## Testing the Purchasing Power Parity (PPP) in West and Central Africa

## Source details

**Canonical URL:** [Testing the Purchasing Power Parity (PPP) in West and Central Africa](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025119-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025119-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025119-print-pdf.pdf.json)

---

### Scope and approach
- Investigates the Purchasing Power Parity (PPP) hypothesis within the CFA currency zone and the Economic Community of West African States (ECOWAS).
- Uses four numeraire currencies: the euro, the US dollar, the Chinese yuan (the renminbi), and the CFA franc.
- Employs multiple methods to test absolute and relative PPP and considers the potential impact of the Balassa-Samuelson effect on long-term real exchange rate trends.
- Sample and data:
  - Monthly data on CPI, CPI inflation, and nominal exchange rates from IMF’s International Financial Statistics (IFS).
  - Sample period: 1986M12-2020M02.
  - Monthly interpolated annual price levels from Penn World Table 10.01 (PWT) combined with monthly CPI via multiplicative chaining.
  - Cross-country averages weighted by each country’s contribution to zone real GDP in 2017 PPP international dollars.
  - All monthly series seasonally adjusted using X-13 ARIMA-SEATS.

### Key findings
- Evidence of PPP variability
  - PPP is supported within the CFA and ECOWAS regions, but empirical support varies with the choice of numeraire and the country grouping analyzed.
  - Evidence for PPP is slightly stronger for the euro compared to other numeraires.
  - Possible explanation: adjustment of price structures of CFA countries (15 out of 21 countries) to the Euro area countries enforced by the longstanding peg.
- Regional heterogeneity
  - Within the CFA zone, stronger support for PPP is found in the WAEMU sub-region compared to the CEMAC sub-region.
  - The discrepancy may be attributed to economic structure differences, notably the significant share of oil-exporting nations in the CEMAC.
- Choice of numeraire matters
  - The euro produced more consistent evidence for PPP than the US dollar and the renminbi.
  - The renminbi emerges post-1994 as the numeraire toward which reversion to PPP is fastest in some tests, reflecting growing trade links with China.
- Adjustment mechanisms to PPP
  - Cointegration and ECM evidence indicate reversion to PPP occurs primarily through adjustments in foreign prices measured in domestic currency rather than through domestic price adjustments.
  - This pattern may reflect adjustments in foreign prices or nominal exchange rates of benchmark currencies that are not pegged, or greater domestic price rigidity.
- Post-devaluation dynamics
  - Evidence for PPP across the studied regions is stronger after the CFA franc devaluation in 1994.
  - The response to shocks causing deviations from PPP averages between about 1.7 and 20 months, depending on currency and regional specifics (interpreted from estimated convergence coefficients).

### Empirical methods and main methodological results
- Univariate tests:
  - Augmented Dickey-Fuller (ADF) unit-root tests applied to GDP-weighted cross-country averages assuming deterministic trends.
  - Cointegration tests (Phillips-Ouliaris) and equilibrium-regression tests for Absolute PPP (β0 = 0 and β1 = 1) and Relative PPP (β1 = 1).
  - Error Correction Models (ECM) for ∆f_i and ∆p_i to identify whether adjustment comes via foreign prices or domestic prices.
- Panel methods:
  - Panel unit-root tests: Levin, Lin, and Chu (LLC); Im, Pesaran and Shin (IPS); Fisher-ADF; Fisher-PP.
  - Panel half-life and reversion-speed calculations via autoregressive specifications and h = ln(0.5)/ln(1 + β).
- Representative methodological conclusions:
  - ADF unit-root tests: limited support for absolute PPP for the full period; euro numeraire supports PPP for CFA and WAEMU; CFA numeraire supports PPP for the wider CFA zone.
  - Cointegration evidence: WAEMU under the euro; CFA zone under CFA franc and euro; CEMAC shows stronger cointegration when RMB numeraire used.
  - ECM results: stronger and faster reversion via foreign-price adjustment (∆f_i) than via domestic-price adjustment (∆p_i); euro displays the fastest reversion speed for most sub-groups except ECOWAS.
  - Panel unit-root tests: LLC tests support PPP for the full sample under any numeraire for all sub-groups except CEMAC when RMB is numeraire; evidence of stationarity supporting PPP is more common after the 1994 devaluation (except for RMB).

### Key numerical statistics (reported exactly)
- Sample and transformation:
  - Sample period: 1986M12-2020M02.
  - Observations reported for many series: 970, 985, 35 (as in source tables).
- Selected summary statistics:
  - Inflation Rate (in log difference): Mean 0.01; Median 0.00; Maximum 0.53; Minimum -0.22; Std. Dev. 0.025; Observations 970.
  - Price Level (in Log): Mean 0.18; Median -0.35; Maximum 4.65; Minimum -2.28; Std. Dev. 1.50; Observations 985.
  - Real Exchange Rate (CFA franc as Numeraire): Mean 1.32; Median -0.17; Maximum 10.27; Minimum -1.77; Std. Dev. 2.76; Observations 985.
  - Real Exchange Rate (euro as Numeraire): Mean -7.87; Median -9.37; Maximum 1.14; Minimum -11.00; Std. Dev. 2.76; Observations 985.
  - Real Exchange Rate (RMB as Numeraire): Mean -7.56; Median -8.99; Maximum 1.45; Minimum -10.48; Std. Dev. 2.76; Observations 985.
  - Real Exchange Rate (USD as Numeraire): Mean -9.53; Median -10.95; Maximum -0.49; Minimum -12.62; Std. Dev. 2.76; Observations 985.
  - Weighted Real GDP Growth Rate (CEMAC Group): Mean 0.03; Median 0.03; Maximum 0.10; Minimum -0.04; Std. Dev. 0.03; Observations 35.
  - Weighted Real GDP Growth Rate (CFA Group): Mean 0.03; Median 0.04; Maximum 0.06; Minimum -0.01; Std. Dev. 0.02; Observations 35.
  - Weighted Real GDP Growth Rate (ECOWAS Group): Mean 0.04; Median 0.04; Maximum 0.09; Minimum -0.01; Std. Dev. 0.02; Observations 35.
  - Weighted Real GDP Growth Rate (WAEMU Group): Mean 0.04; Median 0.04; Maximum 0.07; Minimum 0.00; Std. Dev. 0.02; Observations 35.
- Trade openness noted as "around 30 percent of GDP."
- Reported range of estimated speeds of convergence: from less than (minus) 0.05 to slightly below (minus) 0.6 (implying reversal to PPP may take between 1.7 and 20 months in response to a 1 percent foreign-price shock).

