## _wp08287

## Source details

**Canonical URL:** [_wp08287](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08287.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08287.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08287.pdf.json)

---

### Definition of region and sample
- West African region comprises Benin, Burkina Faso, Cameroon, Cape Verde, Chad, Central African Republic, Republic of Congo, Côte d’Ivoire, Gabon, Guinea, Equatorial Guinea, São Tomé and Príncipe, Niger, Nigeria, Mali, Mauritania, Togo, Ghana, Liberia, Sierra Leone, Senegal, and the Gambia.
- Time horizon for growth-turnaround identification set as five years (n=4).

### Stylized facts on growth performance
- Many episodes of short-run growth in West Africa are often followed by large episodes of negative growth.
- Long-run stagnation and economic volatility characterize West African economies, with wide variation across countries and country groups (WAEMU, CEMAC, Others).
- Since 2000 growth has been systematically high across the Western African region as a whole due to improving terms of trade and better macroeconomic policies.
- Frequency of turnaround episodes (1960–2006): 59 periods of growth acceleration, 51 periods of deceleration, and 25 periods of growth collapse.

### Real GDP per capita growth rates in Western Africa (1960–2005)
- WAEMU: 1961-69 -> 1.9180; 1970-79 -> 0.584; 1980-89 -> -1.258; 1990-99 -> -1.136; 2000-05 -> 0.237
- CEMAC: 1961-69 -> 1.155; 1970-79 -> 2.095; 1980-89 -> 0.907; 1990-99 -> 1.811; 2000-05 -> 2.591
- Other: 1961-69 -> 1.951; 1970-79 -> 0.987; 1980-89 -> -0.889; 1990-99 -> 0.033; 2000-05 -> 2.201

### Methodology for identifying growth turnarounds
- Growth rate at time t over n years defined as least squares growth rate of GDP per capita from t to t+n; change in growth rate Δg defined as g_{t+n, t+n} − g_{t, t+n} (source notation preserved).
- Growth acceleration year t conditions:
  - (i) Growth is rapid, g_{t+n,t} ≥ 3.0 pp a.
  - (ii) Growth accelerates, Δg_{t+n,t} ≥ 0.2 pp a.
  - (iii) Post-acceleration output is above pre-acceleration output: max_{i∈{t,...,t+n}} y_i ≥ y_{t-1}.
- Growth deceleration year t conditions:
  - (i) Growth is slow, 0.0 ≤ g_{t+n,t} ≤ 0.5 pp a.
  - (ii) Growth decelerates, Δg_{t+n,t} ≤ −0.2 pp a.
  - (iii) Post-acceleration output is still above pre-acceleration output: max_{i∈{t,...,t+n}} y_i ≥ y_{t-1}.
- Growth collapse episode conditions:
  - (i) Growth is negative, g_{t+n,t} < 0 pp a.
  - (ii) Growth decelerates, Δg_{t+n,t} ≤ −0.2 pp a.
  - (iii) Post-acceleration output is below pre-acceleration output: max_{i∈{t,...,t+n}} y_i ≤ y_{t-1}.
- Event variables are dummies set to 1 in the year the event occurs and in the next four years.
- Time-variants specified in year t as percent change in average value from period [t-5,t] to [t,t+5].
- Random effects probit model used for pooled cross-section and time-series structure (Maddala, 1987). Statistical significance threshold for interpretation: above 90 percent.

### Hypothesized drivers of turnarounds (categories)
- External shocks: terms of trade, natural disasters, remittances, world demand, foreign aid.
- Political changes: regime change, civil wars, local conflicts.
- Economic reforms: trade liberalization, labor market reforms, pension reforms, IMF-WB structural adjustment programs.
- Africa-specific variables: malaria, geography (distance from markets, coastal share of population), ethnic fractionalization, rainfall, nonfuel primary exporter status, fuel exporters, WAEMU/CEMAC membership.

### External shocks — findings and data notes
- Terms of trade:
  - Defined as export price divided by import price (World Development Indicators, 2008).
  - Terms-of-trade changes are very volatile for individual countries; WAEMU and CEMAC separately often exhibit annual terms-of-trade changes larger than 10 percent.
  - Average percent change in terms of trade for the region shows less volatility due to diversification effects across countries.
- Remittances:
  - Remittances in SSA grew steadily between 1973 and 2005.
  - Remittances are described as "unrequited", stable over time, and tend to smooth incomes; data likely underreported (WDI).
- Foreign aid:
  - Aid rose dramatically since 2002, categorized as Humanitarian Aid, Current Expenditure Aid, Capital Expenditure Aid (OECD DAC).
  - Capital expenditure and current expenditure aid expected to have positive effects on growth; humanitarian aid may have negative correlation due to selection bias.
- World demand:
  - World real GDP growth included as determinant (World Development Indicators / World Economic Outlook).

### Political and institutional variables — measurement and expected effects
- Conflict: years of conflict identified using the Armed Conflict Dataset Book; "war" defined as at least 1,000 battle deaths a year.
- Governance: four ICRG variables used (corruption, bureaucratic quality, accountability, law and order) plus two Polity IV indices (Polity 2 Index and years without major regime change).
- Legal origin: distinction between common law (England) and civil law (France, Spain, Portugal) considered.
- Monetary institutions: WAEMU and CEMAC countries share CFA franc zone arrangements; WAEMU mainly oil importers, CEMAC mainly oil producers.

### Economic reforms and macro variables
- Economic liberalization years (Wacziarg and Welch dataset) for sample countries include Benin 1990; Burkina Faso 1998; Cameroon 1993; Cape Verde 1991; Cote d'Ivoire 1994; Gambia, The 1985; Ghana 1985; Guinea 1986; Guinea-Bissau 1987; Mali 1988; Mauritania 1995; Niger 1994; Sierra Leone 2001.
- Financial reform proxied by share of domestic credit-to-GDP (WDI); potential endogeneity noted.
- Macroeconomic instability proxied by CPI inflation (WDI).

