## Appendix A: The CFA Franc Zone

## Source details

**Canonical URL:** [Appendix A: The CFA Franc Zone](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp06236.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp06236.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp06236.pdf.json)

---

### Background and scope
- The CFA Franc arrangement dates back to the mid-1940s and is among the longest standing fixed exchange rate regimes worldwide.
- The CFA Franc zone includes France and two monetary unions in Central and West Africa:
  - WAEMU: Benin, Burkina Faso, Côte d’Ivoire, Guinea-Bissau, Mali, Niger, Senegal, and Togo.
  - CEMAC: Cameroon, Central African Republic, Chad, Republic of the Congo, Equatorial Guinea, and Gabon.
- Each monetary union has a separate note-issuing central bank: La Banque Centrale des États de l’Afrique de l’Ouest (BCEAO) and La Banque des États de l’Afrique Centrale (BEAC). The Comorian franc is also pegged to the euro.

### Recent exchange-rate developments and competitiveness
- Over the past two years the CFA franc appreciated—along with the euro to which it is pegged—by more than 25 percent in nominal terms vis-à-vis the U.S. dollar, putting additional pressure on the region’s competitiveness.
- The 1994 CFA franc devaluation was 50 percent vis-à-vis the French franc and is identified as a major historical adjustment restoring external competitiveness.
- Cumulative REER movements:
  - CEMAC: appreciated cumulatively by about 33 percent through December 2000 and by a further 16 percent from January 2001 to December 2005; by December 2005 REER_C was at 87 percent of its pre-devaluation level.
  - WAEMU: appreciated cumulatively by about 22 percent through December 2000 and by a further 12 percent from January 2001 to September 2005; by December 2004 REER_W was at 76 percent of its pre-devaluation level.
- Cross-country REER variation by December 2005:
  - WAEMU: Senegal at 56 percent and Benin at 91 percent of pre-devaluation levels.
  - CEMAC: Equatorial Guinea at 115 percent and Gabon at 70 percent of pre-devaluation levels.

### Analytical approach and methodology
- Equilibrium assessment: fundamentals equilibrium exchange rate (FEER) approach based on Edwards (1989) and Johansen (1995) cointegration methodology.
- Rationale: distinguish whether REER movements represent misalignment or shifts in the equilibrium real effective exchange rate (EREER) driven by changes in economic fundamentals.
- Variables (log form) in vector Yt (6×1): Real effective exchange rate, Terms-of-trade, Government consumption, Technological progress, Investment, Openness.
- Permanent/transitory decomposition methods applied: Hodrick-Prescott (HP) filter (illustrative) and Gonzalo-Granger (GG) decomposition (theoretically appealing).

### Data and stylized facts
- Sample: annual observations for the period 1970-2005 for both CEMAC and WAEMU.
- Variables:
  - LREER (ln REER), LTTT (ln terms-of-trade), LCGR / LNCGR (ln government consumption/GDP), LPROD (ln real GDP per capita relative to trading partners), LOPEN (ln openness/GDP), LNIR (ln investment/GDP).
  - Dummy variables for 1994 devaluation and outliers.
- Stylized patterns:
  - Persistent decline in real GDP per capita relative to trading partners from mid-1970s through end of sample.
  - Investment increased starting in the 1990s.
  - Terms-of-trade volatile with average increases in the 2000s (oil for CEMAC; cotton, cocoa, gold for WAEMU).
  - CEMAC surge in foreign direct investment in 2000-2003 associated with oil-related construction in Chad and Equatorial Guinea.
  - Government consumption: CEMAC roughly constant until about 1990 then slight decline; WAEMU continuous decline since mid-1980s.
  - Region-level population and GDP (2003):
    - WAEMU eight countries: total population 76 million and combined GDP US $ 37 billion.
    - CEMAC six countries: total population 34 million and combined GDP US$ 28 billion.

### Econometric methodology and tests
- Stationarity:
  - ADF tests: cannot reject unit root in levels for all variables; strongly reject unit root in first differences → variables are I(1).
  - Example ADF statistics on first differences (reported with significance):
    - Dln(REER) ADF statistic 0: -6.09 0.00 ***
    - Dln(TTT) ADF statistic 0: -5.70 0.00 ***
    - Dln(NCGR) ADF statistic 0: -7.12 0.00 ***
    - Dln(NIR) ADF statistic 0: -6.10 0.00 ***
    - Dln(PROD) ADF statistic 0: -4.91 0.00 ***
    - Dln(OPEN) ADF statistic 0: -5.76 0.00 ***
- VAR/VECM setup:
  - CEMAC VAR includes impulse dummies for 1994, 1976, 1978, 1985, and 2001; simplified to VAR(2).
  - WAEMU VAR includes impulse dummies for 1994, 1974 & 1979, and 2003; simplified to VAR(1).
  - VAR residual diagnostics: no rejections at 5 percent level for serial correlation, normality, heteroskedasticity; VARs empirically well-specified.
  - Recursive estimation and Chow tests indicate VAR stability at the 1 percent significance level.
- Cointegration:
  - Johansen trace and max-eigen tests indicate presence of one cointegrating vector for each sample (CEMAC and WAEMU).
  - Interpretation: stable long-run relationship between REER and fundamentals.