### Half-life and reversion-speed highlights (Annex VI exact entries, selected)
- CFA (first block)
  - US Dollar: beta -0.25; t-statistic -1.03; half_life, in months 2.40; reversion speed, in years 0.20
  - Euro: beta 0.10; t-statistic 0.95; half_life, in months -7.19; reversion speed, in years -0.60
  - CFA: beta 1.07; t-statistic 5.31; half_life, in months -0.95; reversion speed, in years -0.08
  - Chinese RMB: beta -0.05; t-statistic -0.57; half_life, in months 13.59; reversion speed, in years 1.13
- WAEMU (first block)
  - US Dollar: beta -0.77; t-statistic -2.09; half_life, in months 0.48; reversion speed, in years 0.04
  - Euro: beta 0.38; t-statistic 1.56; half_life, in months -2.16; reversion speed, in years -0.18
  - CFA: beta 0.98; t-statistic 9.47; half_life, in months -1.01; reversion speed, in years -0.08
  - Chinese RMB: beta -0.01; t-statistic -0.09; half_life, in months 88.59; reversion speed, in years 7.38
- ECOWAS (first block)
  - US Dollar: beta -0.04; t-statistic -0.32; half_life, in months 17.81; reversion speed, in years 1.48
  - Euro: beta -0.21; t-statistic -1.32; half_life, in months 3.00; reversion speed, in years 0.25
  - CFA: beta -0.57; t-statistic -3.14; half_life, in months 0.83; reversion speed, in years 0.07
  - Chinese RMB: beta -0.06; t-statistic -0.74; half_life, in months 11.77; reversion speed, in years 0.98
- Multiple additional blocks in Annex VI report many exact beta, t-statistic, half_life (in months) and reversion speed (in years) values for CFA, WAEMU, CEMAC, and ECOWAS under each numeraire (presented verbatim in the source).

### Country-level cointegration and assessments (Annex IV exact-format examples)
- Country-specific cointegration regressions present β0, Absolute PPP (p-value), β1, Relative PPP (p-value), Assessment (None/Relative/Absolute), and PO Test (p-value).
- Example entries (as presented in the source):
  - Burkina Faso (CFA numeraire) — Beta 0: 0.886426; Absolute PPP p-value: 0.0000; Beta 1: 1.394138; Relative PPP p-value: 0.0000; Assessment: None; PO Test: 0.3848
  - Central African Republic (CFA numeraire) — Beta 0: 0.145963; Absolute PPP p-value: 0.0000; Beta 1: 1.042603; Relative PPP p-value: 0.4267; Assessment: Relative; PO Test: 0.6122
  - Burkina Faso (Euro numeraire) — Beta 0: 10.02081; Absolute PPP p-value: 0.0000; Beta 1: 1.340309; Relative PPP p-value: 0.0000; Assessment: None; PO Test: 0.3958
  - Burkina Faso (RMB numeraire) — Beta 0: 9.966358; Absolute PPP p-value: 0.0000; Beta 1: 1.634396; Relative PPP p-value: 0.0000; Assessment: None; PO Test: 0.3056
  - Burkina Faso (USD numeraire) — Beta 0: 11.82155; Absolute PPP p-value: 0.0000; Beta 1: 1.503783; Relative PPP p-value: 0.0000; Assessment: None; PO Test: 0.4316
- Country-level assessments identify cases with "None", "Relative", or "Absolute" PPP and report Philips-Ouliaris (PO) test p-values for cointegration.

### Policy recommendations
- Regional monetary policies should critically consider the implications of currency choice for fostering economic stability and convergence.
- Reforms to enhance domestic goods market flexibility may be needed to support price alignment and adjustment within monetary unions.
- The burgeoning significance of the renminbi presents a viable candidate for future monetary frameworks in the region and should be evaluated alongside other potential numeraire currencies.
- Choice of numeraire should factor in:
  - degree of price rigidities relative to trading partners;
  - currency-specific frictions (price-setting, currency of invoicing);
  - overall trade openness and trade restrictions affecting partners asymmetrically;
  - proximity and strength of economic relations with the numeraire country;
  - exchange rate regimes’ degrees of flexibility and volatility.
- Policymakers should weigh the loss of national monetary and exchange rate policy space against gains from union membership, using evidence on PPP prevalence as one indicator of potential costs.
- Suggested monitoring and further study:
  - effects of geoeconomic fragmentation on PPP;
  - implications of digital currencies on PPP and monetary arrangements;
  - continued monitoring of numeraire-specific frictions and invoice-currency choice impacts on PPP and real exchange rate dynamics;
  - role of structural characteristics (oil dependence, distance from ports, factor mobility) on price dispersion and adjustment.

*IMF Working Paper — Testing the Purchasing Power Parity (PPP) in West and Central Africa (wpiea2025119-print-pdf)*

### Executive Summary ......................................................................................................

### Executive Summary

### Scope and approach
- Investigates the Purchasing Power Parity (PPP) hypothesis within the CFA currency zone and the Economic Community of West African States (ECOWAS).
- Uses four numeraire currencies: the euro, the US dollar, the Chinese yuan (the renminbi), and the CFA franc.
- Employs multiple methods to test absolute and relative PPP and considers the potential impact of the Balassa-Samuelson effect on long-term real exchange rate trends.

### Key findings
- Evidence of PPP variability
  - PPP is supported within the CFA and ECOWAS regions, but empirical support varies with the choice of numeraire and the country grouping analyzed.
  - Evidence for PPP is slightly stronger for the euro compared to other numeraires.
  - Possible explanation: adjustment of price structures of CFA countries (15 out of 21 countries) to the Euro area countries enforced by the longstanding peg.
- Regional heterogeneity
  - Within the CFA zone, stronger support for PPP is found in the WAEMU sub-region compared to the CEMAC sub-region.
  - The discrepancy may be attributed to economic structure differences, notably the significant share of oil-exporting nations in the CEMAC.
- Choice of numeraire matters
  - The euro produced more consistent evidence for PPP than the US dollar and the renminbi.
  - This suggests the stability and recognition of the euro may enhance the effectiveness of currency arrangements and price convergence.
- Adjustment mechanisms to PPP
  - Cointegration tests indicate reversion to PPP occurs primarily through adjustments in foreign prices measured in domestic currency rather than through domestic price adjustments.
  - This pattern may reflect adjustments in foreign prices or the nominal exchange rates of benchmark currencies that are not pegged (euro, dollar, renminbi), or a greater degree of price rigidity within the regions studied.
- Post-devaluation dynamics
  - Evidence for PPP across the studied regions is stronger after the CFA franc devaluation in 1994.
  - The response to shocks causing deviations from PPP averages between about 2 and 20 months, depending on currency and regional specifics.
  - After the 1994 devaluation, the renminbi is the numeraire currency toward which reversion to PPP is fastest, reflecting the growing importance of China in trade flows with the region.

*IMF Working Paper — Executive Summary (wpiea2025119-print-pdf)*

### 6. Policy Recommendations: The paper suggests that regional monetary policies should consider the

### 6. Policy Recommendations: The paper suggests that regional monetary policies should consider the implications of currency choice critically, especially as they pertain to fostering economic stability and convergence.

