2021, to be repaid over five years after a one-year grace period.
- Final extension (April 2021): covered debt service through December 2021.
- Formal expiry: end of December 2021.
- Eligibility: open to all IDA countries and least developed countries (LDCs) with no arrears vis-à-vis the IMF or World Bank — in practice 73 eligible countries (72 IDA countries plus Angola).
- Conditionality for beneficiaries:
  - Formal request to creditors.
  - Participation in, or at least a request for, an IMF financing arrangement (including IMF emergency facilities).
  - Commit to use freed-up resources to address the COVID-19 crisis subject to IMF–World Bank fiscal monitoring.
  - Disclose public debt composition.
  - Respect IMF and World Bank limits on contracting new non-concessional debt during the suspension period.
- G20 called on private creditors to participate “on comparable terms”; private creditors ultimately did not provide debt service suspension.
- MDBs were kept outside the DSSI perimeter and did not provide debt service relief.

### Implementation outcomes and key statistics
- 48 out of the 73 eligible countries made requests to take part in the DSSI.
  - Note: this number includes Uganda (signed a Paris Club deal but paid original debt service) and Malawi (appears to have requested but not received relief).
- Estimated suspended debt service between May 2020 and December 2021: USD 8.9 billion.
  - USD 3.4 billion falling due in 2020.
  - USD 5.5 billion falling due in 2021 (World Bank, 2022).
- Earlier projected total (for comparison): USD 12.9 billion; Brautigam and Huang (2023) report a higher total of USD 13.1 billion due to alternative inclusions (e.g., USD 4.1 billion savings for Angola from voluntary reprofiling by China Development Bank and ICBC).
- Private creditor participation: largely absent; very few debtor countries requested private creditor participation (reported requests: Grenada, Chad, Zambia), without success.
- Example of administrative burden: Pakistan had signed 93 deferral agreements with 21 creditors by June 2022.
- Rating actions: some countries (e.g., Ethiopia, Pakistan, Cameroon, Senegal, Cote d’Ivoire) were placed under review by Moody’s when they requested DSSI support; these reviews concluded without downgrades once it was clear private creditors would not join.

### Rationale, limits, and incentives of the Common Framework
- Introduced in November 2020 (with the DSSI’s first extension and Paris Club approval) as the Common Framework for Debt Treatments beyond the DSSI.
- Purpose: facilitate timely and orderly debt treatments for DSSI-eligible countries, on a case-by-case basis and at the debtor’s request.
- Debt treatment design and envelope: determined based on parameters of a full-fledged IMF-supported programme including conditionality, an accompanying IMF–World Bank DSA, and a collective assessment by participating official bilateral creditors.
- Creditor coordination: bilateral creditors coordinate and negotiate via an ad hoc creditor committee.
- Possible outcomes under the Common Framework: range from short-term reprofiling to deep restructuring with large NPV reductions or nominal write-offs, but the term sheet restricts write-offs to “the most difficult cases” and requires due consideration of creditors’ domestic approval procedures.
- Comparable treatment: debtor countries agreeing key parameters with participating official bilateral creditors are formally required to seek at least as favourable treatment from other official bilateral and private creditors; this requirement is binding under the Common Framework.
- MDBs: not expected to participate in Common Framework treatments but asked to explore options to help meet longer-term financing needs while protecting their credit ratings.

### Common Framework case progress (as of writing)
- Four requests received by G20 and Paris Club creditors:
  - Chad (January 2021)
  - Ethiopia (February 2021)
  - Zambia (February 2021)
  - Ghana (January 2023)
- Implementation status highlights:
  - Chad: first Common Framework participant to conclude a debt treatment (November 2022); involved a limited reprofiling of debt service; official bilateral creditors and largest private creditor (Glencore) committed to reconvene should downside risks materialize (IMF, 2023).
  - Zambia: in June 2023 Zambia and its official bilateral creditors reached agreement on a debt treatment expected to result in a substantial NPV reduction; exact interest rates and maturity extensions contingent on a future joint IMF–World Bank assessment of Zambia’s debt-carrying capacity.

### Benefits, costs, and participation incentives (summary)
- Primary DSSI benefit: temporary debt service savings; value increases with the share of scheduled debt service owed to participating official bilateral creditors and where liquidity pressures are acute.
- Common Framework potential benefits: broader resource availability through debt service relief, potential debt stock relief to address debt overhang, and restoring debt sustainability depending on treatment depth.
- Perceived costs of participation:
  - Reputational costs, potentially involving credit rating downgrade and higher future borrowing costs.
  - Stigma from requesting IMF assistance (example: Vanuatu withdrew to avoid IMF financing).
  - Administrative costs related to reconciliation and new legal documentation across many creditors (example: Pakistan’s 93 deferral agreements).
  - Extra reporting obligations to IMF and World Bank.
- Common Framework costs are likely higher than DSSI costs due to typically deeper restructuring, larger rating impacts, and more complex negotiations with an official creditor committee and private creditors.