### Estimation results: long-run coefficients (cointegrating vector) and short-run feedback
- Long-run cointegrating estimates (format: coefficient and reported significance as in source):
  - ln(terms of trade): CEMAC 0.70 *** (9.04) ; WAEMU 0.58 *** (5.16)
  - ln(government consumption): CEMAC 0.41 *** (2.71) ; WAEMU 0.69 *** (13.24)
  - ln(technological progress): CEMAC 0.59 *** (15.60) ; WAEMU 0.26 *** (5.08)
  - ln(investment): CEMAC -0.21 ** (2.50) ; WAEMU -0.28 *** (4.35)
  - ln(openness): CEMAC -0.24 ** (2.36) ; WAEMU -0.18 *** (2.72)
  - Constant: CEMAC 1.57 ; WAEMU 1.42
- Interpretation as elasticities (effect of a 1 percent increase):
  - CEMAC:
    - terms-of-trade → 0.70 percent appreciation of the REER
    - government consumption → 0.41 percent appreciation
    - technological progress → 0.59 percent appreciation
    - investment → 0.21 percent depreciation
    - openness → 0.24 percent depreciation
  - WAEMU:
    - terms-of-trade → 0.58 percent appreciation
    - government consumption → 0.69 percent appreciation
    - technological progress → 0.26 percent appreciation
    - investment → 0.28 percent depreciation
    - openness → 0.18 percent depreciation
- Short-run feedback coefficients (D[..]) with t-statistics (selected):
  - D[ln(real effective exchange rate)]: CEMAC -0.12 ** (1.96) ; WAEMU -0.24 *** (2.93)
  - D[ln(terms of trade)]: CEMAC 0.26 (1.45) ; WAEMU -0.15 (0.99)
  - D[ln(government consumption)]: CEMAC 0.2 (1.36) ; WAEMU 0.51 *** (6.13)
  - D[ln(technological progress)]: CEMAC -0.2 (1.19) ; WAEMU 0.09 (0.44)
  - D[ln(investment)]: CEMAC 0.33 *** (3.22) ; WAEMU 0.05 (0.75)
  - D[ln(openness)]: CEMAC -0.06 (0.35) ; WAEMU 0.05 (0.30)

### Dynamics, speed of adjustment, and impulse responses
- Half-life of deviation from equilibrium:
  - CEMAC: 5.6
  - WAEMU: 2.9
- Error-correction interpretation:
  - CEMAC: on average about 0.12 percent of the gap is eliminated every year → about half the gap would be closed within 5.6 years in the absence of further shocks.
  - WAEMU: on average about 0.24 percent of the gap is eliminated every year → about half the gap would be closed within 2.9 years in the absence of further shocks.
- Impulse response results:
  - Accumulated impulse responses align with long-run coefficient signs:
    - Positive investment and openness shocks → long-run depreciating effect on REER.
    - Positive terms-of-trade, government consumption, and productivity shocks → long-run appreciating effect on REER.
  - WAEMU non-accumulated impulse responses stabilize in about half the time it takes CEMAC responses to stabilize, consistent with faster WAEMU adjustment.

### Misalignment measurement and historical episodes
- Methods: HP filter (illustrative) and Gonzalo-Granger (GG) decomposition used to derive permanent component and EREER; GG used to identify statistically significant misalignments.
- Historical pattern (1985–2005):
  - Both WAEMU and CEMAC REERs were overvalued prior to 1994; supported the 1994 devaluation.
  - After 1994, both REERs generally remained above equilibrium for remainder of sample because of changing fundamentals.
  - CEMAC: REER temporarily exceeded equilibrium in 1999 and during 2001-04, with statistically significant misalignments in those episodes.
  - WAEMU: no statistically significant misalignments after the devaluation until a short period in 2003-04.
  - In 2005 both CEMAC and WAEMU REERs were slightly above estimated long-run equilibrium levels, but none of these overvaluations were statistically significant.

### Contributions to recent appreciation (2000–2005)
- WAEMU (2001-2005):
  - REER appreciated by about 12 percent.
  - Increases in terms-of-trade and government consumption accounted for an appreciation of the equilibrium real exchange rate in the order of about 9 percent each.
  - Increases in investment and openness and decreases in the productivity index contributed to REER depreciations of 2, 1, and 3 percent, respectively.
- CEMAC (2001-2005):
  - REER appreciated by 13 percent.
  - Decomposition: appreciation of about 23 percent from increases in terms-of-trade; depreciation caused by government consumption and productivity decreases (2 and 7 percent, respectively) and openness increase (about 1 percent).

### Policy-relevant implications and conclusions
- The long-run behavior of REERs for WAEMU and CEMAC can be largely explained by terms-of-trade, government consumption, investment, openness and productivity.
- Evidence points to overvaluation before 1994 (supporting devaluation), followed by recent real appreciations that generated short episodes of temporary overvaluation.
- As of 2005, no statistically significant misalignment detected for either region — implying no immediate exchange rate adjustment needed.
- Given overall sound macroeconomic policies, strong commodity prices, increasing reserve levels, limited capital flows, and cautiously optimistic market assessments, there are no immediate macroeconomic imbalances that would call for correction through an exchange rate adjustment.
- Fixed exchange rate regimes can be sustainable if deviations from long-term equilibrium rates are small and mean-reverting; one-sided, sustained significant misalignments could necessitate exchange rate action in addition to demand management.

