## sipea2026019

## Source details

**Canonical URL:** [sipea2026019](https://www.imf.org/-/media/files/publications/selected-issues-papers/2026/english/sipea2026019.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/selected-issues-papers/2026/english/sipea2026019.pdf.md)
- [Structured JSON version](/-/media/files/publications/selected-issues-papers/2026/english/sipea2026019.pdf.json)

---

### A. Interest Rate Channel of Monetary Policy
- Purpose: Estimate interest rate pass-through of the policy rate to bank lending rates in Cabo Verde over January 2001–May 2025, and test for asymmetries in adjustment to increases vs decreases in the policy rate.
- Recent tightening cycle:
  - May 2023–Oct 2025: duration 30 months; total increase 225bps; number of hikes 6.
  - The May 2023 tightening is the steepest over the past two decades and reflects normalization after COVID-19 easing.
- Observed behavior and market context:
  - Lending rates and long-term rates have not adjusted noticeably to recent policy rate increases.
  - Emigrant and resident deposit rates have not increased significantly; emigrant deposit rate premium has dissipated.
  - Emigrant deposits constitute an average of 30 percent of total deposits in the banking system.
  - The average spread between the average lending rate and the policy rate is 113 basis points higher during policy easing phases than during policy tightening phases.
  - Possible explanations for incomplete/asymmetric pass-through: market concentration, risk premium and consumer search costs, excess liquidity in the banking system, and high funding costs for banks.
- Monetary policy cycles (summary):
  - Jan 2001–April 2002: duration 15 months; total increase 100bps; number of hikes 1.
  - May 2007–Dec 2009: duration 32 months; total increase 175bps; number of hikes 5.
  - Feb 2012–Feb 2014: duration 26 months; total increase 150bps; number of hikes 1.
  - May 2023–Oct 2025: duration 30 months; total increase 225bps; number of hikes 6.
- Interest rate betas and cycle heterogeneity:
  - Unconditional betas show complete pass-through in 2001 tightening, limited pass-through in 2012, and negative betas in 2007 and the current tightening cycle (i.e., lending rates declined despite tightening).
  - Interpretation: policy rate may act more as a signal than a binding cost for banks, especially because many banks do not rely on BCV refinancing.

### B. Interest-Rate Pass-Through (Econometric Evidence)
- Data and methods:
  - Monthly aggregate BCV data, January 2001–May 2025.
  - Two models estimated: error-correction model (ECM) via Engle and Granger, and local projections (LP) for horizons one to six months.
  - Benchmark controls: reserve requirements ratio (RRR), CPI inflation, crisis dummy (GFC and COVID). Additional controls tested: NPLs, excess liquidity, loan-to-deposit ratio (LDR).
- Main quantitative findings (full sample, benchmark specification):
  - A 1-percentage-point policy rate change leads to:
    - 0.19–0.28 increase in average lending rates (medium- to long-run).
    - 0.13–0.23 increase in long-term lending rates (medium- to long-run).
  - These ranges reflect results across ECM and LP models (LP horizon six months ahead).
- Robustness with additional controls:
  - Adding NPLs, excess liquidity, and LDR (discarding the first 2001 tightening episode) yields:
    - Average lending rate pass-through: 0.17-0.3.
    - Long-term rate pass-through: 0.11−0.15.
  - Conclusion: limited pass-through is robust to alternative specifications.
- Banking structure relevant to pass-through:
  - Cabo Verde has eight banks; three dominate the system.
  - Top banks’ shares (as of December 2024):
    - BCA: 27.34% of total assets; 25.09% share of total credit; 30.40% share of total deposits; classified as High D-SIB.
    - CAIXA (CECV): 26.41% of total assets; 29.04% share of total credit; 30.09% share of total deposits; classified as High D-SIB.
    - BIA: 11.14% of total assets; 11.81% share of total credit; 10.96% share of total deposits; Slight systemic importance.
  - Two banks are classified as High D-SIB; others classified as Slight systemic importance or Non D-SIB.
  - Bank concentration can reduce competition and weaken transmission of policy rates to bank rates.

### C. Asymmetric Pass-through During Loosening and Tightening Cycles
- Asymmetry in pass-through:
  - Pass-through is larger during tightening episodes than during easing episodes, but remains limited in absolute terms.
  - LP model (long-term lending rate, six-month horizon):
    - Tightening episodes: pass-through 0.35.
    - Easing episodes: pass-through 0.17.
  - Hence, pass-through during tightening is roughly twice as large as during easing.
- Heterogeneity across tightening episodes:
  - Larger tightening pass-through appears driven by April 2001–April 2002 and January 2012–February 2014 hiking cycles.
  - May 2007–Dec 2009 and the current May 2023 cycle appear broadly insensitive to monetary policy (limited transmission).
- Interest rate beta ranges (average across cycles):
  - Tightening cycles: pass-through falls into a range of -0.75−0.39 percentage points.
  - Loosening cycles: pass-through falls into a range of -0.49−0.66 percentage points.
- Interpretation and implications:
  - The asymmetric and generally limited pass-through highlights challenges for monetary policy effectiveness in Cabo Verde.
  - Factors potentially reducing pass-through include excess liquidity (so banks do not rely on BCV funding), market concentration, and non-market-based pricing or strategic behavior aimed at preserving market share.

