## References (wpiea2019197-print-pdf)

## Source details

**Canonical URL:** [References (wpiea2019197-print-pdf)](https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019197-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2019/wpiea2019197-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2019/wpiea2019197-print-pdf.pdf.json)

---

### I. Introduction — context and research question
- Global inflation trended down from the peak of over 40 percent in the 1990s to an average of 3.5 percent over the past decade.
- Caribbean average inflation moved from 17.4 percent to 4.5 percent over the same period.
- Research question: whether outsourcing the effective conduct of monetary policy via fixed exchange rate regimes (prevalent in the Caribbean) delivers low and stable inflation, assessed through an index of monetary autonomy.
- Sample: panel of 15 Caribbean countries over the period 1980–2017; nine out of 15 Caribbean countries in the sample maintain pegged exchange rates relative to the U.S. dollar.
- Econometric approaches: panel fixed effects and system Generalized Method of Moments (GMM).

### II. Key findings and empirical results
- Core empirical result:
  - Monetary independence has a statistically significant effect on inflation (CPI), even after controlling for macroeconomic and financial factors.
  - Direction: greater monetary independence is associated with lower inflation in the sample of Caribbean countries over 1980–2017.
  - Robustness: relationship holds across alternative specifications and estimation methodologies.
- Policy implication:
  - Does not imply abolishing fixed exchange rate regimes.
  - Stresses need to better align monetary policy stance with domestic conditions and improve channels of monetary transmission.

### III. Literature links — theoretical and empirical background
- Two literature strands: determinants of inflation; central bank independence and monetary independence.
- Determinants of inflation listed: policy preferences, income level, trade openness, fiscal deficits, labor-market institutions, exchange rate regime type, political/institutional factors.
- Legal central bank independence literature: mixed empirical evidence on whether greater legal independence consistently reduces inflation.
- Trilemma literature: monetary independence, exchange rate stability, and financial openness cannot be simultaneously achieved; pegged regimes tend to reduce monetary independence.
- Prior regional findings: mixed; some studies find fixed exchange rates lead countries to follow base-country interest rates more closely; other studies find capital mobility disciplines inflation; de Mendonça and Veiga (2017) find no statistically significant effect of monetary independence on inflation for Latin America.

### IV. Data and measurement
- Data sources: IMF’s International Financial Statistics (IFS), IMF World Economic Outlook (WEO), World Bank World Development Indicators (WDI).
- Dependent variable:
  - Inflation measured by the annual rate of change in the CPI, transformed as the logarithm of (1+inflation) to minimize heteroscedasticity.
- Control variables included:
  - logarithm of real GDP per capita,
  - output gap (obtained via Hodrick-Prescott (HP) filter),
  - logarithm of trade openness (share of international trade in GDP),
  - Chinn-Ito financial openness/integration index,
  - change in the logarithm of the terms-of-trade index,
  - money supply growth,
  - real effective exchange rate (REER),
  - fiscal balance.
- HP filter details:
  - weight (λ) adopted: 6.25 after experimenting with a range of smoothing parameters.
  - end-point problem mitigation: series extended through 2024 using projections before applying the HP filter.

### V. Monetary independence index — definition and properties
- Index used: Aizenman, Chinn, and Ito (2010) definition (reciprocal of the annual correlation between monthly short-term interest rates of the home country and the base country).
- Formula as presented (text preserved):
  - 푀푀푀푀 = 1 −
    푐푐푐푐푐푐푐푐 �푖푖
    푐푐
    ,푖푖
    푗푗
    � − (−1)
    1 − (−1)
- Interpretation and properties:
  - Base country defined as the country with which a home country’s monetary policy is most closely linked (Shambaugh, 2004); for Caribbean countries, the U.S. is the base country.
  - Index ranges between 0 and 1.
  - Higher index value means lower correlation of interest rates and thus greater monetary policy independence.
- Caveat and control:
  - Correlation between home and base-country short-term interest rates may reflect business cycle synchronization rather than lack of monetary independence.
  - Business cycle synchronization controlled for by estimating the correlation between output gaps in the home country and the U.S. using three-year moving averages (t–1, t, t+1).