### Selected quantitative diagnostics and representative figures
- WAEMU period percentage changes (Table C1a):
  - Real effective exchange rate Jan 1994–Dec 1998: 35.4
  - Real effective exchange rate Jan 1999–Dec 2000: -8.9
  - Real effective exchange rate Jan 2001–Dec 2005: 9.4
- CEMAC period percentage changes (Table C1b):
  - Real effective exchange rate Jan 1994–Dec 1998: 51.2
  - Real effective exchange rate Jan 1999–Dec 2000: -11.3
  - Real effective exchange rate Jan 2001–Dec 2005: 13.4
- VAR reduction test examples (Table C3):
  - CEMAC VAR(2) to VAR(1): F(36,42) = 2.07, p-value = 0.012 **
  - WAEMU VAR(2) to VAR(1): F(36,51) = 1.53, p-value = 0.080
- Diagnostic examples (Table C4):
  - CEMAC AR 1-2 test for LREER: F(2,13) = 0.81, p-value = 0.467
  - WAEMU AR 1-2 test for LREER: F(2,23) = 0.09, p-value = 0.915
  - CEMAC Normality LREER: Chi^2(2) = 5.8, p-value = 0.054
  - WAEMU Normality LREER: Chi^2(2) = 0.21, p-value = 0.899
- Johansen cointegration excerpts (Table C5):
  - CEMAC Trace test interpretation: 1 cointegrating eqn(s) at the 0.05 level.
  - WAEMU Trace test interpretation: 1 cointegrating eqn(s) at the 0.05 level.
  - WAEMU Max-Eigen test note: None: Eigenvalue = 0.835, Statistic = 0.79, p-value = 0.00 *** (Max-eigen interpretation: 1 cointegrating eqn(s) at the 0.01 level).

*Source: Appendix A: The CFA Franc Zone (from the provided IMF content unit).*

### Appendix A: The CFA Franc Zone .........................................................................................

### Appendix A: The CFA Franc Zone

### Background and scope
- The CFA Franc arrangement dates back to the mid-1940s and is among the longest standing fixed exchange rate regimes worldwide.
- The CFA Franc zone includes France and two monetary unions in Central and West Africa:
  - WAEMU: Benin, Burkina Faso, Côte d’Ivoire, Guinea-Bissau, Mali, Niger, Senegal, and Togo.
  - CEMAC: Cameroon, Central African Republic, Chad, Republic of the Congo, Equatorial Guinea, and Gabon.

### Recent exchange-rate developments and competitiveness
- Over the past two years the CFA franc appreciated—along with the euro to which it is pegged—by more than 25 percent in nominal terms vis-à-vis the U.S. dollar, putting additional pressure on the region’s competitiveness.
- This appreciation has renewed interest in the prospects of and the outlook for the CFA franc.

### Analytical approach and methodology
- The paper assesses actual REER movements relative to their long-run equilibrium values using the fundamentals equilibrium exchange rate (FEER) approach based on the Edwards (1989) model and the Johansen (1995) cointegration methodology.
- Rationale for the fundamentals approach: it helps distinguish whether REER movements represent misalignment or shifts in the equilibrium real effective exchange rate (EREER) driven by changes in economic fundamentals.

### Key empirical findings
- The proposed fundamentals account for most of the fluctuation of the real effective exchange rates.
- Estimated effects of fundamentals on the REER:
  - Increases in the terms of trade cause the exchange rate to appreciate.
  - Increases in government consumption cause the exchange rate to appreciate.
  - Productivity improvements cause the exchange rate to appreciate.
  - Increases in investment lead to a depreciation.
  - Increases in openness lead to a depreciation.
- Misalignment at end-2005:
  - Both the WAEMU and CEMAC real exchange rates were slightly more appreciated than their estimated long-run equilibrium levels at end-2005.
  - The estimated misalignments are not statistically significant.
- Dynamics and adjustment:
  - A feedback effect is identified for both CEMAC and WAEMU, indicating reversion to the time-varying long-run equilibrium following a shock.
  - The speed of reversion is about two times faster in WAEMU than in CEMAC.

### Structure of the paper (organization)
- Section II: background, brief literature review, and empirical formulation of the FEER.
- Section III: econometric analysis and results (enumeration in the source outlines appendices and tables for unit roots, cointegration, diagnostics, impulse responses, and stability tests).

*Source: Appendix A: The CFA Franc Zone (from the provided IMF content unit).*

### Appendix A provides some background and key CFA Franc zone dates. Also, see Hadjimichael and Galy

### _wp06236 - Appendix A provides some background and key CFA Franc zone dates. Also, see Hadjimichael and Galy

### Background and literature
- CFA Franc zone composition: WAEMU, CEMAC, Comoros, and France form the Franc zone. Each monetary union has a separate note-issuing central bank: La Banque Centrale des États de l’Afrique de l’Ouest (BCEAO) and La Banque des États de l’Afrique Centrale (BEAC). The Comorian franc is also pegged to the euro.
- Relevant prior work: Hadjimichael and Galy (1997) and Masson and Pattillo (2004).
- Approaches to estimating equilibrium real exchange rates discussed:
  - Uncovered interest parity (UIP) and purchasing power parity (PPP).
  - Fundamental equilibrium exchange rate (FEER), underlying internal-external balance (UIEB), desired equilibrium exchange rate (DEER), natural real exchange rate (NATREX).
  - Behavioral equilibrium exchange rate (BEER).
- Empirical regularities: UIP better explains adjustment paths than levels; PPP not strongly supported empirically; combining UIP and PPP via cointegration (CHEER) can raise speed of convergence estimates.

### FEER model specification and predicted signs
- FEER defines equilibrium as simultaneous attainment of internal and external balance (Edwards (1989) framework).
- Key modeled fundamentals and expected sign on REER:
  - Terms-of-trade of goods: expected sign positive (wealth effect → REER appreciation).
  - Government consumption as share of GDP: expected sign ambiguous (depends tradable vs nontradable composition).
  - Degree of trade controls/restrictions (proxied by openness): expected sign negative (liberalization → depreciation).
  - Productivity (Balassa-Samuelson effect): expected sign positive (higher tradable productivity → REER appreciation).
  - Investment: expected sign ambiguous a priori; given high import content in developing countries, expected sign negative (investment ↑ → REER depreciation).
- Misalignment measurement requires decomposing fundamentals into permanent and transitory components (methods: Beveridge-Nelson, moving average, Hodrick-Prescott, Quah, Casa, Gonzalo-Granger).

