## 1. Interrupted Time Series Analysis

## Source details

**Canonical URL:** [1. Interrupted Time Series Analysis](https://www.imf.org/-/media/files/publications/wp/2026/english/wpiea2026077-source-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2026/english/wpiea2026077-source-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2026/english/wpiea2026077-source-pdf.pdf.json)

---

### Methodology
- Regression specification:
  - Y_t = β0 + β1 Time_t + β2 Treatment_t + β3 Post_t + ε_t
  - Y_t denotes the monthly average lead time.
  - Time_t is a time trend.
  - Treatment_t is a binary variable equal to 1 after SUPRES adoption and 0 otherwise.
  - Post_t = 0 before SUPRES and counts the number of months since implementation.
  - Interpretation:
    - β1 captures the pre-SUPRES time trend.
    - β2 estimates the immediate change in level after SUPRES adoption.
    - β3 reflects the change in trend between pre- and post-SUPRES periods.
- Data handling:
  - Cases where lead days are negative are excluded (these occur when IMAS used its own resources to pay benefits because TSA disbursements were late).

### Main results: lead times and variability
- Aggregate findings:
  - Avancemos: lead time decreased from approximately 9.3 days to just over 2 days.
  - Prestación alimentaria / Atención a familias (DESAF): lead time fell from around 13 days to under 3 days on average.
  - The range between minimum and maximum monthly lead times narrowed after SUPRES, indicating improved predictability.
- Interpretation:
  - Improved predictability facilitates more accurate cash forecasting and reduces the need for precautionary liquidity buffers.
  - Regression results confirm statistically significant reductions in lead time for both programs.

### Regression estimates (Table 3)
- Dependent variable = Lead time (in days)
- Avancemos:
  - Time: 0.002 (0.035)
  - Treatment: -6.921*** (1.455)
  - Post: -0.013 (0.048)
  - Constant: 9.294*** (0.844)
  - Observations: 84
  - R-squared: 0.495
- Atención a Familias / Prestación Alimentaria (DESAF):
  - Time: -0.015 (0.034)
  - Treatment: -8.246*** (1.342)
  - Post: -0.056** (0.056)
  - Constant: 12.912*** (1.369)
  - Observations: 84
  - R-squared: 0.431
- Notes:
  - Robust standard errors in parentheses.
  - *** p<0.01, ** p<0.05, * p<0.1

### Program-specific dynamics and interpretation
- Avancemos:
  - No significant time trend before or after SUPRES; the 7-day reduction (from 9.3 to just over 2 days) appears as a one-time shift consistent with a “big bang” migration of all beneficiaries onto SUPRES.
  - Mechanism: prior multi-step payment chain caused staged Treasury releases; SUPRES enabled validation of payment files directly in SUPRES and direct TSA-to-beneficiary execution, keeping failed payments in the Treasury account and aligning outflows with actual payments.
- DESAF:
  - ITS estimates an immediate reduction of 8.25 days in lead times upon adoption.
  - Post-adoption time trend is significantly negative, indicating continued gradual reductions after the initial drop and a total realized improvement of approximately 9 days.
  - A temporary increase in lead time immediately before DESAF integration (July 2023) was driven by BNCR processing delays due to a transaction surge; issue was subsequently resolved.

### Disaggregation by transaction size
- Pre-SUPRES pattern:
  - Reverse U-shaped relationship between transaction size and lead time: mid-range payments (around ₡500–₡2,000) experienced longer lead times; very small and very large transfers cleared more quickly.
- Post-SUPRES pattern:
  - Automation and direct integration standardized workflows: lead times across transaction-size bins converged to approximately two to three days.
  - The size–time curvature flattened markedly, indicating structural process improvements rather than mere compositional shifts in transaction sizes.
- Complementary reforms:
  - SINIRUBE reforms were introduced during the same period but focus on beneficiary identification; SINIRUBE has not yet achieved full interoperability with SUPRES, so reductions in lead time are attributed to SUPRES.