### Real option interpretation of debtor participation choices
- Participation decisions can be modelled as exercising a real (dividend-paying call) option: a debtor exercises the option (participates) to incur uncertain costs (I) for uncertain benefits (V), with the option to wait (value C) to gather information.
- Optimal timing: uncertainty can justify waiting even when current net present value V−I is positive; the decision rule becomes exercise only when V−I ≥ C.
- Opportunity costs of waiting (q) are crucial: without positive q, decision-makers could wait indefinitely.
- Key option value drivers and directional impacts on option value (C):
  - V (Underlying asset price): Present value of expected DSSI/CF benefits — impact on C: positive.
  - I (Strike price): Present value of expected DSSI/CF costs — impact on C: negative.
  - σ (Volatility): Uncertainty about benefits and/or costs — impact on C: positive.
  - t (Time to maturity): Time left before DSSI expires — impact on C: positive.
  - r (Risk-free rate): Risk-free rate — impact on C: positive.
  - q (Dividend yield): Opportunity costs of holding out from DSSI/CF — impact on C: negative.
- Comparative implications:
  - Reductions in expected costs or uncertainty lower the waiting value and prompt earlier participation.
  - Increased uncertainty or longer time-to-expiry increase the value of waiting and may delay participation.
- Graphical intuition: NPV (V−I) and option value (C) curves determine thresholds for immediate exercise versus waiting; changes in parameters shift these curves and shift exercise points.

*Source: IMF working paper — section 2.1 "An overview of the DSSI and Common Framework" (wpiea2023187-print-pdf).*

### 2.3    A qualitative validation

### 2.3 A qualitative validation

### Real-option framing and anecdotal support
- The paper’s real option characterisation of debtor decisions on DSSI or Common Framework participation is supported by press statements from participating and non-participating countries.
- Key real option value drivers highlighted in statements: expected benefits, expected costs (notably reputational costs), and uncertainty.

### DSSI (non-)participation — illustrative country cases
- Benin (April 2020): Romuald Wadagni argued DSSI solutions would “further tarnish the reputation of our governments and jeopardize their access to future financing... A moratorium... could even be considered... as an event of default by private creditors... In addition to sanctions from rating agencies for defaulting on loan repayments, all our countries’ efforts to improve both the business climate and the perception of risk. . . would be simply destroyed”.
- Rwanda: President Paul Kagame preferred larger emergency financing over a debt moratorium: “We shouldn’t be looking for excuses to cancel debt for its own sake. If there is another idea that would achieve the same results, that is welcome. Stimulus is stimulus, no matter the mechanism”. Both Benin and Rwanda had relatively recent Eurobond access (first-time issuances in 2019 and 2013, respectively).
- Pakistan: Early applicant to DSSI intending to use savings for COVID-related purposes; Ministry of Finance emphasised it “would not request a payment freeze from the private sector”.
- Senegal: Minister Amadou Hott stated “there is no need to force any participation from private creditors... Our priority is to maintain our relationship with private creditors that are key long-term partners to bridge our financing gap”.
- Kenya: Initially ruled out participation due to potential credit rating downgrade; Treasury Secretary Ukur Yatani (May 2020): “We fear we might unnecessarily create a crisis... The G20 debt relief initiative does not offer optimal benefit given the structure of Kenya’s debt portfolio... Kenya is taking a cautious approach of seeking debt relief from bilateral creditors to safeguard its sovereign credit rating”. By November 2020 Yatani switched: “We have been reluctant in the past... But now after getting a bit of assurance that it is a matter that can be managed, we are now strongly considering joining the arrangement”. Kenya joined the DSSI for its second round in January 2021.
- Guinea-Bissau: Switched from holding out in the first DSSI round to participating in the second round; possibly related to a new IMF–World Bank DSA showing an increase in risk of debt distress from moderate to high.

### Common Framework (non-)participation — illustrative country cases
- Chad (January 2021): First country to request Common Framework support. Context: heavy external debt burden, high oil dependence, oil prices plunged. Chad had secured an IMF staff-level agreement on a new four-year programme and had not issued publicly traded external debt nor had a credit rating, limiting reputational-cost concerns. Much external debt was owed to Glencore; Chad completed HIPC in 2015.
- Zambia (February 2021): Had already defaulted on multiple Eurobonds and deferred payments under the DSSI and China Development Bank; Minister Bwalya Ng’andu: “Zambia is committed to transparency and equal treatment of all creditors in the restructuring process... Our application to benefit from the G20 Common Framework will hopefully reassure all creditors of our commitment to such treatment”.
- Ethiopia (February 2021): Signed up to reduce debt vulnerabilities and lower impact of debt distress. Ministry of Finance downplayed private creditor implications: “It would be a fair burden-sharing between all our official bilateral creditors and then, based on that, we will look at whether we need to reach out to private creditors, which is very unlikely”.
- Observed effect: slow progress and uncertain prospects of early Common Framework cases deterred other eligible debtors from applying. Mauritania’s Economy Minister Ousmane Mamoudou Kane (November 2021): “In principle, the prerequisites of the Common Framework are attractive... [but] we need to better understand what to expect from the framework and to see what the successful examples are”. Ghana eventually requested Common Framework treatment in January 2023 after concerns about long delays.