### Econometric methodology
- VAR/VECM framework using Johansen maximum likelihood procedure to test for cointegration among REER and fundamentals.
- Vector Yt is a 6×1 vector: [Real effective exchange rate, Terms-of-trade, Government consumption, Technological progress, Investment, Openness] (all variables in natural logarithm).
- If rank(π) = r with 0 < r < 6, π = αβ′ and VECM form used; α columns represent speeds of adjustment; β′ contains cointegrating vector(s).
- Permanent/transitory decomposition methods applied:
  - Hodrick-Prescott (HP) filter (illustrative; choice of λ affects results).
  - Gonzalo-Granger (GG) decomposition (theoretically appealing: transitory component does not Granger-cause permanent component; provides observable permanent component and allows error bands).

### Data
- Samples: annual observations for the period 1970-2005 for both CEMAC and WAEMU.
- Variables: LREER (natural logarithm of REER), LTTT (ln terms-of-trade), LCGR (ln government consumption/GDP), LPROD (ln real GDP per capita relative to trading partners), LOPEN (ln openness/GDP), LNIR (ln investment/GDP). Dummy variables for 1994 devaluation and outliers.
- Stylized patterns:
  - 1994 CFA franc devaluation by 50 percent vis-à-vis the French franc.
  - Post-1994: steady appreciation of REERs.
  - Cumulative REER movements:
    - CEMAC: REER_C appreciated cumulatively by about 33 percent through December 2000 and by a further 16 percent from January 2001 to December 2005; by December 2005 REER_C was at 87 percent of its pre-devaluation level.
    - WAEMU: REER_W appreciated cumulatively by about 22 percent through December 2000 and by a further 12 percent from January 2001 to September 2005; by December 2004 REER_W was at 76 percent of its pre-devaluation level.
  - Cross-country REER variation by December 2005 (WAEMU): Senegal at 56 percent and Benin at 91 percent of pre-devaluation levels.
  - Cross-country REER variation (CEMAC): Equatorial Guinea at 115 percent and Gabon at 70 percent of pre-devaluation levels.
  - Persistent decline in real GDP per capita relative to trading partners from mid-1970s through end of sample.
  - Investment increased starting in the 1990s; terms-of-trade volatile with average increases in the 2000s (oil for CEMAC; cotton, cocoa, gold for WAEMU).
  - CEMAC surge in foreign direct investment in 2000-2003 associated with oil-related construction in Chad and Equatorial Guinea.
  - Government consumption patterns: CEMAC roughly constant until about 1990 then slight decline; WAEMU continuous decline since mid-1980s.
  - Region-level population and GDP (2003):
    - WAEMU eight countries: total population 76 million and combined GDP US $ 37 billion.
    - CEMAC six countries: total population 34 million and combined GDP US$ 28 billion.

### Time-series properties and model selection
- Augmented Dickey-Fuller (ADF) tests:
  - Cannot reject unit root in levels for all variables.
  - Strongly reject unit root in first differences → variables are I(1).
- VAR specification and impulse dummies:
  - CEMAC VAR includes impulse dummies for 1994, 1976, 1978, 1985, and 2001.
  - WAEMU VAR includes impulse dummies for 1994, 1974 & 1979, and 2003.
  - Economic justifications: devaluation, oil price shocks, Côte d’Ivoire crisis, country-specific GDP shocks, surge in FDI and terms-of-trade increases.
- Lag-order selection and residual diagnostics:
  - Simplification tests led to VAR(2) for CEMAC and VAR(1) for WAEMU.
  - VAR residual diagnostics: no rejections at 5 percent level for serial correlation, normality, heteroskedasticity; VARs empirically well-specified.
  - Recursive estimation and Chow tests indicate VAR stability at the 1 percent significance level.
- Cointegration tests:
  - Trace statistics reject the null of zero cointegrating vectors for both CEMAC and WAEMU.
  - Tests indicate presence of one cointegrating vector for each sample.

### Estimation results: long-run and short-run relationships (selected estimates)
- Estimates of the cointegrating relationships (β coefficients) and reported t-statistics (format: coefficient [t-statistic]):
  - ln(terms of trade): CEMAC 0.70 *** (9.04) ; WAEMU 0.58 *** (5.16)
  - ln(government consumption): CEMAC 0.41 *** (2.71) ; WAEMU 0.69 *** (13.24)
  - ln(technological progress): CEMAC 0.59 *** (15.60) ; WAEMU 0.26 *** (5.08)
  - ln(investment): CEMAC -0.21 ** (2.50) ; WAEMU -0.28 *** (4.35)
  - ln(openness): CEMAC -0.24 ** (2.36) ; WAEMU -0.18 *** (2.72)
  - Constant: CEMAC 1.57 ; WAEMU 1.42
  - Significance notation preserved exactly as reported: **, ***.
- Estimates of the short-term feedback coefficients (D[..]) with t-statistics:
  - D[ln(real effective exchange rate)]: CEMAC -0.12 ** (1.96) ; WAEMU -0.24 *** (2.93)
  - D[ln(terms of trade)]: CEMAC 0.26 (1.45) ; WAEMU -0.15 (0.99)
  - D[ln(government consumption)]: CEMAC 0.2 (1.36) ; WAEMU 0.51 *** (6.13)
  - D[ln(technological progress)]: CEMAC -0.2 (1.19) ; WAEMU 0.09 (0.44)
  - D[ln(investment)]: CEMAC 0.33 *** (3.22) ; WAEMU 0.05 (0.75)
  - D[ln(openness)]: CEMAC -0.06 (0.35) ; WAEMU 0.05 (0.30)
- Half-life of deviation from equilibrium:
  - CEMAC: 5.6
  - WAEMU: 2.9