### Fiscal implications and opportunity-cost quantification
- Estimated cumulative opportunity cost of idle cash (cost to the National Treasury of not being able to utilize funds during lead time):
  - Avancemos (January 2018 to August 2022): ₡ 828 million (approximately USD 1.5 million).
  - Atención a Familias / Prestación Alimentaria (January 2018 to July 2023): ₡ 1,410 million (approximately UDS 2.8 million).
  - Total: approximately USD 4.3 million.
- Implementation costs:
  - Total implementation costs totalize approximately USD 600,000.
  - Even under conservative assumptions, annual benefits exceed estimated one‑off and transitional costs.
- Operational indicators after SUPRES:
  - Average monthly transaction amount of each social program decreased by one-third.
  - Number of monthly transactions increased by 300%.
- Illustrative social trade-offs:
  - Avancemos caseload: approximately 421,000 beneficiaries per year; estimated 80 percent are upper‑secondary students.
  - Equivalent annual savings from ₡ 828 million (2018–Aug‑2022) map to roughly ₡ 177 million per full year.
  - Program benefit levels: ₡40,000 per month for upper secondary student; ₡18,000 per month for lower school student.
  - These savings correspond to approximately 370–820 additional full‑year beneficiaries each year during 2018–2021.
- Caveats:
  - Estimates may overstate efficiency gains because they do not account for Treasury’s real-time borrowing needs or actual government borrowing/debt issuance activities. A more conservative estimate, adjusted for fiscal balance and debt issuance, is provided in Annex I and still confirms materiality of savings.

---

### Digital Treasury Reform and Fiscal Efficiency

### Fiscal impacts and opportunity cost savings
- Estimated opportunity cost savings for the Treasury during 2018–2024: exceeding USD 4.3 million.
- Reform total implementation cost: about USD 600,000.
- Pre‑SUPRES opportunity cost in 2020 corresponded to roughly 1.1 percent of total domestic short‑term interest payments (zero‑coupon bonds + T‑bills).
- Annual interest paid on T‑bills: ₡4.9 billion in 2019, ₡782 million in 2020.
- SUPRES reduced average lead times for key social programs (Avancemos, Atencion a Familias, Prestacion Alimentaria) from 9-13 days to 2-3 days.

### Yearly fiscal and SUPRES metrics (selected)
- 2018: SUPRES Total Opportunity Cost (₡ mn) 279.5; Zero Coupon Bonds & T-bill Interest - Domestic (₡ mn) 70,879.0; Overall Fiscal Balance (% of GDP) -5.6; Gross Public Debt (% of GDP) 52%
- 2019: SUPRES Total Opportunity Cost (₡ mn) 584.8; Zero Coupon Bonds & T-bill Interest - Domestic (₡ mn) 61,479.4; Overall Fiscal Balance (% of GDP) -6.6; Gross Public Debt (% of GDP) 56%
- 2020: SUPRES Total Opportunity Cost (₡ mn) 323.6; Zero Coupon Bonds & T-bill Interest - Domestic (₡ mn) 29,721.8; Overall Fiscal Balance (% of GDP) -8.3; Gross Public Debt (% of GDP) 66%
- 2021: SUPRES Total Opportunity Cost (₡ mn) 339.7; Zero Coupon Bonds & T-bill Interest - Domestic (₡ mn) 15,538.0; Overall Fiscal Balance (% of GDP) -5; Gross Public Debt (% of GDP) 67%
- 2022: SUPRES Total Opportunity Cost (₡ mn) 623.7; Zero Coupon Bonds & T-bill Interest - Domestic (₡ mn) 18,516.4; Overall Fiscal Balance (% of GDP) -2.7; Gross Public Debt (% of GDP) 61%
- 2023: SUPRES Total Opportunity Cost (₡ mn) 184.4; Zero Coupon Bonds & T-bill Interest - Domestic (₡ mn) 25,055.7; Overall Fiscal Balance (% of GDP) -3.2; Gross Public Debt (% of GDP) 60%

### Coverage limitations and potential scale gains
- Current SUPRES coverage in 2023: platform handled payments for approximately 439,188 beneficiaries.
  - IMAS beneficiaries on SUPRES: about 421,002 (roughly 80 percent of IMAS’s beneficiaries).
- IMAS social programs: 0.44 percent of GDP; the analyzed programs represent approximately 0.33 percent of GDP (programs covering 80% of IMAS expenditure).
- Total social spending: 1.7% of GDP; overall government expenditure: approximately 19% of GDP.
- Major social programs not yet on SUPRES and their shares:
  - CCSS non-contributory pensions: 0.43 percent of GDP; 146,905 beneficiaries.
  - Ministry of Education social programs: 0.35 percent of GDP; ~981,000 beneficiaries.
  - BANHVI housing subsidies: 0.20 percent of GDP.
- Combined, these non‑migrated programs represent nearly 1 percent of GDP in social expenditures not yet benefiting from SUPRES efficiencies.
- Implication: expanding SUPRES to include these programs would significantly amplify fiscal and treasury operational gains.