### Survival analysis — objectives and event definition
- Empirical strategy: survival (time-to-event) analysis where the event is the request by an eligible debtor to participate in the original DSSI.
- Objective: explain timing of requests based on proxies for real option value drivers — perceived benefits, perceived costs, uncertainty, and opportunity costs of waiting.

### Sample construction and key sample statistics
- Initial eligible sample: 73 DSSI-eligible countries.
- First-leg participation: 43 participated in the first leg of the DSSI (covering debt service due between May and December 2020).
- Exclusions: five small economies excluded (all DSSI non-participants) due to lack of detailed World Bank debt service data: Kiribati, Marshall Islands, Micronesia, South Sudan and Tuvalu. Note: South Sudan was already in debt distress; the other four were at high risk of distress.
- For 36 out of 43 DSSI participants the exact date of the request to the Paris Club was obtained.
- Countries dropped for lack of timing information: Afghanistan, Burundi, Fiji, The Gambia, Tonga — these participated according to World Bank data but did not file a request with the Paris Club; timing unknown. Uganda and Malawi participation status not entirely clear.
- Censoring window for the analysis: launch date set to 15 April 2020 and censored at 15 December 2020.
- Highest pre-DSSI credit rating in the sample: BB-/Ba3 (Bangladesh).

### Variables used as proxies for real-option drivers
- Benefits (DSSI temporary debt relief): (potential) debt service savings as a percentage of GDP for 2020 from the World Bank’s DSSI overview table (estimated as of April 2021). Higher savings should increase attractiveness of participation.
- Creditor-share variables (from World Bank’s International Debt Statistics, 2021 edition):
  - Shares of total external debt service projected to be paid to official bilateral creditors (excluding China) in 2020.
  - Shares of total external debt service projected to be paid to Chinese (official and non-official) creditor agencies in 2020.
  - Share of external debt service projected to be paid to international bondholders in 2020.
- Debt-distress dummy: 0 for eligible countries deemed at low risk of debt distress in the latest pre-DSSI IMF–World Bank DSA; 1 for countries at moderate/high risk of or already in distress.
- Credit-rating dummy: based on average of S&P, Moody’s and Fitch; value 0 for unrated or rating below B levels; value 1 for countries with a B-level rating.
- IMF-arrangement dummy: value 1 if country had already received COVID-19-related emergency financing (Rapid Credit Facility and/or Rapid Financing Instrument) or was involved in a regular IMF arrangement as of 1 May 2020; reflects cost/time/stigma of requesting an IMF arrangement (a formal DSSI participation requirement).

### Methodology overview
- Non-parametric Kaplan-Meier estimator used to plot Kaplan-Meier failure functions F(t) = 1 − S(t), where S(t) = product over event times of (n_j − d_j)/n_j; t_j denotes day of DSSI request for country j, d_j number of countries requesting at t_j, and n_j number at risk just prior to t_j.
- Log-rank tests used to test equality of survivor/failure functions across groups.
- Multivariate analysis: semi-parametric Cox proportional hazard models (hazard h(t) = h0(t) e^{Xβ}), estimated by maximum likelihood.
- Parametric alternative proportional hazard models estimated: Weibull (baseline hazard h0(t) = p t^{p−1}, p estimated), exponential (Weibull with p = 1, constant hazard), and Gompertz (h0(t) = e^{γ t}, γ estimated; γ = 0 collapses to exponential).

### Descriptive survival-analysis findings
- Kaplan-Meier failure functions indicate:
  - Countries with above-median external debt service savings were more likely, at almost any time, to request DSSI support than those with below-median savings.
  - Countries at moderate/high risk of or in debt distress were quicker to request support than those at low risk.
  - Pre-DSSI involvement in an IMF arrangement was associated with earlier requests; countries without such arrangements exhibited substantially longer delays.
  - Higher shares of external debt service going to official bilateral or Chinese creditors were associated with quicker action.
  - Countries with top-quartile shares of debt service to international bondholders or with relatively good (B-level) credit ratings were slightly more reluctant to request DSSI participation.
- Log-rank tests show differences in panels (a), (b) and (c) of Figure 3 are statistically significant at the 10% level or lower.