### Key empirical findings and implications
- There is a stable long-run relationship between REER and identified fundamentals in both CEMAC and WAEMU (one cointegrating vector).
- Long-run determinants: terms-of-trade, government consumption, technological progress, investment, and openness all enter the cointegrating vector with expected signs in most cases (terms-of-trade, government consumption, and productivity positive; investment and openness negative).
- Short-run adjustment:
  - Feedback coefficients indicate that shocks to the REER correct toward equilibrium with estimated half-lives of 5.6 years (CEMAC) and 2.9 years (WAEMU).
  - Short-run effects of government consumption and investment are significant in some specifications, indicating plausible short-term channels for REER dynamics.
- Policy-relevant observations:
  - The 1994 devaluation (50 percent) restored external competitiveness; subsequent REER appreciation partly reflects euro strengthening and commodity price movements.
  - Liberalization of capital inflows (proxied by BFDIR) shows no robust net long-run effect on REER (estimated BFDIR coefficients: WAEMU 0.012 with t-statistic 1.17; CEMAC 0.018 with t-statistic 1.42), suggesting offsetting channels.
  - Structural factors (commodity dependence, shallow/segmented financial markets, limited increase in internal trade and factor mobility) shape regional REER dynamics and resilience of the exchange arrangement.

*Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp06236.pdf*

### 1. Three asterisks and two asterisks,denote statistical significance at the 0.01 and

### _wp06236 - 1. Three asterisks and two asterisks,denote statistical significance at the 0.01 and

### Cointegration estimation — long-run relationships
- Estimated cointegration equations for the dependent variable ln(Real Effective Exchange Rate) are consistent with economic theory; all estimated coefficients (elasticities) have the expected signs and are strongly significant.
- Long-run sign and interpretation of fundamentals (both CEMAC and WAEMU samples):
  - Terms-of-trade: positively correlated with the REER (improvement → appreciation via possible wealth effect).
  - Government consumption: positive (appreciating) impact on the REER (implying government spending concentrated on nontradables).
  - Technological progress (proxied by relative real GDP per capita): positive long-term impact (Ballassa-Samuelson effect).
  - Investment: negatively correlated with the REER (investment increases spending towards traded goods → depreciation).
  - Openness: increases associated with depreciation of the REER (through increases in imports).

### Estimated elasticities / marginal impacts (effect of a 1 percent increase)
- CEMAC:
  - terms-of-trade → 0.70 percent appreciation of the REER
  - government consumption (share of GDP) → 0.41 percent appreciation of the REER
  - technological progress → 0.59 percent appreciation of the REER
  - investment (share of GDP) → 0.21 percent depreciation of the REER
  - openness → 0.24 percent depreciation of the REER
- WAEMU:
  - terms-of-trade → 0.58 percent appreciation of the REER
  - government consumption (share of GDP) → 0.69 percent appreciation of the REER
  - technological progress → 0.26 percent appreciation of the REER
  - investment → 0.28 percent depreciation of the REER
  - openness → 0.18 percent depreciation of the REER

### Short-run dynamics and weak exogeneity
- Feedback coefficients for the cointegrating vectors (short-run relationships) indicate some fundamentals are not weakly exogenous; when deviations from long-run equilibrium occur these variables jointly respond and move the system back to equilibrium.
- The feedback coefficient for the DLREER equation is negative and significantly different from zero for both CEMAC and WAEMU, suggesting stability of the error correction mechanism.

### Misalignment episodes and equilibrium REER estimation
- Two methods used to derive permanent component of fundamentals: HP filter (illustrative) and GG decomposition (theoretically attractive). EREERs computed from long-term components of fundamentals; GG decomposition used to identify statistically significant misalignments.
- Historical pattern (1985–2005, HP and GG results similar, especially for CEMAC):
  - Both WAEMU and CEMAC REERs were overvalued prior to 1994 (CEMAC: actual REER well above equilibrium; WAEMU: less so), supporting the 1994 CFA devaluation as warranted.
  - After 1994, both REERs generally remained above equilibrium for the remainder of the sample because of changes in fundamentals differing across regions.
  - CEMAC: REER temporarily exceeded equilibrium in 1999 and during 2001-04, with statistically significant misalignments in those episodes.
  - WAEMU: no statistically significant misalignments after the devaluation until a short period in 2003-04.
  - In 2005 both CEMAC and WAEMU REERs were slightly above estimated long-run equilibrium levels, but none of these overvaluations were statistically significant — implying both REERs were broadly in line with long-run equilibrium at end-2005.

### Contributions to recent appreciation (2000–2005)
- WAEMU (2001-2005):
  - REER appreciated by about 12 percent.
  - Increases in terms-of-trade and government consumption accounted for an appreciation of the equilibrium real exchange rate in the order of about 9 percent each.
  - Increases in investment and openness and decreases in the productivity index contributed to REER depreciations of 2, 1, and 3 percent, respectively.
- CEMAC (2001-2005):
  - REER appreciated by 13 percent.
  - Decomposition: appreciation of about 23 percent from increases in terms-of-trade; depreciation caused by government consumption and productivity decreases (2 and 7 percent, respectively) and openness increase (about 1 percent).