### VI. Methodology notes relevant to interpretation
- Estimation techniques: panel fixed effects; system GMM (Arellano and Bover (1995); Blundell and Bond (1998)), one-step version used to address endogeneity and dynamic panel biases.
- Dependent variable transformation: log(1+inflation).
- Frequency and period: annual observations covering 1980–2017.
- Monetary independence index frequency distribution and time series evolution presented in figures (1980–2017).
- Standard errors: robust, clustered at country level.
- Identification and instruments:
  - Instruments: first differences instrumented by lags of levels; levels instrumented by lags of first differences.
  - Address instrument proliferation and over-identification following Roodman (2009).
  - Validation: AR(1) and AR(2) tests; Hansen J-test for overidentifying restrictions.

### Baseline estimation results (selected coefficients and diagnostics)
- Monetary independence coefficient range across specifications: -0.68 to -0.90.
- Specific reported coefficients (Table 1): -0.623***, -0.758*, -0.680***, -0.904*** (standard errors [0.350], [0.380], [0.273], [0.266]).
- Lagged inflation coefficients: 0.458*** and 0.250*** (standard errors [0.044] and [0.048]).
- Selected control variable estimates (examples):
  - Financial openness: -0.616** and -0.615*** (standard errors [0.280] and [0.182]).
  - Real GDP per capita: -0.748** and -0.400** (standard errors [0.300] and [0.178]).
  - Business cycle synchronization: 0.044 and 0.074 (standard errors [0.048] and [0.047]).
  - Trade openness: 0.496 and 0.427** (standard errors [0.343] and [0.210]).
  - Terms-of-trade: -1.852*** and -1.967*** (standard errors [0.498] and [0.269]).
  - Money growth: 1.161*** and 0.860*** (standard errors [0.311] and [0.231]).
- Model diagnostics and sample details:
  - Number of observations: 499, 454, 467, 429 (across reported columns).
  - Number of countries: 15 (all columns).
  - Country FE: Yes.
  - Year FE: Yes.
  - Adj R^2: 0.01 and 0.25 (reported).
  - AR1 p-value: 0.000, 0.000.
  - AR2 p-value: 0.562, 0.984.
  - Hansen J-test p-value: 0.043, 0.022.
- Note: *, **, *** denote significance at the 10%, 5%, and 1% levels, respectively.

### Robustness checks (Appendix Table 3 highlights)
- Robustness strategies employed:
  - Truncate sample at 5th and 95th percentiles.
  - Use 3-year nonoverlapping intervals (3-year averages).
  - Add controls: REER, fiscal balance.
  - Use interaction MI*FO (monetary independence × financial openness).
  - Replace country-specific terms-of-trade with international oil prices.
- Selected robustness coefficient examples:
  - Inflation_{t-1}: 0.257***, 0.041, 0.187***, 0.185***, 0.257*** (standard errors [0.052], [0.092], [0.059], [0.057], [0.047]).
  - Monetary independence: -0.383*, -0.890**, -1.091***, -0.975*** (standard errors [0.207], [0.394], [0.330], [0.259]).
  - Financial openness: -0.121, -0.516*, -0.070*, -0.639*** (standard errors [0.185], [0.293], [0.372], [0.181]).
  - Terms-of-trade (when included): -1.002***, -2.556***, -2.449***, -2.493*** (standard errors [0.209], [0.600], [0.362], [0.364]).
  - Money growth: 0.046*, 1.909***, 0.674***, 0.827***, 0.756*** (standard errors [0.306], [0.655], [0.284], [0.279], [0.232]).
  - REER: -0.007** (standard error [0.004]).
  - Fiscal balance: 0.002 (standard error [0.012]).
  - MI*FO: -0.547** (standard error [0.469]).
  - Oil Price (used instead of terms-of-trade): 0.941*** (standard error [0.126]).
- Robustness diagnostics:
  - Number of observations across robustness columns: 342, 144, 289, 283, 433.
  - Number of countries: 15, 15, 14, 15, 15.
  - AR1 p-values: 0.000 (all five reported).
  - AR2 p-values: 0.973, 0.431, 0.466, 0.462, 0.675.
  - Hansen J-test p-values: 0.208, 0.000, 0.072, 0.071, 0.036.
- Interpretation: Monetary independence maintains a significant negative effect on inflation across robustness checks, though magnitudes vary.