*Source: wpiea2023187-print-pdf — 2.3 A qualitative validation (IMF).*

### 3.4    Estimation results and discussion

### 3.4    Estimation results and discussion

### Baseline: multivariate Cox proportional hazard model — main findings
- Reported coefficients are exponentiated ('hazard ratios' / 'relative hazards'). Hazard ratios:
  - larger than one imply a higher hazard (shorter survival time / earlier DSSI request);
  - smaller than one imply a lower hazard (longer time before DSSI request).
- Key qualitative results (multivariate Cox broadly confirms Kaplan-Meier univariate findings):
  - Faster DSSI requests associated with:
    - higher debt service savings;
    - larger exposures to willing bilateral creditors (and to a lesser extent Chinese creditors);
    - having to request an IMF arrangement first was less of a hurdle.
  - Slower / less likely DSSI requests associated with:
    - larger exposures to bondholders;
    - better pre-DSSI credit ratings.
- Model diagnostics:
  - Schoenfeld residuals: the null hypothesis of proportional hazards cannot be rejected.
  - Cox-Snell residuals: overall model fit tests do not indicate problems.
- Baseline Cox estimation results (hazard ratios; robust standard errors in parentheses; significance codes from table):
  - Column (1) — Cox (excludes Bhutan):
    - Debt service savings: 2.2377 (0.5956) ***
    - Risk of external debt distress: 2.1039 (1.1897) †
    - IMF arrangement: 2.3450 (0.8538) **
    - Number of countries: 60
    - Number of DSSI requests: 36
    - Log pseudo-likelihood: -126.6483
    - Akaike information criterion: 259.2965
    - Baysesian information criterion: 265.5796
  - Column (2) — Cox (excludes Bhutan and Haiti):
    - Debt service savings: 2.3002 (1.1222) *
    - Risk of external debt distress: 2.0798 (1.0950) †
    - IMF arrangement: 3.2423 (1.2699) ***
    - Share of bilateral debt service: 1.0301 (0.0162) *
    - Share of Chinese debt service: 1.0098 (0.0159)
    - Share of bonded debt service: 0.9744 (0.0138) *
    - Credit rating: 0.7119 (0.2405)
    - Number of countries: 59
    - Number of DSSI requests: 36
    - Log pseudo-likelihood: -120.2649
    - Akaike information criterion: 252.5297
    - Baysesian information criterion: 264.9949
  - Column (3) — Cox (excludes Bhutan and Haiti; alternative specification):
    - Debt service savings: 1.4224 (0.4988)
    - Risk of external debt distress: 2.2714 (1.1555) †
    - IMF arrangement: 3.0887 (1.2859) ***
    - Number of countries: 59
    - Number of DSSI requests: 36
    - Log pseudo-likelihood: -121.0731
    - Akaike information criterion: 254.1462
    - Baysesian information criterion: 266.6114
- Notes on significance markers: ***p<0.01, **p<0.05, *p<0.1, †p<0.2.

### Alternative estimators: Exponential, Weibull and Gompertz proportional hazard models
- Parametric alternatives produce qualitatively similar hazard ratios to the baseline Cox model; often with higher statistical significance.
- Additional model diagnostics:
  - Weibull shape parameter p not significantly different from 1.
  - Gompertz auxiliary parameter γ not significantly different from 0.
  - These results explain similarity to the exponential model.
- Selected hazard ratios from Table 3 (columns grouped by model family; robust standard errors in parentheses):
  - Exponential (columns (1)-(3)):
    - Debt service savings: 2.8530 (0.8517) ***; 2.9993 (1.4566) **; 1.6946 (0.5873) †
    - Risk of external debt distress: 2.4478 (1.5586) †; 2.4310 (1.4155) †; 2.5330 (1.4225) *
    - IMF arrangement: 2.9091 (1.3404) **; 4.0686 (1.7889) ***; 3.8955 (1.8141) ***
  - Weibull (columns (4)-(6)):
    - Debt service savings: 2.6086 (0.7419) ***; 3.0720 (1.5218) **; 1.7150 (0.6106) †
    - Risk of external debt distress: 2.3637 (1.4865) †; 2.4515 (1.4640) †; 2.5466 (1.4562) †
    - IMF arrangement: 2.6795 (1.1885) **; 4.1508 (2.0068) ***; 3.9673 (1.9883) ***
    - Shape parameter p: 0.8657; 1.0263; 1.0233 (not significantly different from 1)
  - Gompertz (columns (7)-(9)):
    - Debt service savings: 2.0914 (0.4924) ***; 2.1790 (0.8393) **; 1.4054 (0.3981)
    - Risk of external debt distress: 2.0947 (1.1998) †; 2.1008 (1.1268) †; 2.2601 (1.1576) †
    - IMF arrangement: 2.1841 (0.8051) **; 3.0234 (1.1921) ***; 2.8732 (1.1829) **
    - Auxiliary parameter γ: -0.0117; -0.0088; -0.0090 (not significantly different from 0)
- Model sample and fit indicators (representative):
  - Number of countries: 60 or 59 depending on column.
  - Number of DSSI requests: 36 (for all reported columns).
  - Log pseudo-likelihood and information criteria provided per column in Table 3.