### Administrative and operational efficiency — survey evidence
- Survey administered online between February and March 2025 to staff processing social assistance payments using SUPRES; 56 respondents covering key departments and most IMAS regional offices.
- Key survey findings:
  - 65% reported reduction in time required to process social benefit payments.
  - 75% responded positively regarding SUPRES contribution to their efficiency by automating payment processes.
  - 64% reported reduced administrative errors and improved accuracy in data.
  - 70% agreed SUPRES enhanced transparency and traceability in payment processing while reducing system fragmentation.
  - 73% believed SUPRES had already contributed positively to improving targeting of social programs.
  - 77% confident SUPRES has potential to continue to drive improvements.
- Operational trade-offs and challenges identified:
  - SUPRES overnight validation can extend some processes from one day to up to two days — trade-off between auditability/control and operational flexibility.
  - Reported issues: added technical complexity, slow/inefficient technical support, opacity of certain operational information (e.g., contingency claims spread across multiple unreconciled files).
  - Common recommendations from users: include all social benefits in SUPRES, step up system capabilities to collect/manage large datasets, introduce immediate transfer functionalities for emergencies, strengthen IT stability, and promote beneficiary participation.

### Financial inclusion, payment predictability, and system resilience
- Pre-SUPRES: IMAS opened simplified accounts for beneficiaries at Banco Nacional de Costa Rica (BNCR), limiting beneficiary choice and sometimes creating transaction costs when beneficiaries moved funds to other banks.
- SUPRES using the central bank’s SINPE platform: beneficiaries can receive transfers at any financial institution of their choice.
- Measured changes (2023–2024):
  - Total number of social benefit transactions increased by 3% between 2023 and 2024.
  - Number of transactions directed to bank accounts outside BNCR rose by 122% (number of transactions executed; not number of beneficiaries or monetary value).
- Benefits cited:
  - Greater beneficiary autonomy and financial empowerment.
  - Diversification across banks reduces concentration and operational risks, fosters competition among financial institutions.
  - Multi‑bank payment model strengthens resilience of social transfer delivery and aligns with international best practices.
- Predictability:
  - SUPRES validates payment information before execution and channels transfers directly from the TSA through the national payment system, reducing administrative delays and narrowing the range of payment dates within each cycle.
  - More predictable timing plausibly eases short‑term liquidity stress for beneficiaries (direct measurement not possible due to data limitations).

### Institutional enablers, IMF role, and implementation lessons
- Key enablers of SUPRES:
  - Functioning Treasury Single Account (TSA), a national payments infrastructure (SINPE), and an integrated beneficiary registry (SINIRUBE) enabling automated validation.
  - Legal underpinnings: service-level agreements and a presidential decree helped institutionalize the platform.
- IMF contribution:
  - Technical assistance and collaboration under the Extended Fund Facility (EFF) program.
  - Support included a PFM-focused hackathon, roadmap development, and inclusion of SUPRES as a structural benchmark under the EFF program.
- Implementation lessons and future priorities:
  - Expand SUPRES coverage to remaining large social programs to amplify fiscal and operational gains.
  - Continue improving legal framework to support platform expansion.
  - Enhance interoperability, particularly with SINIRUBE, for integrated and data-driven fiscal operations and better targeting.
  - Invest in system stability, long‑term operations, maintenance, cybersecurity, user training, and beneficiary engagement to preserve sustainability and enable rapid scaling in emergencies.
  - Establish clearer baselines and monitoring frameworks to collect granular data needed to assess specific impacts (e.g., financial inclusion outcomes or reductions in bank commission fees).

---

### Annex I. Opportunity Cost Estimates Robustness

### Baseline methodology
- Baseline approach:
  - Applied the weighted average interest rate of outstanding public debt, as reported by the Ministry of Finance, to the daily cash balance that could have been retained longer in the TSA due to reduced lead times.
  - Captures a broad fiscal benchmark for the cost of borrowing.