### Data dispersion, stationarity, and sample properties
- Sample: 15 Caribbean countries, period 1980–2017.
- Panel unit root test: Im-Pesaran-Shin (2003) procedure.
- Stationarity: Variables are stationary after logarithmic transformation or upon first differencing (results available upon request).
- Summary statistics (Appendix Table 2 key values):
  - Consumer price inflation (Obs. 554): Mean 7.21, Std. Dev. 12.2, Min. -11.4, Max. 142.8.
  - Monetary independence index (Obs. 516): Mean 0.50, Std. Dev. 0.10, Min. 0.1, Max. 0.9.
  - Real GDP per capita (Obs. 544): Mean 6,137.6, Std. Dev. 10,043.9, Min. 1,327.0, Max. 52,2874.9.
  - Output gap (Obs. 570): Mean 0.0, Std. Dev. 2.5, Min. -14.3, Max. 11.0.
  - Business cycle synchronization (Obs. 540): Mean -0.2, Std. Dev. 0.7, Min. -1.0, Max. 1.0.
  - Trade openness (Obs. 542): Mean 104.13, Std. Dev. 32.2, Min. 31.1, Max. 275.0.
  - Financial openness (Obs. 526): Mean 0.4, Std. Dev. 0.3, Min. 0.0, Max. 1.0.
  - Real effective exchange rate index (Obs. 561): Mean 114.8, Std. Dev. 118.2, Min. 28.9, Max. 1293.6.
  - Terms-of-trade (Obs. 549): Mean 145.4, Std. Dev. 43.2, Min. 55.5, Max. 314.6.
  - Money growth (Obs. 540): Mean 12.9, Std. Dev. 16.6, Min. -10.3, Max. 205.6.
  - Fiscal balance (Obs. 357): Mean 1.8, Std. Dev. 4.4, Min. -9.9, Max. 17.4.
- Note: there is a significant degree of dispersion across Caribbean countries in terms of inflation and overall macroeconomic performance as well as the extent of monetary policy independence.

### Conclusion and policy implications
- Core empirical conclusion:
  - Greater monetary independence is associated with lower consumer price inflation across a panel of 15 Caribbean countries during 1980–2017.
  - Negative effect of monetary independence on inflation is statistically significant across model specifications and estimation techniques (system GMM primary).
- Trade-offs and interpretation for exchange rate regimes:
  - Fixed exchange rate regimes outsource monetary policy to the anchor currency issuer, potentially providing credibility and low, stable inflation.
  - Floating regimes maximize monetary policy flexibility for domestic stabilization.
  - Empirical evidence here: pegged exchange rate regimes across the Caribbean do not necessarily deliver lower inflation for domestic purposes; greater monetary independence improves management of inflation dynamics.
- Policy message:
  - Findings do not mandate abolishing exchange rate pegs.
  - Highlight need to better align monetary policy stance with domestic conditions and to improve monetary transmission channels.

*Source: wpiea2019197-print-pdf (IMF working paper content provided).*

### References .............................................................................................................

### References (wpiea2019197-print-pdf)

### I. Introduction — context and research question
- Global inflation trended down from the peak of over 40 percent in the 1990s to an average of 3.5 percent over the past decade.
- Caribbean average inflation moved from 17.4 percent to 4.5 percent over the same period.
- Research question: whether outsourcing the effective conduct of monetary policy via fixed exchange rate regimes (prevalent in the Caribbean) delivers low and stable inflation, assessed through an index of monetary autonomy.
- Sample: panel of 15 Caribbean countries over the period 1980–2017; nine out of 15 Caribbean countries in the sample maintain pegged exchange rates relative to the U.S. dollar.
- Econometric approaches: panel fixed effects and system Generalized Method of Moments (GMM).