### Speed of adjustment (error-correction interpretation)
- CEMAC:
  - On average about 0.12 percent of the gap is eliminated every year.
  - Implies that, in the absence of further shocks, about half the gap would be closed within 5.6 years.
- WAEMU:
  - On average about 0.24 percent of the gap is eliminated every year.
  - Implies that, in the absence of further shocks, about half the gap would be closed within 2.9 years.
- Larger deviations (e.g., the 1994 devaluation) may take much longer to absorb.
- Comparative context: WAEMU and CEMAC adjustment speeds are described as reasonable relative to literature estimates (examples cited: half-life less than a year for Malawi; 2–2.5 years for South Africa; 3–5 years in some studies).

### Impulse response and dynamic propagation
- Impulse response functions (generalized impulses) trace the effect of a one-time one standard deviation shock to fundamentals on current and future values and on the REER.
- Accumulated impulse responses (CEMAC and WAEMU) align with long-run coefficient signs:
  - Positive investment and openness shocks → long-run depreciating effect on REER.
  - Positive terms-of-trade, government consumption, and productivity shocks → long-run appreciating effect on REER.
- WAEMU non-accumulated impulse responses stabilize in about half the time it takes CEMAC responses to stabilize, consistent with the faster WAEMU adjustment speed.

### Policy-relevant implications and conclusions
- The long-run behavior of REERs for WAEMU and CEMAC can be largely explained by terms-of-trade, government consumption, investment, openness and productivity.
- Evidence points to overvaluation before 1994 (supporting devaluation), followed by recent real appreciations that generated short episodes of temporary overvaluation.
- As of 2005, no statistically significant misalignment detected for either region — no immediate exchange rate adjustment needed.
- Given overall sound macroeconomic policies, strong commodity prices, increasing reserve levels, limited capital flows, and cautiously optimistic market assessments, there are no immediate macroeconomic imbalances that would call for correction through an exchange rate adjustment.
- Fixed exchange rate regimes can be sustainable if deviations from long-term equilibrium rates are small and mean-reverting; one-sided, sustained significant misalignments could necessitate exchange rate action in addition to demand management.

### Selected quantitative tables and measures (representative figures from appendices)
- WAEMU period percentage changes (Table C1a):
  - Real effective exchange rate Jan 1994–Dec 1998: 35.4
  - Real effective exchange rate Jan 1999–Dec 2000: -8.9
  - Real effective exchange rate Jan 2001–Dec 2005: 9.4
- CEMAC period percentage changes (Table C1b):
  - Real effective exchange rate Jan 1994–Dec 1998: 51.2
  - Real effective exchange rate Jan 1999–Dec 2000: -11.3
  - Real effective exchange rate Jan 2001–Dec 2005: 13.4
- Unit-root / ADF table excerpts (appendix C) report, for example:
  - Dln(REER) ADF statistic 0: -6.09 0.00 *** (with 1% level -3.63, 5% level -2.95, 10% level -2.61)
  - Dln(TTT) ADF statistic 0: -5.70 0.00 *** (levels as above)
  - Dln(NCGR) ADF statistic 0: -7.12 0.00 *** (levels as above)
  - Dln(NIR) ADF statistic 0: -6.10 0.00 *** (levels as above)
  - Dln(PROD) ADF statistic 0: -4.91 0.00 *** (levels as above)
  - Dln(OPEN) ADF statistic 0: -5.76 0.00 *** (levels as above)

*Source: IMF staff analysis as presented in the supplied chapter content.*

### 1. D denotes the difference operator.

### _wp06236 - 1. D denotes the difference operator.

### Unit Root Tests (Augmented Dickey-Fuller)
- Definition and equation:
  - D denotes the difference operator.
  - For a given variable x, the augmented Dickey-Fuller equation has the following form: (equation shown in source) where εt is a white noise disturbance.
  - The augmented Dickey-Fuller statistic, t-ADF, is the t-ratio on π. The statistic tests the null hypothesis of a unit root in x, i.e. π = 0, against the alternative of stationarity.
- Reporting conventions:
  - The table reports the number of lags on the dependent variable, p, chosen using the Schwartz information criterion.
  - The table also reports the critical values at the 1 percent and 5 percent significance levels.
  - The symbols ** and *** denote rejection of the null hypothesis at the 5 percent and 1 percent critical values, respectively.

### VAR Model Reduction Tests (Table C3)
- CEMAC model reductions and test statistics:
  - VAR(3) to VAR(2): F(36,15) = 1.14, p-value = 0.407
  - VAR(3) to VAR(1): F(72,22) = 1.64, p-value = 0.096
  - VAR(2) to VAR(1): F(36,42) = 2.07, p-value = 0.012 **
- WAEMU model reductions and test statistics:
  - VAR(3) to VAR(2): F(36,24) = 1.38, p-value = 0.205
  - VAR(3) to VAR(1): F(72,33) = 1.56, p-value = 0.081
  - VAR(2) to VAR(1): F(36,51) = 1.53, p-value = 0.080
- Notes on model composition:
  - CEMAC VARs include the variables: LREER, LTTT, LNCGR, LNIR, LPROD, LOPEN, a constant, and five dummy variables for 1994, 2001, 1978, 1976, and 1985.
  - WAEMU VARs include the variables: LREER, LTTT, LNCGR, LNIR, LPROD, LOPEN, a constant, and three dummy variables for 1994, 1974-1979, and 2003.
- Interpretation:
  - The F statistic tests the null hypothesis indicated by the model to the right of the arrow against the maintained hypothesis given by the model to the left of the arrow. Tail probabilities associated with the values of the F statistic are reported under p-Value.
  - The symbol ** denotes rejection of the null hypothesis at the 5 percent critical value.