### Alternative estimation approaches
- Conditional Estimate:
  - Applies the same weighted average interest rate but only for months in which the government faced a fiscal deficit.
- Short-Term Debt Benchmark:
  - Matches monthly lead time of disbursement with the maturity structure of short-term debt issuance and applies the interest rate of short-term instruments (e.g., Letras del Tesoro) to estimated idle balances.
  - Data limitation: short-term debt issuance data are available only up to August 2023.

### Key robustness findings
- Consistency across methods:
  - All three approaches yield consistent trends and confirm that the SUPRES reform generated meaningful fiscal savings by allowing the Treasury to retain liquidity longer.
- Magnitudes reported:
  - The weighted average rate (baseline) yields the highest cumulative opportunity cost, amounting to over ₡ 1,000 million by mid-2024.
  - The conditional estimate is lower than the weighted average rate but remains significant given frequent deficits.
  - The short-term debt rate yields the most conservative cumulative estimate; with available data up to August 2023 it arrives with a cumulative value just below ₡ 700 million by August 2023.
- Caveats:
  - None of the methods precisely replicate real-time government borrowing behavior or cash balances, but together they provide a credible range of potential savings from faster and more transparent fund disbursement enabled by SUPRES.

*Source: wpiea2026077-source-pdf - 1. Interrupted Time Series Analysis; Digital Treasury Reform and Fiscal Efficiency; Annex I. Opportunity Cost Estimates Robustness*

### 1. Interrupted Time Series Analysis

### Interrupted Time Series Analysis

### Methodology
- Interrupted time series (ITS) regression specification:
  - Y_t = β0 + β1 Time_t + β2 Treatment_t + β3 Post_t + ε_t
  - Y_t denotes the monthly average lead time.
  - Time_t is a time trend.
  - Treatment_t is a binary variable equal to 1 after SUPRES adoption and 0 otherwise.
  - Post_t = 0 before SUPRES and counts the number of months since implementation.
  - Interpretation:
    - β1 captures the pre-SUPRES time trend.
    - β2 estimates the immediate change in level after SUPRES adoption.
    - β3 reflects the change in trend between pre- and post-SUPRES periods.
- Cases where lead days are negative are excluded (these occur when IMAS used its own resources to pay benefits because TSA disbursements were late).

### Main results: lead times and variability
- Aggregate findings:
  - Avancemos: lead time decreased from approximately 9.3 days to just over 2 days.
  - Prestación alimentaria / Atención a familias (DESAF): lead time fell from around 13 days to under 3 days on average.
  - The range between minimum and maximum monthly lead times narrowed after SUPRES, indicating improved predictability.
- Interpretation:
  - Improved predictability facilitates more accurate cash forecasting and reduces the need for precautionary liquidity buffers.
  - Regression results confirm statistically significant reductions in lead time for both programs.

### Regression estimates (Table 3)
- Dependent variable = Lead time (in days)
- Avancemos:
  - Time: 0.002 (0.035)
  - Treatment: -6.921*** (1.455)
  - Post: -0.013 (0.048)
  - Constant: 9.294*** (0.844)
  - Observations: 84
  - R-squared: 0.495
- Atención a Familias / Prestación Alimentaria (DESAF):
  - Time: -0.015 (0.034)
  - Treatment: -8.246*** (1.342)
  - Post: -0.056** (0.056)
  - Constant: 12.912*** (1.369)
  - Observations: 84
  - R-squared: 0.431
- Notes:
  - Robust standard errors in parentheses.
  - *** p<0.01, ** p<0.05, * p<0.1

### Program-specific dynamics and interpretation
- Avancemos:
  - No significant time trend before or after SUPRES; the 7-day reduction (from 9.3 to just over 2 days) appears as a one-time shift consistent with a “big bang” migration of all beneficiaries onto SUPRES.
  - Box 1 mechanism: prior multi-step payment chain caused staged Treasury releases; SUPRES enabled validation of payment files directly in SUPRES and direct TSA-to-beneficiary execution, keeping failed payments in the Treasury account and aligning outflows with actual payments.
- DESAF:
  - ITS estimates an immediate reduction of 8.25 days in lead times upon adoption.
  - Post-adoption time trend is significantly negative, indicating continued gradual reductions after the initial drop and a total realized improvement of approximately 9 days.
  - A temporary increase in lead time immediately before DESAF integration (July 2023) was driven by BNCR processing delays due to a transaction surge; issue was subsequently resolved.