### Key policy recommendations
- Regional monetary policies should critically consider the implications of currency choice for fostering economic stability and convergence.
- Reforms to enhance domestic goods market flexibility may be needed to support price alignment and adjustment within monetary unions.
- The burgeoning significance of the renminbi presents a viable candidate for future monetary frameworks in the region and should be evaluated alongside other potential numeraire currencies.
- Choice of numeraire should factor in:
  - degree of price rigidities relative to trading partners;
  - currency-specific frictions (price-setting, currency of invoicing);
  - overall trade openness and trade restrictions affecting partners asymmetrically;
  - proximity and strength of economic relations with the numeraire country;
  - exchange rate regimes’ degrees of flexibility and volatility.
- Policymakers should weigh the loss of national monetary and exchange rate policy space against gains from union membership, using evidence on PPP prevalence as one indicator of potential costs.

### Findings relevant for policy design and monetary arrangements
- Tests indicate that PPP holds in the CFA and ECOWAS country-groups, but evidence varies considerably by:
  - methods used;
  - numeraire currency considered;
  - country-groups.
- Evidence for PPP is more common:
  - (i) (slightly) for the euro (15 cases);
  - (ii) for the CFA zone taken as a whole (16 cases);
  - (iii) within the CFA zone, for the WAEMU rather than the CEMAC;
  - (iv) when using panel data rather than averaged series of real exchange rates across country groupings.
- Convergence to PPP occurs mainly from adjustment in foreign prices expressed in domestic currency.
- Trade openness in the sample is noted as "around 30 percent of GDP."
- Price and inflation dispersion dynamics:
  - 휎휎-convergence in inflation improved (decline in 휎휎휋휋) around the late 1990s, supported by convergence pacts and policy reforms in unions;
  - Price dispersion (휎휎푝푝) rose from mid-1980s to mid-1990s, then flattened in the WAEMU from mid-1990s indicating preliminary relative PPP in WAEMU;
  - Price dispersion in CEMAC is at least half smaller than in WAEMU but is noisier, reflecting oil-dependence and passthrough of shocks;
  - In ECOWAS, prices continued diverging after the 1990s albeit at a slower pace; inflation has converged in all four zones.
- The Balassa-Samuelson effect can introduce trends in real exchange rates when productivity growth differs between tradable sectors, and tests explicitly account for this possibility.

### Implications for monetary unions and numeraire choice
- Prevalence of PPP within and across countries may indicate lower costs of forming or maintaining a currency union, all else equal.
- Loss of national policy space in a union is less costly if business cycles and price dynamics are symmetric and if PPP tendencies facilitate adjustment through domestic prices.
- The numeraire choice matters empirically and economically; evidence of stronger PPP with a particular numeraire (e.g., the euro in these tests) supports considering that currency as an anchor, conditional on political and other considerations.

### Areas for further study and monitoring
- Further studies should expand analysis of:
  - effects of geoeconomic fragmentation on PPP in these economies;
  - implications of digital currencies on PPP and monetary arrangements.
- Continued monitoring of:
  - how numeraire-specific frictions and invoice currency choice affect PPP tests and real exchange rate dynamics;
  - the role of structural characteristics (e.g., oil dependence, distance from ports, factor mobility) on price dispersion and adjustment.

*IMF Working Paper — Testing the Purchasing Power Parity (PPP) in West and Central Africa*

### conclusions in several ways. Some argue that this result is an artefact of the period; studies of longer periods

### Testing the Purchasing Power Parity (PPP) in West and Central Africa

### Literature and interpretation of PPP
- Studies give mixed conclusions on PPP:
  - Some argue observed rejections are an artefact of the period; studies of longer periods give stronger evidence for long-run PPP, especially during the floating rate period of the 1920s (Lothian and Taylor, 1996).
  - Others argue real exchange rates are determined by “fundamentals” such as productivity and demand factors; if these are non-stationary, PPP may appear violated over short-time periods.
  - An “efficient markets” or ex-ante variant posits the real exchange rate follows a random walk and criticizes traditional tests for simultaneity bias (Phylaktis and Kassimatis, 1994).
- Traditional tests:
  - Regression of the nominal exchange rate against the ratio of domestic to foreign prices (Nagayasu, 1998) typically rejects PPP (e.g., Roll, 1979; Frenkel, 1980; Cumby and Obstfeld, 1984) but may be biased if series are integrated.
  - Cointegration tests (Engle and Granger (1987) spirit) test whether nominal exchange rate and domestic/foreign price ratio are I(1) and cointegrated; if so PPP is claimed to hold.
- Frequency and sample length effects:
  - Short- or medium-length time-series often find PPP does not hold (e.g., Roll, 1979; Mishkin, 1984; Piggot and Sweeney, 1985).
  - Longer samples typically support PPP (Abuaf and Jorion, 1990; Froot and Rogoff, 1994; Lothian and Taylor, 1996).
  - High frequency data (e.g., monthly) typically do not support long-term PPP (McNown and Wallace, 1989; Taylor, 1988; Corbae and Ouliaris, 1998).
  - Low-frequency data and cointegration techniques usually support long-term convergence to PPP (Edison, 1987; Kim, 1990).

### Unit-root, cointegration, and panel evidence
- Univariate testing developments:
  - Early ADF tests often fail to find long-run PPP due mainly to low power in small samples.
  - Long-horizon data (up to 200 years) show stronger rejections of the unit root hypothesis but mix fixed and floating regimes.
  - More powerful tests (DF-GLS of Elliott, Rothenberg and Stock (1996)) yield more rejections, often at weak (10 percent) significance (Cheung and Lai, 2000).
  - Covariate-augmented tests find stronger rejections (Elliott and Pesavento, 2004; Amara and Papell, 2005).
- Speed of convergence estimates:
  - Rabe and Waddle (2020) find half-life deviations from PPP have fallen by approximately 2 years between 1960 and 2015 and estimate an average rate of PPP convergence of 3 years.
- Intranational and currency-union evidence:
  - PPP is more likely to hold within currency unions or single countries due to absence of nominal exchange rate fluctuations, higher market integration, lower trade barriers and transport costs, and smaller measurement errors.
  - Sustained deviations explained by structural/technological differences (Kravis and Lipsey, 1983) or resource discoveries (Buiter and Miller, 1981).
- Panel studies:
  - Mixed results: early support (Frankel and Rose, 1996; Jorion and Sweeney, 1996) but weaker evidence when accounting for serial or contemporaneous correlation (Papell, 1997; O’Connell, 1998).
  - Post-1973 quarterly data panels with USD numeraire through 1998 tend to support PPP for developed countries (Higgins and Zakrajšek, 2000; Wu and Wu, 2001; Papell, 2005).
  - For less developed countries, panel unit root tests provide limited support; results vary (Phylaktis and Kassimatis, 1994; Oh, 1996; Holmes, 2001).