### II. Key findings and empirical results
- Core empirical result: monetary independence has a statistically significant effect on inflation as measured by the consumer price index (CPI), even after controlling for macroeconomic and financial factors.
- Direction of effect: greater monetary independence is associated with lower inflation in the sample of Caribbean countries over 1980–2017.
- Robustness: the relationship is robust to alternative specifications and estimation methodologies.
- Policy implication highlighted: the result does not imply abolishing fixed exchange rate regimes but stresses the need to better align monetary policy stance with domestic conditions and improve channels of monetary transmission.

### III. Literature links — theoretical and empirical background
- Two literature strands: determinants of inflation; central bank independence and monetary independence.
- Determinants of inflation include: policy preferences, income level, trade openness, fiscal deficits, labor-market institutions, exchange rate regime type, and political/institutional factors.
- Legal central bank independence literature: mixed empirical evidence on whether greater legal independence consistently reduces inflation.
- Trilemma literature: monetary independence, exchange rate stability, and financial openness cannot be simultaneously achieved; pegged regimes tend to reduce monetary independence.
- Prior regional findings: mixed. Some studies find fixed exchange rates lead countries to follow base-country interest rates more closely; other studies find capital mobility disciplines inflation; de Mendonça and Veiga (2017) find no statistically significant effect of monetary independence on inflation for Latin America.

### IV. Data and measurement
- Data sources: IMF’s International Financial Statistics (IFS), IMF World Economic Outlook (WEO), World Bank World Development Indicators (WDI).
- Dependent variable: inflation measured by the annual rate of change in the CPI, transformed as the logarithm of (1+inflation) to minimize heteroscedasticity.
- Control variables included:
  - logarithm of real GDP per capita (proxy for level of economic development),
  - output gap (to capture cyclical conditions; obtained via Hodrick-Prescott (HP) filter),
  - logarithm of trade openness (share of international trade in GDP),
  - Chinn-Ito financial openness/integration index,
  - change in the logarithm of the terms-of-trade index,
  - money supply growth,
  - real effective exchange rate (REER),
  - fiscal balance.
- HP filter details:
  - weight (λ) adopted: 6.25 after experimenting with a range of smoothing parameters.
  - end-point problem mitigation: series extended through 2024 using projections before applying the HP filter.

### V. Monetary independence index — definition and properties
- Monetary independence index used: Aizenman, Chinn, and Ito (2010) definition (reciprocal of the annual correlation between monthly short-term interest rates of the home country and the base country).
- Formula as presented:
  - 푀푀푀푀 = 1 −
    푐푐푐푐푐푐푐푐 �푖푖
    푐푐
    ,푖푖
    푗푗
    � − (−1)
    1 − (−1)
  - (formula text preserved as in source)
- Interpretation and properties:
  - Base country is the country with which a home country’s monetary policy is most closely linked (Shambaugh, 2004); for Caribbean countries, the U.S. is the base country.
  - By construction, the minimum and maximum values are 0 and 1, respectively.
  - Higher index value means lower correlation of interest rates and thus greater monetary policy independence.
- Caveat addressed: correlation between home and base-country short-term interest rates may reflect business cycle synchronization rather than lack of monetary independence. Business cycle synchronization is controlled for by estimating the correlation between output gaps in the home country and the U.S. economy using three-year moving averages (t–1, t, t+1).

### VI. Methodology notes relevant to interpretation
- Estimation techniques: panel fixed effects; system GMM to address potential endogeneity and dynamic panel biases.
- Dependent variable transformation: log(1+inflation) used to reduce heteroscedasticity.
- Frequency and period: annual observations covering 1980–2017.
- Monetary independence index frequency distribution and time series evolution presented in figures (1980–2017).

*Source: wpiea2019197-print-pdf (IMF working paper content provided).*

### 2. There is a significant degree of dispersion across Caribbean countries in terms of inflation and

### 2. There is a significant degree of dispersion across Caribbean countries in terms of inflation and overall macroeconomic performance as well as the extent of monetary policy independence

### Data and stationarity
- Sample: 15 Caribbean countries, period 1980–2017.
- Panel unit root test: Im-Pesaran-Shin (2003) procedure.  
- Stationarity: Variables are stationary after logarithmic transformation or upon first differencing (results available upon request).