### Interest-rate pass-through during tightening and easing episodes (Section 2)
- Tightening episodes (detailed):
  - Expanding the local projection model to estimate pass-through by tightening episode in the same regression shows a larger pass-through during tightening cycles, but this result is driven by the 2001 and 2012 hiking cycles.
  - 2001 episode: led to a full pass-through.
  - 2012 episode: associated with a limited pass-through of around 0.2 after six months.
  - 2007 cycle and the current cycle that started in May 2023: lending rates were broadly insensitive to monetary policy.
- Easing episodes (detailed):
  - Pass-through varies across easing cycles.
  - Jan 2010-Dec 2012 cycle: pass-through is stronger, though it took some time to materialize.
  - March 2014-April 2023 cycle (the longest easing cycle on record): estimated pass-through is practically zero; banks did not decrease lending rates despite cuts in policy rates by the BCV.
- Time-varying pass-through and robustness:
  - Time-varying estimates of lending rate pass-through are consistent with impulse response function estimates.
  - Monthly time-varying estimation indicates the current tightening cycle that started in May 2023 seems to be broadly insensitive to monetary policy.
- Summary conclusions and key estimates:
  - The policy rate pass-through is estimated to be rather limited in Cabo Verde.
  - Medium- to long-run pass-through estimates:
    - Average lending rate: range between 0.19 and 0.28.
    - Long-term rate: range between 0.13 and 0.23.
  - Pass-through is roughly twice as large during tightening episodes compared to easing episodes, although overall levels remain relatively low.
  - Larger pass-through during tightening cycles appears driven by the 2001 and 2012 hiking cycles; the 2007 and the current May 2023 cycles seem broadly insensitive to monetary policy.

*Source: IMF Selected Issues Paper SIP/2026/019, “Interest Rate Pass-Through in Cabo Verde,” completed on January 9, 2026.*

### Section 1

### Interest Rate Pass-Through in Cabo Verde

### A. Interest Rate Channel of Monetary Policy
- Purpose: Estimate interest rate pass-through of the policy rate to bank lending rates in Cabo Verde over January 2001–May 2025, and test for asymmetries in adjustment to increases vs decreases in the policy rate.
- Recent tightening cycle:
  - May 2023–Oct 2025: duration 30 months; total increase 225bps; number of hikes 6.
  - The May 2023 tightening is the steepest over the past two decades and reflects normalization after COVID-19 easing.
- Observed behavior and market context:
  - Lending rates and long-term rates have not adjusted noticeably to recent policy rate increases.
  - Emigrant and resident deposit rates have not increased significantly; emigrant deposit rate premium has dissipated.
  - Emigrant deposits constitute an average of 30 percent of total deposits in the banking system.
  - The average spread between the average lending rate and the policy rate is 113 basis points higher during policy easing phases than during policy tightening phases.
  - Possible explanations for incomplete/asymmetric pass-through: market concentration, risk premium and consumer search costs, excess liquidity in the banking system, and high funding costs for banks.
- Monetary policy cycles (summary):
  - Jan 2001–April 2002: duration 15 months; total increase 100bps; number of hikes 1.
  - May 2007–Dec 2009: duration 32 months; total increase 175bps; number of hikes 5.
  - Feb 2012–Feb 2014: duration 26 months; total increase 150bps; number of hikes 1.
  - May 2023–Oct 2025: duration 30 months; total increase 225bps; number of hikes 6.
- Interest rate betas and cycle heterogeneity:
  - Unconditional betas show complete pass-through in 2001 tightening, limited pass-through in 2012, and negative betas in 2007 and the current tightening cycle (i.e., lending rates declined despite tightening).
  - Interpretation: policy rate may act more as a signal than a binding cost for banks, especially because many banks do not rely on BCV refinancing.