### Data sources, transformations, and sample
- Data and sample:
  - Monthly data on consumer price indices (CPI), CPI inflation, and nominal exchange rates from IMF’s International Financial Statistics (IFS).
  - Sample period: 1986M12-2020M02.
  - Monthly interpolated estimates of annual price level of household consumption from Penn World Table 10.01 (PWT) derived using monthly CPI index and sequential inflation rate.
  - Economic size proxies for weighted averages: GDP in US dollars and in PPP international dollars from IMF’s World Economic Outlook (WEO).
- Converting price indices into price levels:
  - Monthly price level 푃푃푃푃푃푃푖,푚 obtained from an interpolated annual PWT price level at last month of prior year (푃푃푃푃푃푃푖−1,12) and monthly price index푃푃푖,푚 via multiplicative chaining, ensuring:
    - (i) monthly changes of estimated 푃푃푃푃푃푃푖,푚 equal monthly changes in price index 푃푃푖,푚;
    - (ii) annual average of 푃푃푃푃푃푃푖,푚 equals observed annual price level from PWT.
  - For the rest of the paper, “price level” refers to the result from this procedure.
- Weights for cross-country averages:
  - Cross-country averages use weights reflecting each country’s contribution to zone real GDP in 2017 PPP international dollars:
    - 푤푤푖 = 푅푅푅푅푅푅푃푃2017 푝푝푝푝푝푝,푖 / ∑_{m=1}^{14} 푅푅푅푅푅푅푃푃2017 푝푝푝푝푝푝,푚 (example for 14 CFA countries).
- Seasonal adjustment:
  - All monthly data on price levels, inflation, and exchange rates seasonally adjusted using U.S. Census Bureau’s X-13 seasonal adjustment tools (X-13 ARIMA-SEATS).
- Prices and real exchange rates used in PPP tests:
  - Transformations and definitions (domestic price 푝푝푖푖푖푖, foreign price 푝푝푖푖푖푖∗, domestic value of foreign price 푓푓푖, real exchange rate 푞푞푖푖푖푖, nominal exchange rate 푠푠푖푖푖푖) are:
    - 푓푓푖 = 푝푝푖푖푖푖∗ + 푠푠푖푖푖푖
    - 푞푞푖푖푖푖 = 푝푝푖푖푖푖 − 푝푝푖푖푖푖∗ − 푠푠푖푖푖푖
    - 푠푠푖푖푖푖 = 푝푝푖푖푖푖 − 푝푝푖푖푖푖∗ − 푞푞푖푖푖푖
    - 푓푓푖 = 푝푝푖푖푖푖 − 푞푞푖푖푖푖
- Numeraires considered:
  - Four numeraire currencies: the US dollar (USD), the euro (or French franc until 1998), the Chinese Yuan (RMB), and the CFA franc.
  - PPP implication: deviations of domestic prices from foreign prices measured in a common currency should be transitory; real exchange rate 푞푞 should be stationary.

### PPP test cases and sample split
- Three cases for PPP test-equation:
  - Case 1 (CFA zone, CFA numeraire): 푞푞푖푖푖푖 = 푝푝푖푖푖푖 − 푝푝푖푖푖푖푗푗 (nominal exchange rate effect disappears).
  - Case 2 (CFA zone, non-CFA numeraire): 푞푞푖푖푖푖 = 푝푝푖푖푖푖 − 푝푝푖푖푖푖푗푗 − 푠푠𝑖 ∈ 𝐶𝐹𝐴,𝑖 (only relevant nominal exchange rate is between CFA franc and numeraire).
  - Case 3 (non-CFA countries): equation (5) applies as is.
- Sample split to analyze 1994 CFA Franc devaluation:
  - Full sample: 1986M01 to 2020M02.
  - Pre-devaluation: 1986M01 to 1994M05.
  - Post-devaluation: 1994M06 to 2020M02.

### Summary statistics (selected exact values)
- Inflation Rate (in log difference): Mean 0.01; Median 0.00; Maximum 0.53; Minimum -0.22; Std. Dev. 0.025; Observations 970.
- Price Level (in Log): Mean 0.18; Median -0.35; Maximum 4.65; Minimum -2.28; Std. Dev. 1.50; Observations 985.
- Real Exchange Rate (CFA franc as Numeraire): Mean 1.32; Median -0.17; Maximum 10.27; Minimum -1.77; Std. Dev. 2.76; Observations 985.
- Real Exchange Rate (euro as Numeraire): Mean -7.87; Median -9.37; Maximum 1.14; Minimum -11.00; Std. Dev. 2.76; Observations 985.
- Real Exchange Rate (RMB as Numeraire): Mean -7.56; Median -8.99; Maximum 1.45; Minimum -10.48; Std. Dev. 2.76; Observations 985.
- Real Exchange Rate (USD as Numeraire): Mean -9.53; Median -10.95; Maximum -0.49; Minimum -12.62; Std. Dev. 2.76; Observations 985.
- Weighted Real GDP Growth Rate (CEMAC Group): Mean 0.03; Median 0.03; Maximum 0.10; Minimum -0.04; Std. Dev. 0.03; Observations 35.
- Weighted Real GDP Growth Rate (CFA Group): Mean 0.03; Median 0.04; Maximum 0.06; Minimum -0.01; Std. Dev. 0.02; Observations 35.
- Weighted Real GDP Growth Rate (ECOWAS Group): Mean 0.04; Median 0.04; Maximum 0.09; Minimum -0.01; Std. Dev. 0.02; Observations 35.
- Weighted Real GDP Growth Rate (WAEMU Group): Mean 0.04; Median 0.04; Maximum 0.07; Minimum 0.00; Std. Dev. 0.02; Observations 35.

*IMF Working Papers — Testing the Purchasing Power Parity (PPP) in West and Central Africa*

### 1. Univariate Analysis

### 1. Univariate Analysis

### Overview
- If absolute PPP holds, deviations of price levels from the numeraire price level should be temporary when prices are measured in the same currency. For relative PPP, that condition only needs to be valid for inflation rates, not necessarily prices.
- In both cases, the real exchange rate should converge to a constant (or be trend-stationary if fundamentals are driving real appreciation à la BalassaMB-Samuelson).
- Univariate PPP tests are performed for currency unions: ECOWAS, WAEMU, CEMAC, and the CFA zones, using successively the US dollar (USD), the euro, the renminbi (RMB), and the CFA franc as numeraire currencies.

### GDP-Weighted Average Real Exchange Rate (Figure 4) — descriptive findings
- WAEMU and CFA zones show some tendency toward convergence of the real exchange rate after the 1994 devaluation of the CFA franc when using the euro and the CFA franc as numeraire.
- Using the renminbi as anchor makes WAEMU and CEMAC real exchange rate dynamics noisier locally, with a long-term tendency toward depreciation.
- ECOWAS shows a clear upward trend of the average real exchange rate toward its level before the 1998 depreciation of the Naira; ECOWAS dynamics largely follow Nigeria, which accounts for more than 60% of ECOWAS GDP.
- The ECOWAS real exchange rate appreciates when the CFA is used as numeraire.