### Africa-specific factors and data usage
- Rainfall: volatility of rainfall affects agricultural output; rainfall data from NASA used both in absolute terms and as share of GDP.
- Ethnic fractionalization: probability that two randomly selected individuals belong to different ethnic groups; values range 0 to 1; used as time-invariant.
- Distance/landlockedness and coastal population share used to capture geography and trade cost effects.
- Malaria: share of population affected by malaria from Gallup and Sachs used; malaria considered largely exogenous and negatively correlated with growth in macro regressions.
- Primary-exporter classification: nonfuel primary exporters vs fuel exporters considered.

### Econometric results — main statistically robust findings
- External shocks:
  - Improvements in the terms of trade and accelerations in world demand increase the probability of growth accelerations.
  - Contractions in the terms of trade are related to growth collapses.
  - Aid effect on growth acceleration appears insignificant; aid negatively correlated with growth collapses.
  - Remittances do not have a statistically significant effect on turnaround periods.
- Political/institutional variables and reforms:
  - Results vary across specifications; interpretation constrained to coefficients with significance above 90 percent.
- Multicollinearity and specification notes:
  - WAEMU/CEMAC and malaria generally dropped from probit regressions due to multicollinearity.
  - Robustness checks include specifications with political/institutional variables as levels, category-by-category regressions, and interactions of political/institutional variables with ethnic fractionalization.

### Practical implications for policymakers (short run)
- External environment matters: favorable terms-of-trade changes and stronger world demand materially increase the probability of growth accelerations, but governments cannot directly control these external shocks.
- Reduce vulnerability to external shocks by fostering development of higher-value-added production to reduce exposure to volatile raw-good terms of trade.
- Aid policy: aggregate aid alone is not a clear driver of accelerations; differentiation by aid type (capital, current, humanitarian) matters.
- Remittances: no robust statistical evidence that remittances drive growth turnarounds; remittances may smooth income rather than induce sustained turnarounds.
- Institutional and political stability: conflict episodes, weak governance, and corruption impede sustained growth and responses to shocks.
- Economic reforms: trade liberalization and broader economic liberalization episodes are part of the toolkit for creating conditions conducive to accelerations; sequencing and complementarities with institutional improvements matter.

### Appendix — Political and institutional factors (summary)
- Conflict:
  - “War Ends” and “War Begins” dropped due to multicollinearity.
  - Simple correlation between GDP growth and a conflict dummy shows a strong and negative statistical relation (Appendix Table 5).
- Polity 2:
  - Polity 2 alone does not have a statistically significant effect on growth turnarounds.
  - Interaction: Polity 2 interacted with ethnic fractionalization shows a positive and statistically significant effect on growth accelerations; democracy has a more positive effect in less fractionalized countries.
- Political stability:
  - Years without political interruption positively related to growth accelerations; benefits of stable political environments have outweighed potential harms in Western Africa.
- Governance indices:
  - Reductions in corruption decreased probabilities of moderate deceleration but do not change likelihood of growth acceleration or collapse robustly.
  - Bureaucratic quality, accountability, and law and order show effects in some specifications but are not robust across all specifications.
- Colonial heritage:
  - Being a former British colony does not, by itself, increase the probability of a turnaround.

### Appendix — Economic reforms, geography, and other factors (summary)
- Inflation: does not seem to be an important determinant of turnarounds in Western Africa.
- Economic liberalization: positive, significant, and robust impact on probability of growth acceleration; no episodes of liberalization followed by collapse in the data.
- Credit: change in Domestic Credit-to-GDP negatively correlated with likelihood of growth collapses; interpretation consistent with financial disintermediation following collapses.
- Rainfall: not robustly related to growth turnarounds in probit results.
- Geography/population:
  - Greater share of population within 100 km of coast and longitude closer to the coast associated with higher chances of growth acceleration.
  - Larger coastal population share also associated with increased likelihood of growth collapse in several specifications, implying higher output volatility for coastal economies.
- Ethnic fractionalization: generally not robustly related to turnarounds; one specification links higher fractionalization to collapses but result is not robust.

### Key probit coefficients (selected, preserve exact reported values and significance)
- External shock:
  - Terms of Trade Change: 0.01* (Accelerations), -0.05*** (Collapses) (Table 4 summary)
  - Change in Aid-Funded Expenditures: -0.02** (Collapses) (Table 4 summary)
  - Change in World Demand Growth: 1.02* (Accelerations) (Table 4 summary)
- Political and Institutional Factors (selected):
  - Change in Polity 2 Index: 0.15 (Accelerations) (Table 4 summary)
  - Years without political interruption: 0.14** (Accelerations); -0.04* (Decelerations) (Table 4 summary)
  - Change in Corruption Index: 0.80 (Accelerations); -0.33* (Decelerations) (Table 4 summary)
  - Change in Bureaucratic Quality Index: -1.15** (Accelerations) (Table 4 summary)
  - Change in Accountability Index: -0.50*** (Collapses) (Table 4 summary)
  - Change in Law and Order: 0.58*** (Collapses) (Table 4 summary)
- Economic Reforms:
  - Economic Liberalization: 3.381*** (Accelerations) (Table 4 summary)
  - Change in Domestic Credit-to-GDP: -0.01** (Collapses) (Table 4 summary)
  - Structural Adjustment Program: -1.42 (Accelerations); -7.70 (Decelerations); -7.74 (Collapses) (Table 4 summary)
- Other (selected):
  - Output Gap: -8.73 (Accelerations), 10.98** (Collapses) (Table 4 summary)
  - Rainfall Variation: -13.13 (Accelerations) (Table 4 summary)
  - Primary Exporters (nonfuel): -2.93** (Accelerations) (Table 4 summary)
  - Longitude: 0.09* (Accelerations), -0.08*** (Decelerations) (Table 4 summary)
  - Population within 100 km of coast: 7.12** (Accelerations), 5.36*** (Collapses) (Table 4 summary)
- Sample and fit (Table 4 summary):
  - Observations: 111 (Accelerations), 201 (Decelerations), 201 (Collapses)
  - Number of nid: 9 (Accelerations), 13 (Decelerations), 13 (Collapses)
  - Pseudo R2: 0.54 (Accelerations), 0.23 (Decelerations), 0.36 (Collapses)
  - Significance notation: *** p<0.01, ** p<0.05, * p<0.1