### Diagnostic Tests for Residuals (Table C4)
- CEMAC (VAR includes 2 lags on each variable; constant; five dummies):
  - AR 1-2 test:
    - LREER: F(2,13) = 0.81, p-value = 0.467
    - LTTT: F(2,13) = 1.21, p-value = 0.329
    - LNCGR: F(2,13) = 0.57, p-value = 0.580
    - LNIR: F(2,13) = 2.25, p-value = 0.145
    - LPROD: F(2,13) = 0.99, p-value = 0.397
    - LOPEN: F(2,13) = 1.98, p-value = 0.178
  - Normality test:
    - LREER: Chi^2(2) = 5.8, p-value = 0.054
    - LTTT: Chi^2(2) = 0.33, p-value = 0.849
    - LNCGR: Chi^2(2) = 1.07, p-value = 0.587
    - LNIR: Chi^2(2) = 5.18, p-value = 0.075
    - LPROD: Chi^2(2) = 0.53, p-value = 0.766
    - LOPEN: Chi^2(2) = 2.65, p-value = 0.266
  - Heteroscedasticity test:
    - LREER: Chi^2(24) = 21.25, p-value = 0.624
    - LTTT: Chi^2(24) = 18.01, p-value = 0.803
    - LNCGR: Chi^2(24) = 24.87, p-value = 0.413
    - LNIR: Chi^2(24) = 24.28, p-value = 0.446
    - LPROD: Chi^2(24) = 24.91, p-value = 0.411
    - LOPEN: Chi^2(24) = 21.78, p-value = 0.593
- WAEMU (VAR includes 1 lag on each variable; constant; three dummies):
  - AR 1-2 test:
    - LREER: F(2,23) = 0.09, p-value = 0.915
    - LTTT: F(2,23) = 3.04, p-value = 0.067
    - LNCGR: F(2,23) = 1.75, p-value = 0.196
    - LNIR: F(2,23) = 0.57, p-value = 0.573
    - LPROD: F(2,23) = 1.97, p-value = 0.163
    - LOPEN: F(2,23) = 1.83, p-value = 0.184
  - Normality test:
    - LREER: Chi^2(2) = 0.21, p-value = 0.899
    - LTTT: Chi^2(2) = 2.6, p-value = 0.273
    - LNCGR: Chi^2(2) = 2.8, p-value = 0.247
    - LNIR: Chi^2(2) = 5.82, p-value = 0.054
    - LPROD: Chi^2(2) = 1.93, p-value = 0.381
    - LOPEN: Chi^2(2) = 0.8, p-value = 0.671
  - Heteroscedasticity test:
    - LREER: Chi^2(12) = 7.88, p-value = 0.795
    - LTTT: Chi^2(12) = 19.95, p-value = 0.068
    - LNCGR: Chi^2(12) = 5.03, p-value = 0.957
    - LNIR: Chi^2(12) = 25.48, p-value = 0.051
    - LPROD: Chi^2(12) = 8.54, p-value = 0.742
    - LOPEN: Chi^2(12) = 12.98, p-value = 0.371
- Notes:
  - CEMAC VAR includes 2 lags on each variable (LREER, LTTT, LNCGR, LNIR, LPROD, LOpen), a constant, and five dummy variables for 1994, 2001, 1978, 1976, and 1985.
  - WAEMU VAR includes a single lag on each variable (LREER, LTTT, LNCGR, LNIR, LPROD, LOpen), a constant, and three dummy variables for 1994, 1974-1979, and 2003.

### Johansen Cointegration Tests (Table C5)
- CEMAC Trace test (Eigenvalues and statistics; p-values):
  - None: Eigenvalue = 0.8599, Statistic = 9.79, p-value = 0.02 **
  - At most 1: Eigenvalue = 0.645, Statistic = 5.63, p-value = 0.40
  - At most 2: Eigenvalue = 0.503, Statistic = 2.34, p-value = 0.60
  - At most 3: Eigenvalue = 0.401, Statistic = 6.57, p-value = 0.68
  - At most 4: Eigenvalue = 0.164, Statistic = 4.71, p-value = 0.84
  - At most 5: Eigenvalue = 0.03, Statistic = 0.63, p-value = 0.43
  - Interpretation: The Trace test indicates 1 cointegrating eqn(s) at the 0.05 level.
- CEMAC Max-Eigen test:
  - None: Eigenvalue = 0.85, Statistic = 4.16, p-value = 0.01 **
  - At most 1: Eigenvalue = 0.64, Statistic = 3.29, p-value = 0.52
  - At most 2: Eigenvalue = 0.50, Statistic = 15.77, p-value = 0.69
  - At most 3: Eigenvalue = 0.40, Statistic = 11.86, p-value = 0.57
  - At most 4: Eigenvalue = 0.16, Statistic = 4.08, p-value = 0.84
  - At most 5: Eigenvalue = 0.03, Statistic = 0.63, p-value = 0.43
  - Interpretation: The Max-eigenvalue test indicates 1 cointegrating eqn(s) at the 0.05 level.
- WAEMU Trace test:
  - None: Eigenvalue = 0.839, Statistic = 9.82, p-value = 0.02 **
  - At most 1: Eigenvalue = 0.544, Statistic = 9.03, p-value = 0.68
  - At most 2: Eigenvalue = 0.382, Statistic = 6.73, p-value = 0.86
  - At most 3: Eigenvalue = 0.261, Statistic = 2.78, p-value = 0.90
  - At most 4: Eigenvalue = 0.124, Statistic = 4.17, p-value = 0.88
  - At most 5: Eigenvalue = 0.01, Statistic = 0.31, p-value = 0.58
  - Interpretation: The Trace test indicates 1 cointegrating eqn(s) at the 0.05 level.
- WAEMU Max-Eigen test:
  - None: Eigenvalue = 0.835, Statistic = 0.79, p-value = 0.00 ***
  - At most 1: Eigenvalue = 0.542, Statistic = 2.30, p-value = 0.60
  - At most 2: Eigenvalue = 0.381, Statistic = 3.95, p-value = 0.82
  - At most 3: Eigenvalue = 0.268, Statistic = 8.60, p-value = 0.86
  - At most 4: Eigenvalue = 0.123, Statistic = 3.87, p-value = 0.87
  - At most 5: Eigenvalue = 0.01, Statistic = 0.31, p-value = 0.58
  - Interpretation: The Max-eigenvalue test indicates 1 cointegrating eqn(s) at the 0.01 level.
- Notes:
  - CEMAC VAR includes 2 lags on each variable (LREER, LTTT, LNCGR, LNIR, LPROD, LOpen), a constant, and five dummy variables for 1994, 2001, 1978, 1976, and 1985.
  - WAEMU VAR includes a single lag on each variable (LREER, LTTT, LNCGR, LNIR, LPROD, LOpen), a constant, and three dummy variables for 1994, 1974-1979, and 2003.
  - ** (***) denotes rejection of the null hypothesis at the 5% (1%) significance level.