### Disaggregation by transaction size
- Pre-SUPRES pattern:
  - Reverse U-shaped relationship between transaction size and lead time: mid-range payments (around ₡500–₡2,000) experienced longer lead times; very small and very large transfers cleared more quickly.
- Post-SUPRES pattern:
  - Automation and direct integration standardized workflows: lead times across transaction-size bins converged to approximately two to three days.
  - The size–time curvature flattened markedly, indicating structural process improvements rather than mere compositional shifts in transaction sizes.
- Complementary reforms:
  - SINIRUBE reforms were introduced during the same period but focus on beneficiary identification; SINIRUBE has not yet achieved full interoperability with SUPRES, so reductions in lead time are attributed to SUPRES.

### Fiscal implications and opportunity-cost quantification
- Estimated cumulative opportunity cost of idle cash (cost to the National Treasury of not being able to utilize funds during lead time):
  - Avancemos (January 2018 to August 2022): ₡ 828 million (approximately USD 1.5 million).
  - Atención a Familias / Prestación Alimentaria (January 2018 to July 2023): ₡ 1,410 million (approximately UDS 2.8 million).
  - Total: approximately USD 4.3 million.
- Implementation costs:
  - Total implementation costs totalize approximately USD 600,000.
  - Even under conservative assumptions, annual benefits exceed estimated one‑off and transitional costs.
- Operational indicators after SUPRES:
  - Average monthly transaction amount of each social program decreased by one-third.
  - Number of monthly transactions increased by 300%.
- Illustrative social trade-offs:
  - Avancemos caseload: approximately 421,000 beneficiaries per year; estimated 80 percent are upper‑secondary students.
  - Equivalent annual savings from ₡ 828 million (2018–Aug‑2022) map to roughly ₡ 177 million per full year.
  - Program benefit levels: ₡40,000 per month for upper secondary student; ₡18,000 per month for lower school student.
  - These savings correspond to approximately 370–820 additional full‑year beneficiaries each year during 2018–2021.
- Caveats:
  - Estimates may overstate efficiency gains because they do not account for Treasury’s real-time borrowing needs or actual government borrowing/debt issuance activities. A more conservative estimate, adjusted for fiscal balance and debt issuance, is provided in Annex I (not reproduced here) and still confirms materiality of savings.

### Costa Rica fiscal context (Box 2)
- Fiscal context 2018–2023:
  - Overall fiscal deficits ranged from 5.6 to 8.3 percent of GDP (2018–2020).
  - Gross public debt rose from 52 to 66 percent of GDP before stabilizing at 60 percent by 2023.
  - Short-term domestic debt instruments (zero-coupon bonds and Treasury bills) combined interest payments:
    - ₡70.9 billion in 2018
    - ₡61.5 billion in 2019
    - ₡29.7 billion in 2020
- Pre-SUPRES opportunity cost associated with payment lead times:
  - ₡280 million in 2018
  - ₡585 million in 2019
  - ₡324 in 2020

*Source: wpiea2026077-source-pdf - 1. Interrupted Time Series Analysis*

### 0.9 percent in 2019, and 1.1 percent in 2020. Although modest relative to the broader short-term interest

### Digital Treasury Reform and Fiscal Efficiency

### Fiscal impacts and opportunity cost savings
- Estimated opportunity cost savings for the Treasury during 2018–2024: exceeding USD 4.3 million.
- Reform total implementation cost: about USD 600,000.
- Pre‑SUPRES opportunity cost in 2020 corresponded to roughly 1.1 percent of total domestic short‑term interest payments (zero‑coupon bonds + T‑bills).
- Annual interest paid on T‑bills: ₡4.9 billion in 2019, ₡782 million in 2020.
- SUPRES reduced average lead times for key social programs (Avancemos, Atencion a Familias, Prestacion Alimentaria) from 9-13 days to 2-3 days.