### Appendix Tables — selected mean comparisons (Appendix Tables 1.a–1.c)
- Acceleration vs Non-Acceleration (selected means)
  - Terms of Trade Change (annual, %): Non-Acceleration Years 7.99131.676
  - Change in Aid-Funded K Expenditures (annual, %): Non-Acceleration Years 52.240, Acceleration Years 121.389
  - Change in Remittances (annual, %): Non-Acceleration Years 145.623, Acceleration Years 42.848
  - Economic Liberalization (%): Non-Acceleration Years 10.412, Acceleration Years 30.986
  - Change in Domestic Credit-to-GDP (annual, %): Non-Acceleration Years 2.323, Acceleration Years 15.165
  - Ethnic Fractionalization (%): Non-Acceleration Years 72.695, Acceleration Years 73.776
- Deceleration vs Non-Deceleration (selected means)
  - Terms of Trade Change (annual, %): Non-Deceleration Years 12.214, Deceleration Years 18.340
  - Population within 100 km of coast (%): Non-Deceleration Years 37.679, Deceleration Years 41.302
  - Longitude (Degrees): Non-Deceleration Years -1.625, Deceleration Years -2.796
- Collapse vs Non-Collapse (selected means)
  - Terms of Trade Change (annual, %): Non-Collapse Years 15.631, Collapse Years -6.077
  - Change in Aid-Funded K Expenditures (annual, %): Non-Collapse Years 68.829, Collapse Years 41.640
  - Change in Remittances (annual, %): Non-Collapse Years 77.811, Collapse Years 584.285
  - Years without political interruption: Non-Collapse Years 11.564, Collapse Years 24.658
  - Change in Domestic Credit-to-GDP (annual, %): Non-Collapse Years 7.631, Collapse Years -13.363
  - Ethnic Fractionalization (%): Non-Collapse Years 72.518, Collapse Years 77.471
- Note: In Appendix Tables 1.a–1.c, means in bold are statistically different at their 95% confidence intervals.

### Appendix Table 2 — Political variables specified in levels (selected coefficients)
- Terms of Trade Change: 0.01* (Accelerations); -0.053*** (Collapses)
- Change in World Demand Growth: 1.123*** (Accelerations)
- Change in Polity 2 Index: 0.454** (Accelerations)
- Years without political interruption: 0.155** (Accelerations); -0.070* (Decelerations)
- Corruption Index: 0.89 (Accelerations); -1.732** (Decelerations)
- Bureaucratic Quality Index: -1.143** (Accelerations); 1.204** (Decelerations)
- Economic Liberalization: 3.374*** (Accelerations)
- Change in Domestic Credit-to-GDP: -0.019** (Collapses)
- Population within 100 km of coast: 7.706*** (Accelerations); 5.132*** (Collapses)
- Model statistics:
  - Observations: 111 (Accelerations); 126 (Decelerations); 201 (Collapses)
  - Pseudo R2: 0.56 (Accelerations); 0.37 (Decelerations); 0.38 (Collapses)

### Appendix Table 3 (category regressions) — highlights
- Appendix Table 3a (Accelerations): Economic Liberalization: 0.623***; Change in Domestic Credit-to-GDP: 0.001***
- Appendix Table 3b (Decelerations): Terms of Trade Change: 0.014**; Change in Accountability Index: 0.007**
- Appendix Table 3c (Collapses): Terms of Trade Change: -0.031***; Change in Remittances: 0.001**; Population within 100 km of coast: 1.797**; Ethnic Fractionalization: 0.031**

### Appendix Table 4 — Political change variables interacted with ethnic fractionalization (selected)
- Change in Polity 2 Index: 1.662** (Accelerations); 12.48** (Decelerations)
- Years without political interruption: 0.030* (Accelerations); -0.05** (Collapses)
- Change in Accountability Index: -0.199** (Accelerations); -0.572** (Decelerations)
- Interaction terms (Ethnic Fractionalization × variable) show several statistically significant moderating effects (e.g., Interaction of Ethnic Fractionalization with Change in Polity 2 Index: -0.022** for Accelerations).

### Appendix Table 5 — GDP growth in conflict vs non-conflict years
- Mean GDP growth:
  - Non-Conflict Years: 4.0969773; 3.572638; 4.621316
  - Conflict Years: -4.034829; -7.792785; -0.2768723
- Source: Armed Conflict Dataset Book (2005)
- 95% Confidence Interval reported in source.

### Conclusion and distilled policy implications
- Empirical findings:
  - Growth accelerations in Western Africa have been frequent, but decelerations and collapses are also frequent; collapses have erased acceleration gains in several cases.
  - Accelerations most likely triggered by increases in terms of trade, rising world growth, and economic liberalization; occurred more often during long-standing regimes and in coastal countries.
  - Decelerations mainly determined by duration of political regime; less likely when corruption is more effectively controlled.
  - Collapses linked to negative terms of trade shocks, declines in foreign aid, worsening accountability, improvements in law and order, and reduction in domestic credit; more frequent in countries closer to the sea.
- Policy implications:
  - Political stabilization and market-oriented reforms can significantly increase chances of acceleration.
  - Long-run policies promoting higher-value-added products could reduce exposure to commodity price volatility.
  - Evidence supports strengthening the financial system, avoiding policies that fend off foreign aid, and controlling corruption, while recognizing endogeneity concerns for some variables.