### Cointegration Variables and Time Series Figures (Figures C1–C3)
- CEMAC cointegration variables (plotted series labels and ranges shown in source figures):
  - LREER_C, LTTT_C, LNCGR_C, LNIR_C, LPROD1_C, LOPEN_C with time span 1975–2005 and value ranges as depicted.
- WAEMU cointegration variables (plotted series labels and ranges shown in source figures):
  - LREER_W, LTTT_W, LNCGR_W, LNIR_W, LPROD1_W, LOPEN_W with time span 1975–2005 and value ranges as depicted.
- Exchange rates and relative prices (Figures C2a and C2b):
  - Index 1990=100; series include USD per 10,000 CFAF (left scale), NEER (right scale), REER (right scale), Relative Price Index (right scale); time span Jan-93 through Jan-05.
- Real Effective Exchange Rates by member countries (Figures C3a and C3b):
  - Index 1990=100; CEMAC members plotted: Cameroon, Central African Rep., Chad, Congo, Gabon, CEMAC, Equatorial Guinea; WAEMU members plotted: Benin, Burkina Faso, Côte d'Ivoire, Guinea-Bissau, Niger, Togo, WAEMU, Senegal, Mali; time span Jan-93 through Jan-05.

### Impulse Response Functions and Accumulated Responses (Figures C4a and C4b)
- CEMAC (Figure C4a):
  - Response of LREER_C to Generalized One S.D. Innovations (24 periods shown).
  - Accumulated Response of LREER_C to Generalized One S.D. Innovations (24 periods shown).
  - Variables shown: LREER_C, LTTT_C, LNCGR_C, LPROD1_C, LNIR_C, LOPEN_C.
- WAEMU (Figure C4b):
  - Response of LREER_W to Generalized One S.D. Innovations (24 periods shown).
  - Accumulated Response of LREER_W to Generalized One S.D. Innovations (24 periods shown).
  - Variables shown: LREER_W, LTTT_W2, LNCGR_W, LNIR_W, LPROD1_W, LOPEN_W.

### Model Stability Tests (Recursive Chow Tests) (Figures C5a and C5b)
- CEMAC recursive Chow tests (Figure C5a):
  - Plots shown for Ndn LREERu, Ndn LTTTu, Ndn LNCGRu, Ndn LNIRu, Ndn LPROD1u, Ndn LOpenU, and Ndn CHOWs with 1% critical line; time span 1995–2005.
- WAEMU recursive Chow tests (Figure C5b):
  - Plots shown for Ndn LREERu, Ndn LTTTuu, Ndn LNCGRu, Ndn LNIRu, Ndn LPROD1u, Ndn LOPENu, and Ndn CHOWs with 1% critical line; time span 1990–2005.

### References (selected items as listed)
- Acemoglu, D., S. Johnson, J.A. Robinson, and Y. Thaicharoen, 2003, “Institutional Causes, Macroeconomic Symptoms: Volatility, Crises and Growth,” Journal of Monetary Economics, Vol. 50, pp. 49-123.
- Baxter, M. and R.G. King, 1999, “Measuring Business Cycles: Approximate Band-Pass Filters for Economic Time Series,” Review of Economics and Statistics, Vol. 81, pp. 575-93.
- Johansen, S., 1988, “Statistical Analysis of Cointegrating Vectors,” Journal of Economic Dynamics and Control, Vol. 12, pp. 231-54.
- Johansen, S., 1995, Likelihood-based Inference in Cointegrated Vector Autoregressive Models (United Kingdom: Oxford University Press).
- Masson, P., and C. Pattillo, 2001, Monetary Union in West Africa (ECOWAS)—Is it Desirable and How Could It be Achieved? Occasional Paper No. 204 (Washington: International Monetary Fund).
- (Additional references appear in the source list.)

*Source: _wp06236 - 1. D denotes the difference operator.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp06236.pdf_