- Yearly table of selected fiscal and SUPRES metrics (as reported):
  - 2018: SUPRES Total Opportunity Cost (₡ mn) 279.5; Zero Coupon Bonds & T-bill Interest - Domestic (₡ mn) 70,879.0; Overall Fiscal Balance (% of GDP) -5.6; Gross Public Debt (% of GDP) 52%
  - 2019: SUPRES Total Opportunity Cost (₡ mn) 584.8; Zero Coupon Bonds & T-bill Interest - Domestic (₡ mn) 61,479.4; Overall Fiscal Balance (% of GDP) -6.6; Gross Public Debt (% of GDP) 56%
  - 2020: SUPRES Total Opportunity Cost (₡ mn) 323.6; Zero Coupon Bonds & T-bill Interest - Domestic (₡ mn) 29,721.8; Overall Fiscal Balance (% of GDP) -8.3; Gross Public Debt (% of GDP) 66%
  - 2021: SUPRES Total Opportunity Cost (₡ mn) 339.7; Zero Coupon Bonds & T-bill Interest - Domestic (₡ mn) 15,538.0; Overall Fiscal Balance (% of GDP) -5; Gross Public Debt (% of GDP) 67%
  - 2022: SUPRES Total Opportunity Cost (₡ mn) 623.7; Zero Coupon Bonds & T-bill Interest - Domestic (₡ mn) 18,516.4; Overall Fiscal Balance (% of GDP) -2.7; Gross Public Debt (% of GDP) 61%
  - 2023: SUPRES Total Opportunity Cost (₡ mn) 184.4; Zero Coupon Bonds & T-bill Interest - Domestic (₡ mn) 25,055.7; Overall Fiscal Balance (% of GDP) -3.2; Gross Public Debt (% of GDP) 60%

### Coverage limitations and potential scale gains
- Current SUPRES coverage in 2023: platform handled payments for approximately 439,188 beneficiaries.
  - IMAS beneficiaries on SUPRES: about 421,002 (roughly 80 percent of IMAS’s beneficiaries).
- IMAS social programs: 0.44 percent of GDP; the analyzed programs represent approximately 0.33 percent of GDP (programs covering 80% of IMAS expenditure).
- Total social spending: 1.7% of GDP; overall government expenditure: approximately 19% of GDP.
- Major social programs not yet on SUPRES and their shares:
  - CCSS non-contributory pensions: 0.43 percent of GDP; 146,905 beneficiaries.
  - Ministry of Education social programs: 0.35 percent of GDP; ~981,000 beneficiaries.
  - BANHVI housing subsidies: 0.20 percent of GDP.
- Combined, these non‑migrated programs represent nearly 1 percent of GDP in social expenditures not yet benefiting from SUPRES efficiencies.
- Implication: expanding SUPRES to include these programs would significantly amplify fiscal and treasury operational gains.

### Administrative and operational efficiency — survey evidence
- Survey administered online between February and March 2025 to staff processing social assistance payments using SUPRES; 56 respondents covering key departments and most IMAS regional offices.
- Key survey findings (reported proportions preserved where provided):
  - 65% reported reduction in time required to process social benefit payments.
  - 75% responded positively regarding SUPRES contribution to their efficiency by automating payment processes.
  - 64% reported reduced administrative errors and improved accuracy in data.
  - 70% agreed SUPRES enhanced transparency and traceability in payment processing while reducing system fragmentation.
  - 73% believed SUPRES had already contributed positively to improving targeting of social programs.
  - 77% confident SUPRES has potential to continue to drive improvements.
- Operational trade-offs and challenges identified:
  - SUPRES overnight validation can extend some processes from one day to up to two days — trade-off between auditability/control and operational flexibility.
  - Reported issues: added technical complexity, slow/inefficient technical support, opacity of certain operational information (e.g., contingency claims spread across multiple unreconciled files).
  - Common recommendations from users: include all social benefits in SUPRES, step up system capabilities to collect/manage large datasets, introduce immediate transfer functionalities for emergencies, strengthen IT stability, and promote beneficiary participation.

### Financial inclusion, payment predictability, and system resilience
- Pre-SUPRES: IMAS opened simplified accounts for beneficiaries at Banco Nacional de Costa Rica (BNCR), limiting beneficiary choice and sometimes creating transaction costs when beneficiaries moved funds to other banks.
- With SUPRES using the central bank’s SINPE platform: beneficiaries can receive transfers at any financial institution of their choice.
- Measured changes (2023–2024):
  - Total number of social benefit transactions increased by 3% between 2023 and 2024.
  - Number of transactions directed to bank accounts outside BNCR rose by 122% (number of transactions executed; not number of beneficiaries or monetary value).
- Benefits cited:
  - Greater beneficiary autonomy and financial empowerment.
  - Diversification across banks reduces concentration and operational risks, fosters competition among financial institutions.
  - Multi‑bank payment model strengthens resilience of social transfer delivery and aligns with international best practices.
- Predictability:
  - SUPRES validates payment information before execution and channels transfers directly from the TSA through the national payment system, reducing administrative delays and narrowing the range of payment dates within each cycle.
  - More predictable timing plausibly eases short‑term liquidity stress for beneficiaries (direct measurement not possible due to data limitations).