*Source: _wp08287 (authors’ estimates and exposition as provided in the supplied PDF content).*

### conclusions that policymakers can implement in the short run. Finding for example that investment in human

### _wp08287 - conclusions that policymakers can implement in the short run. Finding for example that investment in human

### Definition of region and sample
- West African region defined as comprising Benin, Burkina Faso, Cameroon, Cape Verde, Chad, Central African Republic, Republic of Congo, Côte d’Ivoire, Gabon, Guinea, Equatorial Guinea, São Tomé and Príncipe, Niger, Nigeria, Mali, Mauritania, Togo, Ghana, Liberia, Sierra Leone, Senegal, and the Gambia.
- Time horizon for growth-turnaround identification set as five years (n=4).

### Key stylized facts on growth performance
- Many episodes of short-run growth have occurred in West Africa but are often followed by large episodes of negative growth.
- Long-run stagnation and economic volatility characterize West African economies, with wide variation across countries and country groups (WAEMU, CEMAC, Others).
- Since 2000 growth has been systematically high across the Western African region as a whole due to improving terms of trade and better macroeconomic policies.
- Frequency of turnaround episodes (1960–2006): 59 periods of growth acceleration, 51 periods of deceleration, and 25 periods of growth collapse.

### Table of real GDP per capita growth rates in Western Africa (1960–2005)
- WAEMU: 1961-69 -> 1.9180; 1970-79 -> 0.584; 1980-89 -> -1.258; 1990-99 -> -1.136; 2000-05 -> 0.237
- CEMAC: 1961-69 -> 1.155; 1970-79 -> 2.095; 1980-89 -> 0.907; 1990-99 -> 1.811; 2000-05 -> 2.591
- Other: 1961-69 -> 1.951; 1970-79 -> 0.987; 1980-89 -> -0.889; 1990-99 -> 0.033; 2000-05 -> 2.201
- Source: IMF World Economic Outlook

### Methodology for identifying growth turnarounds
- Growth rate at time t over n years defined as least squares growth rate of GDP per capita from t to t+n; change in growth rate Δg defined as g_{t+n, t+n} − g_{t, t+n}? (definition expressed in source notation).
- Growth acceleration year t conditions:
  - (i) Growth is rapid, g_{t+n,t} ≥ 3.0 pp a.
  - (ii) Growth accelerates, Δg_{t+n,t} ≥ 0.2 pp a.
  - (iii) Post-acceleration output is above pre-acceleration output: max_{i∈{t,...,t+n}} y_i ≥ y_{t-1} (source notation preserved).
- Growth deceleration year t conditions:
  - (i) Growth is slow, 0.0 ≤ g_{t+n,t} ≤ 0.5 pp a.
  - (ii) Growth decelerates, Δg_{t+n,t} ≤ −0.2 pp a.
  - (iii) Post-acceleration output is still above pre-acceleration output: max_{i∈{t,...,t+n}} y_i ≥ y_{t-1}.
- Growth collapse episode conditions:
  - (i) Growth is negative, g_{t+n,t} < 0 pp a.
  - (ii) Growth decelerates, Δg_{t+n,t} ≤ −0.2 pp a.
  - (iii) Post-acceleration output is below pre-acceleration output: max_{i∈{t,...,t+n}} y_i ≤ y_{t-1}.
- Event variables are dummies set to 1 in the year the event occurs and in the next four years.
- Time-variants specified in year t as percent change in average value from period [t-5,t] to [t,t+5].
- Probit econometric approach: random effects probit model used to account for pooled cross-section and time-series structure (Maddala, 1987). Statistical significance threshold for interpretation: above 90 percent.

### Hypothesized drivers of turnarounds (categories)
- External shocks: terms of trade, natural disasters, remittances, world demand, foreign aid.
- Political changes: regime change, civil wars, local conflicts.
- Economic reforms: trade liberalization, labor market reforms, pension reforms, IMF-WB structural adjustment programs.
- Africa-specific variables: malaria, geography (distance from markets, coastal share of population), ethnic fractionalization, rainfall, nonfuel primary exporter status, fuel exporters, WAEMU/CEMAC membership.

### External shocks — findings and data notes
- Terms of trade:
  - Terms of trade defined as export price divided by import price (World Development Indicators, 2008).
  - Terms-of-trade changes are very volatile for individual countries; WAEMU and CEMAC separately often exhibit annual terms-of-trade changes larger than 10 percent.
  - Average percent change in terms of trade for the region shows less volatility due to diversification effects across countries.
- Remittances:
  - Remittances in SSA grew steadily between 1973 and 2005.
  - Remittances are "unrequited", stable over time, and tend to smooth incomes; data likely underreported (WDI).
- Foreign aid:
  - Aid rose dramatically since 2002, with types categorized as Humanitarian Aid, Current Expenditure Aid, Capital Expenditure Aid (OECD DAC).
  - Not all aid is equal; capital expenditure and current expenditure aid expected to have positive effects on growth, humanitarian aid may have negative correlation due to selection bias.
- World demand:
  - World real GDP growth included as determinant; source World Development Indicators / World Economic Outlook.

### Political and institutional variables — measurement and expected effects
- Conflict: years of conflict identified using the Armed Conflict Dataset Book; "war" defined as at least 1,000 battle deaths a year.
- Governance: four ICRG variables used (corruption, bureaucratic quality, accountability, law and order) plus two Polity IV indices (Polity 2 Index and years without major regime change).
- Legal origin: distinction between common law (England) and civil law (France, Spain, Portugal) considered as potential determinant of institutional quality.
- Monetary institutions: WAEMU and CEMAC countries share CFA franc zone arrangements; WAEMU mainly oil importers, CEMAC mainly oil producers.