### Method 1 — Augmented Dickey-Fuller (ADF) Unit Root Tests (Table 1) — summary findings
- Tests applied to GDP-weighted cross-country averages of the real exchange rate within each currency zone, assuming a deterministic trend.
- Results provide limited support for absolute PPP for the full period; evidence found only:
  - Using the euro as numeraire for CFA and WAEMU.
  - Using the CFA as numeraire for the wider CFA zone.
- For the full period, under any numeraire currency, the real exchange rate for ECOWAS and CEMAC display a unit root (price differentials do not tend to die out).
- PPP for CEMAC and ECOWAS is rejected under any anchor currency; PPP for WAEMU and CFA zones cannot be rejected when the euro is used.
- Before the 1994 devaluation, less evidence for PPP; PPP could be found only for WAEMU using the US dollar as numeraire.

### Unit root p-value patterns and preferred numeraire
- Using p-values as a proxy for distance to PPP, the ranking of numeraire currencies for the CFA zone (best to worst): CFA, euro, US dollar, RMB.
- Diverging results across sub-blocks:
  - CEMAC: stronger rejection of absolute PPP using the euro as numeraire.
  - WAEMU: lowest p-values (stronger support) under the euro.
- Only the US dollar supported PPP for WAEMU during instability preceding the 1994 devaluation.
- Overall: shallow evidence of PPP; choice of numeraire matters.

### Country-specific ADF results (Table 2) — highlights
- Using the US dollar as numeraire, PPP could not be rejected for only two ECOWAS countries: Sierra Leone and Guinea-Bissau (insufficient for ECOWAS as a whole).
- Cote d’Ivoire and Benin drive acceptance of PPP for WAEMU before the devaluation using the US dollar as numeraire.
- The CFA and the euro are the numeraire currencies under which PPP cannot be rejected for the largest number of countries.
- When the euro is the numeraire, PPP holds for:
  - Four WAEMU countries: Burkina Faso, Mali, Senegal, and Guinea Bissau.
  - Three non-WAEMU ECOWAS countries: The Gambia, Ghana, and Guinea.
  - One CEMAC country: Gabon.
- The number of CEMAC countries for which PPP cannot be rejected is largest when the CFA franc is the numeraire.
- Guinea-Bissau is the only country with evidence of PPP under all four numeraire currencies.
- Ten countries show no evidence of PPP for the full period: CAR, Congo, Equatorial Guinea, Cameroon (CEMAC); Cote d’Ivoire, Togo, Benin (WAEMU); Liberia, Nigeria, Guinea (ECOWAS).
- Oil-exporting countries more often display a unit root:
  - Gabon and Chad (CEMAC) are the only oil exporters with deviations of inflation tending to revert to a constant, but only in specific numeraire cases (Gabon: euro; Chad: CFA).
  - Outside the CFA zone, Nigeria displays stationarity of its inflation series when the CFA and the renminbi are anchors (but only after 1994).

### Method 2 — Cointegration Tests (Philips-Ouliaris) (Table 3) — summary
- Cointegration tests check whether sequences of foreign- and domestic-price series are cointegrated for each numeraire.
- p-value < 0.05 rejects the null of no cointegration.
- Full period cointegration evidence:
  - WAEMU under the euro.
  - CFA zone under both the CFA franc and the euro.
- CEMAC shows stronger evidence of cointegration when RMB is the numeraire.
- ECOWAS shows stronger evidence of cointegration when the euro is the numeraire.
- Diverging dynamics between WAEMU and CEMAC confirmed.
- Country-specific note: Ghana presents strong evidence of cointegration when CFA franc or euro are numeraires.

### Cointegration regression tests for Absolute and Relative PPP (Tables 4 and 5) — summary
- Absolute PPP test (Null: β0 = 0 and β1 = 1). Full sample:
  - Absolute PPP rejected for all groupings and numeraire currencies (rejection is very strong).
  - Pre-1994: weak evidence (10% level) of absolute PPP for CEMAC under CFA franc.
- Relative PPP test (Null: β1 = 1). Full sample:
  - Evidence for relative PPP strongest for:
    - WAEMU with RMB as numeraire.
    - CFA with euro as numeraire.
    - WAEMU with US dollar as numeraire.
  - Contrary to unit root and Philips-Ouliaris tests, the weak form of PPP cannot be rejected when USD and RMB are numeraires.
  - After the 1994 devaluation, relative PPP holds only for WAEMU using RMB as numeraire.

### Method 3 — Error Correction Models (ECM) (equations (7) and (8)) — purpose and setup
- ECMs estimated for each currency area under each numeraire to assess how deviations from long-run PPP are reversed:
  - ∆f_i = a0 + a1 * μ̂_{i,t−1}
  - ∆p_i = b0 + b1 * μ̂_{i,t−1}
  - μ̂_{i,t−1} = y_{i,t−1} − β0 − β1 x_{i,t−1} (x = p or f; y = f or p)
- Negative and statistically significant a1 and b1 indicate reversal to long-term equilibrium.

### ECM results — adjustment dynamics (Figures 5–6) — key findings
- ECM for ∆f_i (Figure 5):
  - Evidence of cointegration suggests reversion to PPP mainly through adjustment in foreign prices measured in domestic currency.
  - Large negative estimated values for a1 indicate changes in f_i undo deviations from long-run PPP.
  - For pegged currency areas (CEMAC, WAEMU), adjustment excludes changes in CFA nominal exchange rate against the euro, implying foreign prices drive adjustment (including via euro nominal exchange rate movements).
  - Full sample: reversion to PPP via foreign prices is faster for the CFA zone with the euro as numeraire (facilitated by CFA peg to euro).
  - Within CFA zone, WAEMU shows stronger signs of cointegration and faster reversion than CEMAC.
  - Reversion to equilibrium becomes less significant for both sub-regions when CFA is used as numeraire.
  - Among numeraires, the euro displays the fastest reversion speed for all sub-groups except ECOWAS (which prefers CFA franc by speed of adjustment).
  - Pre- vs. post-1994: before 1994, USD and euro supported faster convergence; after 1994, CFA franc becomes best option for CEMAC and ECOWAS, and RMB supports faster return to PPP for ECOWAS than USD and euro.
- ECM for ∆p_i (Figure 6):
  - Evidence for cointegration is weaker, indicating adjustment toward long-run equilibrium does not occur mainly through domestic prices but rather through foreign prices measured in domestic currency.
  - Negative estimated convergence-speed parameters for the full sample are typically lower in absolute value than those for ∆f_i.
  - Pre-1994, CFA zone and WAEMU show convergence-speed coefficients with the wrong sign (positive) for domestic-price adjustment.
  - Pre-1994, strongest evidence for convergence via domestic prices is in CEMAC (about the same speed as foreign-price adjustment).
  - Post-1994, evidence of convergence via domestic prices exists for all groupings except CEMAC.
- Overall implication: prices in the CFA zone (both WAEMU and CEMAC) and in ECOWAS do not primarily change to eliminate long-run PPP discrepancies; most adjustment comes from foreign prices measured in domestic currency.