### Empirical methodology
- Baseline dynamic inflation model:
  - π_{c,t} = β1 + β2 π_{c,t−1} + β3 MI_{c,t} + β4 X_{c,t} + η_c + μ_t + ε_{c,t}
  - Definitions:
    - π_{c,t}: inflation in country c at time t.
    - MI: monetary independence index.
    - X_{c,t}: vector of controls including real GDP per capita, trade openness, terms-of-trade index, and money supply growth.
    - η_c: time-invariant country-specific effects.
    - μ_t: time effects controlling for common shocks.
    - ε_{c,t}: idiosyncratic error term (zero mean, constant variance).
- Estimation issues and choices:
  - Fixed effects models reported, but lagged dependent variable with country fixed effects induces bias when T is fixed.
  - Primary estimator: system GMM (Arellano and Bover (1995); Blundell and Bond (1998)), one-step version used.
  - Instruments: first differences instrumented by lags of levels; levels instrumented by lags of first differences.
  - Instrument proliferation and over-identification concerns addressed following Roodman (2009).
  - Identification validation: AR(1) and AR(2) tests (p-values reported), Hansen J-test for overidentifying restrictions.
  - Standard errors: robust, clustered at country level.

### Baseline estimation results (Table 1 highlights)
- Main finding on monetary independence:
  - Monetary independence coefficient ranges between -0.68 and -0.90 across specifications.
  - Specific reported coefficients in Table 1: -0.623***, -0.758*, -0.680***, -0.904*** (standard errors [0.350], [0.380], [0.273], [0.266]).
- Lagged inflation:
  - Inflation_{t-1} coefficients: 0.458*** and 0.250*** (standard errors [0.044] and [0.048]).
  - Interpretation: high persistence in inflation dynamics.
- Control variables (selected):
  - Financial openness: -0.616** and -0.615*** (standard errors [0.280] and [0.182]); dampening effect on inflation.
  - Real GDP per capita: -0.748** and -0.400** (standard errors [0.300] and [0.178]); higher development associated with lower inflation.
  - Business cycle synchronization: 0.044 and 0.074 (standard errors [0.048] and [0.047]); not significant.
  - Trade openness: 0.496 and 0.427** (standard errors [0.343] and [0.210]); positive determinant of inflation.
  - Terms-of-trade: -1.852*** and -1.967*** (standard errors [0.498] and [0.269]); increases (lower import prices relative to export prices) reduce inflation.
  - Money growth: 1.161*** and 0.860*** (standard errors [0.311] and [0.231]); positive impact on inflation.
- Model diagnostics and sample:
  - Number of observations: 499, 454, 467, 429 (across reported columns).
  - Number of countries: 15, 15, 15, 15.
  - Country FE: Yes (all).
  - Year FE: Yes (all).
  - Adj R^2: 0.01 and 0.25 (reported).
  - AR1 p-value: 0.000, 0.000.
  - AR2 p-value: 0.562, 0.984.
  - Hansen J-test p-value: 0.043, 0.022.
- Note: Robust standard errors clustered at country level; *, **, *** denote significance at the 10%, 5%, and 1% levels, respectively.