### Economic reforms and macro variables
- Economic liberalization years (Wacziarg and Welch dataset) for sample countries (selected entries):
  - Benin 1990; Burkina Faso 1998; Cameroon 1993; Cape Verde 1991; Cote d'Ivoire 1994; Gambia, The 1985; Ghana 1985; Guinea 1986; Guinea-Bissau 1987; Mali 1988; Mauritania 1995; Niger 1994; Sierra Leone 2001.
- Financial reform proxied by share of domestic credit-to-GDP (WDI); potential endogeneity noted.
- Macroeconomic instability proxied by CPI inflation (WDI).

### Africa-specific factors and data usage
- Rainfall: volatility of rainfall affects agricultural output; rainfall data from NASA used both in absolute terms and as share of GDP.
- Ethnic fractionalization: defined as probability that two randomly selected individuals belong to different ethnic groups; values range 0 to 1; used as time-invariant.
- Distance/landlockedness and coastal population share used to capture geography and trade cost effects.
- Malaria: share of population affected by malaria from Gallup and Sachs used; malaria considered largely exogenous and uniquely negatively correlated with growth in macro regressions.
- Primary-exporter classification: nonfuel primary exporters vs fuel exporters considered in light of the Prebisch-Singer hypothesis and commodity-price dynamics.

### Econometric results (summary of main statistically robust findings)
- External shocks:
  - Improvements in the terms of trade and accelerations in world demand increase the probability of growth accelerations.
  - Contractions in the terms of trade are related to growth collapses.
  - Aid effect on growth acceleration appears insignificant; aid negatively correlated with growth collapses (interpretation: either reductions in aid lead to collapses or collapses prompt donors to reduce aid).
  - Remittances do not have a statistically significant effect on turnaround periods.
- Political/institutional variables and reforms:
  - (Detailed regression coefficients and significance levels are reported in the source tables; interpretation constrained to coefficients with significance above 90 percent as the authors do.)
- Multicollinearity and specification notes:
  - WAEMU/CEMAC and malaria generally dropped from probit regressions due to multicollinearity.
  - Robustness checks include specification with political/institutional variables as levels, category-by-category regressions, and interactions of political/institutional variables with ethnic fractionalization.

### Practical implications for policymakers (policy-relevant conclusions in the short run)
- External environment matters: favorable terms-of-trade changes and stronger world demand materially increase the probability of growth accelerations, but governments cannot directly control these external shocks.
- Policy direction to reduce vulnerability to external shocks:
  - Foster development of higher-value-added production to reduce exposure to volatile raw-good terms of trade.
- Aid policy:
  - Aggregate aid alone is not a clear driver of accelerations; differentiation by aid type (capital, current, humanitarian) matters for expected growth impact.
- Remittances:
  - No robust statistical evidence that remittances drive growth turnarounds; remittances may be more consumption-oriented and smooth income rather than induce sustained turnarounds.
- Institutional and political stability:
  - Political and institutional quality influence growth dynamics; conflict episodes, weak governance, and corruption are impediments to sustained growth and to responding effectively to shocks.
- Economic reforms:
  - Trade liberalization and broader economic liberalization episodes are part of the toolkit for creating conditions conducive to accelerations; sequencing and complementarities with institutional improvements matter.

*Source: _wp08287 (authors’ estimates and exposition as provided in the supplied PDF content).*

### Appendix Table 1a-c present correlation between growth determinants and the occurrence of growth

### _wp08287 - Appendix Table 1a-c present correlation between growth determinants and the occurrence of growth turnarounds

### Political and Institutional Factors
- Probit regressions (Table 4):
  - “War Ends” and “War Begins” are dropped due to multicollinearity.
  - Simple correlation between GDP growth and a conflict dummy shows a strong and negative statistical relation (Appendix Table 5) on a year-to-year basis.
  - Conflicts may immediately affect economic growth but not necessarily in a sustained way that meets growth-turnaround criteria.
- Polity 2 (scale –10 to 10):
  - Polity 2 alone does not have a statistically significant effect on growth turnarounds.
  - Interaction: Polity 2 interacted with ethnic fractionalization shows a positive and statistically significant effect on growth accelerations; the negative sign of the interaction term implies democracy has a more positive effect in less fractionalized countries (see Appendix Table 4).
- Political stability:
  - Periods measured by years without political interruption seem positively related to growth accelerations.
  - In Western Africa, benefits of stable political environments have outweighed potential harms from long regimes (distributional coalitions, corruption, rent-seeking).
- Corruption, bureaucracy, accountability, law and order:
  - Reductions in corruption decreased probabilities of moderate deceleration but corruption does not change likelihood of growth acceleration or collapse.
  - A fall in bureaucratic quality appears positively related to growth acceleration in one probit specification, but this relation is not robust to specification changes.
  - Accountability and Law and Order show statistically significant effects on some growth turnarounds, but coefficient signs are not robust across specifications.
- Colonial heritage:
  - Being a former British colony does not, by itself, increase the probability of a turnaround.

### Economic Reforms
- Macroeconomic stability (inflation):
  - Inflation does not seem to be an important determinant of turnarounds in Western Africa.
- Economic liberalization:
  - Economic liberalization has a positive, significant, and robust impact on the probability of growth acceleration in Western Africa.
  - No episodes of liberalization followed by collapse in the data.
  - Conclusion: liberalizations launched in West Africa made a net positive contribution to recovery after 1980s crises.
- Credit:
  - Change in Domestic Credit-to-GDP is negatively correlated with the likelihood of growth collapses in Western Africa.
  - Interpretation: financial disintermediation more often follows economic collapses than causes them in this region.

### Other Factors
- Rainfall:
  - Probit results do not show a robust relation between rainfall and growth turnarounds; rainfall may explain short-term growth but not longer-term episodes.
- Geography and population distribution:
  - Countries with more population within 100 km of coast have higher chances of growth acceleration.
  - Countries located at a longitude west of the meridian (closer to the coast) have higher probability of growth acceleration.
  - Several specifications show a larger coastal population share also increases likelihood of growth collapse, implying higher output volatility for coastal economies.
- Ethnic fractionalization:
  - Ethnic fractionalization on its own does not have a close relationship with growth turnarounds.
  - One specification (Appendix Table 3c) shows higher ethnic fractionalization significantly associated with collapses; result not robust.