### Estimated speeds of convergence — numerical range and interpretation
- Estimated coefficients for the speed of convergence range from less than (minus) 0.05 to slightly below (minus) 0.6.
- Interpretation: in response to a shock of 1 percent in foreign prices, the reversal to PPP may take between 1.7 and 20 months.

*Source: IMF Working Paper — "Testing the Purchasing Power Parity (PPP) in West and Central Africa", 1. Univariate Analysis, figures and tables as presented in the chapter.*

### 2. Panel Data Analysis

### 2. Panel Data Analysis

### Method 1: Panel Unit Root Tests
- Panel structure: unbalanced panel covering December 1986 to February 2020; tests used:
  - Common unit root process: Levin, Lin, and Chu (LLC).
  - Country-specific unit root processes: Im, Pesaran and Shin (IPS); Fisher-ADF; Fisher-PP.
- Panel AR(1) specification:
  - q_i,t = ρ_i * q_i,t−1 + ε_i,t where i = 1, 2, …, 21 series of real exchange rates.
  - If |ρ_i| < 1 then y_i,t is weakly (trend-) stationary; if |ρ_i| = 1 then y_i,t contains a unit root.
  - Two testing assumptions:
    - ρ_i = ρ for all i (common unit root process — LLC).
    - ρ_i allowed to vary across i (individual unit root processes — IPS, Fisher-ADF, Fisher-PP).
- Key findings from panel unit root analysis (text summary of Table 6):
  - The assumptions on the relationship between unit root processes in the panel matter for conclusions on unit root.
  - Under the CFA franc as anchor, results suggest strong evidence for PPP, and stronger evidence when assuming common unit root processes, except for the CEMAC zone.
  - PPP evidence is mixed for the CEMAC, with weak support for PPP when assuming individual unit root processes.
  - LLC tests support the strong version of PPP for the full sample under any numeraire currency for all sub-groups except for the CEMAC when the renminbi is used as numeraire.
  - For the full period, the relevance of the numeraire currency is diluted: evidence of PPP is found regardless of the numeraire currency (at the 10 percent significance level, this includes the CEMAC).
  - Evidence of stationarity supporting PPP is more common after the CFA franc devaluation, except for the renminbi as numeraire.

### Method 2: Deviations from PPP in a Panel Data Framework (Half-life and Reversion Speed)
- Objective: complement univariate convergence-speed analysis using panel framework; measure persistence of shocks via half-life.
- Half-life definition used:
  - For estimated autoregressive coefficient β from ∆q_i,t = c + β * y_i,t−1 + Σ_{p=1}^k γ_p * ∆q_i,t−p + υ_i,t,
  - Half-life h = ln(0.5) / ln(1 + β).
- Literature context: consensus on half-lives of deviations from PPP around 3-5 years using long-horizon data and univariate methods.
- Key empirical results (text summary of Table 7 and discussion):
  - For the full period:
    - Results reinforce panel unit root conclusion of convergence to PPP for WAEMU under the US dollar and for ECOWAS with the CFA franc as numeraire.
    - Estimated reversion speeds: 0.5 month (WAEMU under US dollar) and 0.8 month (ECOWAS with CFA franc).
  - Sub-periods (before vs after CFA devaluation):
    - Currency zones did not revert to PPP under any numeraire before the CFA devaluation.
    - After the 1994 devaluation, significant convergence to PPP for all zones, except the CEMAC, under the renminbi.
    - Reversion speeds implied by estimated coefficients translate to half-lives ranging from 1.8 months to 2.0 months (post-devaluation, renminbi as numeraire).
    - Additional post-devaluation results: reversal to PPP for ECOWAS using the Euro and CFA franc as anchor, with half-lives of 1.7 month and 0.9 month respectively.
  - Interpretation:
    - Reversion speeds reported are faster than usually reported for advanced economies.
    - Faster reversion may be related to adjustments in foreign prices measured in domestic currency (which may come from changes in nominal exchange rates of numeraire currencies) rather than adjustments in (possibly sticky) domestic prices.

### Cross-zone comparison and drivers
- Within the broad CFA zone:
  - Evidence of PPP is more consistently present for WAEMU than for CEMAC.
  - CEMAC shows little evidence of PPP; possible cause: less diversified economies and large exposure to international oil markets (four out of six CEMAC countries—Chad, the Republic of Congo, Equatorial Guinea, and Gabon—are big oil-producers).
  - While PPP (at least weak form) generally holds at regional level, regional aggregates may mask opposing trends in individual countries.

### Summary of empirical patterns (from section 3 and tables)
- Evidence supporting absolute or relative PPP (summary bullets reproduced from study conclusions):
  - Slightly more common for the euro (15 cases) relative to other numeraire currencies.
  - More common for the CFA zone taken as a whole (16 cases) relative to other country groupings.
  - More easily found for WAEMU than for CEMAC, possibly reflecting CEMAC’s oil-export dependence.
  - Stronger when using panel data relative to time-series methods with averaged series of real exchange rates across country groupings.
  - Convergence to PPP occurs mainly from adjustment in foreign prices, suggesting “imported” monetary conditions for CFA and some ECOWAS countries that peg their currencies.

### Policy implications and recommendations (from Conclusions and Policy Implications)
- Main policy-relevant findings:
  - PPP cannot be rejected (at least in its weak form) regardless of the numeraire currency.
  - Numeraire currency matters: more evidence of PPP for the full period when the Euro is the numeraire.
  - WAEMU shows more consistent PPP evidence than other groupings; CEMAC shows little evidence of PPP, raising questions about homogeneity within the CFA zone for PPP-related purposes.
  - Evidence indicates price dynamics changed after the CFA devaluation, increasing PPP evidence.
  - The renminbi appears as the numeraire toward which reversion to PPP is fastest after the CFA devaluation—reflecting growing importance of China for the region’s economies and suggesting the renminbi as a potential candidate for pegging regional currencies.
- Policy recommendations articulated in the study:
  - Countries should regularly assess price dynamics to update the design of their currency arrangements because the relevance of numeraire is time-sensitive.
  - Reforms to liberalize domestic markets are called for to allow domestic prices to adjust faster to shocks, given evidence that reversion to PPP operates mainly through foreign-price adjustments.

*Source: 2. Panel Data Analysis — IMF Working Paper content unit (wpiea2025119-print-pdf).*

### Annex IV.   Country-Specific Cointegration

### Annex IV.   Country-Specific Cointegration Regressions

### Overview of estimations and presentation
- The Annex provides country-specific cointegration relationships based on equation (6), considering four possible numeraire currencies.
- The cointegration vector includes the (logs of) foreign prices expressed in domestic currency and domestic prices.
- In all tables, the second and fourth columns present the estimated values of β0 and β1, respectively.
- The third and fifth columns (“Absolute PPP” and “Relative PPP”) present p-values associated with Wald tests for the PPP restrictions imposed on the coefficients of the equilibrium regression version of equation (6).
- The country-specific assessment (second to last column) uses:
  - “None” = no version of PPP detected;
  - “Relative” = existence of relative PPP but not absolute;
  - “Absolute” = existence of absolute PPP.
- The last column (“PO test”) presents the p-value of the Philips-Ouliaris cointegration test (H0 is no cointegration).