### Institutional enablers, IMF role, and implementation lessons
- Key enablers of SUPRES:
  - Functioning Treasury Single Account (TSA), a national payments infrastructure (SINPE), and an integrated beneficiary registry (SINIRUBE) enabling automated validation.
  - Legal underpinnings: service-level agreements and a presidential decree helped institutionalize the platform.
- IMF contribution:
  - Technical assistance and collaboration under the Extended Fund Facility (EFF) program.
  - Support included a PFM-focused hackathon, roadmap development, and inclusion of SUPRES as a structural benchmark under the EFF program.
- Implementation lessons and future priorities:
  - Expand SUPRES coverage to remaining large social programs to amplify fiscal and operational gains.
  - Continue improving legal framework to support platform expansion.
  - Enhance interoperability, particularly with SINIRUBE, for integrated and data-driven fiscal operations and better targeting.
  - Invest in system stability, long‑term operations, maintenance, cybersecurity, user training, and beneficiary engagement to preserve sustainability and enable rapid scaling in emergencies.
  - Establish clearer baselines and monitoring frameworks to collect granular data needed to assess specific impacts (e.g., financial inclusion outcomes or reductions in bank commission fees).

*Source: IMF staff analysis of Costa Rica’s SUPRES implementation and IMF‑led survey, as reported in the IMF Working Paper "Digital Treasury Reform and Fiscal Efficiency."*

### Annex I. Opportunity Cost Estimates Robustness

### Annex I. Opportunity Cost Estimates Robustness

### Methodology for Baseline Estimate
- Baseline approach (blue line in Figure A1):
  - Applied the weighted average interest rate of outstanding public debt, as reported by the Ministry of Finance, to the daily cash balance that could have been retained longer in the Treasury Single Account (TSA) due to reduced lead times.
  - Captures a broad fiscal benchmark for the cost of borrowing.

### Alternative Estimation Approaches
- Conditional Estimate (green line):
  - Applies the same weighted average interest rate but only for months in which the government faced a fiscal deficit.
  - Rationale: idle funds impose a tangible fiscal cost principally when borrowing is required to cover shortfalls; months of surplus reduce the fiscal pressure to borrow.
- Short-Term Debt Benchmark (red line):
  - Matches monthly lead time of disbursement with the maturity structure of short-term debt issuance and applies the interest rate of short-term instruments (e.g., Letras del Tesoro) to estimated idle balances.
  - Rationale: reflects the real financing cost if gaps in liquidity were covered through short-term borrowing; particularly relevant where short-term instruments are actively used for cash management.
  - Data limitation: short-term debt issuance data are available only up to August 2023.

### Key Findings and Robustness Results
- Consistency across methods:
  - All three approaches yield consistent trends and confirm that the SUPRES reform generated meaningful fiscal savings by allowing the Treasury to retain liquidity longer.
- Magnitudes reported in the source:
  - The weighted average rate (blue) yields the highest cumulative opportunity cost, amounting to over ₡ 1,000 million by mid-2024.
  - The conditional estimate (green) is lower than the weighted average rate but remains significant, reflecting that the government ran deficits in most months during the analysis period.
  - The short-term debt rate (red) yields the most conservative cumulative estimate; with available data up to August 2023 it arrives with a cumulative value just below ₡ 700 million by August 2023.
- Caveats:
  - None of the methods precisely replicate real-time government borrowing behavior or cash balances, but together they provide a credible range of potential savings from faster and more transparent fund disbursement enabled by SUPRES.

*Source: Annex I. Opportunity Cost Estimates Robustness, Digital Treasury Reform and Fiscal Efficiency: Evaluating Costa Rica’s SUPRES Platform Adoption, Working Paper No. WP/2026/077*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2026/english/wpiea2026077-source-pdf.pdf_