### Robustness: additional variables (extended Cox specifications)
- Strategy: introduce additional variables one-by-one to avoid overspecification given limited sample size.
- Key robustness checks and findings:
  - IMF arrangement split:
    - Column (2) of Table 4 splits IMF arrangement into emergency financing prior to DSSI and regular IMF arrangement at the time.
    - Both emergency and regular arrangements are equally significant drivers of early DSSI requests.
  - Macro-economic controls:
    - Column (3): 2019 log GDP per capita — poorer countries (lower GDP per capita) were quicker to make a DSSI request (GDP per capita (log): 0.6789 (0.1401) *).  
    - Columns (4) and (5): forecasted real GDP growth and forecasted fiscal balance for 2020:
      - Forecasted real GDP growth: 1.1984 (0.0974) **
      - Forecasted fiscal balance: 1.1952 (0.0600) ***
      - Counterintuitive finding: countries with higher expected growth and better expected fiscal balance for 2020 were more likely to participate early.
    - Other macro variables (expected current account balance, inflation) show no statistically significant relations with DSSI participation.
    - Controlling for macro conditions does not meaningfully alter baseline results.
  - COVID-19 health indicators:
    - Adding cumulative confirmed COVID cases per million as of 30 April 2020 (WHO) reduces sample size and does not change baseline results.
    - Introducing the Oxford COVID-19 government response stringency index does not change baseline results.
    - 2019 government health expenditures as a percentage of GDP:
      - Column (6): Government health expenditures: 0.7043 (0.0961) ** — countries that spent less on health were faster to sign up for the DSSI.
    - Limitations: pre-determined, time-invariant survival model not ideal for dynamic pandemic impacts; number of COVID deaths or people vaccinated not feasible to include due to data limitations.
  - Political economy variable:
    - Dummy for chief executive up for re-election between May and December 2020 shows no significant relation with DSSI requests when added to baseline; other estimates remain virtually unchanged.
- Robustness Cox estimation results (selected hazard ratios from Table 4; robust standard errors in parentheses):
  - Column (1) — reproduces baseline (corresponds to Table 2 column (2)):
    - Debt service savings: 2.3002 (1.1222) *
    - Risk of external debt distress: 2.0798 (1.0950) †
    - IMF arrangement: 3.2423 (1.2699) ***
    - Share of bilateral debt service: 1.0301 (0.0162) *
    - Share of Chinese debt service: 1.0098 (0.0159)
    - Share of bonded debt service: 0.9744 (0.0138) *
    - Number of countries: 59
    - Number of DSSI requests: 36
    - Log pseudo-likelihood: -120.2649
  - Column (2) — IMF arrangement split:
    - Debt service savings: 2.9895 (1.4732) **
    - Risk of external debt distress: 1.5424 (0.8698)
    - Emergency IMF arrangement: 2.8542 (0.9473) ***
    - Regular IMF arrangement: 2.6898 (0.9267) ***
    - Share of bonded debt service: 0.9727 (0.0125) **
    - Number of countries: 59
    - Number of DSSI requests: 36
    - Log pseudo-likelihood: -117.5598
  - Column (3) — adds 2019 GDP per capita (log):
    - Debt service savings: 2.5780 (1.2299) **
    - GDP per capita (log): 0.6789 (0.1401) *
    - Number of countries: 59
    - Number of DSSI requests: 36
    - Log pseudo-likelihood: -118.5922
  - Column (4) — adds forecasted real GDP growth:
    - Debt service savings: 2.9257 (1.4815) **
    - Forecasted real GDP growth: 1.1984 (0.0974) **
    - Share of bonded debt service: 0.9505 (0.0195) **
    - Number of countries: 58
    - Number of DSSI requests: 36
    - Log pseudo-likelihood: -116.2400
  - Column (5) — adds forecasted fiscal balance:
    - Debt service savings: 2.6552 (1.4682) *
    - Forecasted fiscal balance: 1.1952 (0.0600) ***
    - Share of bonded debt service: 0.9730 (0.0135) **
    - Number of countries: 58
    - Number of DSSI requests: 36
    - Log pseudo-likelihood: -114.4291
  - Column (6) — adds government health expenditures (2019 % GDP):
    - Debt service savings: 2.7620 (1.4597) *
    - Government health expenditures: 0.7043 (0.0961) **
    - Share of bonded debt service: 0.9606 (0.0155) **
    - Number of countries: 56
    - Number of DSSI requests: 35
    - Log pseudo-likelihood: -112.1060
- Notes on sample exclusions and outliers:
  - Sample excludes DSSI-eligible countries with missing debt service or timing information, and excludes Bhutan and Haiti as outliers in some specifications. Additional exclusions noted per column in Table 4 (e.g., Guyana, Somalia, Kosovo, Yemen) for data availability or outlier reasons.
- Additional estimation results referenced in text are available from the authors upon request.