### Key Probit Coefficients and Statistical Notes (from Table 4)
- External shock:
  - Terms of Trade Change: 0.01* (Accelerations), -0.01 (Decelerations), -0.05*** (Collapses)
  - Change in Aid-Funded Expenditures: 0.00 (Accelerations), 0.00 (Decelerations), -0.02** (Collapses)
  - Change in World Demand Growth: 1.02* (Accelerations)
- Political and Institutional Factors (selected coefficients):
  - Change in Polity 2 Index: 0.15 (Accelerations), -0.08 (Decelerations), -0.06 (Collapses)
  - Years without political interruption: 0.14** (Accelerations), -0.04* (Decelerations), 0.01 (Collapses)
  - Change in Corruption Index: 0.80 (Accelerations), -0.33* (Decelerations), 0.15 (Collapses)
  - Change in Bureaucratic Quality Index: -1.15** (Accelerations)
  - Change in Accountability Index: -0.95 (Accelerations), -0.19 (Decelerations), -0.50*** (Collapses)
  - Change in Law and Order: 0.65 (Accelerations), 0.16 (Decelerations), 0.58*** (Collapses)
  - Former British Colony: -0.41 (Accelerations), -0.90 (Decelerations), 0.01 (Collapses)
- Economic Reforms:
  - Economic Liberalization: 3.381*** (Accelerations), -3.84 (Decelerations)
  - Change in Domestic Credit-to-GDP: 0.02 (Accelerations), 0.00 (Decelerations), -0.01** (Collapses)
  - Structural Adjustment Program: -1.42 (Accelerations), -7.70 (Decelerations), -7.74 (Collapses)
- Other:
  - Output Gap: -8.73 (Accelerations), -2.23 (Decelerations), 10.98** (Collapses)
  - Rainfall Variation: -13.13 (Accelerations)
  - Primary Exporters (nonfuel): -2.93** (Accelerations)
  - Longitude: 0.09* (Accelerations), -0.08*** (Decelerations)
  - Population within 100 km of coast: 7.12** (Accelerations), -1.34 (Decelerations), 5.36*** (Collapses)
  - Ethnic Fractionalization: 0.02 (Accelerations), -0.01 (Decelerations), 0.02 (Collapses)
- Sample and fit:
  - Observations: 111 (Accelerations), 201 (Decelerations), 201 (Collapses)
  - Number of nid: 9 (Accelerations), 13 (Decelerations), 13 (Collapses)
  - Pseudo R2: 0.54 (Accelerations), 0.23 (Decelerations), 0.36 (Collapses)
  - Significance notation: *** p<0.01, ** p<0.05, * p<0.1

### Averages in Appendix Tables 1.a–1.c (selected statistics)
- Appendix Table 1.a — Acceleration vs Non-Acceleration (means)
  - Terms of Trade Change (annual, %): Non-Acceleration Years 7.99131.676
  - Change in Aid-Funded K Expenditures (annual, %): Non-Acceleration Years 52.240, Acceleration Years 121.389
  - Change in Remittances (annual, %): Non-Acceleration Years 145.623, Acceleration Years 42.848
  - Years without political interruption: Non-Acceleration Years 12.731, Acceleration Years 12.319
  - Change in Corruption Index (%): Non-Acceleration Years 6.140, Acceleration Years -8.610
  - Change in Bureaucratic Quality Index (%): Non-Acceleration Years -14.448, Acceleration Years -16.849
  - Economic Liberalization (%): Non-Acceleration Years 10.412, Acceleration Years 30.986
  - Change in Domestic Credit-to-GDP (annual, %): Non-Acceleration Years 2.323, Acceleration Years 15.165
  - Population within 100 km of coast (%): Non-Acceleration Years 18.571, Acceleration Years 16.495
  - Ethnic Fractionalization (%): Non-Acceleration Years 72.695, Acceleration Years 73.776
- Appendix Table 1.b — Deceleration vs Non-Deceleration (means)
  - Terms of Trade Change (annual, %): Non-Deceleration Years 12.214, Deceleration Years 18.340
  - Change in Aid-Funded K Expenditures (annual, %): Non-Deceleration Years 62.518, Deceleration Years 74.002
  - Change in Remittances (annual, %): Non-Deceleration Years 117.625, Deceleration Years 170.303
  - Years without political interruption: Non-Deceleration Years 12.225, Deceleration Years 15.000
  - Change in Polity 2 Index (%): Non-Deceleration Years 8.100, Deceleration Years 3.880
  - Population within 100 km of coast (%): Non-Deceleration Years 37.679, Deceleration Years 41.302
  - Longitude (Degrees): Non-Deceleration Years -1.625, Deceleration Years -2.796
- Appendix Table 1.c — Collapse vs Non-Collapse (means)
  - Terms of Trade Change (annual, %): Non-Collapse Years 15.631, Collapse Years -6.077
  - Change in Aid-Funded K Expenditures (annual, %): Non-Collapse Years 68.829, Collapse Years 41.640
  - Change in Remittances (annual, %): Non-Collapse Years 77.811, Collapse Years 584.285
  - War Beginning (%): Non-Collapse Years 3.661, Collapse Years 5.128
  - Years without political interruption: Non-Collapse Years 11.564, Collapse Years 24.658
  - Change in Corruption Index (%): Non-Collapse Years 1.310, Collapse Years 21.733
  - Change in Accountability Index (%): Non-Collapse Years 35.924, Collapse Years -12.286
  - Change in Domestic Credit-to-GDP (annual, %): Non-Collapse Years 7.631, Collapse Years -13.363
  - Population within 100 km of coast (%): Non-Collapse Years 37.837, Collapse Years 42.897
  - Ethnic Fractionalization (%): Non-Collapse Years 72.518, Collapse Years 77.471
- Note: In Appendix Tables 1.a–1.c, means in bold are statistically different at their 95% confidence intervals.