### B. Interest-Rate Pass-Through (Econometric Evidence)
- Data and methods:
  - Monthly aggregate BCV data, January 2001–May 2025.
  - Two models estimated: error-correction model (ECM) via Engle and Granger, and local projections (LP) for horizons one to six months.
  - Benchmark controls: reserve requirements ratio (RRR), CPI inflation, crisis dummy (GFC and COVID). Additional controls tested: NPLs, excess liquidity, loan-to-deposit ratio (LDR).
- Main quantitative findings (full sample, benchmark specification):
  - A 1-percentage-point policy rate change leads to:
    - 0.19–0.28 increase in average lending rates (medium- to long-run).
    - 0.13–0.23 increase in long-term lending rates (medium- to long-run).
  - These ranges reflect results across ECM and LP models (LP horizon six months ahead).
- Robustness with additional controls:
  - Adding NPLs, excess liquidity, and LDR (discarding the first 2001 tightening episode) yields:
    - Average lending rate pass-through: 0.17-0.3.
    - Long-term rate pass-through: 0.11−0.15.
  - Conclusion: limited pass-through is robust to alternative specifications.
- Banking structure relevant to pass-through:
  - Cabo Verde has eight banks; three dominate the system.
  - Top banks’ shares (as of December 2024):
    - BCA: 27.34% of total assets; 25.09% share of total credit; 30.40% share of total deposits; classified as High D-SIB.
    - CAIXA (CECV): 26.41% of total assets; 29.04% share of total credit; 30.09% share of total deposits; classified as High D-SIB.
    - BIA: 11.14% of total assets; 11.81% share of total credit; 10.96% share of total deposits; Slight systemic importance.
  - Two banks are classified as High D-SIB; others classified as Slight systemic importance or Non D-SIB.
  - Bank concentration can reduce competition and weaken transmission of policy rates to bank rates.

### C. Asymmetric Pass-through During Loosening and Tightening Cycles
- Asymmetry in pass-through:
  - Pass-through is larger during tightening episodes than during easing episodes, but remains limited in absolute terms.
  - LP model (long-term lending rate, six-month horizon):
    - Tightening episodes: pass-through 0.35.
    - Easing episodes: pass-through 0.17.
  - Hence, pass-through during tightening is roughly twice as large as during easing.
- Heterogeneity across tightening episodes:
  - Larger tightening pass-through appears driven by April 2001–April 2002 and January 2012–February 2014 hiking cycles.
  - May 2007–Dec 2009 and the current May 2023 cycle appear broadly insensitive to monetary policy (limited transmission).
- Interest rate beta ranges (average across cycles):
  - Tightening cycles: pass-through falls into a range of -0.75−0.39 percentage points.
  - Loosening cycles: pass-through falls into a range of -0.49−0.66 percentage points.
- Interpretation and implications:
  - The asymmetric and generally limited pass-through highlights challenges for monetary policy effectiveness in Cabo Verde.
  - Factors potentially reducing pass-through include excess liquidity (so banks do not rely on BCV funding), market concentration, and non-market-based pricing or strategic behavior aimed at preserving market share.

*Source: IMF Selected Issues Paper SIP/2026/019, “Interest Rate Pass-Through in Cabo Verde,” completed on January 9, 2026.*

### Section 2

### sipea2026019 - Section 2

### Interest-rate pass-through during tightening episodes
- Expanding the local projection model to estimate pass-through by tightening episode in the same regression shows a larger pass-through during tightening cycles, but this result is driven by the 2001 and 2012 hiking cycles.
- 2001 episode: led to a full pass-through.
- 2012 episode: associated with a limited pass-through of around 0.2 after six months.
- 2007 cycle and the current cycle that started in May 2023: lending rates were broadly insensitive to monetary policy.

### Interest-rate pass-through during easing episodes
- Pass-through varies across easing cycles.
- Jan 2010-Dec 2012 cycle: pass-through is stronger, though it took some time to materialize.
- March 2014-April 2023 cycle (the longest easing cycle on record): estimated pass-through is practically zero; banks did not decrease lending rates despite cuts in policy rates by the BCV.

### Time-varying pass-through and robustness
- Time-varying estimates of lending rate pass-through are consistent with impulse response function estimates.
- Monthly time-varying estimation indicates the current tightening cycle that started in May 2023 seems to be broadly insensitive to monetary policy.

### Summary conclusions and key estimates
- The policy rate pass-through is estimated to be rather limited in Cabo Verde.
- Medium- to long-run pass-through estimates:
  - Average lending rate: range between 0.19 and 0.28.
  - Long-term rate: range between 0.13 and 0.23.
- Pass-through is roughly twice as large during tightening episodes compared to easing episodes, although overall levels remain relatively low.
- Larger pass-through during tightening cycles appears driven by the 2001 and 2012 hiking cycles; the 2007 and the current May 2023 cycles seem broadly insensitive to monetary policy.

*Source: sipea2026019 - Section 2*

---


_Source: https://www.imf.org/-/media/files/publications/selected-issues-papers/2026/english/sipea2026019.pdf_