### Implications signalled by estimation results (link to policy levers)
- Empirical evidence suggests that the following factors materially affect the timing and likelihood of DSSI requests:
  - Magnitude of expected debt service savings.
  - Exposure composition (bilateral, Chinese, bonded).
  - Presence and type of IMF arrangement (emergency vs. regular).
  - Pre-DSSI economic conditions (GDP per capita) and 2020 forecasts.
  - Pre-existing government health spending.
- These drivers map to the real option framework levers discussed subsequently in the chapter:
  - Lever 1 (V): Increase present value of expected benefits from debt treatment.
  - Lever 2 (I): Reduce present value of expected costs from debt treatment.
  - Lever 3 (σ): Reduce uncertainty about benefits and/or costs from debt treatment.
  - Lever 4 (q): Increase opportunity costs of holding out from debt treatment.
  - Lever 6 (t): Narrow window of opportunity for debt treatment request.
  - (Lever 5 r: risk-free interest rate considered exogenous.)

*Source: 3.4 Estimation results and discussion, from the supplied IMF content unit.*

### 4.2    Policy interventions using non-NPV levers

### 4.2    Policy interventions using non-NPV levers

### Uncertainty
- Reducing uncertainty about the expected benefits and costs of debt treatments is a key non-NPV lever to incentivise eligible debtors to come forward earlier.
- Further clarification is needed on debt treatment procedures and indicative timelines for process steps to anchor debtor countries’ expectations (Georgieva and Pazarbasioglu, 2021; Malpass, 2022).
- In July 2022 the Paris Club Secretariat published a compendium of guidelines for debtor countries considering a request for debt treatment with the Paris Club; the compendium describes:
  - i) the preconditions for entering into informal and formal discussions with the Paris Club,
  - ii) the various parties that are expected to participate in Paris Club meetings,
  - iii) the information-sharing requirements, and
  - iv) the indicative timelines for the provision of financing assurances to the IMF by Paris Club creditors, for signing bilateral agreements between the debtor and individual Paris Club creditors, and for the conclusion of negotiations with non-Paris Club official and private creditors.
- The G20 has not agreed a similar set of guidelines for the Common Framework.
- Policy options to reduce uncertainty include:
  - Tighter (but not overly strict) deadlines for key steps (formation of an official creditor committee, financing assurances, completion of a memorandum of understanding).
  - Greater transparency on (and adjustments to where needed) creditors’ domestic processes and approval procedures.
  - Improved debt transparency by both debtors (with technical assistance if necessary) and creditors, and early engagement by the IMF and World Bank on macroeconomic projections and DSAs.
  - Clarification of the perimeter of debt eligible for treatment, including treatment of non-government debt carrying government guarantees, arrears, syndicated loans, collateralisation, and domestic debt treatment implications for external creditor efforts.
  - Continued dialogue with rating agencies and private creditor organisations to shed light on potential costs of participation in terms of credit ratings and market access, with the IMF positioned to act as intermediary or facilitator of debtor-creditor (or debtor-rating agency) dialogue.
- Comparability of treatment:
  - Assessment and enforcement of comparability of treatment is highly contentious and a key source of uncertainty.
  - The Paris Club’s “broad-based approach” compares treatments across creditors based on changes in nominal debt service, the debt reduction in NPV terms, and the extension of the duration of claims; higher creditor efforts on one parameter may compensate for lower efforts on another.
  - Several details of the Paris Club methodology remain unspecified, including the appropriate discount rate for NPV calculations.
  - Quantitative evidence indicates that, on average, Paris Club creditors have faced significantly larger NPV reductions on their claims than private creditors (especially bondholders) in debt restructurings (Schlegl et al., 2019).
  - The G20’s Common Framework references the same three comparability parameters but provides less detail than the Paris Club guidelines, creating opaqueness about who will judge fulfilment of comparability.
- Proposed approaches to comparability and enforcement:
  - World Bank proposal to replace the Paris Club’s multi-factor approach with a standardised and transparent NPV formula using a single market-based discount rate, and to involve private creditors sooner in negotiations.
  - Lazard (2022) proposes a single assessment metric depending on the nature of debt problems: an NPV measure using a fixed 5% discount rate when debt is unsustainable, and a proportional contribution to the financing envelope in case of a temporary liquidity constraint. The IMF would assess whether comparability is satisfactory.
  - Reichert-Facilides (2023) suggests the NPV discount rate be confirmed by a majority vote across all affected creditors together with other restructuring terms.
  - Targeted legislative actions to incentivise private creditor participation discussed by Talero (2022) and Gill and Buchheit (2022) include codifying ‘good faith’ negotiations, immunising sovereign assets from attachment, limiting recovery value for holdout creditors, and retrofitting collective action clauses into existing contracts.
  - Buchheit and Gulati (2023) and Lazard (2023) propose cross-creditor group ‘most favoured creditor’ clauses in restructuring documentation to help enforce comparability across Paris Club bilaterals, non-Paris Club bilaterals, and private creditors.
- Forums and demonstration effects:
  - At the time of writing, debt treatment steps and timelines, debt perimeter, and comparability of treatment were being discussed in the Global Sovereign Debt Roundtable (GSDR), a new forum co-chaired by the IMF, World Bank and the G20 Presidency.
  - The GSDR was launched in February 2023 to build greater common understanding among key stakeholders and to find solutions to current shortcomings.
  - Demonstration effects matter: more efficient handling and actual progress on ongoing debt treatment cases would reduce uncertainty for other debtor candidates and boost confidence to apply.