### Conclusion and Policy Implications
- Empirical findings:
  - Growth accelerations in Western Africa have been frequent, but decelerations and collapses are also frequent; collapses have erased acceleration gains in several cases.
  - Accelerations most likely triggered by increases in terms of trade, rising world growth, and economic liberalization; occurred more often during long-standing regimes and in coastal countries.
  - Decelerations mainly determined by duration of political regime; less likely when corruption is more effectively controlled.
  - Collapses linked to negative terms of trade shocks, declines in foreign aid, worsening accountability, improvements in law and order, and reduction in domestic credit; more frequent in countries closer to the sea.
- Policy implications:
  - Political stabilization and market-oriented reforms can significantly increase chances of acceleration.
  - Long-run policies promoting higher-value-added products could reduce exposure to commodity price volatility.
  - Evidence supports strengthening the financial system, avoiding policies that fend off foreign aid, and controlling corruption, while recognizing endogeneity concerns for some variables.

*Source: _wp08287 - Appendix Table 1a-c present correlation between growth determinants and the occurrence of growth turnarounds*

### Appendix Table 2

### _wp08287 - Appendix Table 2

### Probit Estimation of Growth Accelerations, Decelerations and Collapses: Political Variables Specified in Levels
- External shock
  - Terms of Trade Change: 0.01* (Accelerations); 0.005 (Decelerations); -0.053*** (Collapses)
  - Change in Aid-Funded Expenditures: 0.00 (Accelerations); 0.00 (Decelerations); 0.0021 (Collapses)
  - Change in Remittances: 0.00 (Accelerations)
  - Change in World Demand Growth: 1.123*** (Accelerations); 0.017 (Decelerations); 0.331 (Collapses)
- Political and Institutional Factors
  - War End: 15.46 (Accelerations)
  - Change in Polity 2 Index: 0.454** (Accelerations); -0.23 (Decelerations); -0.042 (Collapses)
  - Years without political interruption: 0.155** (Accelerations); -0.070* (Decelerations); 0.016 (Collapses)
  - Corruption Index: 0.89 (Accelerations); -1.732** (Decelerations); 0.107 (Collapses)
  - Bureaucratic Quality Index: -1.143** (Accelerations); 1.204** (Decelerations); 0.246 (Collapses)
  - Accountability Index: -0.991 (Accelerations); -0.681 (Decelerations); -0.580*** (Collapses)
  - Law and Order: 0.775 (Accelerations); 0.812 (Decelerations); 0.509** (Collapses)
  - Former British Colony: -0.28 (Accelerations); -2.14 (Decelerations); 0.182 (Collapses)
- Economic Reforms
  - Inflation Rate: 0.00 (Accelerations); 0.00 (Decelerations)
  - Economic Liberalization: 3.374*** (Accelerations); -3.826 (Decelerations); -7.19 (Collapses)
  - Change in Domestic Credit-to-GDP: 0.0168 (Accelerations); 0.00 (Decelerations); -0.019** (Collapses)
  - Structural Adjustment Program: -1.529 (Accelerations); -6.14 (Decelerations); -7.48 (Collapses)
- Other
  - Output Gap: -16.59 (Accelerations); -5.517 (Decelerations); 10.55** (Collapses)
  - Rainfall Variation: -12.69 (Accelerations); -20.61* (Decelerations); -0.799 (Collapses)
  - Primary Exporters (nonfuel): -2.940** (Accelerations); 0.919 (Decelerations); -0.063 (Collapses)
  - Longitude: 0.104* (Accelerations); -0.294** (Decelerations); 0.016 (Collapses)
  - Population within 100 km of coast: 7.706*** (Accelerations); -4.501* (Decelerations); 5.132*** (Collapses)
  - Ethnic Fractionalization: 0.0298 (Accelerations); 0.0339 (Decelerations); 0.0153 (Collapses)
- Model statistics
  - Constant: -4.185 (Accelerations); 0.686 (Decelerations); -5.422** (Collapses)
  - Observations: 111 (Accelerations); 126 (Decelerations); 201 (Collapses)
  - Number of nid: 99 (Accelerations); 13 (Decelerations)
  - Pseudo R2: 0.56 (Accelerations); 0.37 (Decelerations); 0.38 (Collapses)
- Significance legend: *** p<0.01, ** p<0.05, * p<0.1
- Note: Standard errors in parentheses (not reproduced here)

---

### Appendix Table 3a

### Probit Estimation of Growth Accelerations by Category of Independent Variable
- External shock
  - Terms of Trade Change: 0.01*
  - Change in Aid-Funded Expenditures: 0.001*
  - Change in Remittances: 0.00
  - Change in World Demand Growth: 0.155*
- Political and Institutional Factors
  - War End: 0.632
  - Change in Polity 2 Index: 0.025
  - Years without political interruption: 0.026*
  - Change in Corruption Index: -0.02***
  - Change in Bureaucratic Quality Index: 0.002
  - Change in Accountability Index: -0.005
  - Change in Law and Order: -0.002
  - Former British Colony: -0.47
- Economic Reforms
  - Inflation Rate: -0.011
  - Economic Liberalization: 0.623***
  - Change in Domestic Credit-to-GDP: 0.001***
  - Structural Adjustment Program: 0.113
- Other
  - Output Gap: -1.044; -9.616*; -6.878***; -3.599*** (multiple model specifications)
  - Rain Volatility: -3.409**
  - Primary Exporters (nonfuel): -0.169
  - Longitude: -0.001
  - Population within 100 km of coast: -0.349
  - Ethnic Fractionalization: 0.003
- Model statistics (by specification)
  - Constant: -0.869***; -1.384***; -0.918***; -0.900*
  - Observations: 258; 151; 295; 721
  - Number of nid: 171; 417; 171
  - Pseudo R2: 0.06; 0.11; 0.09; 0.02
- Significance legend: *** p<0.01, ** p<0.05, * p<0.1