### CFA franc as numeraire — sample entries and structure
- The tables present for each country: Beta 0, Absolute PPP (p-value), Beta 1, Relative PPP (p-value), Assessment, PO Test (p-value).
- Example entries (as presented in the source):
  - Burkina Faso — Beta 0: 0.886426; Absolute PPP p-value: 0.0000; Beta 1: 1.394138; Relative PPP p-value: 0.0000; Assessment: None; PO Test: 0.3848
  - Cameroon — Beta 0: 0.365037; Absolute PPP p-value: 0.0000; Beta 1: 1.268812; Relative PPP p-value: 0.0000; Assessment: None; PO Test: 0.6571
  - Central African Republic — Beta 0: 0.145963; Absolute PPP p-value: 0.0000; Beta 1: 1.042603; Relative PPP p-value: 0.4267; Assessment: Relative; PO Test: 0.6122
- The CFA franc tables include full-sample estimations and subsample estimations (Before CFA Depreciation (1994) and After Depreciation).

### Euro as numeraire — sample entries and structure
- Structure mirrors the CFA franc tables: Beta 0, Absolute PPP p-value, Beta 1, Relative PPP p-value, Assessment, PO Test.
- Example entries (as presented in the source):
  - Burkina Faso — Beta 0: 10.02081; Absolute PPP p-value: 0.0000; Beta 1: 1.340309; Relative PPP p-value: 0.0000; Assessment: None; PO Test: 0.3958
  - Cameroon — Beta 0: 9.485549; Absolute PPP p-value: 0.0000; Beta 1: 1.186707; Relative PPP p-value: 0.0000; Assessment: None; PO Test: 0.4944
  - Central African Republic — Beta 0: 9.300636; Absolute PPP p-value: 0.0000; Beta 1: 0.970540; Relative PPP p-value: 0.6657; Assessment: Relative; PO Test: 0.4412
- Contains full-sample and subsample (Before CFA Depreciation (1994) and After CFA Depreciation) estimations.

### RMB as numeraire — sample entries and structure
- Same presentation format (β0, Absolute PPP p-value, β1, Relative PPP p-value, Assessment, PO Test).
- Example entries (as presented in the source):
  - Burkina Faso — Beta 0: 9.966358; Absolute PPP p-value: 0.0000; Beta 1: 1.634396; Relative PPP p-value: 0.0000; Assessment: None; PO Test: 0.3056
  - Cameroon — Beta 0: 9.359375; Absolute PPP p-value: 0.0000; Beta 1: 1.501963; Relative PPP p-value: 0.0000; Assessment: None; PO Test: 0.1595
  - Central African Republic — Beta 0: 9.113708; Absolute PPP p-value: 0.0000; Beta 1: 1.281647; Relative PPP p-value: 0.0000; Assessment: None; PO Test: 0.1852
- Includes full-sample and subsample estimations.

### US dollar as numeraire — sample entries and structure
- Same presentation format as other numeraires.
- Example entries (as presented in the source):
  - Burkina Faso — Beta 0: 11.82155; Absolute PPP p-value: 0.0000; Beta 1: 1.503783; Relative PPP p-value: 0.0000; Assessment: None; PO Test: 0.4316
  - Cameroon — Beta 0: 11.25865; Absolute PPP p-value: 0.0000; Beta 1: 1.390401; Relative PPP p-value: 0.0000; Assessment: None; PO Test: 0.3456
  - Central African Republic — Beta 0: 11.01629; Absolute PPP p-value: 0.0000; Beta 1: 1.115549; Relative PPP p-value: 0.1493; Assessment: Relative; PO Test: 0.4058
- Contains full-sample and subsample estimations.

### Annex V. Error Correction Model (ECM)

### Overview
- This Annex provides additional results of the estimated ECMs based on equations (7)-(8), considering the four possible numeraire currencies.
- ECM estimations are reported for the full sample and for subsamples (Before CFA Depreciation (1994M06) and After CFA Depreciation).

### Annex VI. Reversion to PPP: Half-Life Estimations

- The Annex reports half-life estimations (and related statistics) for regional aggregates (CFA, WAEMU, CEMAC, ECOWAS) under four numeraires: US Dollar, Euro, CFA, Chinese RMB.
- For each region and numeraire the reported statistics are: beta, t-statistic, half_life, in months, reversion speed, in years.
- Values are reported exactly as in the source. Selected entries (presented in the same sequence as in the source):

CFA (first block)
- US Dollar: beta -0.25; t-statistic -1.03; half_life, in months 2.40; reversion speed, in years 0.20
- Euro: beta 0.10; t-statistic 0.95; half_life, in months -7.19; reversion speed, in years -0.60
- CFA: beta 1.07; t-statistic 5.31; half_life, in months -0.95; reversion speed, in years -0.08
- Chinese RMB: beta -0.05; t-statistic -0.57; half_life, in months 13.59; reversion speed, in years 1.13

WAEMU (first block)
- US Dollar: beta -0.77; t-statistic -2.09; half_life, in months 0.48; reversion speed, in years 0.04
- Euro: beta 0.38; t-statistic 1.56; half_life, in months -2.16; reversion speed, in years -0.18
- CFA: beta 0.98; t-statistic 9.47; half_life, in months -1.01; reversion speed, in years -0.08
- Chinese RMB: beta -0.01; t-statistic -0.09; half_life, in months 88.59; reversion speed, in years 7.38

CEMAC (first block)
- US Dollar: beta 0.28; t-statistic 1.22; half_life, in months -2.79; reversion speed, in years -0.23
- Euro: beta 0.11; t-statistic 1.05; half_life, in months -6.64; reversion speed, in years -0.55
- CFA: beta 0.49; t-statistic 2.60; half_life, in months -1.73; reversion speed, in years -0.14
- Chinese RMB: beta -0.16; t-statistic -0.66; half_life, in months 4.08; reversion speed, in years 0.34

ECOWAS (first block)
- US Dollar: beta -0.04; t-statistic -0.32; half_life, in months 17.81; reversion speed, in years 1.48
- Euro: beta -0.21; t-statistic -1.32; half_life, in months 3.00; reversion speed, in years 0.25
- CFA: beta -0.57; t-statistic -3.14; half_life, in months 0.83; reversion speed, in years 0.07
- Chinese RMB: beta -0.06; t-statistic -0.74; half_life, in months 11.77; reversion speed, in years 0.98