### Opportunity costs and time window
- Increasing the benefits to the debtor of an early request—for example, by introducing a non-retroactive debt service standstill triggered by the request—would increase the opportunity costs of holding out.
- Narrowing the real option time window may be less relevant for the Common Framework’s case-by-case approach because:
  - There is no sunset clause in place for the Common Framework, and no clear reason to introduce one in the absence of a potential successor framework.
  - The DSSI extension in April 2021 served as a time-window lever: the G20 communicated that this would be the final extension and the DSSI would expire at the end of 2021, aiming to incentivise movement toward permanent solutions while providing flexibility and liquidity to countries coping with the COVID crisis (IMF and World Bank, 2021).

### Key empirical and analytical findings (from the paper’s analysis)
- Debtor countries’ decisions on whether and when to participate in sovereign debt relief initiatives can be analyzed through a real options framework: comparing net benefits of participating now versus waiting for later participation with better information.
- Press statements by debtor government officials corroborate the importance of expected benefits, expected costs, and uncertainty about benefits and costs in deliberations on DSSI and Common Framework participation.
- A survival analysis of eligible countries’ (non-)participation in the DSSI finds:
  - Countries that stood to benefit more from DSSI participation—due to higher expected debt service savings or larger exposures to willing bilateral creditors (and to a lesser extent Chinese creditors), or countries facing less of a hurdle in having to request an IMF arrangement first—were quicker to request DSSI support.
  - Conversely, debtors with more to lose in terms of market reputation—proxied by larger exposures to bondholders or better pre-DSSI credit ratings—were less likely to make an early DSSI request.

### Policy implications and recommendations
- NPV-related levers to increase expected benefits or reduce expected costs include:
  - Temporary debt service standstills during debt restructuring negotiations.
  - Increased focus on post-treatment growth.
  - Protection of new financing.
  - Capacity building support.
- Non-NPV levers that reduce the option value of waiting—thereby encouraging earlier participation—include:
  - Clarification of debt treatment procedures and timelines.
  - Clarification of the debt perimeter eligible for treatment.
  - Clarification of the methodology for assessing and enforcing comparability of treatment.
  - Clarification on likely credit rating reactions and continued dialogue with rating agencies.
  - Demonstrable progress on ongoing debt treatment cases to create positive demonstration effects and increase confidence among debtor countries.

*Source: wpiea2023187-print-pdf - 4.2    Policy interventions using non-NPV levers*

### References

### References

### Key literature cited
- Empirical and theoretical studies on sovereign debt restructuring, delays, and renegotiations:
  - Agnello, L., Castro, V., and Sousa, R. M. (2018); Agnello, L., Castro, V., and Sousa, R. M. (2021).
  - Asonuma, T. (2016); Asonuma, T. and Joo, H. (2020); Asonuma, T. and Trebesch, C. (2016).
  - Bai, Y. and Zhang, J. (2012); Benjamin, D. and Wright, M. L. J. (2013, 2019).
  - Bi, R. (2008); Bi, R., Chamon, M., and Zettelmeyer, J. (2016).
  - Graf von Luckner, C., Meyer, J., Reinhart, C. M., and Trebesch, C. (2021); Horn, S., Reinhart, C. M., and Trebesch, C. (2022).
  - Reinhart, C. M. and Trebesch, C. (2016).
- Policy analysis, multilateral frameworks, and practitioner commentaries:
  - IMF (2023); IMF and World Bank (2020, 2021).
  - Brautigam, D. and Huang, Y. (2023); Gill, I. (2022); Gill, I. and Buchheit, L. C. (2022).
  - Lazard (2022, 2023); Setser, B. W. (2023); Malpass, D. R. (2022).
- Technical and methodological references for option valuation and real options:
  - Black, F. and Scholes, M. (1973); Merton, R. C. (1973); Hull, J. C. (2012).
  - Dixit, A. K. and Pindyck, R. S. (1994); Trigeorgis, L. (1996); Damodaran, A. (2005); McDonald, R. and Siegel, D. (1986).
- Databases, working papers, and discussion papers used:
  - Chuku, C. et al. (2023) IMF Working Paper, 23/79.
  - Committeri, M. et al. (2021) ECB Occasional Paper, 262.
  - Cruz, C., Keefer, P., and Scartascini, C. (2021) Database of Political Institutions 2020.