---

### Appendix Table 3b

### Probit Estimation of Growth Decelerations by Category of Independent Variable
- External shock
  - Terms of Trade Change: 0.014**
  - Change in Aid-Funded Expenditures: 0.00
  - Change in Remittances: 0.00
  - Change in World Demand Growth: 0.116
- Political change
  - War End: -5.879
  - Change in Polity 2 Index: -0.069
  - Years without political interruption: -0.002
  - Change in Corruption Index: 0.004
  - Change in Bureaucratic Quality Index: -0.004
  - Change in Accountability Index: 0.007**
  - Change in Law and Order: 0.004
  - Former British Colony: 0.23
- Economic Reforms
  - Inflation Rate: 0.006
  - Economic Reform: -0.378
  - Change in Domestic Credit-to-GDP: -0.001
  - Structural Adjustment Program: -0.619
- Other
  - Output Gap: 4.957; 8.127; 3.384; 3.258** (multiple model specifications)
  - Rainfall Variation: 0.929
  - Primary Exporters (nonfuel): -0.066
  - Longitude: -0.002
  - Population within 100 km of coast: -0.266
  - Ethnic Fractionalization: -0.003
- Model statistics (by specification)
  - Constant: -1.024***; -1.807***; -0.927***; -0.675*
  - Observations: 258; 151; 295; 721
  - Number of nid: 171; 417; 19
  - Pseudo R2: 0.06; 0.13; 0.03; 0.01
- Significance legend: *** p<0.01, ** p<0.05, * p<0.1

---

### Appendix Table 3c

### Probit Estimation of Growth Collapses by Category of Independent Variable
- External shock
  - Terms of Trade Change: -0.031***
  - Change in Aid-Funded Expenditures: 0.00
  - Change in Remittances: 0.001**
  - Change in World Demand Growth: 0.209
- Political change
  - War End: -8.452
  - Change in Polity 2 Index: 0.041
  - Years without political interruption: -0.040*
  - Change in Corruption Index: 0.011
  - Change in Bureaucratic Quality Index: 0.009
  - Change in Accountability Index: -0.011**
  - Change in Law and Order: 0.013*
  - Former British Colony: -0.39
- Economic Reforms
  - Inflation Rate: 0.013
  - Change in Domestic Credit-to-GDP: -0.001
  - Structural Adjustment Program: (not reported)
- Other
  - Output Gap: 12.79**; 0.364; 3.972; 3.221** (multiple model specifications)
  - Rainfall Variation: -6.391
  - Primary Exporters (nonfuel): 0.134
  - Longitude: 0.029*
  - Population within 100 km of coast: 1.797**
  - Ethnic Fractionalization: 0.031**
- Model statistics (by specification)
  - Constant: -2.131***; -1.116**; -1.574***; -4.388***
  - Observations: 258; 151; 295; 721
  - Number of nid: 171; 417; 171
  - Pseudo R2: 0.18; 0.22; 0.08; 0.04
- Significance legend: *** p<0.01, ** p<0.05, * p<0.1

---

### Appendix Table 4

### Probit Estimation of Growth Accelerations, Decelerations and Collapses: Political Change Variables Interacted with Ethnic Fractionalization
- Political and Institutional Factors
  - Change in Polity 2 Index: 1.662** (Accelerations); 12.48** (Decelerations); 0.871 (Collapses)
  - Years without political interruption: 0.030* (Accelerations); -0.04 (Decelerations); -0.05** (Collapses)
  - Change in Corruption Index: -0.112 (Accelerations); -0.647* (Decelerations); -0.005 (Collapses)
  - Change in Bureaucratic Quality Index: 0.051 (Accelerations); -0.611* (Decelerations); 0.001 (Collapses)
  - Change in Accountability Index: -0.199** (Accelerations); -0.572** (Decelerations); 0.263* (Collapses)
  - Change in Law and Order: 0.044 (Accelerations); 0.17 (Decelerations); -0.028 (Collapses)
- Other
  - Output Gap: -2.62 (Accelerations); -8.804 (Decelerations); 4.282 (Collapses)
  - Ethnic Fractionalization: 0.071 (Accelerations); 0.193 (Decelerations); -0.078 (Collapses)
- Interaction of Ethnic Fractionalization with:
  - Change in Polity 2 Index: -0.022** (Accelerations); -0.173** (Decelerations); -0.014 (Collapses)
  - Change in Corruption Index: 0.001 (Accelerations); 0.009* (Decelerations); 0.001 (Collapses)
  - Change in Bureaucratic Quality Index: -0.001 (Accelerations); 0.008* (Decelerations); 0 (Collapses)
  - Change in Accountability Index: 0.003** (Accelerations); 0.008*** (Decelerations); -0.004 (Collapses)
  - Change in Law and Order: -0.001 (Accelerations); -0.002 (Decelerations); 0.001 (Collapses)
- Model statistics
  - Constant: -6.864** (Accelerations); -16.224 (Decelerations); 0.757 (Collapses)
  - Observations: 141 (each model)
  - Number of nid: 13 (each model)
  - Pseudo R2: 0.16 (Accelerations); 0.45 (Decelerations); 0.38 (Collapses)
- Significance legend: *** p<0.01, ** p<0.05, * p<0.1

---

### Appendix Table 5

### GDP Growth in Conflict and Non-Conflict Years
- Mean GDP growth
  - Non-Conflict Years: 4.0969773; 3.572638; 4.621316
  - Conflict Years: -4.034829; -7.792785; -0.2768723
- Source: Armed Conflict Dataset Book (2005)
- 95% Confidence Interval

*Source: _wp08287 - Appendix Tables as provided in the source PDF*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08287.pdf_