CFA (second block)
- US Dollar: beta -0.09; t-statistic -0.42; half_life, in months 7.76; reversion speed, in years 0.65
- Euro: beta 0.27; t-statistic 1.40; half_life, in months -2.89; reversion speed, in years -0.24
- CFA: beta 1.13; t-statistic 3.63; half_life, in months -0.91; reversion speed, in years -0.08
- Chinese RMB: beta 0.26; t-statistic 1.40; half_life, in months -3.02; reversion speed, in years -0.25

WAEMU (second block)
- US Dollar: beta -0.89; t-statistic -1.52; half_life, in months 0.31; reversion speed, in years 0.03
- Euro: beta 0.59; t-statistic 1.75; half_life, in months -1.50; reversion speed, in years -0.13
- CFA: beta 0.96; t-statistic 2.21; half_life, in months -1.03; reversion speed, in years -0.09
- Chinese RMB: beta 0.29; t-statistic 1.86; half_life, in months -2.71; reversion speed, in years -0.23

CEMAC (second block)
- US Dollar: beta 0.21; t-statistic 1.02; half_life, in months -3.59; reversion speed, in years -0.30
- Euro: beta 0.22; t-statistic 1.19; half_life, in months -3.42; reversion speed, in years -0.29
- CFA: beta 0.59; t-statistic 2.61; half_life, in months -1.49; reversion speed, in years -0.12
- Chinese RMB: beta -0.32; t-statistic -0.55; half_life, in months 1.81; reversion speed, in years 0.15

ECOWAS (second block)
- US Dollar: beta -0.26; t-statistic -0.70; half_life, in months 2.26; reversion speed, in years 0.19
- Euro: beta 0.00; t-statistic 0.00; half_life, in months 3569.68; reversion speed, in years 47.47
- CFA: beta -0.53; t-statistic -1.04; half_life, in months 0.83; reversion speed, in years 0.08
- Chinese RMB: beta 0.15; t-statistic 1.20; half_life, in months -4.88; reversion speed, in years -0.41

CFA (third block)
- US Dollar: beta 0.05; t-statistic 0.75; half_life, in months -13.69; reversion speed, in years -1.14
- Euro: beta 0.66; t-statistic 4.13; half_life, in months -1.37; reversion speed, in years -0.11
- CFA: beta 0.30; t-statistic 1.33; half_life, in months -2.67; reversion speed, in years -0.22
- Chinese RMB: beta -0.30; t-statistic -3.54; half_life, in months 1.95; reversion speed, in years 0.16

WAEMU (third block)
- US Dollar: beta 0.03; t-statistic 0.36; half_life, in months -23.14; reversion speed, in years -1.93
- Euro: beta 0.62; t-statistic 5.05; half_life, in months -1.43; reversion speed, in years -0.12
- CFA: beta 0.49; t-statistic 2.86; half_life, in months -1.75; reversion speed, in years -0.15
- Chinese RMB: beta -0.32; t-statistic -3.16; half_life, in months 1.80; reversion speed, in years 0.15

CEMAC (third block)
- US Dollar: beta 0.09; t-statistic 0.72; half_life, in months -8.49; reversion speed, in years -0.71
- Euro: beta 0.68; t-statistic 1.86; half_life, in months -1.33; reversion speed, in years -0.11
- CFA: beta 0.21; t-statistic 0.46; half_life, in months -3.70; reversion speed, in years -0.31
- Chinese RMB: beta -0.26; t-statistic -1.77; half_life, in months 2.32; reversion speed, in years 0.19

ECOWAS (third block)
- US Dollar: beta 0.00; t-statistic -0.02; half_life, in months 313.86; reversion speed, in years 26.16
- Euro: beta -0.33; t-statistic -2.39; half_life, in months 1.71; reversion speed, in years 0.14
- CFA: beta -0.54; t-statistic -3.47; half_life, in months 0.88; reversion speed, in years 0.07
- Chinese RMB: beta -0.29; t-statistic -2.54; half_life, in months 2.02; reversion speed, in years 0.17

*Annex IV–VI tables, estimations, and exact numerical values are presented as in the source document.*

### Annex VII. ECOWAS excluding Nigeria

### Annex VII. ECOWAS excluding Nigeria

### Real Exchange Rate for the ECOWAS zones (excluding Nigeria) using different numeraire currencies
- Summary table of test outcomes for Purchasing Power Parity (Full Sample for ECOWAS, excluding Nigeria).
- Note: Green highlights indicate evidence of PPP.

### Summary of Results of Purchasing Power Parity (Full Sample for ECOWAS, excluding Nigeria)
- Value Attribute columns: US Dollar (USD), Euro (Euro), CFA Franc (CFA Franc), Chinese RMB (Chinese RMB).
- Method 1: ADF test — P-value
  - USDEuroCFA FrancChinese RMB: 0.81 0.67 0.71 0.81
- Method 2a: Co-Integration test — P-value
  - USDEuroCFA FrancChinese RMB: 0.65 0.43 0.41 0.53
- Method 2b: Absolute PPP test using equilibrium regression — F-stat
  - USDEuroCFA FrancChinese RMB: 0.00 0.00 0.00 0.00
- Method 2c: Relative PPP test using equilibrium regression — F-stat
  - USDEuroCFA FrancChinese RMB: 0.00 0.00 0.00 0.00
- Method 3a: Error Correction Model foreign prices adjustment — Coefficient (P-Value)
  - USDEuroCFA FrancChinese RMB: -0.01 -0.017* -0.017* -0.01
- Method 3b: Error Correction Model domestic prices adjustment — Coefficient (P-Value)
  - USDEuroCFA FrancChinese RMB: 0.00 0.00 0.00 0.01***
- Method 4a: Unit Root test with panel data (common unit root) — P-value
  - USDEuroCFA FrancChinese RMB: 0.00 0.00 0.00 0.00
- Method 4b: Unit Root test with panel data (individual unit root) — Summary Results
  - USDEuroCFA FrancChinese RMB: Mixed Yes Yes Yes
- Method 5: Half-life — Coefficient (P-Value)
  - USDEuroCFA FrancChinese RMB: -0.03 -0.10 -0.44*** -0.02

### Key takeaways
- Across the listed methods, many test statistics report exact values of 0.00 or small P-values for unit-root style and equilibrium regression tests.
- Method 3a shows negative adjustment coefficients for foreign prices with significance indicators for Euro and CFA Franc (-0.017* each).
- Method 3b shows a domestic price adjustment coefficient of 0.01*** for Chinese RMB.
- Method 4b indicates heterogeneous results across currencies: "Mixed" for USD and "Yes" for Euro, CFA Franc, and Chinese RMB.
- Method 5 half-life results include a statistically significant coefficient for CFA Franc (-0.44***).

*Source: Testing the Purchasing Power Parity (PPP) in West and Central Africa — Annex VII. ECOWAS excluding Nigeria (Working Paper No. WP/2025/119).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025119-print-pdf.pdf_