### Appendix A — Comparative tables (high-level highlights)
- Table A.1: Comparison of main characteristics of DSSI vs. Common Framework
  - Timeline:
    - DSSI: May 2020 to December 2021 (incl. two extensions)
    - Common Framework: Since November 2020
  - Country eligibility:
    - DSSI: 73 IDA countries or LDCs with no multilateral arrears
    - Common Framework: 73 IDA countries or LDCs with no multilateral arrears
  - Debt treatment:
    - DSSI: Temporary NPV-neutral debt service suspension
    - Common Framework: Case-by-case: negotiated reprofiling, NPV reduction and/or debt write-off
  - Number of participating debtors:
    - 484 (as of July 2023)
- Table A.2: Comparison of main option value drivers for DSSI vs. Common Framework
  - Expected benefits:
    - DSSI: Temporary debt service savings, easing of liquidity pressures
    - Common Framework: Potentially (much) broader benefits, incl. debt service savings, overcoming of debt overhang, regaining of debt sustainability
  - Expected costs:
    - DSSI: (Limited) damage to financial market reputation
    - Common Framework: (Longer-lasting) damage to financial market reputation; Stigma from IMF financing request; Administrative costs from debt data reconciliation and changes in loan terms; Costs related to in-depth negotiations with different creditor groups
  - Opportunity costs:
    - DSSI: Foregone suspension of debt service while holding out
    - Common Framework: Potentially (much) larger, incl. missed investments in case of debt overhang
- Table A.3: Debtor country sample for survival analysis
  - Participants in first phase of DSSI (36) vs. Non-participants (25)
  - Notes:
    - The sample excludes DSSI-eligible countries with no information on debt service or on the timing of their DSSI requests.
    - * Bhutan is excluded from the regressions as an outlier on the debt service savings variable.
    - ** Haiti is excluded from most regressions as an outlier on the share of bilateral debt service variable.

### Appendix B — Option valuation model (conceptual framework and definitions)
- Purpose:
  - Present a conceptual real option framework for the value of requesting debt treatment that goes beyond conventional NPV analysis; not to produce a specific quantitative estimate.
- Rationale for model choice:
  - Black-Scholes model (Black and Scholes, 1973; Merton, 1973) provides a closed-form expression and is useful as a floor estimate for American-style real options, per Damodaran (2005).
  - Classic Black-Scholes assumptions: frictionless markets, continuous trading, constant risk-free rate, asset price follows geometric Brownian motion with volatility parameter σ.
  - Limitations: Black-Scholes prices European options; American options (allowing early exercise) are generally more valuable. Classic Black-Scholes does not allow for dividend payments (opportunity costs) without modification.
  - Note: Bjerksund and Sensland (2002) derive a closed-form approximation for American call options, but it is mathematically more complex.

- Dividend-augmented Black-Scholes call option formula (applied to real option case):
  - C = V e^(−q t) N(d1) − I e^(−r t) N(d2)
  - d1 = [ln(V / I) + (r − q + σ^2 / 2) t] / (σ √t)
  - d2 = [ln(V / I) + (r − q − σ^2 / 2) t] / (σ √t) = d1 − σ √t
  - Variable definitions (as applied here):
    - V: present value of debt treatment benefits
    - I: present value of the costs associated with the debt treatment
    - σ: unpredictability of the debt treatment benefits, measured by the standard deviation of the growth rate of those benefits
    - t: time left until the option of requesting the debt treatment expires
    - r: continuous risk-free rate
    - q: opportunity costs of holding out from the debt treatment
    - N(.): cumulative density function of the standard normal distribution
  - Usage: This formula is used to draw the illustrative option value curves in Figures 1 and 2 in the main text.

### Table B.4 — Option Greeks (sensitivity of option price C to parameters)
- Delta (∆ = ∂C/∂V):
  - Formula and sign: e^(−q t) N(d1) > 0
- Chi (χ = ∂C/∂I):
  - Formula and sign: −e^(−r t) N(d2) < 0
- Vega (V = ∂C/∂σ):
  - Formula and sign: V e^(−q t) √t N′(d1) > 0
- Theta (Θ = −∂C/∂t):
  - Formula and sign: q V e^(−q t) N(d1) − V e^(−q t) σ^2 √t N′(d1) − r I e^(−r t) N(d2) < 0
- Rho (P = ∂C/∂r):
  - Formula and sign: t I e^(−r t) N(d2) > 0
- Psi (Ψ = ∂C/∂q):
  - Formula and sign: −t V e^(−q t) N(d1) < 0
- Note: N′(d) is the derivative of N(d) with respect to d, i.e. the standard normal probability density function.

*Source: wpiea2023187-print-pdf - References.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023187-print-pdf.pdf_