### Robustness checks (Appendix Table 3 highlights)
- Robustness strategies:
  - Truncate sample at 5th and 95th percentiles (truncated sample).
  - Use 3-year nonoverlapping intervals (3-year averages).
  - Add controls: real effective exchange rate (REER), fiscal balance.
  - Use interaction MI*FO (monetary independence × financial openness).
  - Replace country-specific terms-of-trade with international oil prices.
- Selected coefficient estimates (columns correspond to different robustness specifications):
  - Inflation_{t-1}: 0.257***, 0.041, 0.187***, 0.185***, 0.257*** (standard errors [0.052], [0.092], [0.059], [0.057], [0.047]).
  - Monetary independence: -0.383*, -0.890**, -1.091***, -0.975*** (standard errors [0.207], [0.394], [0.330], [0.259]).
  - Financial openness: -0.121, -0.516*, -0.070*, -0.639*** (standard errors [0.185], [0.293], [0.372], [0.181]).
  - Real GDP per capita: -0.055, -0.631**, -0.396, -0.178, -0.511*** (standard errors [0.154], [0.269], [0.279], [0.245], [0.178]).
  - Trade openness: 0.410**, 0.182, 0.272, 0.274, 0.394* (standard errors [0.200], [0.348], [0.292], [0.267], [0.209]).
  - Terms-of-trade (when included): -1.002***, -2.556***, -2.449***, -2.493*** (standard errors [0.209], [0.600], [0.362], [0.364]).
  - Money growth: 0.046*, 1.909***, 0.674***, 0.827***, 0.756*** (standard errors [0.306], [0.655], [0.284], [0.279], [0.232]).
  - REER: -0.007** (standard error [0.004]).
  - Fiscal balance: 0.002 (standard error [0.012]).
  - MI*FO: -0.547** (standard error [0.469]).
  - Oil Price (used instead of terms-of-trade): 0.941*** (standard error [0.126]).
- Robustness diagnostics:
  - Number of observations across robustness columns: 342, 144, 289, 283, 433.
  - Number of countries: 15, 15, 14, 15, 15.
  - AR1 p-values: 0.000 (all five reported).
  - AR2 p-values: 0.973, 0.431, 0.466, 0.462, 0.675.
  - Hansen J-test p-values: 0.208, 0.000, 0.072, 0.071, 0.036.
- Interpretation: Monetary independence maintains a significant negative effect on inflation across robustness checks, though magnitudes vary.

### Summary statistics (Appendix Table 2 key values)
- Consumer price inflation (Obs. 554): Mean 7.21, Std. Dev. 12.2, Min. -11.4, Max. 142.8.
- Monetary independence index (Obs. 516): Mean 0.50, Std. Dev. 0.10, Min. 0.1, Max. 0.9.
- Real GDP per capita (Obs. 544): Mean 6,137.6, Std. Dev. 10,043.9, Min. 1,327.0, Max. 52,2874.9. [as reported]
- Output gap (Obs. 570): Mean 0.0, Std. Dev. 2.5, Min. -14.3, Max. 11.0.
- Business cycle synchronization (Obs. 540): Mean -0.2, Std. Dev. 0.7, Min. -1.0, Max. 1.0.
- Trade openness (Obs. 542): Mean 104.13, Std. Dev. 32.2, Min. 31.1, Max. 275.0.
- Financial openness (Obs. 526): Mean 0.4, Std. Dev. 0.3, Min. 0.0, Max. 1.0.
- Real effective exchange rate index (Obs. 561): Mean 114.8, Std. Dev. 118.2, Min. 28.9, Max. 1293.6.
- Terms-of-trade (Obs. 549): Mean 145.4, Std. Dev. 43.2, Min. 55.5, Max. 314.6.
- Money growth (Obs. 540): Mean 12.9, Std. Dev. 16.6, Min. -10.3, Max. 205.6.
- Fiscal balance (Obs. 357): Mean 1.8, Std. Dev. 4.4, Min. -9.9, Max. 17.4.

### Conclusion and policy implications
- Core empirical conclusion:
  - Greater monetary independence is associated with lower consumer price inflation across a panel of 15 Caribbean countries during 1980–2017.
  - The negative effect of monetary independence on inflation is statistically significant across model specifications and estimation techniques (system GMM primary).
- Trade-offs and interpretation for exchange rate regimes:
  - Fixed exchange rate regimes outsource monetary policy to the anchor currency issuer, potentially providing credibility and low, stable inflation.
  - Floating regimes maximize monetary policy flexibility for domestic stabilization.
  - Empirical evidence here: pegged exchange rate regimes across the Caribbean do not necessarily deliver lower inflation for domestic purposes; greater monetary independence improves management of inflation dynamics.
  - Policy message: findings do not mandate abolishing exchange rate pegs but highlight the need to better align monetary policy stance with domestic conditions and to improve monetary transmission channels.

*Source: wpiea2019197-print-pdf (Chapter/section content provided).

---


_Source: https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019197-print-pdf.pdf_